/usr/share/R/doc/manual/R-exts.html

<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN" "http://www.w3.org/TR/html4/loose.dtd">
<html>
<!-- This manual is for R, version 3.3.3 (2017-03-06).

Copyright (C) 1999-2016 R Core Team

Permission is granted to make and distribute verbatim copies of this
manual provided the copyright notice and this permission notice are
preserved on all copies.

Permission is granted to copy and distribute modified versions of this
manual under the conditions for verbatim copying, provided that the
entire resulting derived work is distributed under the terms of a
permission notice identical to this one.

Permission is granted to copy and distribute translations of this manual
into another language, under the above conditions for modified versions,
except that this permission notice may be stated in a translation
approved by the R Core Team. -->
<!-- Created by GNU Texinfo 6.3, http://www.gnu.org/software/texinfo/ -->
<head>
<title>Writing R Extensions</title>

<meta name="description" content="Writing R Extensions">
<meta name="keywords" content="Writing R Extensions">
<meta name="resource-type" content="document">
<meta name="distribution" content="global">
<meta name="Generator" content="texi2any">
<meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
<link href="#Top" rel="start" title="Top">
<link href="#Function-and-variable-index" rel="index" title="Function and variable index">
<link href="#SEC_Contents" rel="contents" title="Table of Contents">
<style type="text/css">
<!--
a.summary-letter {text-decoration: none}
blockquote.indentedblock {margin-right: 0em}
blockquote.smallindentedblock {margin-right: 0em; font-size: smaller}
blockquote.smallquotation {font-size: smaller}
div.display {margin-left: 3.2em}
div.example {margin-left: 3.2em}
div.lisp {margin-left: 3.2em}
div.smalldisplay {margin-left: 3.2em}
div.smallexample {margin-left: 3.2em}
div.smalllisp {margin-left: 3.2em}
kbd {font-style: oblique}
pre.display {font-family: inherit}
pre.format {font-family: inherit}
pre.menu-comment {font-family: serif}
pre.menu-preformatted {font-family: serif}
pre.smalldisplay {font-family: inherit; font-size: smaller}
pre.smallexample {font-size: smaller}
pre.smallformat {font-family: inherit; font-size: smaller}
pre.smalllisp {font-size: smaller}
span.nolinebreak {white-space: nowrap}
span.roman {font-family: initial; font-weight: normal}
span.sansserif {font-family: sans-serif; font-weight: normal}
ul.no-bullet {list-style: none}
body {
    margin-left: 5%;
    margin-right: 5%;
}

h1 {             
    background: white;
    color: rgb(25%, 25%, 25%);
    font-family: monospace;
    font-size: xx-large;
    text-align: center;
}

h2 {
    background: white;
    color: rgb(40%, 40%, 40%);
    font-family: monospace;
    font-size: x-large;
    text-align: center;
}

h3 {
    background: white;
    color: rgb(40%, 40%, 40%);
    font-family: monospace;
    font-size: large;
}

h4 {
    background: white;
    color: rgb(40%, 40%, 40%);
    font-family: monospace;
}

span.samp {
    font-family: monospace;
}

span.command {
    font-family: monospace;
}

span.option {
    font-family: monospace;
}

span.file {
    font-family: monospace;
}

span.env {
    font-family: monospace;
}

ul {
    margin-top: 0.25ex;
    margin-bottom: 0.25ex;
}

li {
    margin-top: 0.25ex;
    margin-bottom: 0.25ex;
}

p {
    margin-top: 0.6ex;
    margin-bottom: 1.2ex;
}

-->
</style>


</head>

<body lang="en">
<h1 class="settitle" align="center">Writing R Extensions</h1>






















<a name="SEC_Contents"></a>
<h2 class="contents-heading">Table of Contents</h2>

<div class="contents">

<ul class="no-bullet">
  <li><a name="toc-Acknowledgements-1" href="#Acknowledgements">Acknowledgements</a></li>
  <li><a name="toc-Creating-R-packages-1" href="#Creating-R-packages">1 Creating R packages</a>
  <ul class="no-bullet">
    <li><a name="toc-Package-structure-1" href="#Package-structure">1.1 Package structure</a>
    <ul class="no-bullet">
      <li><a name="toc-The-DESCRIPTION-file-1" href="#The-DESCRIPTION-file">1.1.1 The <samp>DESCRIPTION</samp> file</a></li>
      <li><a name="toc-Licensing-1" href="#Licensing">1.1.2 Licensing</a></li>
      <li><a name="toc-Package-Dependencies-1" href="#Package-Dependencies">1.1.3 Package Dependencies</a>
      <ul class="no-bullet">
        <li><a name="toc-Suggested-packages-1" href="#Suggested-packages">1.1.3.1 Suggested packages</a></li>
      </ul></li>
      <li><a name="toc-The-INDEX-file-1" href="#The-INDEX-file">1.1.4 The <samp>INDEX</samp> file</a></li>
      <li><a name="toc-Package-subdirectories-1" href="#Package-subdirectories">1.1.5 Package subdirectories</a></li>
      <li><a name="toc-Data-in-packages-1" href="#Data-in-packages">1.1.6 Data in packages</a></li>
      <li><a name="toc-Non_002dR-scripts-in-packages-1" href="#Non_002dR-scripts-in-packages">1.1.7 Non-R scripts in packages</a></li>
      <li><a name="toc-Specifying-URLs-1" href="#Specifying-URLs">1.1.8 Specifying URLs</a></li>
    </ul></li>
    <li><a name="toc-Configure-and-cleanup-1" href="#Configure-and-cleanup">1.2 Configure and cleanup</a>
    <ul class="no-bullet">
      <li><a name="toc-Using-Makevars-1" href="#Using-Makevars">1.2.1 Using <samp>Makevars</samp></a>
      <ul class="no-bullet">
        <li><a name="toc-OpenMP-support-1" href="#OpenMP-support">1.2.1.1 OpenMP support</a></li>
        <li><a name="toc-Using-pthreads-1" href="#Using-pthreads">1.2.1.2 Using pthreads</a></li>
        <li><a name="toc-Compiling-in-sub_002ddirectories-1" href="#Compiling-in-sub_002ddirectories">1.2.1.3 Compiling in sub-directories</a></li>
      </ul></li>
      <li><a name="toc-Configure-example-1" href="#Configure-example">1.2.2 Configure example</a></li>
      <li><a name="toc-Using-F95-code-1" href="#Using-F95-code">1.2.3 Using F95 code</a></li>
      <li><a name="toc-Using-C_002b_002b11-code-1" href="#Using-C_002b_002b11-code">1.2.4 Using C++11 code</a></li>
    </ul></li>
    <li><a name="toc-Checking-and-building-packages-1" href="#Checking-and-building-packages">1.3 Checking and building packages</a>
    <ul class="no-bullet">
      <li><a name="toc-Checking-packages-1" href="#Checking-packages">1.3.1 Checking packages</a></li>
      <li><a name="toc-Building-package-tarballs-1" href="#Building-package-tarballs">1.3.2 Building package tarballs</a></li>
      <li><a name="toc-Building-binary-packages-1" href="#Building-binary-packages">1.3.3 Building binary packages</a></li>
    </ul></li>
    <li><a name="toc-Writing-package-vignettes-1" href="#Writing-package-vignettes">1.4 Writing package vignettes</a>
    <ul class="no-bullet">
      <li><a name="toc-Encodings-and-vignettes-1" href="#Encodings-and-vignettes">1.4.1 Encodings and vignettes</a></li>
      <li><a name="toc-Non_002dSweave-vignettes-1" href="#Non_002dSweave-vignettes">1.4.2 Non-Sweave vignettes</a></li>
    </ul></li>
    <li><a name="toc-Package-namespaces-1" href="#Package-namespaces">1.5 Package namespaces</a>
    <ul class="no-bullet">
      <li><a name="toc-Specifying-imports-and-exports-1" href="#Specifying-imports-and-exports">1.5.1 Specifying imports and exports</a></li>
      <li><a name="toc-Registering-S3-methods-1" href="#Registering-S3-methods">1.5.2 Registering S3 methods</a></li>
      <li><a name="toc-Load-hooks-1" href="#Load-hooks">1.5.3 Load hooks</a></li>
      <li><a name="toc-useDynLib-1" href="#useDynLib">1.5.4 useDynLib</a></li>
      <li><a name="toc-An-example-1" href="#An-example">1.5.5 An example</a></li>
      <li><a name="toc-Namespaces-with-S4-classes-and-methods-1" href="#Namespaces-with-S4-classes-and-methods">1.5.6 Namespaces with S4 classes and methods</a></li>
    </ul></li>
    <li><a name="toc-Writing-portable-packages-1" href="#Writing-portable-packages">1.6 Writing portable packages</a>
    <ul class="no-bullet">
      <li><a name="toc-PDF-size-1" href="#PDF-size">1.6.1 PDF size</a></li>
      <li><a name="toc-Check-timing-1" href="#Check-timing">1.6.2 Check timing</a></li>
      <li><a name="toc-Encoding-issues-1" href="#Encoding-issues">1.6.3 Encoding issues</a></li>
      <li><a name="toc-Portable-C-and-C_002b_002b-code-1" href="#Portable-C-and-C_002b_002b-code">1.6.4 Portable C and C++ code</a></li>
      <li><a name="toc-Binary-distribution-1" href="#Binary-distribution">1.6.5 Binary distribution</a></li>
    </ul></li>
    <li><a name="toc-Diagnostic-messages-1" href="#Diagnostic-messages">1.7 Diagnostic messages</a></li>
    <li><a name="toc-Internationalization-1" href="#Internationalization">1.8 Internationalization</a>
    <ul class="no-bullet">
      <li><a name="toc-C_002dlevel-messages-1" href="#C_002dlevel-messages">1.8.1 C-level messages</a></li>
      <li><a name="toc-R-messages-1" href="#R-messages">1.8.2 R messages</a></li>
      <li><a name="toc-Preparing-translations-1" href="#Preparing-translations">1.8.3 Preparing translations</a></li>
    </ul></li>
    <li><a name="toc-CITATION-files-1" href="#CITATION-files">1.9 CITATION files</a></li>
    <li><a name="toc-Package-types-1" href="#Package-types">1.10 Package types</a>
    <ul class="no-bullet">
      <li><a name="toc-Frontend-1" href="#Frontend">1.10.1 Frontend</a></li>
    </ul></li>
    <li><a name="toc-Services-1" href="#Services">1.11 Services</a></li>
  </ul></li>
  <li><a name="toc-Writing-R-documentation-files-1" href="#Writing-R-documentation-files">2 Writing R documentation files</a>
  <ul class="no-bullet">
    <li><a name="toc-Rd-format-1" href="#Rd-format">2.1 Rd format</a>
    <ul class="no-bullet">
      <li><a name="toc-Documenting-functions-1" href="#Documenting-functions">2.1.1 Documenting functions</a></li>
      <li><a name="toc-Documenting-data-sets-1" href="#Documenting-data-sets">2.1.2 Documenting data sets</a></li>
      <li><a name="toc-Documenting-S4-classes-and-methods-1" href="#Documenting-S4-classes-and-methods">2.1.3 Documenting S4 classes and methods</a></li>
      <li><a name="toc-Documenting-packages-1" href="#Documenting-packages">2.1.4 Documenting packages</a></li>
    </ul></li>
    <li><a name="toc-Sectioning-1" href="#Sectioning">2.2 Sectioning</a></li>
    <li><a name="toc-Marking-text-1" href="#Marking-text">2.3 Marking text</a></li>
    <li><a name="toc-Lists-and-tables-1" href="#Lists-and-tables">2.4 Lists and tables</a></li>
    <li><a name="toc-Cross_002dreferences-1" href="#Cross_002dreferences">2.5 Cross-references</a></li>
    <li><a name="toc-Mathematics-1" href="#Mathematics">2.6 Mathematics</a></li>
    <li><a name="toc-Figures-1" href="#Figures">2.7 Figures</a></li>
    <li><a name="toc-Insertions-1" href="#Insertions">2.8 Insertions</a></li>
    <li><a name="toc-Indices-1" href="#Indices">2.9 Indices</a></li>
    <li><a name="toc-Platform_002dspecific-documentation" href="#Platform_002dspecific-sections">2.10 Platform-specific documentation</a></li>
    <li><a name="toc-Conditional-text-1" href="#Conditional-text">2.11 Conditional text</a></li>
    <li><a name="toc-Dynamic-pages-1" href="#Dynamic-pages">2.12 Dynamic pages</a></li>
    <li><a name="toc-User_002ddefined-macros-1" href="#User_002ddefined-macros">2.13 User-defined macros</a></li>
    <li><a name="toc-Encoding-1" href="#Encoding">2.14 Encoding</a></li>
    <li><a name="toc-Processing-documentation-files-1" href="#Processing-documentation-files">2.15 Processing documentation files</a></li>
    <li><a name="toc-Editing-Rd-files-1" href="#Editing-Rd-files">2.16 Editing Rd files</a></li>
  </ul></li>
  <li><a name="toc-Tidying-and-profiling-R-code-1" href="#Tidying-and-profiling-R-code">3 Tidying and profiling R code</a>
  <ul class="no-bullet">
    <li><a name="toc-Tidying-R-code-1" href="#Tidying-R-code">3.1 Tidying R code</a></li>
    <li><a name="toc-Profiling-R-code-for-speed-1" href="#Profiling-R-code-for-speed">3.2 Profiling R code for speed</a></li>
    <li><a name="toc-Profiling-R-code-for-memory-use-1" href="#Profiling-R-code-for-memory-use">3.3 Profiling R code for memory use</a>
    <ul class="no-bullet">
      <li><a name="toc-Memory-statistics-from-Rprof-1" href="#Memory-statistics-from-Rprof">3.3.1 Memory statistics from <code>Rprof</code></a></li>
      <li><a name="toc-Tracking-memory-allocations-1" href="#Tracking-memory-allocations">3.3.2 Tracking memory allocations</a></li>
      <li><a name="toc-Tracing-copies-of-an-object-1" href="#Tracing-copies-of-an-object">3.3.3 Tracing copies of an object</a></li>
    </ul></li>
    <li><a name="toc-Profiling-compiled-code-1" href="#Profiling-compiled-code">3.4 Profiling compiled code</a>
    <ul class="no-bullet">
      <li><a name="toc-Linux-1" href="#Linux">3.4.1 Linux</a>
      <ul class="no-bullet">
        <li><a name="toc-sprof" href="#sprof">3.4.1.1 sprof</a></li>
        <li><a name="toc-oprofile-and-operf" href="#oprofile-and-operf">3.4.1.2 oprofile and operf</a></li>
      </ul></li>
      <li><a name="toc-Solaris-1" href="#Solaris">3.4.2 Solaris</a></li>
      <li><a name="toc-macOS-1" href="#macOS">3.4.3 macOS</a></li>
    </ul></li>
  </ul></li>
  <li><a name="toc-Debugging-1" href="#Debugging">4 Debugging</a>
  <ul class="no-bullet">
    <li><a name="toc-Browsing-1" href="#Browsing">4.1 Browsing</a></li>
    <li><a name="toc-Debugging-R-code-1" href="#Debugging-R-code">4.2 Debugging R code</a></li>
    <li><a name="toc-Checking-memory-access-1" href="#Checking-memory-access">4.3 Checking memory access</a>
    <ul class="no-bullet">
      <li><a name="toc-Using-gctorture-1" href="#Using-gctorture">4.3.1 Using gctorture</a></li>
      <li><a name="toc-Using-valgrind-1" href="#Using-valgrind">4.3.2 Using valgrind</a></li>
      <li><a name="toc-Using-the-Address-Sanitizer" href="#Using-Address-Sanitizer">4.3.3 Using the Address Sanitizer</a>
      <ul class="no-bullet">
        <li><a name="toc-Using-the-Leak-Sanitizer" href="#Using-Leak-Sanitizer">4.3.3.1 Using the Leak Sanitizer</a></li>
      </ul></li>
      <li><a name="toc-Using-the-Undefined-Behaviour-Sanitizer" href="#Using-Undefined-Behaviour-Sanitizer">4.3.4 Using the Undefined Behaviour Sanitizer</a></li>
      <li><a name="toc-Other-analyses-with-_0060clang_0027-1" href="#Other-analyses-with-_0060clang_0027">4.3.5 Other analyses with &lsquo;clang&rsquo;</a></li>
      <li><a name="toc-Using-_0060Dr_002e-Memory_0027-1" href="#Using-_0060Dr_002e-Memory_0027">4.3.6 Using &lsquo;Dr. Memory&rsquo;</a></li>
      <li><a name="toc-Fortran-array-bounds-checking-1" href="#Fortran-array-bounds-checking">4.3.7 Fortran array bounds checking</a></li>
    </ul></li>
    <li><a name="toc-Debugging-compiled-code-1" href="#Debugging-compiled-code">4.4 Debugging compiled code</a>
    <ul class="no-bullet">
      <li><a name="toc-Finding-entry-points-in-dynamically-loaded-code" href="#Finding-entry-points">4.4.1 Finding entry points in dynamically loaded code</a></li>
      <li><a name="toc-Inspecting-R-objects-when-debugging" href="#Inspecting-R-objects">4.4.2 Inspecting R objects when debugging</a></li>
    </ul></li>
  </ul></li>
  <li><a name="toc-System-and-foreign-language-interfaces-1" href="#System-and-foreign-language-interfaces">5 System and foreign language interfaces</a>
  <ul class="no-bullet">
    <li><a name="toc-Operating-system-access-1" href="#Operating-system-access">5.1 Operating system access</a></li>
    <li><a name="toc-Interface-functions-_002eC-and-_002eFortran-1" href="#Interface-functions-_002eC-and-_002eFortran">5.2 Interface functions <code>.C</code> and <code>.Fortran</code></a></li>
    <li><a name="toc-dyn_002eload-and-dyn_002eunload-1" href="#dyn_002eload-and-dyn_002eunload">5.3 <code>dyn.load</code> and <code>dyn.unload</code></a></li>
    <li><a name="toc-Registering-native-routines-1" href="#Registering-native-routines">5.4 Registering native routines</a>
    <ul class="no-bullet">
      <li><a name="toc-Speed-considerations-1" href="#Speed-considerations">5.4.1 Speed considerations</a></li>
      <li><a name="toc-Example_003a-converting-a-package-to-use-registration" href="#Converting-a-package-to-use-registration">5.4.2 Example: converting a package to use registration</a></li>
      <li><a name="toc-Linking-to-native-routines-in-other-packages-1" href="#Linking-to-native-routines-in-other-packages">5.4.3 Linking to native routines in other packages</a></li>
    </ul></li>
    <li><a name="toc-Creating-shared-objects-1" href="#Creating-shared-objects">5.5 Creating shared objects</a></li>
    <li><a name="toc-Interfacing-C_002b_002b-code-1" href="#Interfacing-C_002b_002b-code">5.6 Interfacing C++ code</a>
    <ul class="no-bullet">
      <li><a name="toc-External-C_002b_002b-code" href="#External-C_002b_002b-code">5.6.1 External C++ code</a></li>
    </ul></li>
    <li><a name="toc-Fortran-I_002fO-1" href="#Fortran-I_002fO">5.7 Fortran I/O</a></li>
    <li><a name="toc-Linking-to-other-packages-1" href="#Linking-to-other-packages">5.8 Linking to other packages</a>
    <ul class="no-bullet">
      <li><a name="toc-Unix_002dalikes-1" href="#Unix_002dalikes">5.8.1 Unix-alikes</a></li>
      <li><a name="toc-Windows-1" href="#Windows">5.8.2 Windows</a></li>
    </ul></li>
    <li><a name="toc-Handling-R-objects-in-C-1" href="#Handling-R-objects-in-C">5.9 Handling R objects in C</a>
    <ul class="no-bullet">
      <li><a name="toc-Handling-the-effects-of-garbage-collection" href="#Garbage-Collection">5.9.1 Handling the effects of garbage collection</a></li>
      <li><a name="toc-Allocating-storage-1" href="#Allocating-storage">5.9.2 Allocating storage</a></li>
      <li><a name="toc-Details-of-R-types-1" href="#Details-of-R-types">5.9.3 Details of R types</a></li>
      <li><a name="toc-Attributes-1" href="#Attributes">5.9.4 Attributes</a></li>
      <li><a name="toc-Classes-1" href="#Classes">5.9.5 Classes</a></li>
      <li><a name="toc-Handling-lists-1" href="#Handling-lists">5.9.6 Handling lists</a></li>
      <li><a name="toc-Handling-character-data-1" href="#Handling-character-data">5.9.7 Handling character data</a></li>
      <li><a name="toc-Finding-and-setting-variables-1" href="#Finding-and-setting-variables">5.9.8 Finding and setting variables</a></li>
      <li><a name="toc-Some-convenience-functions-1" href="#Some-convenience-functions">5.9.9 Some convenience functions</a>
      <ul class="no-bullet">
        <li><a name="toc-Semi_002dinternal-convenience-functions-1" href="#Semi_002dinternal-convenience-functions">5.9.9.1 Semi-internal convenience functions</a></li>
      </ul></li>
      <li><a name="toc-Named-objects-and-copying-1" href="#Named-objects-and-copying">5.9.10 Named objects and copying</a></li>
    </ul></li>
    <li><a name="toc-Interface-functions-_002eCall-and-_002eExternal-1" href="#Interface-functions-_002eCall-and-_002eExternal">5.10 Interface functions <code>.Call</code> and <code>.External</code></a>
    <ul class="no-bullet">
      <li><a name="toc-Calling-_002eCall-1" href="#Calling-_002eCall">5.10.1 Calling <code>.Call</code></a></li>
      <li><a name="toc-Calling-_002eExternal-1" href="#Calling-_002eExternal">5.10.2 Calling <code>.External</code></a></li>
      <li><a name="toc-Missing-and-special-values-1" href="#Missing-and-special-values">5.10.3 Missing and special values</a></li>
    </ul></li>
    <li><a name="toc-Evaluating-R-expressions-from-C-1" href="#Evaluating-R-expressions-from-C">5.11 Evaluating R expressions from C</a>
    <ul class="no-bullet">
      <li><a name="toc-Zero_002dfinding-1" href="#Zero_002dfinding">5.11.1 Zero-finding</a></li>
      <li><a name="toc-Calculating-numerical-derivatives-1" href="#Calculating-numerical-derivatives">5.11.2 Calculating numerical derivatives</a></li>
    </ul></li>
    <li><a name="toc-Parsing-R-code-from-C-1" href="#Parsing-R-code-from-C">5.12 Parsing R code from C</a>
    <ul class="no-bullet">
      <li><a name="toc-Accessing-source-references-1" href="#Accessing-source-references">5.12.1 Accessing source references</a></li>
    </ul></li>
    <li><a name="toc-External-pointers-and-weak-references-1" href="#External-pointers-and-weak-references">5.13 External pointers and weak references</a>
    <ul class="no-bullet">
      <li><a name="toc-An-example-2" href="#An-external-pointer-example">5.13.1 An example</a></li>
    </ul></li>
    <li><a name="toc-Vector-accessor-functions-1" href="#Vector-accessor-functions">5.14 Vector accessor functions</a></li>
    <li><a name="toc-Character-encoding-issues-1" href="#Character-encoding-issues">5.15 Character encoding issues</a></li>
  </ul></li>
  <li><a name="toc-The-R-API_003a-entry-points-for-C-code" href="#The-R-API">6 The R <acronym>API</acronym>: entry points for C code</a>
  <ul class="no-bullet">
    <li><a name="toc-Memory-allocation-1" href="#Memory-allocation">6.1 Memory allocation</a>
    <ul class="no-bullet">
      <li><a name="toc-Transient-storage-allocation-1" href="#Transient-storage-allocation">6.1.1 Transient storage allocation</a></li>
      <li><a name="toc-User_002dcontrolled-memory-1" href="#User_002dcontrolled-memory">6.1.2 User-controlled memory</a></li>
    </ul></li>
    <li><a name="toc-Error-handling-1" href="#Error-handling">6.2 Error handling</a>
    <ul class="no-bullet">
      <li><a name="toc-Error-handling-from-FORTRAN-1" href="#Error-handling-from-FORTRAN">6.2.1 Error handling from FORTRAN</a></li>
    </ul></li>
    <li><a name="toc-Random-number-generation" href="#Random-numbers">6.3 Random number generation</a></li>
    <li><a name="toc-Missing-and-IEEE-special-values" href="#Missing-and-IEEE-values">6.4 Missing and <acronym>IEEE</acronym> special values</a></li>
    <li><a name="toc-Printing-1" href="#Printing">6.5 Printing</a>
    <ul class="no-bullet">
      <li><a name="toc-Printing-from-FORTRAN-1" href="#Printing-from-FORTRAN">6.5.1 Printing from FORTRAN</a></li>
    </ul></li>
    <li><a name="toc-Calling-C-from-FORTRAN-and-vice-versa-1" href="#Calling-C-from-FORTRAN-and-vice-versa">6.6 Calling C from FORTRAN and vice versa</a></li>
    <li><a name="toc-Numerical-analysis-subroutines-1" href="#Numerical-analysis-subroutines">6.7 Numerical analysis subroutines</a>
    <ul class="no-bullet">
      <li><a name="toc-Distribution-functions-1" href="#Distribution-functions">6.7.1 Distribution functions</a></li>
      <li><a name="toc-Mathematical-functions-1" href="#Mathematical-functions">6.7.2 Mathematical functions</a></li>
      <li><a name="toc-Numerical-Utilities-1" href="#Numerical-Utilities">6.7.3 Numerical Utilities</a></li>
      <li><a name="toc-Mathematical-constants-1" href="#Mathematical-constants">6.7.4 Mathematical constants</a></li>
    </ul></li>
    <li><a name="toc-Optimization-1" href="#Optimization">6.8 Optimization</a></li>
    <li><a name="toc-Integration-1" href="#Integration">6.9 Integration</a></li>
    <li><a name="toc-Utility-functions-1" href="#Utility-functions">6.10 Utility functions</a></li>
    <li><a name="toc-Re_002dencoding-1" href="#Re_002dencoding">6.11 Re-encoding</a></li>
    <li><a name="toc-Allowing-interrupts-1" href="#Allowing-interrupts">6.12 Allowing interrupts</a></li>
    <li><a name="toc-Platform-and-version-information-1" href="#Platform-and-version-information">6.13 Platform and version information</a></li>
    <li><a name="toc-Inlining-C-functions-1" href="#Inlining-C-functions">6.14 Inlining C functions</a></li>
    <li><a name="toc-Controlling-visibility-1" href="#Controlling-visibility">6.15 Controlling visibility</a></li>
    <li><a name="toc-Using-these-functions-in-your-own-C-code" href="#Standalone-Mathlib">6.16 Using these functions in your own C code</a></li>
    <li><a name="toc-Organization-of-header-files-1" href="#Organization-of-header-files">6.17 Organization of header files</a></li>
  </ul></li>
  <li><a name="toc-Generic-functions-and-methods-1" href="#Generic-functions-and-methods">7 Generic functions and methods</a>
  <ul class="no-bullet">
    <li><a name="toc-Adding-new-generics-1" href="#Adding-new-generics">7.1 Adding new generics</a></li>
  </ul></li>
  <li><a name="toc-Linking-GUIs-and-other-front_002dends-to-R-1" href="#Linking-GUIs-and-other-front_002dends-to-R">8 Linking GUIs and other front-ends to R</a>
  <ul class="no-bullet">
    <li><a name="toc-Embedding-R-under-Unix_002dalikes-1" href="#Embedding-R-under-Unix_002dalikes">8.1 Embedding R under Unix-alikes</a>
    <ul class="no-bullet">
      <li><a name="toc-Compiling-against-the-R-library-1" href="#Compiling-against-the-R-library">8.1.1 Compiling against the R library</a></li>
      <li><a name="toc-Setting-R-callbacks-1" href="#Setting-R-callbacks">8.1.2 Setting R callbacks</a></li>
      <li><a name="toc-Registering-symbols-1" href="#Registering-symbols">8.1.3 Registering symbols</a></li>
      <li><a name="toc-Meshing-event-loops-1" href="#Meshing-event-loops">8.1.4 Meshing event loops</a></li>
      <li><a name="toc-Threading-issues-1" href="#Threading-issues">8.1.5 Threading issues</a></li>
    </ul></li>
    <li><a name="toc-Embedding-R-under-Windows-1" href="#Embedding-R-under-Windows">8.2 Embedding R under Windows</a>
    <ul class="no-bullet">
      <li><a name="toc-Using-_0028D_0029COM-1" href="#Using-_0028D_0029COM">8.2.1 Using (D)COM</a></li>
      <li><a name="toc-Calling-R_002edll-directly-1" href="#Calling-R_002edll-directly">8.2.2 Calling R.dll directly</a></li>
      <li><a name="toc-Finding-R_005fHOME-1" href="#Finding-R_005fHOME">8.2.3 Finding R_HOME</a></li>
    </ul></li>
  </ul></li>
  <li><a name="toc-Function-and-variable-index-1" href="#Function-and-variable-index">Function and variable index</a></li>
  <li><a name="toc-Concept-index-1" href="#Concept-index">Concept index</a></li>
</ul>
</div>


<a name="Top"></a>
<div class="header">
<p>
Next: <a href="#Acknowledgements" accesskey="n" rel="next">Acknowledgements</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Writing-R-Extensions"></a>
<h1 class="top">Writing R Extensions</h1>

<p>This is a guide to extending R, describing the process of creating
R add-on packages, writing R documentation, R&rsquo;s system and
foreign language interfaces, and the R <acronym>API</acronym>.
</p>
<p>This manual is for R, version 3.3.3 (2017-03-06).
</p>
<p>Copyright &copy; 1999&ndash;2016 R Core Team
</p>
<blockquote>
<p>Permission is granted to make and distribute verbatim copies of this
manual provided the copyright notice and this permission notice are
preserved on all copies.
</p>
<p>Permission is granted to copy and distribute modified versions of this
manual under the conditions for verbatim copying, provided that the
entire resulting derived work is distributed under the terms of a
permission notice identical to this one.
</p>
<p>Permission is granted to copy and distribute translations of this manual
into another language, under the above conditions for modified versions,
except that this permission notice may be stated in a translation
approved by the R Core Team.
</p></blockquote>



<table summary="" class="menu" border="0" cellspacing="0">
<tr><td align="left" valign="top">&bull; <a href="#Acknowledgements" accesskey="1">Acknowledgements</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Creating-R-packages" accesskey="2">Creating R packages</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Writing-R-documentation-files" accesskey="3">Writing R documentation files</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Tidying-and-profiling-R-code" accesskey="4">Tidying and profiling R code</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Debugging" accesskey="5">Debugging</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#System-and-foreign-language-interfaces" accesskey="6">System and foreign language interfaces</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#The-R-API" accesskey="7">The R API</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Generic-functions-and-methods" accesskey="8">Generic functions and methods</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Linking-GUIs-and-other-front_002dends-to-R" accesskey="9">Linking GUIs and other front-ends to R</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Function-and-variable-index">Function and variable index</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Concept-index">Concept index</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
</table>

<hr>
<a name="Acknowledgements"></a>
<div class="header">
<p>
Next: <a href="#Creating-R-packages" accesskey="n" rel="next">Creating R packages</a>, Previous: <a href="#Top" accesskey="p" rel="prev">Top</a>, Up: <a href="#Top" accesskey="u" rel="up">Top</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Acknowledgements-1"></a>
<h2 class="unnumbered">Acknowledgements</h2>


<p>The contributions to early versions of this manual by Saikat DebRoy
(who wrote the first draft of a guide to using <code>.Call</code> and
<code>.External</code>) and Adrian Trapletti (who provided information on the
C++ interface) are gratefully acknowledged.
</p>
<hr>
<a name="Creating-R-packages"></a>
<div class="header">
<p>
Next: <a href="#Writing-R-documentation-files" accesskey="n" rel="next">Writing R documentation files</a>, Previous: <a href="#Acknowledgements" accesskey="p" rel="prev">Acknowledgements</a>, Up: <a href="#Top" accesskey="u" rel="up">Top</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Creating-R-packages-1"></a>
<h2 class="chapter">1 Creating R packages</h2>
<a name="index-Packages"></a>
<a name="index-Creating-packages"></a>

<p>Packages provide a mechanism for loading optional code, data and
documentation as needed.  The R distribution itself includes about 30
packages.
</p>
<p>In the following, we assume that you know the <code>library()</code> command,
including its <code>lib.loc</code> argument, and we also assume basic
knowledge of the <code>R CMD INSTALL</code> utility.  Otherwise, please
look at R&rsquo;s help pages on
</p>
<div class="example">
<pre class="example">?library
?INSTALL
</pre></div>

<p>before reading on.
</p>
<p>For packages which contain code to be compiled, a computing environment
including a number of tools is assumed; the &ldquo;R Installation and
Administration&rdquo; manual describes what is needed for each OS.
</p>
<p>Once a source package is created, it must be installed by
the command <code>R CMD INSTALL</code>.
See <a href="http://cran.r-project.org/doc/manuals/R-admin.html#Add_002don-packages">Add-on-packages</a> in <cite>R Installation and Administration</cite>.
</p>
<p>Other types of extensions are supported (but rare): See <a href="#Package-types">Package types</a>.
</p>
<p>Some notes on terminology complete this introduction.  These will help
with the reading of this manual, and also in describing concepts
accurately when asking for help.
</p>
<p>A <em>package</em> is a directory of files which extend R, a
<em>source package</em> (the master files of a package), or a tarball
containing the files of a source package, or an <em>installed</em>
package, the result of running <code>R CMD INSTALL</code> on a source
package.  On some platforms (notably macOS and Windows) there are also
<em>binary packages</em>, a zip file or tarball containing the files of an
installed package which can be unpacked rather than installing from
sources.
</p>
<p>A package is <strong>not</strong><a name="DOCF1" href="#FOOT1"><sup>1</sup></a> a
<em>library</em>.  The latter is used in two senses in R documentation.
</p>
<ul>
<li> A directory into which packages are installed, e.g.
<samp>/usr/lib/R/library</samp>: in that sense it is sometimes referred to as
a <em>library directory</em> or <em>library tree</em> (since the library is
a directory which contains packages as directories, which themselves
contain directories).

</li><li> That used by the operating system, as a shared, dynamic or static
library or (especially on Windows) a DLL, where the second L stands for
&lsquo;library&rsquo;.  Installed packages may contain compiled code in what is
known on Unix-alikes as a <em>shared object</em> and on Windows as a DLL.
The concept of a <em>shared library</em> (<em>dynamic library</em> on macOS)
as a collection of compiled code to which a package might link is also
used, especially for R itself on some platforms. On most platforms
these concepts are interchangeable (shared objects and DLLs can both be
loaded into the R process and be linked against), but macOS
distinguishes between shared objects (extension <samp>.so</samp>) and dynamic
libraries (extension <samp>.dylib</samp>).

</li></ul>

<p>There are a number of well-defined operations on source packages.
</p>
<ul>
<li> The most common is <em>installation</em> which takes a source package and
installs it in a library using <code>R CMD INSTALL</code> or
<code>install.packages</code>.

</li><li> Source packages can be <em>built</em>.  This involves taking a source
directory and creating a tarball ready for distribution, including
cleaning it up and creating PDF documentation from any <em>vignettes</em>
it may contain.  Source packages (and most often tarballs) can be
<em>checked</em>, when a test installation is done and tested (including
running its examples); also, the contents of the package are tested in
various ways for consistency and portability.

</li><li> <em>Compilation</em> is not a correct term for a package.  Installing a
source package which contains C, C++ or Fortran code will involve
compiling that code.  There is also the possibility of &lsquo;byte&rsquo; compiling
the R code in a package (using the facilities of package
<strong>compiler</strong>): already base and recommended packages are normally
byte-compiled and this can be specified for other packages.  So
<em>compiling</em> a package may come to mean byte-compiling its R
code.

</li><li> It used to be unambiguous to talk about <em>loading</em> an installed
package using <code>library()</code>, but since the advent of package
namespaces this has been less clear: people now often talk about
<em>loading</em> the package&rsquo;s namespace and then <em>attaching</em> the
package so it becomes visible on the search path.  Function
<code>library</code> performs both steps, but a package&rsquo;s namespace can be
loaded without the package being attached (for example by calls like
<code>splines::ns</code>).

</li></ul>

<p>The concept of <em>lazy loading</em> of code or data is mentioned at
several points.  This is part of the installation, always selected for
R code but optional for data.  When used the R objects of the
package are created at installation time and stored in a database in the
<samp>R</samp> directory of the installed package, being loaded into the
session at first use.  This makes the R session start up faster and
use less (virtual) memory.
(For technical details,
see <a href="http://cran.r-project.org/doc/manuals/R-ints.html#Lazy-loading">Lazy loading</a> in <cite>R Internals</cite>.)
</p>
<a name="index-CRAN"></a>
<p><acronym>CRAN</acronym> is a network of WWW sites holding the R distributions
and contributed code, especially R packages.  Users of R are
encouraged to join in the collaborative project and to submit their own
packages to <acronym>CRAN</acronym>: current instructions are linked from
<a href="https://CRAN.R-project.org/banner.shtml#submitting">https://CRAN.R-project.org/banner.shtml#submitting</a>.
</p>

<table summary="" class="menu" border="0" cellspacing="0">
<tr><td align="left" valign="top">&bull; <a href="#Package-structure" accesskey="1">Package structure</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Configure-and-cleanup" accesskey="2">Configure and cleanup</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Checking-and-building-packages" accesskey="3">Checking and building packages</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Writing-package-vignettes" accesskey="4">Writing package vignettes</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Package-namespaces" accesskey="5">Package namespaces</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Writing-portable-packages" accesskey="6">Writing portable packages</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Diagnostic-messages" accesskey="7">Diagnostic messages</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Internationalization" accesskey="8">Internationalization</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#CITATION-files" accesskey="9">CITATION files</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Package-types">Package types</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Services">Services</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
</table>

<hr>
<a name="Package-structure"></a>
<div class="header">
<p>
Next: <a href="#Configure-and-cleanup" accesskey="n" rel="next">Configure and cleanup</a>, Previous: <a href="#Creating-R-packages" accesskey="p" rel="prev">Creating R packages</a>, Up: <a href="#Creating-R-packages" accesskey="u" rel="up">Creating R packages</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Package-structure-1"></a>
<h3 class="section">1.1 Package structure</h3>
<a name="index-Package-structure"></a>

<p>The sources of an R package consists of a subdirectory containing a
files <samp>DESCRIPTION</samp> and <samp>NAMESPACE</samp>, and the subdirectories
<samp>R</samp>, <samp>data</samp>, <samp>demo</samp>, <samp>exec</samp>, <samp>inst</samp>,
<samp>man</samp>, <samp>po</samp>, <samp>src</samp>, <samp>tests</samp>, <samp>tools</samp> and
<samp>vignettes</samp> (some of which can be missing, but which should not be
empty).  The package subdirectory may also contain files <samp>INDEX</samp>,
<samp>configure</samp>, <samp>cleanup</samp>, <samp>LICENSE</samp>, <samp>LICENCE</samp> and
<samp>NEWS</samp>.  Other files such as <samp>INSTALL</samp> (for non-standard
installation instructions), <samp>README</samp>/<samp>README.md</samp><a name="DOCF2" href="#FOOT2"><sup>2</sup></a>, or <samp>ChangeLog</samp> will be ignored by R, but may
be useful to end users.  The utility <code>R CMD build</code> may add files
in a <samp>build</samp> directory (but this should not be used for other
purposes).
</p>
<p>Except where specifically mentioned,<a name="DOCF3" href="#FOOT3"><sup>3</sup></a> packages should not contain
Unix-style &lsquo;hidden&rsquo; files/directories (that is, those whose name starts
with a dot).
</p>
<p>The <samp>DESCRIPTION</samp> and <samp>INDEX</samp> files are described in the
subsections below.  The <samp>NAMESPACE</samp> file is described in the
section on <a href="#Package-namespaces">Package namespaces</a>.
</p>
<a name="index-configure-file"></a>
<a name="index-cleanup-file"></a>

<p>The optional files <samp>configure</samp> and <samp>cleanup</samp> are (Bourne)
shell scripts which are, respectively, executed before and (if option
<samp>--clean</samp> was given) after installation on Unix-alikes, see
<a href="#Configure-and-cleanup">Configure and cleanup</a>.  The analogues on Windows are
<samp>configure.win</samp> and <samp>cleanup.win</samp>.
</p>
<p>For the conventions for files <samp>NEWS</samp> and <samp>ChangeLog</samp> in the
<acronym>GNU</acronym> project see
<a href="https://www.gnu.org/prep/standards/standards.html#Documentation">https://www.gnu.org/prep/standards/standards.html#Documentation</a>.
</p>
<p>The package subdirectory should be given the same name as the package.
Because some file systems (e.g., those on Windows and by default on OS
X) are not case-sensitive, to maintain portability it is strongly
recommended that case distinctions not be used to distinguish different
packages.  For example, if you have a package named <samp>foo</samp>, do not
also create a package named <samp>Foo</samp>.
</p>
<p>To ensure that file names are valid across file systems and supported
operating systems, the <acronym>ASCII</acronym> control characters as well as the
characters &lsquo;<samp>&quot;</samp>&rsquo;, &lsquo;<samp>*</samp>&rsquo;, &lsquo;<samp>:</samp>&rsquo;, &lsquo;<samp>/</samp>&rsquo;, &lsquo;<samp>&lt;</samp>&rsquo;, &lsquo;<samp>&gt;</samp>&rsquo;,
&lsquo;<samp>?</samp>&rsquo;, &lsquo;<samp>\</samp>&rsquo;, and &lsquo;<samp>|</samp>&rsquo; are not allowed in file names.  In
addition, files with names &lsquo;<samp>con</samp>&rsquo;, &lsquo;<samp>prn</samp>&rsquo;, &lsquo;<samp>aux</samp>&rsquo;,
&lsquo;<samp>clock$</samp>&rsquo;, &lsquo;<samp>nul</samp>&rsquo;, &lsquo;<samp>com1</samp>&rsquo; to &lsquo;<samp>com9</samp>&rsquo;, and &lsquo;<samp>lpt1</samp>&rsquo;
to &lsquo;<samp>lpt9</samp>&rsquo; after conversion to lower case and stripping possible
&ldquo;extensions&rdquo; (e.g., &lsquo;<samp>lpt5.foo.bar</samp>&rsquo;), are disallowed.  Also, file
names in the same directory must not differ only by case (see the
previous paragraph).  In addition, the basenames of &lsquo;<samp>.Rd</samp>&rsquo; files may
be used in URLs and so must be <acronym>ASCII</acronym> and not contain <code>%</code>.
For maximal portability filenames should only contain only
<acronym>ASCII</acronym> characters not excluded already (that is
<code>A-Za-z0-9._!#$%&amp;+,;=@^(){}'[]</code> &mdash; we exclude space as many
utilities do not accept spaces in file paths): non-English alphabetic
characters cannot be guaranteed to be supported in all locales.  It
would be good practice to avoid the shell metacharacters
<code>(){}'[]$~</code>: <code>~</code> is also used as part of &lsquo;8.3&rsquo; filenames on
Windows.  In addition, packages are normally distributed as tarballs,
and these have a limit on path lengths: for maximal portability 100
bytes.
</p>
<p>A source package if possible should not contain binary executable files:
they are not portable, and a security risk if they are of the
appropriate architecture.  <code>R CMD check</code> will warn about
them<a name="DOCF4" href="#FOOT4"><sup>4</sup></a> unless they are listed (one filepath per line) in a file
<samp>BinaryFiles</samp> at the top level of the package.  Note that
<acronym>CRAN</acronym> will not accept submissions containing binary files
even if they are listed.
</p>
<p>The R function <code>package.skeleton</code> can help to create the
structure for a new package: see its help page for details.
</p>
<table summary="" class="menu" border="0" cellspacing="0">
<tr><td align="left" valign="top">&bull; <a href="#The-DESCRIPTION-file" accesskey="1">The DESCRIPTION file</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Licensing" accesskey="2">Licensing</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Package-Dependencies" accesskey="3">Package Dependencies</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#The-INDEX-file" accesskey="4">The INDEX file</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Package-subdirectories" accesskey="5">Package subdirectories</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Data-in-packages" accesskey="6">Data in packages</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Non_002dR-scripts-in-packages" accesskey="7">Non-R scripts in packages</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Specifying-URLs" accesskey="8">Specifying URLs</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
</table>

<hr>
<a name="The-DESCRIPTION-file"></a>
<div class="header">
<p>
Next: <a href="#Licensing" accesskey="n" rel="next">Licensing</a>, Previous: <a href="#Package-structure" accesskey="p" rel="prev">Package structure</a>, Up: <a href="#Package-structure" accesskey="u" rel="up">Package structure</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="The-DESCRIPTION-file-1"></a>
<h4 class="subsection">1.1.1 The <samp>DESCRIPTION</samp> file</h4>
<a name="index-DESCRIPTION-file"></a>

<p>The <samp>DESCRIPTION</samp> file contains basic information about the package
in the following format:
</p>
<blockquote>
<table summary="" class="cartouche" border="1"><tr><td>
<div class="smallexample">
<pre class="smallexample">Package: pkgname
Version: 0.5-1
Date: 2015-01-01
Title: My First Collection of Functions
Authors@R: c(person(&quot;Joe&quot;, &quot;Developer&quot;, role = c(&quot;aut&quot;, &quot;cre&quot;),
                     email = &quot;Joe.Developer@some.domain.net&quot;),
              person(&quot;Pat&quot;, &quot;Developer&quot;, role = &quot;aut&quot;),
              person(&quot;A.&quot;, &quot;User&quot;, role = &quot;ctb&quot;,
                     email = &quot;A.User@whereever.net&quot;))
Author: Joe Developer [aut, cre],
  Pat Developer [aut],
  A. User [ctb]
Maintainer: Joe Developer &lt;Joe.Developer@some.domain.net&gt;
Depends: R (&gt;= 3.1.0), nlme
Suggests: MASS
Description: A (one paragraph) description of what
  the package does and why it may be useful.
License: GPL (&gt;= 2)
URL: https://www.r-project.org, http://www.another.url
BugReports: https://pkgname.bugtracker.url
</pre></div>
</td></tr></table>
</blockquote>

<p>The format is that of a version of a &lsquo;Debian Control File&rsquo; (see the help
for &lsquo;<samp>read.dcf</samp>&rsquo; and
<a href="https://www.debian.org/doc/debian-policy/ch-controlfields.html">https://www.debian.org/doc/debian-policy/ch-controlfields.html</a>:
R does not require encoding in UTF-8 and does not support comments
starting with &lsquo;<samp>#</samp>&rsquo;).  Fields start with an <acronym>ASCII</acronym> name
immediately followed by a colon: the value starts after the colon and a
space.  Continuation lines (for example, for descriptions longer than
one line) start with a space or tab.  Field names are case-sensitive:
all those used by R are capitalized.
</p>
<p>For maximal portability, the <samp>DESCRIPTION</samp> file should be written
entirely in <acronym>ASCII</acronym> &mdash; if this is not possible it must contain
an &lsquo;<samp>Encoding</samp>&rsquo; field (see below).
</p>
<p>Several optional fields take <em>logical values</em>: these can be
specified as &lsquo;<samp>yes</samp>&rsquo;, &lsquo;<samp>true</samp>&rsquo;, &lsquo;<samp>no</samp>&rsquo; or &lsquo;<samp>false</samp>&rsquo;:
capitalized values are also accepted.
</p>
<p>The &lsquo;<samp>Package</samp>&rsquo;, &lsquo;<samp>Version</samp>&rsquo;, &lsquo;<samp>License</samp>&rsquo;, &lsquo;<samp>Description</samp>&rsquo;,
&lsquo;<samp>Title</samp>&rsquo;, &lsquo;<samp>Author</samp>&rsquo;, and &lsquo;<samp>Maintainer</samp>&rsquo; fields are mandatory,
all other fields are optional.  Fields &lsquo;<samp>Author</samp>&rsquo; and
&lsquo;<samp>Maintainer</samp>&rsquo; can be auto-generated from &lsquo;<samp>Authors@R</samp>&rsquo;, and may
be omitted if the latter is provided: however if they are not
<acronym>ASCII</acronym> we recommend that they are provided.
</p>
<p>The mandatory &lsquo;<samp>Package</samp>&rsquo; field gives the name of the package.  This
should contain only (<acronym>ASCII</acronym>) letters, numbers and dot, have at
least two characters and start with a letter and not end in a dot.  If
it needs explaining, this should be done in the &lsquo;<samp>Description</samp>&rsquo; field
(and not the &lsquo;<samp>Title</samp>&rsquo; field).
</p>
<p>The mandatory &lsquo;<samp>Version</samp>&rsquo; field gives the version of the package.
This is a sequence of at least <em>two</em> (and usually three)
non-negative integers separated by single &lsquo;<samp>.</samp>&rsquo; or &lsquo;<samp>-</samp>&rsquo;
characters.  The canonical form is as shown in the example, and a
version such as &lsquo;<samp>0.01</samp>&rsquo; or &lsquo;<samp>0.01.0</samp>&rsquo; will be handled as if it
were &lsquo;<samp>0.1-0</samp>&rsquo;.  It is <strong>not</strong> a decimal number, so for example
<code>0.9 &lt; 0.75</code> since <code>9 &lt; 75</code>.
</p>
<p>The mandatory &lsquo;<samp>License</samp>&rsquo; field is discussed in the next subsection.
</p>
<p>The mandatory &lsquo;<samp>Title</samp>&rsquo; field should give a <em>short</em> description
of the package.  Some package listings may truncate the title to 65
characters.  It should use <em>title case</em> (that is, use capitals for
the principal words: <code>tools::toTitleCase</code> can help you with this),
not use any markup, not have any continuation lines, and not end in a
period (unless part of &hellip;).  Do not repeat the package name: it is
often used prefixed by the name.  Refer to other packages and external
software in single quotes, and to book titles (and similar) in double
quotes.
</p>
<p>The mandatory &lsquo;<samp>Description</samp>&rsquo; field should give a
<em>comprehensive</em> description of what the package does.  One can use
several (complete) sentences, but only one paragraph. It should be
intelligible to all the intended readership (e.g. for a <acronym>CRAN</acronym>
package to all <acronym>CRAN</acronym> users).  It is good practice not to start
with the package name, &lsquo;This package&rsquo; or similar.  As with the
&lsquo;<samp>Title</samp>&rsquo; field, double quotes should be used for quotations
(including titles of books and articles), and single quotes for
non-English usage, including names of other packages and external
software.  This field should also be used for explaining the package
name if necessary.  URLs should be enclosed in angle brackets, e.g.
&lsquo;<samp>&lt;https://www.r-project.org&gt;</samp>&rsquo;: see also <a href="#Specifying-URLs">Specifying URLs</a>.
</p>
<p>The mandatory &lsquo;<samp>Author</samp>&rsquo; field describes who wrote <em>the
package</em>.  It is a plain text field intended for human readers, but not
for automatic processing (such as extracting the email addresses of all
listed contributors: for that use &lsquo;<samp>Authors@R</samp>&rsquo;).  Note that all
significant contributors must be included: if you wrote an R wrapper
for the work of others included in the <samp>src</samp> directory, you are not
the sole (and maybe not even the main) author.
</p>
<p>The mandatory &lsquo;<samp>Maintainer</samp>&rsquo; field should give a <em>single</em> name
followed by a <em>valid</em> (RFC 2822) email address in angle brackets.  It
should not end in a period or comma.  This field is what is reported by
the <code>maintainer</code> function and used by <code>bug.report</code>.  For a
<acronym>CRAN</acronym> package it should be a <em>person</em>, not a mailing list
and not a corporate entity: do ensure that it is valid and will remain
valid for the lifetime of the package.
</p>
<p>Note that the <em>display name</em> (the part before the address in angle
brackets) should be enclosed in double quotes if it contains
non-alphanumeric characters such as comma or period.  (The current
standard, RFC 5322, allows periods but RFC 2822 did not.)
</p>
<p>Both &lsquo;<samp>Author</samp>&rsquo; and &lsquo;<samp>Maintainer</samp>&rsquo; fields can be omitted if a
suitable &lsquo;<samp>Authors@R</samp>&rsquo; field is given.  This field can be used to
provide a refined and machine-readable description of the package
&ldquo;authors&rdquo; (in particular specifying their precise <em>roles</em>), via
suitable R code. It should create an object of class <code>&quot;person&quot;</code>,
by either a call to <code>person</code> or a series of calls (one per
&ldquo;author&rdquo;) concatenated by <code>c()</code>: see the example
<samp>DESCRIPTION</samp> file above.  The roles can include &lsquo;<samp>&quot;aut&quot;</samp>&rsquo;
(author) for full authors, &lsquo;<samp>&quot;cre&quot;</samp>&rsquo; (creator) for the package
maintainer, and &lsquo;<samp>&quot;ctb&quot;</samp>&rsquo; (contributor) for other contributors,
&lsquo;<samp>&quot;cph&quot;</samp>&rsquo; (copyright holder), among others.  See <code>?person</code> for
more information.  Note that no role is assumed by default.
Auto-generated package citation information takes advantage of this
specification. The &lsquo;<samp>Author</samp>&rsquo; and &lsquo;<samp>Maintainer</samp>&rsquo; fields are
auto-generated from it if needed when building<a name="DOCF5" href="#FOOT5"><sup>5</sup></a> or installing.
</p>
<a name="index-COPYRIGHTS"></a>
<p>An optional &lsquo;<samp>Copyright</samp>&rsquo; field can be used where the copyright
holder(s) are not the authors.  If necessary, this can refer to an
installed file: the convention is to use file <samp>inst/COPYRIGHTS</samp>.
</p>
<p>The optional &lsquo;<samp>Date</samp>&rsquo; field gives the <em>release date</em> of the
current version of the package.  It is strongly recommended<a name="DOCF6" href="#FOOT6"><sup>6</sup></a> to use the &lsquo;<samp>yyyy-mm-dd</samp>&rsquo; format conforming to the ISO
8601 standard.
</p>
<p>The &lsquo;<samp>Depends</samp>&rsquo;, &lsquo;<samp>Imports</samp>&rsquo;, &lsquo;<samp>Suggests</samp>&rsquo;, &lsquo;<samp>Enhances</samp>&rsquo;,
&lsquo;<samp>LinkingTo</samp>&rsquo; and &lsquo;<samp>Additional_repositories</samp>&rsquo; fields are discussed
in a later subsection.
</p>
<p>Dependencies external to the R system should be listed in the
&lsquo;<samp>SystemRequirements</samp>&rsquo; field, possibly amplified in a separate
<samp>README</samp> file.
</p>
<p>The &lsquo;<samp>URL</samp>&rsquo; field may give a list of <acronym>URL</acronym>s
separated by commas or whitespace, for example the homepage of the
author or a page where additional material describing the software can
be found.  These <acronym>URL</acronym>s are converted to active hyperlinks in
<acronym>CRAN</acronym> package listings.  See <a href="#Specifying-URLs">Specifying URLs</a>.
</p>
<p>The &lsquo;<samp>BugReports</samp>&rsquo; field may contain a single <acronym>URL</acronym> to which
bug reports about the package should be submitted.  This <acronym>URL</acronym>
will be used by <code>bug.report</code> instead of sending an email to the
maintainer.  A browser is opened for a &lsquo;<samp>http://</samp>&rsquo; or &lsquo;<samp>https://</samp>&rsquo;
<acronym>URL</acronym> (and others are ignored).
</p>
<p>Base and recommended packages (i.e., packages contained in the R
source distribution or available from <acronym>CRAN</acronym> and recommended to
be included in every binary distribution of R) have a &lsquo;<samp>Priority</samp>&rsquo;
field with value &lsquo;<samp>base</samp>&rsquo; or &lsquo;<samp>recommended</samp>&rsquo;, respectively.  These
priorities must not be used by other packages.
</p>
<p>A &lsquo;<samp>Collate</samp>&rsquo; field can be used for controlling the collation order
for the R code files in a package when these are processed for
package installation.  The default is to collate according to the
&lsquo;<samp>C</samp>&rsquo; locale.  If present, the collate specification must list
<em>all</em> R code files in the package (taking possible OS-specific
subdirectories into account, see <a href="#Package-subdirectories">Package subdirectories</a>) as a
whitespace separated list of file paths relative to the <samp>R</samp>
subdirectory.
Paths containing white space or quotes need to be quoted.  An
OS-specific collation field (&lsquo;<samp>Collate.unix</samp>&rsquo; or
&lsquo;<samp>Collate.windows</samp>&rsquo;) will be used in preference to &lsquo;<samp>Collate</samp>&rsquo;.
</p>
<p>The &lsquo;<samp>LazyData</samp>&rsquo; logical field controls whether the R datasets use
lazy-loading.  A &lsquo;<samp>LazyLoad</samp>&rsquo; field was used in versions prior to
2.14.0, but now is ignored.
</p>
<p>The &lsquo;<samp>KeepSource</samp>&rsquo; logical field controls if the package code is sourced
using <code>keep.source = TRUE</code> or <code>FALSE</code>: it might be needed
exceptionally for a package designed to always be used with
<code>keep.source = TRUE</code>.
</p>
<p>The &lsquo;<samp>ByteCompile</samp>&rsquo; logical field controls if the package code is to
be byte-compiled on installation: the default is currently not to, so
this may be useful for a package known to benefit particularly from
byte-compilation (which can take quite a long time and increases the
installed size of the package).  It is used for the recommended
packages, as they are byte-compiled when R is installed and for
consistency should be byte-compiled when updated. This can be overridden
by installing with flag <samp>--no-byte-compile</samp>.
</p>
<p>The &lsquo;<samp>ZipData</samp>&rsquo; logical field was used to control whether the automatic
Windows build would zip up the data directory or not prior to R
2.13.0: it is now ignored.
</p>
<p>The &lsquo;<samp>Biarch</samp>&rsquo; logical field is used on Windows to select the
<code>INSTALL</code> option <samp>--force-biarch</samp> for this package.
(Introduced in R 3.0.0.)
</p>
<p>The &lsquo;<samp>BuildVignettes</samp>&rsquo; logical field can be set to a false value to
stop <code>R CMD build</code> from attempting to build the vignettes, as
well as preventing<a name="DOCF7" href="#FOOT7"><sup>7</sup></a>  <code>R CMD check</code> from testing
this.  This should only be used exceptionally, for example if the PDFs
include large figures which are not part of the package sources (and
hence only in packages which do not have an Open Source license).
</p>
<p>The &lsquo;<samp>VignetteBuilder</samp>&rsquo; field names (in a comma-separated list)
packages that provide an engine for building vignettes.  These may
include the current package, or ones listed in &lsquo;<samp>Depends</samp>&rsquo;,
&lsquo;<samp>Suggests</samp>&rsquo; or &lsquo;<samp>Imports</samp>&rsquo;. The <strong>utils</strong> package is always
implicitly appended.  See <a href="#Non_002dSweave-vignettes">Non-Sweave vignettes</a> for
details.
</p>
<p>If the <samp>DESCRIPTION</samp> file is not entirely in <acronym>ASCII</acronym> it
should contain an &lsquo;<samp>Encoding</samp>&rsquo; field specifying an encoding.  This is
used as the encoding of the <samp>DESCRIPTION</samp> file itself and of the
<samp>R</samp> and <samp>NAMESPACE</samp> files, and as the default encoding of
<samp>.Rd</samp> files.  The examples are assumed to be in this encoding when
running <code>R CMD check</code>, and it is used for the encoding of the
<code>CITATION</code> file.  Only encoding names <code>latin1</code>, <code>latin2</code>
and <code>UTF-8</code> are known to be portable.  (Do not specify an encoding
unless one is actually needed: doing so makes the package <em>less</em>
portable.  If a package has a specified encoding, you should run
<code>R CMD build</code> etc in a locale using that encoding.)
</p>
<p>The &lsquo;<samp>NeedsCompilation</samp>&rsquo; field should be set to <code>&quot;yes&quot;</code> if the
package contains code which to be compiled, otherwise <code>&quot;no&quot;</code> (when
the package could be installed from source on any platform without
additional tools).  This is used by <code>install.packages(type =
&quot;both&quot;)</code> in R &gt;= 2.15.2 on platforms where binary packages are the
norm: it is normally set by <code>R CMD build</code> or the repository
assuming compilation is required if and only if the package has a
<samp>src</samp> directory.
</p>
<p>The &lsquo;<samp>OS_type</samp>&rsquo; field specifies the OS(es) for which the
package is intended.  If present, it should be one of <code>unix</code> or
<code>windows</code>, and indicates that the package can only be installed
on a platform with &lsquo;<samp>.Platform$OS.type</samp>&rsquo; having that value.
</p>
<p>The &lsquo;<samp>Type</samp>&rsquo; field specifies the type of the package:
see <a href="#Package-types">Package types</a>.
</p>
<p>One can add subject classifications for the content of the package using
the fields &lsquo;<samp>Classification/ACM</samp>&rsquo; or &lsquo;<samp>Classification/ACM-2012</samp>&rsquo;
(using the Computing Classification System of the Association for
Computing Machinery, <a href="http://www.acm.org/about/class/">http://www.acm.org/about/class/</a>; the former refers
to the 1998 version), &lsquo;<samp>Classification/JEL</samp>&rsquo; (the Journal of Economic
Literature Classification System,
<a href="https://www.aeaweb.org/econlit/jelCodes.php">https://www.aeaweb.org/econlit/jelCodes.php</a>, or
&lsquo;<samp>Classification/MSC</samp>&rsquo; or &lsquo;<samp>Classification/MSC-2010</samp>&rsquo; (the
Mathematics Subject Classification of the American Mathematical Society,
<a href="http://www.ams.org/msc/">http://www.ams.org/msc/</a>; the former refers to the 2000 version).
The subject classifications should be comma-separated lists of the
respective classification codes, e.g., &lsquo;<samp>Classification/ACM: G.4,
H.2.8, I.5.1</samp>&rsquo;.
</p>
<p>A &lsquo;<samp>Language</samp>&rsquo; field can be used to indicate if the package
documentation is not in English: this should be a comma-separated list
of standard (not private use or grandfathered) IETF language tags as
currently defined by RFC 5646
(<a href="https://tools.ietf.org/html/rfc5646">https://tools.ietf.org/html/rfc5646</a>, see also
<a href="https://en.wikipedia.org/wiki/IETF_language_tag">https://en.wikipedia.org/wiki/IETF_language_tag</a>), i.e., use
language subtags which in essence are 2-letter ISO 639-1
(<a href="https://en.wikipedia.org/wiki/ISO_639-1">https://en.wikipedia.org/wiki/ISO_639-1</a>) or 3-letter ISO
639-3 (<a href="https://en.wikipedia.org/wiki/ISO_639-3">https://en.wikipedia.org/wiki/ISO_639-3</a>) language
codes.
</p>
<p>An &lsquo;<samp>RdMacros</samp>&rsquo; field can be used to hold a comma-separated list of
packages from which the current package will import <samp>Rd</samp> macro
definitions.  These will be imported after the system macros, in the
order listed in the &lsquo;<samp>RdMacros</samp>&rsquo; field, before any macro definitions
in the current package are loaded.  Macro definitions in individual
<samp>.Rd</samp> files in the <samp>man</samp> directory are loaded last, and are
local to later parts of that file.  In case of duplicates, the last
loaded definition will be used<a name="DOCF8" href="#FOOT8"><sup>8</sup></a>  Both <code>R CMD
Rd2pdf</code> and <code>R CMD Rdconv</code> have an optional flag
<samp>--RdMacros=pkglist</samp>.  The option is also a comma-separated list
of package names, and has priority over the value given in
<samp>DESCRIPTION</samp>.  Packages using <samp>Rd</samp> macros should depend on
R 3.2.0 or later.
</p>
<blockquote>
<p><b>Note:</b> There should be no &lsquo;<samp>Built</samp>&rsquo; or &lsquo;<samp>Packaged</samp>&rsquo; fields, as these are
added by the package management tools.
</p></blockquote>

<p>There is no restriction on the use of other fields not mentioned here
(but using other capitalizations of these field names would cause
confusion).  Fields <code>Note</code>, <code>Contact</code> (for contacting the
authors/developers) and <code>MailingList</code> are in common use. Some
repositories (including <acronym>CRAN</acronym> and R-forge) add their own
fields.
</p>


<hr>
<a name="Licensing"></a>
<div class="header">
<p>
Next: <a href="#Package-Dependencies" accesskey="n" rel="next">Package Dependencies</a>, Previous: <a href="#The-DESCRIPTION-file" accesskey="p" rel="prev">The DESCRIPTION file</a>, Up: <a href="#Package-structure" accesskey="u" rel="up">Package structure</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Licensing-1"></a>
<h4 class="subsection">1.1.2 Licensing</h4>

<p>Licensing for a package which might be distributed is an important but
potentially complex subject.
</p>
<p>It is very important that you include license information!  Otherwise,
it may not even be legally correct for others to distribute copies of
the package, let alone use it.
</p>
<p>The package management tools use the concept of
&lsquo;free or open source software&rsquo;
(FOSS, e.g., <a href="https://en.wikipedia.org/wiki/FOSS">https://en.wikipedia.org/wiki/FOSS</a>)
licenses: the idea being that some users of R and its packages want
to restrict themselves to such software.  Others need to ensure that
there are no restrictions stopping them using a package, e.g.
forbidding commercial or military use.  It is a central tenet of FOSS
software that there are no restrictions on users nor usage.
</p>
<p>Do not use the &lsquo;<samp>License</samp>&rsquo; field for information on copyright
holders: if needed, use a &lsquo;<samp>Copyright</samp>&rsquo; field.
</p>
<p>The mandatory &lsquo;<samp>License</samp>&rsquo; field in the <samp>DESCRIPTION</samp> file should
specify the license of the package in a standardized form.  Alternatives
are indicated <em>via</em> vertical bars.  Individual specifications must
be one of
</p><ul>
<li> One of the &ldquo;standard&rdquo; short specifications
<div class="example">
<pre class="example">GPL-2 GPL-3 LGPL-2 LGPL-2.1 LGPL-3 AGPL-3 Artistic-2.0
BSD_2_clause BSD_3_clause MIT
</pre></div>
<p>as made available <em>via</em> <a href="https://www.R-project.org/Licenses/">https://www.R-project.org/Licenses/</a> and
contained in subdirectory <samp>share/licenses</samp> of the R source or home
directory.
</p></li><li> The names or abbreviations of other licenses contained in the license
data base in file <samp>share/licenses/license.db</samp> in the R source or
home directory, possibly (for versioned licenses) followed by a version
restriction of the form &lsquo;<samp>(<var>op</var> <var>v</var>)</samp>&rsquo; with &lsquo;<samp><var>op</var></samp>&rsquo; one of
the comparison operators &lsquo;<samp>&lt;</samp>&rsquo;, &lsquo;<samp>&lt;=</samp>&rsquo;, &lsquo;<samp>&gt;</samp>&rsquo;, &lsquo;<samp>&gt;=</samp>&rsquo;,
&lsquo;<samp>==</samp>&rsquo;, or &lsquo;<samp>!=</samp>&rsquo; and &lsquo;<samp><var>v</var></samp>&rsquo; a numeric version specification
(strings of non-negative integers separated by &lsquo;<samp>.</samp>&rsquo;), possibly
combined <em>via</em> &lsquo;<samp>,</samp>&rsquo; (see below for an example).  For versioned
licenses, one can also specify the name followed by the version, or
combine an existing abbreviation and the version with a &lsquo;<samp>-</samp>&rsquo;.

<p>Abbreviations <code>GPL</code> and <code>LGPL</code> are ambiguous and usually taken
to mean any version of the license: but it is better not to use them.
</p></li><li> One of the strings &lsquo;<samp>file LICENSE</samp>&rsquo; or &lsquo;<samp>file LICENCE</samp>&rsquo; referring
to a file named <samp>LICENSE</samp> or <samp>LICENCE</samp> in the package (source
and installation) top-level directory.
</li><li> The string &lsquo;<samp>Unlimited</samp>&rsquo;, meaning that there are no restrictions on
distribution or use other than those imposed by relevant laws (including
copyright laws).
</li></ul>

<p>If a package license <em>restricts</em> a base license (where permitted,
e.g., using GPL-3 or AGPL-3 with an attribution clause), the additional
terms should be placed in file <samp>LICENSE</samp> (or <samp>LICENCE</samp>), and
the string &lsquo;<samp>+ file LICENSE</samp>&rsquo; (or &lsquo;<samp>+ file LICENCE</samp>&rsquo;,
respectively) should be appended to the corresponding individual license
specification.  Note that several commonly used licenses do not permit
restrictions: this includes GPL-2 and hence any specification which
includes it.
</p>
<p>Examples of standardized specifications include
</p><div class="example">
<pre class="example">License: GPL-2
License: LGPL (&gt;= 2.0, &lt; 3) | Mozilla Public License
License: GPL-2 | file LICENCE
License: GPL (&gt;= 2) | BSD_3_clause + file LICENSE
License: Artistic-2.0 | AGPL-3 + file LICENSE
</pre></div>
<p>Please note in particular that &ldquo;Public domain&rdquo; is not a valid license,
since it is not recognized in some jurisdictions.
</p>
<p>Please ensure that the license you choose also covers any dependencies
(including system dependencies) of your package: it is particularly
important that any restrictions on the use of such dependencies are
evident to people reading your <samp>DESCRIPTION</samp> file.
</p>
<p>Fields &lsquo;<samp>License_is_FOSS</samp>&rsquo; and &lsquo;<samp>License_restricts_use</samp>&rsquo; may be
added by repositories where information cannot be computed from the name
of the license.  &lsquo;<samp>License_is_FOSS: yes</samp>&rsquo; is used for licenses which
are known to be FOSS, and &lsquo;<samp>License_restricts_use</samp>&rsquo; can have values
&lsquo;<samp>yes</samp>&rsquo; or &lsquo;<samp>no</samp>&rsquo; if the <samp>LICENSE</samp> file is known to restrict
users or usage, or known not to.  These are used by, e.g., the
<code>available.packages</code> filters.
</p>

<a name="index-LICENSE-file"></a>
<a name="index-LICENCE-file"></a>
<p>The optional file <samp>LICENSE</samp>/<samp>LICENCE</samp> contains a copy of the
license of the package.  To avoid any confusion only include such a file
if it is referred to in the &lsquo;<samp>License</samp>&rsquo; field of the
<samp>DESCRIPTION</samp> file.
</p>
<p>Whereas you should feel free to include a license file in your
<em>source</em> distribution, please do not arrange to <em>install</em> yet
another copy of the <acronym>GNU</acronym> <samp>COPYING</samp> or <samp>COPYING.LIB</samp>
files but refer to the copies on
<a href="https://www.R-project.org/Licenses/">https://www.R-project.org/Licenses/</a> and included in the R
distribution (in directory <samp>share/licenses</samp>).  Since files named
<samp>LICENSE</samp> or <samp>LICENCE</samp> <em>will</em> be installed, do not use
these names for standard license files.  To include comments about the
licensing rather than the body of a license, use a file named something
like <samp>LICENSE.note</samp>.
</p>
<p>A few &ldquo;standard&rdquo; licenses are rather license templates which need
additional information to be completed <em>via</em> &lsquo;<samp>+ file LICENSE</samp>&rsquo;.
</p>
<hr>
<a name="Package-Dependencies"></a>
<div class="header">
<p>
Next: <a href="#The-INDEX-file" accesskey="n" rel="next">The INDEX file</a>, Previous: <a href="#Licensing" accesskey="p" rel="prev">Licensing</a>, Up: <a href="#Package-structure" accesskey="u" rel="up">Package structure</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Package-Dependencies-1"></a>
<h4 class="subsection">1.1.3 Package Dependencies</h4>

<p>The &lsquo;<samp>Depends</samp>&rsquo; field gives a comma-separated list of package names
which this package depends on.  Those packages will be attached before
the current package when <code>library</code> or <code>require</code> is called.
Each package name may be optionally followed by a comment in parentheses
specifying a version requirement.  The comment should contain a
comparison operator, whitespace and a valid version number,
e.g. &lsquo;<samp>MASS (&gt;= 3.1-20)</samp>&rsquo;.
</p>
<p>The &lsquo;<samp>Depends</samp>&rsquo; field can also specify a dependence on a certain
version of R &mdash; e.g., if the package works only with R
version 3.0.0 or later, include &lsquo;<samp>R (&gt;= 3.0.0)</samp>&rsquo; in the
&lsquo;<samp>Depends</samp>&rsquo; field.  You can also require a certain SVN revision for
R-devel or R-patched, e.g. &lsquo;<samp>R (&gt;= 2.14.0), R (&gt;= r56550)</samp>&rsquo;
requires a version later than R-devel of late July 2011 (including
released versions of 2.14.0).
</p>
<p>It makes no sense to declare a dependence on <code>R</code> without a version
specification, nor on the package <strong>base</strong>: this is an R package
and package <strong>base</strong> is always available.
</p>
<p>A package or &lsquo;<samp>R</samp>&rsquo; can appear more than once in the &lsquo;<samp>Depends</samp>&rsquo;
field, for example to give upper and lower bounds on acceptable versions.
</p>
<p>Both <code>library</code> and the R package checking facilities use this
field: hence it is an error to use improper syntax or misuse the
&lsquo;<samp>Depends</samp>&rsquo; field for comments on other software that might be
needed.  The R <code>INSTALL</code> facilities check if the version of
R used is recent enough for the package being installed, and the list
of packages which is specified will be attached (after checking version
requirements) before the current package.
</p>
<p>The &lsquo;<samp>Imports</samp>&rsquo; field lists packages whose namespaces are imported
from (as specified in the <samp>NAMESPACE</samp> file) but which do not need
to be attached.  Namespaces accessed by the &lsquo;<samp>::</samp>&rsquo; and &lsquo;<samp>:::</samp>&rsquo;
operators must be listed here, or in &lsquo;<samp>Suggests</samp>&rsquo; or &lsquo;<samp>Enhances</samp>&rsquo;
(see below).  Ideally this field will include all the standard packages
that are used, and it is important to include S4-using packages (as
their class definitions can change and the <samp>DESCRIPTION</samp> file is
used to decide which packages to re-install when this happens).
Packages declared in the &lsquo;<samp>Depends</samp>&rsquo; field should not also be in the
&lsquo;<samp>Imports</samp>&rsquo; field.  Version requirements can be specified and are
checked when the namespace is loaded (since R &gt;= 3.0.0).
</p>
<p>The &lsquo;<samp>Suggests</samp>&rsquo; field uses the same syntax as &lsquo;<samp>Depends</samp>&rsquo; and
lists packages that are not necessarily needed.  This includes packages
used only in examples, tests or vignettes (see <a href="#Writing-package-vignettes">Writing package vignettes</a>), and packages loaded in the body of functions.  E.g.,
suppose an example<a name="DOCF9" href="#FOOT9"><sup>9</sup></a> from
package <strong>foo</strong> uses a dataset from package <strong>bar</strong>. Then it is not
necessary to have <strong>bar</strong> use <strong>foo</strong> unless one wants to execute
all the examples/tests/vignettes: it is useful to have <strong>bar</strong>, but
not necessary.  Version requirements can be specified, and will be used
by <code>R CMD check</code>.
</p>
<p>Finally, the &lsquo;<samp>Enhances</samp>&rsquo; field lists packages &ldquo;enhanced&rdquo; by the
package at hand, e.g., by providing methods for classes from these
packages, or ways to handle objects from these packages (so several
packages have &lsquo;<samp>Enhances: chron</samp>&rsquo; because they can handle datetime
objects from <a href="https://CRAN.R-project.org/package=chron"><strong>chron</strong></a> even though they prefer R&rsquo;s native
datetime functions).  Version requirements can be specified, but are
currently not used.  Such packages cannot be required to check the
package: any tests which use them must be conditional on the presence
of the package.  (If your tests use e.g. a dataset from another
package it should be in &lsquo;<samp>Suggests</samp>&rsquo; and not &lsquo;<samp>Enhances</samp>&rsquo;.)
</p>
<p>The general rules are
</p>
<ul>
<li> A package should be listed in only one of these fields.
</li><li> Packages whose namespace only is needed to load the package using
<code>library(<var>pkgname</var>)</code> should be listed in the &lsquo;<samp>Imports</samp>&rsquo; field
and not in the &lsquo;<samp>Depends</samp>&rsquo; field.  Packages listed in <code>imports</code>
or <code>importFrom</code> directives in the <samp>NAMESPACE</samp> file should
almost always be in &lsquo;<samp>Imports</samp>&rsquo; and not &lsquo;<samp>Depends</samp>&rsquo;.
</li><li> Packages that need to be attached to successfully load the package using
<code>library(<var>pkgname</var>)</code> must be listed in the &lsquo;<samp>Depends</samp>&rsquo;
field.
</li><li> All packages that are needed<a name="DOCF10" href="#FOOT10"><sup>10</sup></a> to successfully run <code>R CMD check</code> on the
package must be listed in one of &lsquo;<samp>Depends</samp>&rsquo; or &lsquo;<samp>Suggests</samp>&rsquo; or
&lsquo;<samp>Imports</samp>&rsquo;.  Packages used to run examples or tests conditionally
(e.g. <em>via</em> <code>if(require(<var>pkgname</var>))</code>) should be listed
in &lsquo;<samp>Suggests</samp>&rsquo; or &lsquo;<samp>Enhances</samp>&rsquo;.  (This allows checkers to ensure
that all the packages needed for a complete check are installed.)
</li></ul>

<p>In particular, packages providing &ldquo;only&rdquo; data for examples or
vignettes should be listed in &lsquo;<samp>Suggests</samp>&rsquo; rather than &lsquo;<samp>Depends</samp>&rsquo;
in order to make lean installations possible.
</p>
<p>Version dependencies in the &lsquo;<samp>Depends</samp>&rsquo; and &lsquo;<samp>Imports</samp>&rsquo; fields are
used by <code>library</code> when it loads the package, and
<code>install.packages</code> checks versions for the &lsquo;<samp>Depends</samp>&rsquo;,
&lsquo;<samp>Imports</samp>&rsquo; and (for <code>dependencies = TRUE</code>) &lsquo;<samp>Suggests</samp>&rsquo;
fields.
</p>
<p>It is increasingly important that the information in these fields is
complete and accurate: it is for example used to compute which packages
depend on an updated package and which packages can safely be installed
in parallel.
</p>
<p>This scheme was developed before all packages had namespaces (R
2.14.0 in October 2011), and good practice changed once that was in
place.
</p>
<p>Field &lsquo;<samp>Depends</samp>&rsquo; should nowadays be used rarely, only for packages
which are intended to be put on the search path to make their facilities
available to the end user (and not to the package itself): for example
it makes sense that a user of package <a href="https://CRAN.R-project.org/package=latticeExtra"><strong>latticeExtra</strong></a> would want
the functions of package <a href="https://CRAN.R-project.org/package=lattice"><strong>lattice</strong></a> made available.
</p>
<p>Almost always packages mentioned in &lsquo;<samp>Depends</samp>&rsquo; should also be
imported from in the <samp>NAMESPACE</samp> file: this ensures that any needed
parts of those packages are available when some other package imports
the current package.
</p>
<p>The &lsquo;<samp>Imports</samp>&rsquo; field should not contain packages which are not
imported from (<em>via</em> the <samp>NAMESPACE</samp> file or <code>::</code> or
<code>:::</code> operators), as all the packages listed in that field need to
be installed for the current package to be installed.  (This is checked
by <code>R CMD check</code>.)
</p>
<p>R code in the package should call <code>library</code> or <code>require</code>
only exceptionally.  Such calls are never needed for packages listed in
&lsquo;<samp>Depends</samp>&rsquo; as they will already be on the search path.  It used to
be common practice to use <code>require</code> calls for packages listed in
&lsquo;<samp>Suggests</samp>&rsquo; in functions which used their functionality, but
nowadays it is better to access such functionality <em>via</em> <code>::</code>
calls.
</p>
<p>A package that wishes to make use of header files in other packages needs
to declare them as a comma-separated list in the field &lsquo;<samp>LinkingTo</samp>&rsquo;
in the <samp>DESCRIPTION</samp> file.  For example
</p>
<div class="example">
<pre class="example">LinkingTo: link1, link2
</pre></div>

<p>As from R 3.0.2 the &lsquo;<samp>LinkingTo</samp>&rsquo; field can have a version
requirement which is checked at installation.  (In earlier versions of
R it would cause the specification to be ignored.)
</p>
<p>Specifying a package in &lsquo;<samp>LinkingTo</samp>&rsquo; suffices if these are C++
headers containing source code or static linking is done at
installation: the packages do not need to be (and usually should not be)
listed in the &lsquo;<samp>Depends</samp>&rsquo; or &lsquo;<samp>Imports</samp>&rsquo; fields.  This includes
<acronym>CRAN</acronym> package <a href="https://CRAN.R-project.org/package=BH"><strong>BH</strong></a> and almost all users of
<a href="https://CRAN.R-project.org/package=RcppArmadillo"><strong>RcppArmadillo</strong></a> and <a href="https://CRAN.R-project.org/package=RcppEigen"><strong>RcppEigen</strong></a>.
</p>
<p>For another use of &lsquo;<samp>LinkingTo</samp>&rsquo; see <a href="#Linking-to-native-routines-in-other-packages">Linking to native routines in other packages</a>.
</p>
<p>The &lsquo;<samp>Additional_repositories</samp>&rsquo; field is a comma-separated list of
repository URLs where the packages named in the other fields may be
found.  It is currently used by <code>R CMD check</code> to check that the
packages can be found, at least as source packages (which can be
installed on any platform).
</p>
<table summary="" class="menu" border="0" cellspacing="0">
<tr><td align="left" valign="top">&bull; <a href="#Suggested-packages" accesskey="1">Suggested packages</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
</table>

<hr>
<a name="Suggested-packages"></a>
<div class="header">
<p>
Previous: <a href="#Package-Dependencies" accesskey="p" rel="prev">Package Dependencies</a>, Up: <a href="#Package-Dependencies" accesskey="u" rel="up">Package Dependencies</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Suggested-packages-1"></a>
<h4 class="subsubsection">1.1.3.1 Suggested packages</h4>

<p>Note that someone wanting to run the examples/tests/vignettes may not
have a suggested package available (and it may not even be possible to
install it for that platform).  The recommendation used to be to make
their use conditional <em>via</em> <code>if(require(&quot;<var>pkgname</var>&quot;))</code>):
this is fine if that conditioning is done in examples/tests/vignettes.
</p>
<p>However, using <code>require</code> for conditioning <em>in package code</em> is
not good practice as it alters the search path for the rest of the
session and relies on functions in that package not being masked by
other <code>require</code> or <code>library</code> calls.  It is better practice to
use code like
</p><div class="example">
<pre class="example">   if (requireNamespace(&quot;rgl&quot;, quietly = TRUE)) {
      rgl::plot3d(...)
   } else {
      ## do something else not involving rgl.
   }
</pre></div>
<p>Note the use of <code>rgl::</code> as that object would not necessarily be
visible (and if it is, it need not be the one from that namespace:
<code>plot3d</code> occurs in several other packages).  If the intention is to
give an error if the suggested package is not available, simply use
e.g. <code>rgl::plot3d</code>.
</p>
<p>Note that the recommendation to use suggested packages conditionally in
tests does also apply to packages used to manage test suites: a
notorious example was <a href="https://CRAN.R-project.org/package=testthat"><strong>testthat</strong></a> which in version 1.0.0 contained
illegal C++ code and hence could not be installed on standards-compliant
platforms.
</p>
<p>As noted above, packages in &lsquo;<samp>Enhances</samp>&rsquo; <em>must</em> be used
conditionally and hence objects within them should always be accessed
<em>via</em> <code>::</code>.
</p>
<hr>
<a name="The-INDEX-file"></a>
<div class="header">
<p>
Next: <a href="#Package-subdirectories" accesskey="n" rel="next">Package subdirectories</a>, Previous: <a href="#Package-Dependencies" accesskey="p" rel="prev">Package Dependencies</a>, Up: <a href="#Package-structure" accesskey="u" rel="up">Package structure</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="The-INDEX-file-1"></a>
<h4 class="subsection">1.1.4 The <samp>INDEX</samp> file</h4>
<a name="index-INDEX-file"></a>

<p>The optional file <samp>INDEX</samp> contains a line for each sufficiently
interesting object in the package, giving its name and a description
(functions such as print methods not usually called explicitly might not
be included).  Normally this file is missing and the corresponding
information is automatically generated from the documentation sources
(using <code>tools::Rdindex()</code>) when installing from source.
</p>
<p>The file is part of the information given by <code>library(help =
<var>pkgname</var>)</code>.
</p>
<p>Rather than editing this file, it is preferable to put customized
information about the package into an overview help page
(see <a href="#Documenting-packages">Documenting packages</a>) and/or a vignette (see <a href="#Writing-package-vignettes">Writing package vignettes</a>).
</p>
<hr>
<a name="Package-subdirectories"></a>
<div class="header">
<p>
Next: <a href="#Data-in-packages" accesskey="n" rel="next">Data in packages</a>, Previous: <a href="#The-INDEX-file" accesskey="p" rel="prev">The INDEX file</a>, Up: <a href="#Package-structure" accesskey="u" rel="up">Package structure</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Package-subdirectories-1"></a>
<h4 class="subsection">1.1.5 Package subdirectories</h4>
<a name="index-Package-subdirectories"></a>

<p>The <samp>R</samp> subdirectory contains R code files, only.  The code
files to be installed must start with an <acronym>ASCII</acronym> (lower or upper
case) letter or digit and have one of the extensions<a name="DOCF11" href="#FOOT11"><sup>11</sup></a> <samp>.R</samp>,
<samp>.S</samp>, <samp>.q</samp>, <samp>.r</samp>, or <samp>.s</samp>.  We recommend using
<samp>.R</samp>, as this extension seems to be not used by any other software.
It should be possible to read in the files using <code>source()</code>, so
R objects must be created by assignments.  Note that there need be no
connection between the name of the file and the R objects created by
it.  Ideally, the R code files should only directly assign R
objects and definitely should not call functions with side effects such
as <code>require</code> and <code>options</code>.  If computations are required to
create objects these can use code &lsquo;earlier&rsquo; in the package (see the
&lsquo;<samp>Collate</samp>&rsquo; field) plus functions in the &lsquo;<samp>Depends</samp>&rsquo; packages
provided that the objects created do not depend on those packages except
<em>via</em> namespace imports.
</p>
<p>Two exceptions are allowed: if the <samp>R</samp> subdirectory contains a file
<samp>sysdata.rda</samp> (a saved image of one or more R objects: please
use suitable compression as suggested by <code>tools::resaveRdaFiles</code>,
and see also the &lsquo;<samp>SysDataCompression</samp>&rsquo; <samp>DESCRIPTION</samp> field.)
this will be lazy-loaded into the namespace environment &ndash; this is
intended for system datasets that are not intended to be user-accessible
<em>via</em> <code>data</code>.  Also, files ending in &lsquo;<samp>.in</samp>&rsquo; will be
allowed in the <samp>R</samp> directory to allow a <samp>configure</samp> script to
generate suitable files.
</p>
<p>Only <acronym>ASCII</acronym> characters (and the control characters tab,
formfeed, LF and CR) should be used in code files.  Other characters are
accepted in comments<a name="DOCF12" href="#FOOT12"><sup>12</sup></a>, but then the comments may not
be readable in e.g. a UTF-8 locale.  Non-<acronym>ASCII</acronym> characters in
object names will normally<a name="DOCF13" href="#FOOT13"><sup>13</sup></a> fail when the package is installed.  Any byte will
be allowed in a quoted character string but <code>\uxxxx</code> escapes should
be used for non-<acronym>ASCII</acronym> characters.  However,
non-<acronym>ASCII</acronym> character strings may not be usable in some locales
and may display incorrectly in others.
</p>

<a name="index-library_002edynam"></a>
<p>Various R functions in a package can be used to initialize and
clean up.  See <a href="#Load-hooks">Load hooks</a>.
</p>
<p>The <samp>man</samp> subdirectory should contain (only) documentation files
for the objects in the package in <em>R documentation</em> (Rd) format.
The documentation filenames must start with an <acronym>ASCII</acronym> (lower or
upper case) letter or digit and have the extension <samp>.Rd</samp> (the
default) or <samp>.rd</samp>.  Further, the names must be valid in
&lsquo;<samp>file://</samp>&rsquo; URLs, which means<a name="DOCF14" href="#FOOT14"><sup>14</sup></a>
they must be entirely <acronym>ASCII</acronym> and not contain &lsquo;<samp>%</samp>&rsquo;.
See <a href="#Writing-R-documentation-files">Writing R documentation files</a>, for more information.  Note that
all user-level objects in a package should be documented; if a package
<var>pkg</var> contains user-level objects which are for &ldquo;internal&rdquo; use
only, it should provide a file <samp><var>pkg</var>-internal.Rd</samp> which
documents all such objects, and clearly states that these are not meant
to be called by the user.  See e.g. the sources for package <strong>grid</strong>
in the R distribution.  Note that packages which use internal objects
extensively should not export those objects from their namespace, when
they do not need to be documented (see <a href="#Package-namespaces">Package namespaces</a>).
</p>
<p>Having a <samp>man</samp> directory containing no documentation files may give
an installation error.
</p>
<p>The <samp>man</samp> subdirectory may contain a subdirectory named <samp>macros</samp>;
this will contain source for user-defined Rd macros.
(See <a href="#User_002ddefined-macros">User-defined macros</a>.)  These use the Rd format, but may
not contain anything but macro definitions, comments and whitespace.
</p>
<p>The <samp>R</samp> and <samp>man</samp> subdirectories may contain OS-specific
subdirectories named <samp>unix</samp> or <samp>windows</samp>.
</p>
<p>The sources and headers for the compiled code are in <samp>src</samp>, plus
optionally a file <samp>Makevars</samp> or <samp>Makefile</samp>.  When a package is
installed using <code>R CMD INSTALL</code>, <code>make</code> is used to control
compilation and linking into a shared object for loading into R.
There are default <code>make</code> variables and rules for this
(determined when R is configured and recorded in
<samp><var>R_HOME</var>/etc<var>R_ARCH</var>/Makeconf</samp>), providing support for C,
C++, FORTRAN 77, Fortran 9x<a name="DOCF15" href="#FOOT15"><sup>15</sup></a>, Objective C and Objective
C++<a name="DOCF16" href="#FOOT16"><sup>16</sup></a> with associated extensions <samp>.c</samp>, <samp>.cc</samp> or
<samp>.cpp</samp>, <samp>.f</samp>, <samp>.f90</samp> or <samp>.f95</samp>, <samp>.m</samp>, and
<samp>.mm</samp>, respectively.  We recommend using <samp>.h</samp> for headers,
also for C++<a name="DOCF17" href="#FOOT17"><sup>17</sup></a> or Fortran 9x include files.  (Use of extension <samp>.C</samp> for
C++ is no longer supported.)  Files in the <samp>src</samp> directory should
not be hidden (start with a dot), and hidden files will under some
versions of R be ignored.
</p>
<p>It is not portable (and may not be possible at all) to mix all these
languages in a single package, and we do not support using both C++ and
Fortran 9x. Because R itself uses it, we know that C and FORTRAN 77
can be used together and mixing C and C++ seems to be widely successful.
</p>
<p>If your code needs to depend on the platform there are certain defines
which can used in C or C++.  On all Windows builds (even 64-bit ones)
&lsquo;<samp>_WIN32</samp>&rsquo; will be defined: on 64-bit Windows builds also
&lsquo;<samp>_WIN64</samp>&rsquo;, and on macOS &lsquo;<samp>__APPLE__</samp>&rsquo; is defined.<a name="DOCF18" href="#FOOT18"><sup>18</sup></a>
</p>
<p>The default rules can be tweaked by setting macros<a name="DOCF19" href="#FOOT19"><sup>19</sup></a> in a file
<samp>src/Makevars</samp> (see <a href="#Using-Makevars">Using Makevars</a>).  Note that this mechanism
should be general enough to eliminate the need for a package-specific
<samp>src/Makefile</samp>.  If such a file is to be distributed, considerable
care is needed to make it general enough to work on all R platforms.
If it has any targets at all, it should have an appropriate first target
named &lsquo;<samp>all</samp>&rsquo; and a (possibly empty) target &lsquo;<samp>clean</samp>&rsquo; which
removes all files generated by running <code>make</code> (to be used by
&lsquo;<samp>R CMD INSTALL --clean</samp>&rsquo; and &lsquo;<samp>R CMD INSTALL --preclean</samp>&rsquo;).
There are platform-specific file names on Windows:
<samp>src/Makevars.win</samp> takes precedence over <samp>src/Makevars</samp> and
<samp>src/Makefile.win</samp> must be used.  Some <code>make</code> programs
require makefiles to have a complete final line, including a newline.
</p>
<p>A few packages use the <samp>src</samp> directory for purposes other than
making a shared object (e.g. to create executables).  Such packages
should have files <samp>src/Makefile</samp> and <samp>src/Makefile.win</samp>
(unless intended for only Unix-alikes or only Windows).
</p>
<p>In very special cases packages may create binary files other than the
shared objects/DLLs in the <samp>src</samp> directory.  Such files will not be
installed in a multi-architecture setting since <code>R CMD INSTALL
--libs-only</code> is used to merge multiple sub-architectures and it only
copies shared objects/DLLs.  If a package wants to install other
binaries (for example executable programs), it should provide an R
script <samp>src/install.libs.R</samp> which will be run as part of the
installation in the <code>src</code> build directory <em>instead of</em> copying
the shared objects/DLLs. The script is run in a separate R
environment containing the following variables: <code>R_PACKAGE_NAME</code>
(the name of the package), <code>R_PACKAGE_SOURCE</code> (the path to the
source directory of the package), <code>R_PACKAGE_DIR</code> (the path of the
target installation directory of the package), <code>R_ARCH</code> (the
arch-dependent part of the path, often empty), <code>SHLIB_EXT</code> (the
extension of shared objects) and <code>WINDOWS</code> (<code>TRUE</code> on Windows,
<code>FALSE</code> elsewhere).  Something close to the default behavior could
be replicated with the following <samp>src/install.libs.R</samp> file:
</p>
<div class="example">
<pre class="example">files &lt;- Sys.glob(paste0(&quot;*&quot;, SHLIB_EXT))
dest &lt;- file.path(R_PACKAGE_DIR, paste0('libs', R_ARCH))
dir.create(dest, recursive = TRUE, showWarnings = FALSE)
file.copy(files, dest, overwrite = TRUE)
if(file.exists(&quot;symbols.rds&quot;))
    file.copy(&quot;symbols.rds&quot;, dest, overwrite = TRUE)
</pre></div>
<p>On the other hand, executable programs could be installed along the
lines of
</p><div class="example">
<pre class="example">execs &lt;- c(&quot;one&quot;, &quot;two&quot;, &quot;three&quot;)
if(WINDOWS) execs &lt;- paste0(execs, &quot;.exe&quot;)
if ( any(file.exists(execs)) ) {
  dest &lt;- file.path(R_PACKAGE_DIR,  paste0('bin', R_ARCH))
  dir.create(dest, recursive = TRUE, showWarnings = FALSE)
  file.copy(execs, dest, overwrite = TRUE)
}
</pre></div>

<p>Note the use of architecture-specific subdirectories of <samp>bin</samp> where
needed.
</p>
<p>The <samp>data</samp> subdirectory is for data files: See <a href="#Data-in-packages">Data in packages</a>.
</p>
<p>The <samp>demo</samp> subdirectory is for R scripts (for running <em>via</em>
<code>demo()</code>) that demonstrate some of the functionality of the
package.  Demos may be interactive and are not checked automatically, so
if testing is desired use code in the <samp>tests</samp> directory to achieve
this.  The script files must start with a (lower or upper case) letter
and have one of the extensions <samp>.R</samp> or <samp>.r</samp>.  If present, the
<samp>demo</samp> subdirectory should also have a <samp>00Index</samp> file with one
line for each demo, giving its name and a description separated by a tab
or at least three spaces. (This index file is not generated
automatically.)  Note that a demo does not have a specified encoding and
so should be an <acronym>ASCII</acronym> file (see <a href="#Encoding-issues">Encoding issues</a>).  As from
R 3.0.0 <code>demo()</code> will use the package encoding if there is one,
but this is mainly useful for non-<acronym>ASCII</acronym> comments.
</p>
<a name="index-_002eRinstignore-file"></a>
<p>The contents of the <samp>inst</samp> subdirectory will be copied recursively
to the installation directory.  Subdirectories of <samp>inst</samp> should not
interfere with those used by R (currently, <samp>R</samp>, <samp>data</samp>,
<samp>demo</samp>, <samp>exec</samp>, <samp>libs</samp>, <samp>man</samp>, <samp>help</samp>,
<samp>html</samp> and <samp>Meta</samp>, and earlier versions used <samp>latex</samp>,
<samp>R-ex</samp>).  The copying of the <samp>inst</samp> happens after <samp>src</samp>
is built so its <samp>Makefile</samp> can create files to be installed.  To
exclude files from being installed, one can specify a list of exclude
patterns in file <samp>.Rinstignore</samp> in the top-level source directory.
These patterns should be Perl-like regular expressions (see the help for
<code>regexp</code> in R for the precise details), one per line, to be
matched case-insensitively<a name="DOCF20" href="#FOOT20"><sup>20</sup></a>
against the file and directory paths, e.g. <samp>doc/.*[.]png$</samp> will
exclude all PNG files in <samp>inst/doc</samp> based on the extension.
</p>
<p>Note that with the exceptions of <samp>INDEX</samp>,
<samp>LICENSE</samp>/<samp>LICENCE</samp> and <samp>NEWS</samp>, information files at the
top level of the package will <em>not</em> be installed and so not be
known to users of Windows and macOS compiled packages (and not seen
by those who use <code>R CMD INSTALL</code> or <code>install.packages</code>
on the tarball).  So any information files you wish an end user to see
should be included in <samp>inst</samp>.  Note that if the named exceptions
also occur in <samp>inst</samp>, the version in <samp>inst</samp> will be that seen
in the installed package.
</p>
<a name="index-CITATION"></a>
<a name="index-citation"></a>
<a name="index-NEWS_002eRd"></a>
<a name="index-news"></a>
<p>Things you might like to add to <samp>inst</samp> are a <samp>CITATION</samp> file
for use by the <code>citation</code> function, and a <samp>NEWS.Rd</samp> file for
use by the <code>news</code> function.  See its help page for the specific
format restrictions of the <samp>NEWS.Rd</samp> file.
</p>
<a name="index-AUTHORS"></a>
<a name="index-COPYRIGHTS-1"></a>
<p>Another file sometimes needed in <samp>inst</samp> is <samp>AUTHORS</samp> or
<samp>COPYRIGHTS</samp> to specify the authors or copyright holders when this
is too complex to put in the <samp>DESCRIPTION</samp> file.
</p>
<p>Subdirectory <samp>tests</samp> is for additional package-specific test code,
similar to the specific tests that come with the R distribution.
Test code can either be provided directly in a <samp>.R</samp> file, or
<em>via</em> a <samp>.Rin</samp> file containing code which in turn creates the
corresponding <samp>.R</samp> file (e.g., by collecting all function objects
in the package and then calling them with the strangest arguments).  The
results of running a <samp>.R</samp> file are written to a <samp>.Rout</samp> file.
If there is a corresponding<a name="DOCF21" href="#FOOT21"><sup>21</sup></a> <samp>.Rout.save</samp> file, these two are
compared, with differences being reported but not causing an error.  The
directory <samp>tests</samp> is copied to the check area, and the tests are
run with the copy as the working directory and with <code>R_LIBS</code> set to
ensure that the copy of the package installed during testing will be
found by <code>library(<var>pkg_name</var>)</code>.  Note that the package-specific
tests are run in a vanilla R session without setting the
random-number seed, so tests which use random numbers will need to set
the seed to obtain reproducible results (and it can be helpful to do so
in all cases, to avoid occasional failures when tests are run).
</p>
<p>If directory <samp>tests</samp> has a subdirectory <samp>Examples</samp> containing
a file <code><var>pkg</var>-Ex.Rout.save</code>, this is compared to the output
file for running the examples when the latter are checked.  Reference
output should be produced without having the <samp>--timings</samp> option
set (and note that <samp>--as-cran</samp> sets it).
</p>
<p>Subdirectory <samp>exec</samp> could contain additional executable scripts the
package needs, typically scripts for interpreters such as the shell,
Perl, or Tcl.  NB: only files (and not directories) under <samp>exec</samp> are
installed (and those with names starting with a dot are ignored), and
they are all marked as executable (mode <code>755</code>, moderated by
&lsquo;<samp>umask</samp>&rsquo;) on POSIX platforms.  Note too that this is not suitable
for executable <em>programs</em> since some platforms (including Windows)
support multiple architectures using the same installed package
directory.
</p>
<p>Subdirectory <samp>po</samp> is used for files related to <em>localization</em>:
see <a href="#Internationalization">Internationalization</a>.
</p>
<p>Subdirectory <samp>tools</samp> is the preferred place for auxiliary files
needed during configuration, and also for sources need to re-create
scripts (e.g. M4 files for <code>autoconf</code>).
</p>

<hr>
<a name="Data-in-packages"></a>
<div class="header">
<p>
Next: <a href="#Non_002dR-scripts-in-packages" accesskey="n" rel="next">Non-R scripts in packages</a>, Previous: <a href="#Package-subdirectories" accesskey="p" rel="prev">Package subdirectories</a>, Up: <a href="#Package-structure" accesskey="u" rel="up">Package structure</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Data-in-packages-1"></a>
<h4 class="subsection">1.1.6 Data in packages</h4>

<p>The <samp>data</samp> subdirectory is for data files, either to be made
available <em>via</em> lazy-loading or for loading using <code>data()</code>.
(The choice is made by the &lsquo;<samp>LazyData</samp>&rsquo; field in the
<samp>DESCRIPTION</samp> file: the default is not to do so.)  It should not be
used for other data files needed by the package, and the convention has
grown up to use directory <samp>inst/extdata</samp> for such files.
</p>
<p>Data files can have one of three types as indicated by their extension:
plain R code (<samp>.R</samp> or <samp>.r</samp>), tables (<samp>.tab</samp>,
<samp>.txt</samp>, or <samp>.csv</samp>, see <code>?data</code> for the file formats, and
note that <samp>.csv</samp> is <strong>not</strong> the standard<a name="DOCF22" href="#FOOT22"><sup>22</sup></a> CSV format), or
<code>save()</code> images (<samp>.RData</samp> or <samp>.rda</samp>).  The files should
not be hidden (have names starting with a dot).  Note that R code
should be &ldquo;self-sufficient&rdquo; and not make use of extra functionality
provided by the package, so that the data file can also be used without
having to load the package or its namespace.
</p>
<p>Images (extensions <samp>.RData</samp><a name="DOCF23" href="#FOOT23"><sup>23</sup></a> or <samp>.rda</samp>) can contain
references to the namespaces of packages that were used to create them.
Preferably there should be no such references in data files, and in any
case they should only be to packages listed in the <code>Depends</code> and
<code>Imports</code> fields, as otherwise it may be impossible to install the
package.  To check for such references, load all the images into a
vanilla R session, and look at the output of
<code>loadedNamespaces()</code>.
</p>
<p>If your data files are large and you are not using &lsquo;<samp>LazyData</samp>&rsquo; you
can speed up installation by providing a file <samp>datalist</samp> in the
<samp>data</samp> subdirectory.  This should have one line per topic that
<code>data()</code> will find, in the format &lsquo;<samp>foo</samp>&rsquo; if <code>data(foo)</code>
provides &lsquo;<samp>foo</samp>&rsquo;, or &lsquo;<samp>foo: bar bah</samp>&rsquo; if <code>data(foo)</code> provides
&lsquo;<samp>bar</samp>&rsquo; and &lsquo;<samp>bah</samp>&rsquo;.  <code>R CMD build</code> will automatically add
a <samp>datalist</samp> file to <samp>data</samp> directories of over 1Mb, using the
function <code>tools::add_datalist</code>.
</p>
<p>Tables (<samp>.tab</samp>, <samp>.txt</samp>, or <samp>.csv</samp> files) can be
compressed by <code>gzip</code>, <code>bzip2</code> or <code>xz</code>,
optionally with additional extension <samp>.gz</samp>, <samp>.bz2</samp> or
<samp>.xz</samp>.
</p>
<p>If your package is to be distributed, do consider the resource
implications of large datasets for your users: they can make packages
very slow to download and use up unwelcome amounts of storage space, as
well as taking many seconds to load.  It is normally best to distribute
large datasets as <samp>.rda</samp> images prepared by <code>save(, compress =
TRUE)</code> (the default).  Using <code>bzip2</code> or <code>xz</code> compression
will usually reduce the size of both the package tarball and the
installed package, in some cases by a factor of two or more.
</p>
<p>Package <strong>tools</strong> has a couple of functions to help with data images:
<code>checkRdaFiles</code> reports on the way the image was saved, and
<code>resaveRdaFiles</code> will re-save with a different type of compression,
including choosing the best type for that particular image.
</p>
<p>Some packages using &lsquo;<samp>LazyData</samp>&rsquo; will benefit from using a form of
compression other than <code>gzip</code> in the installed lazy-loading
database.  This can be selected by the <samp>--data-compress</samp> option
to <code>R CMD INSTALL</code> or by using the &lsquo;<samp>LazyDataCompression</samp>&rsquo;
field in the <samp>DESCRIPTION</samp> file.  Useful values are <code>bzip2</code>,
<code>xz</code> and the default, <code>gzip</code>.  The only way to discover which
is best is to try them all and look at the size of the
<samp><var>pkgname</var>/data/Rdata.rdb</samp> file.
</p>
<p>Lazy-loading is not supported for very large datasets (those which when
serialized exceed 2GB, the limit for the format on 32-bit platforms and
all platforms prior to R 3.0.0).
</p>
<p>The analogue for <samp>sysdata.rda</samp> is field &lsquo;<samp>SysDataCompression</samp>&rsquo;:
the default is <code>xz</code> for files bigger than 1MB otherwise
<code>gzip</code>.
</p>
<hr>
<a name="Non_002dR-scripts-in-packages"></a>
<div class="header">
<p>
Next: <a href="#Specifying-URLs" accesskey="n" rel="next">Specifying URLs</a>, Previous: <a href="#Data-in-packages" accesskey="p" rel="prev">Data in packages</a>, Up: <a href="#Package-structure" accesskey="u" rel="up">Package structure</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Non_002dR-scripts-in-packages-1"></a>
<h4 class="subsection">1.1.7 Non-R scripts in packages</h4>

<p>Code which needs to be compiled (C, C++, FORTRAN, Fortran 95 &hellip;)
is included in the <samp>src</samp> subdirectory and discussed elsewhere in
this document.
</p>
<p>Subdirectory <samp>exec</samp> could be used for scripts for interpreters such
as the shell, BUGS, JavaScript, Matlab, Perl, php (<a href="https://CRAN.R-project.org/package=amap"><strong>amap</strong></a>),
Python or Tcl (<a href="https://CRAN.R-project.org/package=Simile"><strong>Simile</strong></a>), or even R.  However, it seems more
common to use the <samp>inst</samp> directory, for example
<samp>WriteXLS/inst/Perl</samp>, <samp>NMF/inst/m-files</samp>,
<samp>RnavGraph/inst/tcl</samp>, <samp>RProtoBuf/inst/python</samp> and
<samp>emdbook/inst/BUGS</samp> and <samp>gridSVG/inst/js</samp>.
</p>
<p>Java code is a special case: except for very small programs,
<samp>.java</samp> files should be byte-compiled (to a <samp>.class</samp> file) and
distributed as part of a <samp>.jar</samp> file: the conventional location for
the <samp>.jar</samp> file(s) is <samp>inst/java</samp>.  It is desirable (and
required under an Open Source license) to make the Java source files
available: this is best done in a top-level <samp>java</samp> directory in the
package&mdash;the source files should not be installed.
</p>
<p>If your package requires one of these interpreters or an extension then
this should be declared in the &lsquo;<samp>SystemRequirements</samp>&rsquo; field of its
<samp>DESCRIPTION</samp> file.  (Users of Java most often do so <em>via</em>
<a href="https://CRAN.R-project.org/package=rJava"><strong>rJava</strong></a>, when depending on/importing that suffices.)
</p>
<p>Windows and Mac users should be aware that the Tcl extensions
&lsquo;<samp>BWidget</samp>&rsquo; and &lsquo;<samp>Tktable</samp>&rsquo; which are currently included with the
R for Windows and in the macOS installers <em>are</em> extensions and do
need to be declared for users of other platforms (and that
&lsquo;<samp>Tktable</samp>&rsquo; is less widely available than it used to be, including
not in the main repositories for major Linux distributions).
</p>
<p>&lsquo;<samp>BWidget</samp>&rsquo; needs to be installed by the user on other OSes.  This is
fairly easy to do: first find the Tcl/Tk search path:
</p>
<div class="example">
<pre class="example">library(tcltk)
strsplit(tclvalue('auto_path'), &quot; &quot;)[[1]]
</pre></div>

<p>then download the sources from
<a href="http://sourceforge.net/projects/tcllib/files/BWidget/">http://sourceforge.net/projects/tcllib/files/BWidget/</a> and
at the command line run something like
</p>
<div class="example">
<pre class="example">tar xf bwidget-1.9.8.tar.gz
sudo mv bwidget-1.9.8 /usr/local/lib
</pre></div>

<p>substituting a location on the Tcl/Tk search path for <samp>/usr/local/lib</samp> if
needed.
</p>
<hr>
<a name="Specifying-URLs"></a>
<div class="header">
<p>
Previous: <a href="#Non_002dR-scripts-in-packages" accesskey="p" rel="prev">Non-R scripts in packages</a>, Up: <a href="#Package-structure" accesskey="u" rel="up">Package structure</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Specifying-URLs-1"></a>
<h4 class="subsection">1.1.8 Specifying URLs</h4>

<p>URLs in many places in the package documentation will be converted to
clickable hyperlinks in at least some of their renderings.  So care is
needed that their forms are correct and portable.
</p>
<p>The full URL should be given, including the scheme (often &lsquo;<samp>http://</samp>&rsquo;
or &lsquo;<samp>https://</samp>&rsquo;) and a final &lsquo;<samp>/</samp>&rsquo; for references to directories.
</p>
<p>Spaces in URLs are not portable and how they are handled does vary by
HTTP server and by client.  There should be no space in the host part of
an &lsquo;<samp>http://</samp>&rsquo; URL, and spaces in the remainder should be encoded,
with each space replaced by &lsquo;<samp>%20</samp>&rsquo;.
</p>
<p>Other characters may benefit from being encoded: see the help on
<code>URLencode()</code>.
</p>
<p>The canonical URL for a <acronym>CRAN</acronym> package is
</p><div class="example">
<pre class="example">https://cran.r-project.org/package=<var>pkgname</var>
</pre></div>

<p>and not a version starting
&lsquo;<samp>http://cran.r-project.org/web/packages/<var>pkgname</var></samp>&rsquo;.
</p>
<hr>
<a name="Configure-and-cleanup"></a>
<div class="header">
<p>
Next: <a href="#Checking-and-building-packages" accesskey="n" rel="next">Checking and building packages</a>, Previous: <a href="#Package-structure" accesskey="p" rel="prev">Package structure</a>, Up: <a href="#Creating-R-packages" accesskey="u" rel="up">Creating R packages</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Configure-and-cleanup-1"></a>
<h3 class="section">1.2 Configure and cleanup</h3>

<p>Note that most of this section is specific to Unix-alikes: see the
comments later on about the Windows port of R.
</p>
<p>If your package needs some system-dependent configuration before
installation you can include an executable (Bourne<a name="DOCF24" href="#FOOT24"><sup>24</sup></a>) shell script
<samp>configure</samp> in your package which (if present) is executed by
<code>R CMD INSTALL</code> before any other action is performed.  This can be
a script created by the Autoconf mechanism, but may also be a script
written by yourself.  Use this to detect if any nonstandard libraries
are present such that corresponding code in the package can be disabled
at install time rather than giving error messages when the package is
compiled or used.  To summarize, the full power of Autoconf is available
for your extension package (including variable substitution, searching
for libraries, etc.).
</p>
<p>Under a Unix-alike only, an executable (Bourne shell) script
<samp>cleanup</samp> is executed as the last thing by <code>R CMD INSTALL</code> if
option <samp>--clean</samp> was given, and by <code>R CMD build</code> when
preparing the package for building from its source.
</p>
<p>As an example consider we want to use functionality provided by a (C or
FORTRAN) library <code>foo</code>.  Using Autoconf, we can create a configure
script which checks for the library, sets variable <code>HAVE_FOO</code> to
<code>TRUE</code> if it was found and to <code>FALSE</code> otherwise, and then
substitutes this value into output files (by replacing instances of
&lsquo;<samp>@HAVE_FOO@</samp>&rsquo; in input files with the value of <code>HAVE_FOO</code>).
For example, if a function named <code>bar</code> is to be made available by
linking against library <code>foo</code> (i.e., using <samp>-lfoo</samp>), one
could use
</p>
<div class="example">
<pre class="example">AC_CHECK_LIB(foo, <var>fun</var>, [HAVE_FOO=TRUE], [HAVE_FOO=FALSE])
AC_SUBST(HAVE_FOO)
......
AC_CONFIG_FILES([foo.R])
AC_OUTPUT
</pre></div>

<p>in <samp>configure.ac</samp> (assuming Autoconf 2.50 or later).
</p>
<p>The definition of the respective R function in <samp>foo.R.in</samp> could be
</p>
<div class="example">
<pre class="example">foo &lt;- function(x) {
    if(!@HAVE_FOO@)
      stop(&quot;Sorry, library ‘foo’ is not available&quot;)
    ...
</pre></div>

<p>From this file <code>configure</code> creates the actual R source file
<samp>foo.R</samp> looking like
</p>
<div class="example">
<pre class="example">foo &lt;- function(x) {
    if(!FALSE)
      stop(&quot;Sorry, library ‘foo’ is not available&quot;)
    ...
</pre></div>

<p>if library <code>foo</code> was not found (with the desired functionality).
In this case, the above R code effectively disables the function.
</p>
<p>One could also use different file fragments for available and missing
functionality, respectively.
</p>
<p>You will very likely need to ensure that the same C compiler and
compiler flags are used in the <samp>configure</samp> tests as when compiling
R or your package.  Under a Unix-alike, you can achieve this by
including the following fragment early in <samp>configure.ac</samp>
(<em>before</em> calling <code>AC_PROG_CC</code>)
</p>
<div class="example">
<pre class="example">: ${R_HOME=`R RHOME`}
if test -z &quot;${R_HOME}&quot;; then
  echo &quot;could not determine R_HOME&quot;
  exit 1
fi
CC=`&quot;${R_HOME}/bin/R&quot; CMD config CC`
CFLAGS=`&quot;${R_HOME}/bin/R&quot; CMD config CFLAGS`
CPPFLAGS=`&quot;${R_HOME}/bin/R&quot; CMD config CPPFLAGS`
</pre></div>

<p>(Using &lsquo;<samp>${R_HOME}/bin/R</samp>&rsquo; rather than just &lsquo;<samp>R</samp>&rsquo; is necessary
in order to use the correct version of R when running the script as
part of <code>R CMD INSTALL</code>, and the quotes since &lsquo;<samp>${R_HOME}</samp>&rsquo;
might contain spaces.)
</p>
<p>If your code does load checks then you may also need
</p><div class="example">
<pre class="example">LDFLAGS=`&quot;${R_HOME}/bin/R&quot; CMD config LDFLAGS`
</pre></div>

<p>and packages written with C++ need to pick up the details for the C++
compiler and switch the current language to C++ by something like
</p><div class="example">
<pre class="example">CXX=`&quot;${R_HOME}/bin/R&quot; CMD config CXX`
CXXFLAGS=`&quot;${R_HOME}/bin/R&quot; CMD config CXXFLAGS`
AC_LANG(C++)
</pre></div>

<p>The latter is important, as for example C headers may not be available
to C++ programs or may not be written to avoid C++ name-mangling.
</p>
<a name="index-R-CMD-config"></a>
<p>You can use <code>R CMD config</code> for getting the value of the basic
configuration variables, and also the header and library flags necessary
for linking a front-end executable program against R, see <kbd>R CMD
config --help</kbd> for details.
</p>
<p>To check for an external BLAS library using the <code>ACX_BLAS</code> macro
from the official Autoconf Macro Archive, one can simply do
</p>
<div class="example">
<pre class="example">F77=`&quot;${R_HOME}/bin/R&quot; CMD config F77`
AC_PROG_F77
FLIBS=`&quot;${R_HOME}/bin/R&quot; CMD config FLIBS`
ACX_BLAS([], AC_MSG_ERROR([could not find your BLAS library], 1))
</pre></div>

<p>Note that <code>FLIBS</code> as determined by R must be used to ensure that
FORTRAN 77 code works on all R platforms.  Calls to the Autoconf macro
<code>AC_F77_LIBRARY_LDFLAGS</code>, which would overwrite <code>FLIBS</code>, must
not be used (and hence e.g. removed from <code>ACX_BLAS</code>).  (Recent
versions of Autoconf in fact allow an already set <code>FLIBS</code> to
override the test for the FORTRAN linker flags.)
</p>

<p><strong>N.B.</strong>: If the <code>configure</code> script creates files, e.g.
<samp>src/Makevars</samp>, you do need a <code>cleanup</code> script to remove
them.  Otherwise <code>R CMD build</code> may ship the files that are
created.  For example, package <a href="https://CRAN.R-project.org/package=RODBC"><strong>RODBC</strong></a> has
</p>
<div class="example">
<pre class="example">#!/bin/sh

rm -f config.* src/Makevars src/config.h
</pre></div>

<p>As this example shows, <code>configure</code> often creates working files
such as <samp>config.log</samp>.
</p>
<p>If your configure script needs auxiliary files, it is recommended that
you ship them in a <samp>tools</samp> directory (as R itself does).
</p>
<p>You should bear in mind that the configure script will not be used on
Windows systems.  If your package is to be made publicly available,
please give enough information for a user on a non-Unix-alike platform
to configure it manually, or provide a <samp>configure.win</samp> script to be
used on that platform.  (Optionally, there can be a <samp>cleanup.win</samp>
script.  Both should be shell scripts to be executed by <code>ash</code>,
which is a minimal version of Bourne-style <code>sh</code>.)  When
<samp>configure.win</samp> is run the environment variables <code>R_HOME</code>
(which uses &lsquo;<samp>/</samp>&rsquo; as the file separator), <code>R_ARCH</code> and Use
<code>R_ARCH_BIN</code> will be set.  Use <code>R_ARCH</code> to decide if this is a
64-bit build (its value there is &lsquo;<samp>/x64</samp>&rsquo;) and to install DLLs to the
correct place (<samp>${R_HOME}/libs${R_ARCH}</samp>).  Use
<code>R_ARCH_BIN</code> to find the correct place under the <samp>bin</samp>
directory, e.g. <samp>${R_HOME}/bin${R_ARCH_BIN}/Rscript.exe</samp>.
</p>
<p>In some rare circumstances, the configuration and cleanup scripts need
to know the location into which the package is being installed.  An
example of this is a package that uses C code and creates two shared
object/DLLs.  Usually, the object that is dynamically loaded by R
is linked against the second, dependent, object.  On some systems, we
can add the location of this dependent object to the object that is
dynamically loaded by R.  This means that each user does not have to
set the value of the <code>LD_LIBRARY_PATH</code> (or equivalent) environment
variable, but that the secondary object is automatically resolved.
Another example is when a package installs support files that are
required at run time, and their location is substituted into an R
data structure at installation time.
<a name="index-R_005fLIBRARY_005fDIR"></a>
<a name="index-R_005fPACKAGE_005fDIR"></a>
<a name="index-R_005fPACKAGE_005fNAME"></a>
The names of the top-level library directory (i.e., specifiable
<em>via</em> the &lsquo;<samp>-l</samp>&rsquo; argument) and the directory of the package
itself are made available to the installation scripts <em>via</em> the two
shell/environment variables <code>R_LIBRARY_DIR</code> and <code>R_PACKAGE_DIR</code>.
Additionally, the name of the package (e.g. &lsquo;<samp>survival</samp>&rsquo; or
&lsquo;<samp>MASS</samp>&rsquo;) being installed is available from the environment variable
<code>R_PACKAGE_NAME</code>.  (Currently the value of <code>R_PACKAGE_DIR</code> is
always <code>${R_LIBRARY_DIR}/${R_PACKAGE_NAME}</code>, but this used not to
be the case when versioned installs were allowed.  Its main use is in
<samp>configure.win</samp> scripts for the installation path of external
software&rsquo;s DLLs.)  Note that the value of <code>R_PACKAGE_DIR</code> may
contain spaces and other shell-unfriendly characters, and so should be
quoted in makefiles and configure scripts.
</p>
<p>One of the more tricky tasks can be to find the headers and libraries of
external software.  One tool which is increasingly available on
Unix-alikes (but not by default on macOS) to do this is
<code>pkg-config</code>.  The <samp>configure</samp> script will need to test for
the presence of the command itself (see for example package
<a href="https://CRAN.R-project.org/package=Cairo"><strong>Cairo</strong></a>), and if present it can be asked if the software is
installed, of a suitable version and for compilation/linking flags by
e.g.
</p>
<div class="example">
<pre class="example">$ pkg-config --exists ‘QtCore &gt;= 4.0.0’  # check the status
$ pkg-config --modversion QtCore
4.7.1
$ pkg-config --cflags QtCore
-DQT_SHARED -I/usr/include/QtCore
$ pkg-config --libs QtCore
-lQtCore
</pre></div>

<p>Note that <code>pkg-config --libs</code> gives the information
required to link against the default version of that library (usually
the dynamic one), and <code>pkg-config --static</code> is needed if the
static library is to be used.
</p>
<p>Sometimes the name by which the software is known to
<code>pkg-config</code> is not what one might expect (e.g.
&lsquo;<samp>gtk+-2.0</samp>&rsquo; even for 2.22).  To get a complete list use
</p>
<div class="example">
<pre class="example">pkg-config --list-all | sort
</pre></div>

<table summary="" class="menu" border="0" cellspacing="0">
<tr><td align="left" valign="top">&bull; <a href="#Using-Makevars" accesskey="1">Using Makevars</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Configure-example" accesskey="2">Configure example</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Using-F95-code" accesskey="3">Using F95 code</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Using-C_002b_002b11-code" accesskey="4">Using C++11 code</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
</table>

<hr>
<a name="Using-Makevars"></a>
<div class="header">
<p>
Next: <a href="#Configure-example" accesskey="n" rel="next">Configure example</a>, Previous: <a href="#Configure-and-cleanup" accesskey="p" rel="prev">Configure and cleanup</a>, Up: <a href="#Configure-and-cleanup" accesskey="u" rel="up">Configure and cleanup</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Using-Makevars-1"></a>
<h4 class="subsection">1.2.1 Using <samp>Makevars</samp></h4>

<table summary="" class="menu" border="0" cellspacing="0">
<tr><td align="left" valign="top">&bull; <a href="#OpenMP-support" accesskey="1">OpenMP support</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Using-pthreads" accesskey="2">Using pthreads</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Compiling-in-sub_002ddirectories" accesskey="3">Compiling in sub-directories</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
</table>

<p>Sometimes writing your own <samp>configure</samp> script can be avoided by
supplying a file <samp>Makevars</samp>: also one of the most common uses of a
<samp>configure</samp> script is to make <samp>Makevars</samp> from
<samp>Makevars.in</samp>.
</p>
<p>A <samp>Makevars</samp> file is a makefile and is used as one of several
makefiles by <code>R CMD SHLIB</code> (which is called by <code>R CMD
INSTALL</code> to compile code in the <samp>src</samp> directory).  It should be
written if at all possible in a portable style, in particular (except
for <samp>Makevars.win</samp>) without the use of GNU extensions.
</p>
<p>The most common use of a <samp>Makevars</samp> file is to set additional
preprocessor options (for example include paths) for C/C++ files
<em>via</em> <code>PKG_CPPFLAGS</code>, and additional compiler flags by setting
<code>PKG_CFLAGS</code>, <code>PKG_CXXFLAGS</code>, <code>PKG_FFLAGS</code> or
<code>PKG_FCFLAGS</code>, for C, C++, FORTRAN or Fortran 9x respectively
(see <a href="#Creating-shared-objects">Creating shared objects</a>).
</p>
<p><strong>N.B.</strong>: Include paths are preprocessor options, not compiler
options, and <strong>must</strong> be set in <code>PKG_CPPFLAGS</code> as otherwise
platform-specific paths (e.g. &lsquo;<samp>-I/usr/local/include</samp>&rsquo;) will take
precedence.
</p>
<p><samp>Makevars</samp> can also be used to set flags for the linker, for
example &lsquo;<samp>-L</samp>&rsquo; and &lsquo;<samp>-l</samp>&rsquo; options, <em>via</em> <code>PKG_LIBS</code>.
</p>
<p>When writing a <samp>Makevars</samp> file for a package you intend to
distribute, take care to ensure that it is not specific to your
compiler: flags such as <samp>-O2 -Wall -pedantic</samp> (and all other
<samp>-W</samp> flags: for the Oracle compilers these are used to pass
arguments to compiler phases) are all specific to GCC.
</p>
<p>Also, do not set variables such as <code>CPPFLAGS</code>, <code>CFLAGS</code> etc.:
these should be settable by users (sites) through appropriate personal
(site-wide) <samp>Makevars</samp> files.
See <a href="http://cran.r-project.org/doc/manuals/R-admin.html#Customizing-package-compilation">Customizing package compilation</a> in <cite>R Installation and Administration</cite>,
</p>
<p>There are some macros<a name="DOCF25" href="#FOOT25"><sup>25</sup></a>  which are set whilst configuring the
building of R itself and are stored in
<samp><var>R_HOME</var>/etc<var>R_ARCH</var>/Makeconf</samp>.  That makefile is included
as a <samp>Makefile</samp> <em>after</em> <samp>Makevars[.win]</samp>, and the macros
it defines can be used in macro assignments and make command lines in
the latter.  These include
</p>
<dl compact="compact">
<dt><code>FLIBS</code></dt>
<dd><a name="index-FLIBS"></a>
<p>A macro containing the set of libraries need to link FORTRAN code.  This
may need to be included in <code>PKG_LIBS</code>: it will normally be included
automatically if the package contains FORTRAN source files.
</p>
</dd>
<dt><code>BLAS_LIBS</code></dt>
<dd><a name="index-BLAS_005fLIBS"></a>
<p>A macro containing the BLAS libraries used when building R.  This may
need to be included in <code>PKG_LIBS</code>.  Beware that if it is empty then
the R executable will contain all the double-precision and
double-complex BLAS routines, but no single-precision nor complex
routines.  If <code>BLAS_LIBS</code> is included, then <code>FLIBS</code> also needs
to be<a name="DOCF26" href="#FOOT26"><sup>26</sup></a> included following it, as most BLAS
libraries are written at least partially in FORTRAN.
</p>
</dd>
<dt><code>LAPACK_LIBS</code></dt>
<dd><a name="index-LAPACK_005fLIBS"></a>
<p>A macro containing the LAPACK libraries (and paths where appropriate)
used when building R.  This may need to be included in
<code>PKG_LIBS</code>.  It may point to a dynamic library <code>libRlapack</code>
which contains the main double-precision LAPACK routines as well as
those double-complex LAPACK routines needed to build R, or it may
point to an external LAPACK library, or may be empty if an external BLAS
library also contains LAPACK.
</p>
<p>[<code>libRlapack</code> includes all the double-precision LAPACK routines
which were current in 2003: a list of which routines are included is in
file <samp>src/modules/lapack/README</samp>. Note that an external LAPACK/BLAS
library need not do so, as some were &lsquo;deprecated&rsquo; (and not compiled by
default) in LAPACK 3.6.0 in late 2015.]
</p>
<p>For portability, the macros <code>BLAS_LIBS</code> and <code>FLIBS</code> should
always be included <em>after</em> <code>LAPACK_LIBS</code> (and in that order).
</p>
</dd>
<dt><code>SAFE_FFLAGS</code></dt>
<dd><a name="index-SAFE_005fFFLAGS"></a>
<p>A macro containing flags which are needed to circumvent
over-optimization of FORTRAN code: it is typically &lsquo;<samp>-g -O2
-ffloat-store</samp>&rsquo; on &lsquo;<samp>ix86</samp>&rsquo; platforms using <code>gfortran</code>.
Note that this is <strong>not</strong> an additional flag to be used as part of
<code>PKG_FFLAGS</code>, but a replacement for <code>FFLAGS</code>, and that it is
intended for the FORTRAN 77 compiler &lsquo;<samp>F77</samp>&rsquo; and not necessarily for
the Fortran 90/95 compiler &lsquo;<samp>FC</samp>&rsquo;.  See the example later in this
section.
</p></dd>
</dl>

<a name="index-OBJECTS"></a>
<p>Setting certain macros in <samp>Makevars</samp> will prevent <code>R CMD
SHLIB</code> setting them: in particular if <samp>Makevars</samp> sets
&lsquo;<samp>OBJECTS</samp>&rsquo; it will not be set on the <code>make</code> command line.
This can be useful in conjunction with implicit rules to allow other
types of source code to be compiled and included in the shared object.
It can also be used to control the set of files which are compiled,
either by excluding some files in <samp>src</samp> or including some files in
subdirectories.  For example
</p>
<div class="example">
<pre class="example">OBJECTS = 4dfp/endianio.o 4dfp/Getifh.o R4dfp-object.o
</pre></div>


<p>Note that <samp>Makevars</samp> should not normally contain targets, as it is
included before the default makefile and <code>make</code> will call the
first target, intended to be <code>all</code> in the default makefile.  If you
really need to circumvent that, use a suitable (phony) target <code>all</code>
before any actual targets in <samp>Makevars.[win]</samp>: for example package
<a href="https://CRAN.R-project.org/package=fastICA"><strong>fastICA</strong></a> used to have
</p>
<div class="example">
<pre class="example">PKG_LIBS = @BLAS_LIBS@

SLAMC_FFLAGS=$(R_XTRA_FFLAGS) $(FPICFLAGS) $(SHLIB_FFLAGS) $(SAFE_FFLAGS)

all: $(SHLIB)

slamc.o: slamc.f
	$(F77) $(SLAMC_FFLAGS) -c -o slamc.o slamc.f
</pre></div>

<p>needed to ensure that the LAPACK routines find some constants without
infinite looping.  The Windows equivalent was
</p>
<div class="example">
<pre class="example">all: $(SHLIB)

slamc.o: slamc.f
	$(F77) $(SAFE_FFLAGS) -c -o slamc.o slamc.f
</pre></div>

<p>(since the other macros are all empty on that platform, and R&rsquo;s
internal BLAS was not used).  Note that the first target in
<samp>Makevars</samp> will be called, but for back-compatibility it is best
named <code>all</code>.
</p>
<p>If you want to create and then link to a library, say using code in a
subdirectory, use something like
</p>
<div class="example">
<pre class="example">.PHONY: all mylibs

all: $(SHLIB)
$(SHLIB): mylibs

mylibs:
	(cd subdir; $(MAKE))
</pre></div>

<p>Be careful to create all the necessary dependencies, as there is no
guarantee that the dependencies of <code>all</code> will be run in a
particular order (and some of the <acronym>CRAN</acronym> build machines use
multiple CPUs and parallel makes).  In particular,
</p>
<div class="example">
<pre class="example">all: mylibs
</pre></div>

<p>does <strong>not</strong> suffice.
</p>
<p>Note that on Windows it is required that <samp>Makevars[.win]</samp> does
create a DLL: this is needed as it is the only reliable way to ensure
that building a DLL succeeded.  If you want to use the <samp>src</samp>
directory for some purpose other than building a DLL, use a
<samp>Makefile.win</samp> file.
</p>
<p>It is sometimes useful to have a target &lsquo;<samp>clean</samp>&rsquo; in <samp>Makevars</samp>
or <samp>Makevars.win</samp>: this will be used by <code>R CMD build</code> to
clean up (a copy of) the package sources.  When it is run by
<code>build</code> it will have fewer macros set, in particular not
<code>$(SHLIB)</code>, nor <code>$(OBJECTS)</code> unless set in the file itself.
It would also be possible to add tasks to the target &lsquo;<samp>shlib-clean</samp>&rsquo;
which is run by <code>R CMD INSTALL</code> and <code>R CMD SHLIB</code> with
options <samp>--clean</samp> and <samp>--preclean</samp>.
</p>
<p>If you want to run R code in <samp>Makevars</samp>, e.g. to find
configuration information, please do ensure that you use the correct
copy of <code>R</code> or <code>Rscript</code>: there might not be one in the path
at all, or it might be the wrong version or architecture.  The correct
way to do this is <em>via</em>
</p>
<div class="example">
<pre class="example">&quot;$(R_HOME)/bin$(R_ARCH_BIN)/Rscript&quot; <var>filename</var>
&quot;$(R_HOME)/bin$(R_ARCH_BIN)/Rscript&quot; -e ‘<var>R expression</var>’
</pre></div>

<p>where <code>$(R_ARCH_BIN)</code> is only needed currently on Windows.
</p>
<p>Environment or make variables can be used to select different macros for
32- and 64-bit code, for example (GNU <code>make</code> syntax, allowed on
Windows)
</p>
<div class="example">
<pre class="example">ifeq &quot;$(WIN)&quot; &quot;64&quot;
PKG_LIBS = <var>value for 64-bit Windows</var>
else
PKG_LIBS = <var>value for 32-bit Windows</var>
endif
</pre></div>

<p>On Windows there is normally a choice between linking to an import
library or directly to a DLL.  Where possible, the latter is much more
reliable: import libraries are tied to a specific toolchain, and in
particular on 64-bit Windows two different conventions have been
commonly used.  So for example instead of
</p>
<div class="example">
<pre class="example">PKG_LIBS = -L$(XML_DIR)/lib -lxml2
</pre></div>

<p>one can use
</p>
<div class="example">
<pre class="example">PKG_LIBS = -L$(XML_DIR)/bin -lxml2
</pre></div>

<p>since on Windows <code>-lxxx</code> will look in turn for
</p>
<div class="example">
<pre class="example">libxxx.dll.a
xxx.dll.a
libxxx.a
xxx.lib
libxxx.dll
xxx.dll
</pre></div>

<p>where the first and second are conventionally import libraries, the
third and fourth often static libraries (with <code>.lib</code> intended for
Visual C++), but might be import libraries.  See for example
<a href="https://sourceware.org/binutils/docs-2.20/ld/WIN32.html#WIN32">https://sourceware.org/binutils/docs-2.20/ld/WIN32.html#WIN32</a>.
</p>
<p>The fly in the ointment is that the DLL might not be named
<samp>libxxx.dll</samp>, and in fact on 32-bit Windows there is a
<samp>libxml2.dll</samp> whereas on one build for 64-bit Windows the DLL is
called <samp>libxml2-2.dll</samp>.  Using import libraries can cover over
these differences but can cause equal difficulties.
</p>
<p>If static libraries are available they can save a lot of problems with
run-time finding of DLLs, especially when binary packages are to be
distributed and even more when these support both architectures.  Where
using DLLs is unavoidable we normally arrange (<em>via</em>
<samp>configure.win</samp>) to ship them in the same directory as the package
DLL.
</p>
<hr>
<a name="OpenMP-support"></a>
<div class="header">
<p>
Next: <a href="#Using-pthreads" accesskey="n" rel="next">Using pthreads</a>, Previous: <a href="#Using-Makevars" accesskey="p" rel="prev">Using Makevars</a>, Up: <a href="#Using-Makevars" accesskey="u" rel="up">Using Makevars</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="OpenMP-support-1"></a>
<h4 class="subsubsection">1.2.1.1 OpenMP support</h4>

<a name="index-OpenMP"></a>

<p>There is some support for packages which wish to use
OpenMP<a name="DOCF27" href="#FOOT27"><sup>27</sup></a>.  The
<code>make</code> macros
</p>
<div class="example">
<pre class="example">SHLIB_OPENMP_CFLAGS
SHLIB_OPENMP_CXXFLAGS
SHLIB_OPENMP_FCFLAGS
SHLIB_OPENMP_FFLAGS
</pre></div>

<p>are available for use in <samp>src/Makevars</samp> or <samp>src/Makevars.win</samp>.
Include the appropriate macro in <code>PKG_CFLAGS</code>, <code>PKG_CPPFLAGS</code>
and so on, and also in <code>PKG_LIBS</code>.  C/C++ code that needs to be
conditioned on the use of OpenMP can be used inside <code>#ifdef
_OPENMP</code>: note that some toolchains used for R (including that of
macOS and some others using <code>clang</code><a name="DOCF28" href="#FOOT28"><sup>28</sup></a>) have no OpenMP
support at all, not even <samp>omp.h</samp>.
</p>
<p>For example, a package with C code written for OpenMP should have in
<samp>src/Makevars</samp> the lines
</p>
<div class="example">
<pre class="example">PKG_CFLAGS = $(SHLIB_OPENMP_CFLAGS)
PKG_LIBS = $(SHLIB_OPENMP_CFLAGS)
</pre></div>

<p>Note that the macro <code>SHLIB_OPENMP_CXXFLAGS</code> applies to the default
C++ compiler and not necessarily to the C++11 compiler: users of the
latter should do their own <code>configure</code> checks (an example is
available in CRAN package <a href="https://CRAN.R-project.org/package=ARTP2"><strong>ARTP2</strong></a>).
</p>
<p>Some care is needed when compilers are from different families which may
use different OpenMP runtimes (e.g. <code>clang</code> <em>vs</em> GCC
including <code>gfortran</code>, although it is currently possible to use
the <code>clang</code> runtime with GCC but not <em>vice versa</em>).  For a
package with Fortran 77 code using OpenMP the appropriate lines are
</p>
<div class="example">
<pre class="example">PKG_FFLAGS = $(SHLIB_OPENMP_FFLAGS)
PKG_LIBS = $(SHLIB_OPENMP_CFLAGS)
</pre></div>

<p>as the C compiler will be used to link the package code (and there is no
guarantee that this will work everywhere).  (This does not apply to
Fortran 9x code, where <code>SHLIB_OPENMP_FCFLAGS</code> should be used in
both <code>PKG_FCFLAGS</code> and <code>PKG_LIBS</code>.)
</p>
<p>For portability, any C/C++ code using the <code>omp_*</code> functions should
include the <samp>omp.h</samp> header: some compilers (but not all) include it
when OpenMP mode is switched on (e.g. <em>via</em> flag
<samp>-fopenmp</samp>).
</p>
<p>There is nothing<a name="DOCF29" href="#FOOT29"><sup>29</sup></a>  to say what
version of OpenMP is supported: version 3.1 (and much of 4.0) is
supported by recent versions<a name="DOCF30" href="#FOOT30"><sup>30</sup></a> of the
Linux, Windows and Solaris platforms, but portable packages cannot
assume that end users have recent versions.<a name="DOCF31" href="#FOOT31"><sup>31</sup></a>  macOS currently uses Apple builds of <code>clang</code>
with no OpenMP support (even if invoked as <code>gcc</code> and despite the
<code>man</code> page including the flag <samp>-fopenmp</samp> for that
command).  <a href="http://www.openmp.org/resources/openmp-compilers">http://www.openmp.org/resources/openmp-compilers</a>
gives some idea of what compilers support what versions.
</p>
<p>The performance of OpenMP varies substantially between platforms.  The
Windows implementation has substantial overheads<a name="DOCF32" href="#FOOT32"><sup>32</sup></a>, so is only beneficial if quite substantial tasks are run in
parallel.  Also, on Windows new threads are started with the
default<a name="DOCF33" href="#FOOT33"><sup>33</sup></a> FPU control
word, so computations done on OpenMP threads will not make use of
extended-precision arithmetic which is the default for the main process.
</p>
<p>Calling any of the R API from threaded code is &lsquo;for experts only&rsquo;:
they will need to read the source code to determine if it is
thread-safe.  In particular, code which makes use of the stack-checking
mechanism must not be called from threaded code.
</p>
<p>Packages are not standard-alone programs, and an R process could
contain more than one OpenMP-enabled package as well as other components
(for example, an optimized BLAS) making use of OpenMP.  So careful
consideration needs to be given to resource usage.  OpenMP works with
parallel regions, and for most implementations the default is to use as
many threads as &lsquo;CPUs&rsquo; for such regions.  Parallel regions can be
nested, although it is common to use only a single thread below the
first level.  The correctness of the detected number of &lsquo;CPUs&rsquo; and the
assumption that the R process is entitled to use them all are both
dubious assumptions.  The best way to limit resources is to limit the
overall number of threads available to OpenMP in the R process: this
can be done via environment variable <code>OMP_THREAD_LIMIT</code>, where
implemented.<a name="DOCF34" href="#FOOT34"><sup>34</sup></a>  Alternatively, the
number of threads per region can be limited by the environment variable
<code>OMP_NUM_THREADS</code> or API call <code>omp_set_num_threads</code>, or,
better, for the regions in your code as part of their
specification. E.g. R uses
</p><div class="example">
<pre class="example">#pragma omp parallel for num_threads(nthreads) &hellip;
</pre></div>
<p>That way you only control your own code and not that of other OpenMP users.
</p>
<hr>
<a name="Using-pthreads"></a>
<div class="header">
<p>
Next: <a href="#Compiling-in-sub_002ddirectories" accesskey="n" rel="next">Compiling in sub-directories</a>, Previous: <a href="#OpenMP-support" accesskey="p" rel="prev">OpenMP support</a>, Up: <a href="#Using-Makevars" accesskey="u" rel="up">Using Makevars</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Using-pthreads-1"></a>
<h4 class="subsubsection">1.2.1.2 Using pthreads</h4>

<p>There is no direct support for the POSIX threads (more commonly known as
<code>pthreads</code>): by the time we considered adding it several packages
were using it unconditionally so it seems that nowadays it is
universally available on POSIX operating systems (hence not Windows).
</p>
<p>For reasonably recent versions of <code>gcc</code> and <code>clang</code> the
correct specification is
</p>
<div class="example">
<pre class="example">PKG_CPPFLAGS = -pthread
PKG_LIBS = -pthread
</pre></div>

<p>(and the plural version is also accepted on some systems/versions).  For
other platforms the specification is
</p>
<div class="example">
<pre class="example">PKG_CPPFLAGS = -D_REENTRANT
PKG_LIBS = -lpthread
</pre></div>
<p>(and note that the library name is singular).  This is what
<samp>-pthread</samp> does on all known current platforms (although earlier
versions of OpenBSD used a different library name).
</p>
<p>For a tutorial see
<a href="https://computing.llnl.gov/tutorials/pthreads/">https://computing.llnl.gov/tutorials/pthreads/</a>.
</p>
<p>POSIX threads are not normally used on Windows, which has its own native
concepts of threads.  However, there are two projects implementing
<code>pthreads</code> on top of Windows, <code>pthreads-w32</code> and
<code>winpthreads</code> (part of the MinGW-w64 project).
</p>
<p>Whether Windows toolchains implement <code>pthreads</code> is up to the
toolchain provider.  A <code>make</code> variable
<code>SHLIB_PTHREAD_FLAGS</code> is available: this should be included in both
<code>PKG_CPPFLAGS</code> (or the Fortran or F9x equivalents) and
<code>PKG_LIBS</code>.
</p>
<p>The presence of a working <code>pthreads</code> implementation cannot be
unambiguously determined without testing for yourself: however, that
&lsquo;<samp>_REENTRANT</samp>&rsquo; is defined<a name="DOCF35" href="#FOOT35"><sup>35</sup></a> in C/C++ code is a good indication.
</p>
<p>See also the comments on thread-safety and performance under OpenMP: on
all known R platforms OpenMP is implemented <em>via</em>
<code>pthreads</code> and the known performance issues are in the latter.
</p>
<hr>
<a name="Compiling-in-sub_002ddirectories"></a>
<div class="header">
<p>
Previous: <a href="#Using-pthreads" accesskey="p" rel="prev">Using pthreads</a>, Up: <a href="#Using-Makevars" accesskey="u" rel="up">Using Makevars</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Compiling-in-sub_002ddirectories-1"></a>
<h4 class="subsubsection">1.2.1.3 Compiling in sub-directories</h4>

<p>Package authors fairly often want to organize code in sub-directories of
<samp>src</samp>, for example if they are including a separate piece of
external software to which this is an R interface.
</p>
<p>One simple way is simply to set <code>OBJECTS</code> to be all the objects
that need to be compiled, including in sub-directories.  For example,
<acronym>CRAN</acronym> package <a href="https://CRAN.R-project.org/package=RSiena"><strong>RSiena</strong></a> has
</p>
<div class="smallexample">
<pre class="smallexample">SOURCES = $(wildcard data/*.cpp network/*.cpp utils/*.cpp model/*.cpp model/*/*.cpp model/*/*/*.cpp)

OBJECTS = siena07utilities.o siena07internals.o siena07setup.o siena07models.o $(SOURCES:.cpp=.o)
</pre></div>

<p>One problem with that approach is that unless GNU make extensions are
used, the source files need to be listed and kept up-to-date.  As in the
following from <acronym>CRAN</acronym> package <a href="https://CRAN.R-project.org/package=lossDev"><strong>lossDev</strong></a>:
</p>
<div class="smallexample">
<pre class="smallexample">OBJECTS.samplers = samplers/ExpandableArray.o samplers/Knots.o \
  samplers/RJumpSpline.o samplers/RJumpSplineFactory.o \
  samplers/RealSlicerOV.o samplers/SliceFactoryOV.o samplers/MNorm.o
OBJECTS.distributions = distributions/DSpline.o \
  distributions/DChisqrOV.o distributions/DTOV.o \
  distributions/DNormOV.o distributions/DUnifOV.o distributions/RScalarDist.o
OBJECTS.root = RJump.o

OBJECTS = $(OBJECTS.samplers) $(OBJECTS.distributions) $(OBJECTS.root)
</pre></div>

<p>Where the subdirectory is self-contained code with a suitable makefile,
the best approach is something like
</p>
<div class="smallexample">
<pre class="smallexample">PKG_LIBS = -LCsdp/lib -lsdp $(LAPACK_LIBS) $(BLAS_LIBS) $(FLIBS)

$(SHLIB): Csdp/lib/libsdp.a

Csdp/lib/libsdp.a:      
	@(cd Csdp/lib &amp;&amp; $(MAKE) libsdp.a \
	  CC=&quot;$(CC)&quot; CFLAGS=&quot;$(CFLAGS) $(CPICFLAGS)&quot; AR=&quot;$(AR)&quot; RANLIB=&quot;$(RANLIB)&quot;)
</pre></div>

<p>Note the quotes: the macros can contain spaces, e.g. <code>CC = &quot;gcc
-m64 -std=gnu99&quot;</code>.  Several authors have forgotten about parallel makes:
the static library in the subdirectory must be made before the shared
object (<code>$(SHLIB)</code>) and so the latter must depend on the former.
Others forget the need<a name="DOCF36" href="#FOOT36"><sup>36</sup></a> for
position-independent code.
</p>
<p>We really do not recommend using <samp>src/Makefile</samp> instead of
<samp>src/Makevars</samp>, and as the example above shows, it is not
necessary.
</p>
<hr>
<a name="Configure-example"></a>
<div class="header">
<p>
Next: <a href="#Using-F95-code" accesskey="n" rel="next">Using F95 code</a>, Previous: <a href="#Using-Makevars" accesskey="p" rel="prev">Using Makevars</a>, Up: <a href="#Configure-and-cleanup" accesskey="u" rel="up">Configure and cleanup</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Configure-example-1"></a>
<h4 class="subsection">1.2.2 Configure example</h4>

<p>It may be helpful to give an extended example of using a
<samp>configure</samp> script to create a <samp>src/Makevars</samp> file: this is
based on that in the <a href="https://CRAN.R-project.org/package=RODBC"><strong>RODBC</strong></a> package.
</p>
<p>The <samp>configure.ac</samp> file follows: <samp>configure</samp> is created from
this by running <code>autoconf</code> in the top-level package directory
(containing <samp>configure.ac</samp>).
</p>
<blockquote>
<div class="smallexample">
<pre class="smallexample">AC_INIT([RODBC], 1.1.8) dnl package name, version

dnl A user-specifiable option
odbc_mgr=&quot;&quot;
AC_ARG_WITH([odbc-manager],
	    AC_HELP_STRING([--with-odbc-manager=MGR],
			   [specify the ODBC manager, e.g. odbc or iodbc]),
	    [odbc_mgr=$withval])

if test &quot;$odbc_mgr&quot; = &quot;odbc&quot; ; then
  AC_PATH_PROGS(ODBC_CONFIG, odbc_config)
fi

dnl Select an optional include path, from a configure option
dnl or from an environment variable.
AC_ARG_WITH([odbc-include],
	    AC_HELP_STRING([--with-odbc-include=INCLUDE_PATH],
			   [the location of ODBC header files]),
	    [odbc_include_path=$withval])
RODBC_CPPFLAGS=&quot;-I.&quot;
if test [ -n &quot;$odbc_include_path&quot; ] ; then
   RODBC_CPPFLAGS=&quot;-I. -I${odbc_include_path}&quot;
else
  if test [ -n &quot;${ODBC_INCLUDE}&quot; ] ; then
     RODBC_CPPFLAGS=&quot;-I. -I${ODBC_INCLUDE}&quot;
  fi
fi

dnl ditto for a library path
AC_ARG_WITH([odbc-lib],
	    AC_HELP_STRING([--with-odbc-lib=LIB_PATH],
			   [the location of ODBC libraries]),
	    [odbc_lib_path=$withval])
if test [ -n &quot;$odbc_lib_path&quot; ] ; then
   LIBS=&quot;-L$odbc_lib_path ${LIBS}&quot;
else
  if test [ -n &quot;${ODBC_LIBS}&quot; ] ; then
     LIBS=&quot;-L${ODBC_LIBS} ${LIBS}&quot;
  else
    if test -n &quot;${ODBC_CONFIG}&quot;; then
      odbc_lib_path=`odbc_config --libs | sed s/-lodbc//`
      LIBS=&quot;${odbc_lib_path} ${LIBS}&quot;
    fi
  fi
fi

dnl Now find the compiler and compiler flags to use
: ${R_HOME=`R RHOME`}
if test -z &quot;${R_HOME}&quot;; then
  echo &quot;could not determine R_HOME&quot;
  exit 1
fi
CC=`&quot;${R_HOME}/bin/R&quot; CMD config CC`
CPP=`&quot;${R_HOME}/bin/R&quot; CMD config CPP`
CFLAGS=`&quot;${R_HOME}/bin/R&quot; CMD config CFLAGS`
CPPFLAGS=`&quot;${R_HOME}/bin/R&quot; CMD config CPPFLAGS`
AC_PROG_CC
AC_PROG_CPP


if test -n &quot;${ODBC_CONFIG}&quot;; then
  RODBC_CPPFLAGS=`odbc_config --cflags`
fi
CPPFLAGS=&quot;${CPPFLAGS} ${RODBC_CPPFLAGS}&quot;

dnl Check the headers can be found
AC_CHECK_HEADERS(sql.h sqlext.h)
if test &quot;${ac_cv_header_sql_h}&quot; = no ||
   test &quot;${ac_cv_header_sqlext_h}&quot; = no; then
   AC_MSG_ERROR(&quot;ODBC headers sql.h and sqlext.h not found&quot;)
fi

dnl search for a library containing an ODBC function
if test [ -n &quot;${odbc_mgr}&quot; ] ; then
  AC_SEARCH_LIBS(SQLTables, ${odbc_mgr}, ,
      AC_MSG_ERROR(&quot;ODBC driver manager ${odbc_mgr} not found&quot;))
else
  AC_SEARCH_LIBS(SQLTables, odbc odbc32 iodbc, ,
      AC_MSG_ERROR(&quot;no ODBC driver manager found&quot;))
fi

dnl for 64-bit ODBC need SQL[U]LEN, and it is unclear where they are defined.
AC_CHECK_TYPES([SQLLEN, SQLULEN], , , [# include &lt;sql.h&gt;])
dnl for unixODBC header
AC_CHECK_SIZEOF(long, 4)

dnl substitute RODBC_CPPFLAGS and LIBS
AC_SUBST(RODBC_CPPFLAGS)
AC_SUBST(LIBS)
AC_CONFIG_HEADERS([src/config.h])
dnl and do substitution in the src/Makevars.in and src/config.h
AC_CONFIG_FILES([src/Makevars])
AC_OUTPUT
</pre></div>
</blockquote>

<p>where <samp>src/Makevars.in</samp> would be simply
</p>
<blockquote>
<div class="example">
<pre class="example">PKG_CPPFLAGS = @RODBC_CPPFLAGS@
PKG_LIBS = @LIBS@
</pre></div>
</blockquote>

<p>A user can then be advised to specify the location of the ODBC driver
manager files by options like (lines broken for easier reading)
</p>
<div class="example">
<pre class="example">R CMD INSTALL \
  --configure-args='--with-odbc-include=/opt/local/include \
  --with-odbc-lib=/opt/local/lib --with-odbc-manager=iodbc' \
  RODBC
</pre></div>

<p>or by setting the environment variables <code>ODBC_INCLUDE</code> and
<code>ODBC_LIBS</code>.
</p>
<hr>
<a name="Using-F95-code"></a>
<div class="header">
<p>
Next: <a href="#Using-C_002b_002b11-code" accesskey="n" rel="next">Using C++11 code</a>, Previous: <a href="#Configure-example" accesskey="p" rel="prev">Configure example</a>, Up: <a href="#Configure-and-cleanup" accesskey="u" rel="up">Configure and cleanup</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Using-F95-code-1"></a>
<h4 class="subsection">1.2.3 Using F95 code</h4>

<p>R assumes that source files with extension <samp>.f</samp> are FORTRAN 77,
and passes them to the compiler specified by &lsquo;<samp>F77</samp>&rsquo;.  On most but
not all platforms that compiler will accept Fortran 90/95 code: some
platforms have a separate Fortran 90/95 compiler and a few (by now quite
rare<a name="DOCF37" href="#FOOT37"><sup>37</sup></a>) platforms have no Fortran
90/95 support.
</p>
<p>This means that portable packages need to be written in correct
FORTRAN 77, which will also be valid Fortran 95.  See
<a href="https://developer.R-project.org/Portability.html">https://developer.R-project.org/Portability.html</a> for reference
resources.  In particular, <em>free source form</em> F95 code is not
portable.
</p>
<p>On some systems an alternative F95 compiler is available: from the
<code>gcc</code> family this might be <code>gfortran</code> or <code>g95</code>.
Configuring R will try to find a compiler which (from its name)
appears to be a Fortran 90/95 compiler, and set it in macro &lsquo;<samp>FC</samp>&rsquo;.
Note that it does not check that such a compiler is fully (or even
partially) compliant with Fortran 90/95.  Packages making use of Fortran
90/95 features should use file extension <samp>.f90</samp> or <samp>.f95</samp> for
the source files: the variable <code>PKG_FCFLAGS</code> specifies any special
flags to be used.  There is no guarantee that compiled Fortran 90/95
code can be mixed with any other type of compiled code, nor that a build
of R will have support for such packages.
</p>
<p>Some (but not) all compilers specified by the &lsquo;<samp>FC</samp>&rsquo; macro will
accept Fortran 2003 or 2008 code: such code should still use file
extension <samp>.f90</samp> or <samp>.f95</samp>.  For platforms using
<code>gfortran</code>, you may need to include <samp>-std=f2003</samp> or
<samp>-std=f2008</samp> in <code>PKG_FCFLAGS</code>: the default is &lsquo;GNU Fortran&rsquo;,
Fortran 95 with non-standard extensions.  The Oracle <code>f95</code>
compiler &lsquo;accepts some Fortran 2003 features&rsquo;.
</p>
<p>Modern versions of Fortran support modules, whereby compiling one source
file creates a module file which is then included in others. (Module
files typically have a <samp>.mod</samp> extension: they do depend on the
compiler used and so should never be included in a package.)  This
creates a dependence which <code>make</code> will not know about and often
causes installation with a parallel make to fail.  Thus it is necessary
to add explicit dependencies to <samp>src/Makevars</samp> to tell
<code>make</code> the constraints on the order of compilation.  For
example, if file <samp>iface.f90</samp> creates a module &lsquo;<samp>iface</samp>&rsquo; used by
files <samp>cmi.f90</samp> and <samp>dmi.f90</samp> then <samp>src/Makevars</samp> needs
to contain something like
</p>
<div class="example">
<pre class="example">cmi.o dmi.o: iface.o
</pre></div>

<hr>
<a name="Using-C_002b_002b11-code"></a>
<div class="header">
<p>
Previous: <a href="#Using-F95-code" accesskey="p" rel="prev">Using F95 code</a>, Up: <a href="#Configure-and-cleanup" accesskey="u" rel="up">Configure and cleanup</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Using-C_002b_002b11-code-1"></a>
<h4 class="subsection">1.2.4 Using C++11 code</h4>

<p>R can be built without a C++ compiler although one is available (but
not necessarily installed) on all known R platforms.  For full
portability across platforms, all that can be assumed is approximate
support for the C++98 standard (the widely used <code>g++</code> deviates
considerably from the standard).  Some compilers have a concept of
&lsquo;C++03&rsquo; (&lsquo;essentially a bug fix&rsquo;) or &lsquo;C++ Technical Report 1&rsquo; (TR1), an
optional addition to the &lsquo;C++03&rsquo; revision which was published in 2007.
A revised standard was published in 2011 and compilers with pretty much
complete implementations are becoming available.  C++11 added all of the
C99 features which are not otherwise implemented in C++, and C++
compilers commonly accept C99 extensions to C++98.  A minor
update<a name="DOCF38" href="#FOOT38"><sup>38</sup></a>
to C++11 (C++14) was published in December 2014.
</p>
<p>What standard a C++ compiler aims to support can be hard to determine:
the value<a name="DOCF39" href="#FOOT39"><sup>39</sup></a> of <code>__cplusplus</code>
may help but some compilers use it to denote a standard which is
partially supported and some the latest standard which is (almost) fully
supported.  As from version 6, <code>g++</code> defaults to C++14 (with GNU
extensions): earlier versions aim to support C++03 with many extensions
(including support for TR1).  <code>clang</code> with its
native<a name="DOCF40" href="#FOOT40"><sup>40</sup></a>  <code>libc++</code> headers and library
include many C++11 features, and does not support TR1.
</p>
<p>Since version 3.1.0, R has provided support for C++11 in packages in
addition to C++98.  This support is not uniform across platforms as it
depends on the capabilities of the compiler (see below).  When R is
configured, it will determine whether the C++ compiler supports C++11
and which compiler flags, if any, are required to enable C++11 support.
For example, recent versions of <code>g++</code> or <code>clang++</code>
accept the compiler flag <samp>-std=c++11</samp>, and earlier versions
support a flag <samp>-std=c++0x</samp>, but the latter only provided partial
support for the C++11 standard (it later became a deprecated synonym for
<samp>-std=c++11</samp>).
</p>
<p>In order to use C++11 code in a package, the package&rsquo;s <samp>Makevars</samp>
file (or <samp>Makevars.win</samp> on Windows) should include the line
</p>
<div class="example">
<pre class="example">CXX_STD = CXX11
</pre></div>
<p>Compilation and linking will then be done with the C++11 compiler.  If
any other value is given to the &lsquo;<samp>CXX_STD</samp>&rsquo; macro it will be ignored.
(Further options may become available in the future as the C++ standard
evolves.)
</p>
<p>Packages without a <samp>Makevars</samp> file may specify that they require
C++11 by including &lsquo;<samp>C++11</samp>&rsquo; in the &lsquo;<samp>SystemRequirements</samp>&rsquo; field
of the <samp>DESCRIPTION</samp> file, e.g.
</p>
<div class="example">
<pre class="example">SystemRequirements: C++11
</pre></div>

<p>If a package does have a <samp>Makevars[.win]</samp> file then setting the
make variable &lsquo;<samp>CXX_STD</samp>&rsquo; is preferred, as it allows <code>R CMD
SHLIB</code> to work correctly in the package&rsquo;s <samp>src</samp> directory.
</p>
<p>The C++11 compiler will be used systematically by R for all C++ code
if the environment variable <code>USE_CXX1X</code> is defined (with any
value). Hence this environment variable should be defined when invoking
<code>R CMD SHLIB</code> in the absence of a <samp>Makevars</samp> file (or
<samp>Makevars.win</samp> on Windows) if a C++11 compiler is required.
</p>
<p>Further control over compilation of C++11 code can be obtained by
specifying the macros &lsquo;<samp>CXX1X</samp>&rsquo; and &lsquo;<samp>CXX1XSTD</samp>&rsquo; when R is
configured<a name="DOCF41" href="#FOOT41"><sup>41</sup></a>, or in a personal or site <samp>Makevars</samp> file.
See <a href="http://cran.r-project.org/doc/manuals/R-admin.html#Customizing-package-compilation">Customizing package compilation</a> in <cite>R Installation and Administration</cite>.
If C++11 support is not available then these macros are both empty; if
it is available by default, &lsquo;<samp>CXX1X</samp>&rsquo; defaults to &lsquo;<samp>CXX</samp>&rsquo; and
&lsquo;<samp>CXX1XSTD</samp>&rsquo; is empty . Otherwise, &lsquo;<samp>CXX1X</samp>&rsquo; defaults to the same
value as the C++ compiler &lsquo;<samp>CXX</samp>&rsquo; and the flag &lsquo;<samp>CXX1XSTD</samp>&rsquo;
defaults to <samp>-std=c++11</samp> or <samp>-std=c++0x</samp> (the latter on
Windows prior to R 3.3.0).  It is possible to specify &lsquo;<samp>CXX1X</samp>&rsquo; to
be a distinct compiler just for C++11&ndash;using packages, e.g.
<code>g++</code> on Solaris.  Note however that different C++ compilers
(and even different versions of the same compiler) often differ in their
ABI so their outputs can rarely be mixed. By setting &lsquo;<samp>CXX1XSTD</samp>&rsquo; it
is also possible to choose a different dialect of the standard such as
<samp>-std=gnu++11</samp>, or enable support for the 2014 revision using
something like <samp>-std=c++14</samp> or <samp>-std=c++1y</samp>.
</p>
<p>As noted above, support for C++11 varies across platforms: on some
platforms, it may be possible or necessary to select a different
compiler for C++11, <em>via</em> personal or site <samp>Makevars</samp> files.
Note too that support for C++11 may include C++14 support, so that
including emulation of C++14 features should be done by something like
</p><div class="example">
<pre class="example">#if __cplusplus &gt;= 201402L
using std::make_unique;
#else
// your emulation
#endif
</pre></div>

<p>There is no guarantee that C++11 can be used in a package in combination
with any other compiled language (even C), as the C++11 compiler may be
incompatible with the native compilers for the platform.  (There are
known problems mixing C++11 with Fortran.)
</p>
<p>If a package using C++11 has a <code>configure</code> script it is
essential that it selects the correct compiler, <em>via</em> something like
</p>
<div class="example">
<pre class="example">CXX1X=`&quot;${R_HOME}/bin/R&quot; CMD config CXX1X`
CXX1XSTD=`&quot;${R_HOME}/bin/R&quot; CMD config CXX1XSTD`
CXX=&quot;${CXX1X} ${CXX1XSTD}&quot;
CXXFLAGS=`&quot;${R_HOME}/bin/R&quot; CMD config CXX1XFLAGS`
AC_LANG(C++)
</pre></div>

<p>(paying attention to all the quotes required).
</p>
<p>If you want to compile C++11 code in a subdirectory, make sure you pass
down the macros to specify that compiler, e.g. in <samp>src/Makevars</samp>
</p><div class="example">
<pre class="example">sublibs:
         @(cd libs &amp;&amp; $(MAKE) \
            CXX=&quot;$(CXX1X) $(CXX1XSTD)&quot; CXXFLAGS=&quot;$(CXX1XFLAGS) $(CXX1XPICFLAGS)&quot;)
</pre></div>

<hr>
<a name="Checking-and-building-packages"></a>
<div class="header">
<p>
Next: <a href="#Writing-package-vignettes" accesskey="n" rel="next">Writing package vignettes</a>, Previous: <a href="#Configure-and-cleanup" accesskey="p" rel="prev">Configure and cleanup</a>, Up: <a href="#Creating-R-packages" accesskey="u" rel="up">Creating R packages</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Checking-and-building-packages-1"></a>
<h3 class="section">1.3 Checking and building packages</h3>

<p>Before using these tools, please check that your package can be
installed (which checked it can be loaded).  <code>R CMD check</code> will
<em>inter alia</em> do this, but you may get more detailed error messages
doing the install directly.
</p>
<table summary="" class="menu" border="0" cellspacing="0">
<tr><td align="left" valign="top">&bull; <a href="#Checking-packages" accesskey="1">Checking packages</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Building-package-tarballs" accesskey="2">Building package tarballs</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Building-binary-packages" accesskey="3">Building binary packages</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
</table>

<p>If your package specifies an encoding in its <samp>DESCRIPTION</samp> file,
you should run these tools in a locale which makes use of that encoding:
they may not work at all or may work incorrectly in other locales
(although UTF-8 locales will most likely work).
</p>
<blockquote>
<p><b>Note:</b> <code>R CMD check</code> and <code>R CMD build</code> run R processes with
<samp>--vanilla</samp> in which none of the user&rsquo;s startup files are read.
If you need <code>R_LIBS</code> set (to find packages in a non-standard
library) you can set it in the environment: also you can use the check
and build environment files (as specified by the environment variables
<code>R_CHECK_ENVIRON</code> and <code>R_BUILD_ENVIRON</code>; if unset,
files<a name="DOCF42" href="#FOOT42"><sup>42</sup></a>  <samp>~/.R/check.Renviron</samp> and
<samp>~/.R/build.Renviron</samp> are used) to set environment variables when
using these utilities.
</p></blockquote>

<blockquote>
<p><b>Note to Windows users:</b> <code>R CMD build</code> may make use of the Windows toolset (see the &ldquo;R
Installation and Administration&rdquo; manual) if present and in your path,
and it is required for packages which need it to install (including
those with <samp>configure.win</samp> or <samp>cleanup.win</samp> scripts or a
<samp>src</samp> directory) and e.g. need vignettes built.
</p>
<p>You may need to set the environment variable <code>TMPDIR</code> to point to a
suitable writable directory with a path not containing spaces &ndash; use
forward slashes for the separators.  Also, the directory needs to be on
a case-honouring file system (some network-mounted file systems are
not).
</p></blockquote>


<hr>
<a name="Checking-packages"></a>
<div class="header">
<p>
Next: <a href="#Building-package-tarballs" accesskey="n" rel="next">Building package tarballs</a>, Previous: <a href="#Checking-and-building-packages" accesskey="p" rel="prev">Checking and building packages</a>, Up: <a href="#Checking-and-building-packages" accesskey="u" rel="up">Checking and building packages</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Checking-packages-1"></a>
<h4 class="subsection">1.3.1 Checking packages</h4>
<a name="index-Checking-packages"></a>

<a name="index-R-CMD-check"></a>
<p>Using <code>R CMD check</code>, the R package checker, one can test whether
<em>source</em> R packages work correctly.  It can be run on one or
more directories, or compressed package <code>tar</code> archives with
extension <samp>.tar.gz</samp>, <samp>.tgz</samp>, <samp>.tar.bz2</samp> or
<samp>.tar.xz</samp>.
</p>
<p>It is strongly recommended that the final checks are run on a
<code>tar</code> archive prepared by <code>R CMD build</code>.
</p>
<p>This runs a series of checks, including
</p>
<ol>
<li> The package is installed.  This will warn about missing cross-references
and duplicate aliases in help files.

</li><li> The file names are checked to be valid across file systems and supported
operating system platforms.

</li><li> The files and directories are checked for sufficient permissions
(Unix-alikes only).

</li><li> The files are checked for binary executables, using a suitable version
of <code>file</code> if available<a name="DOCF43" href="#FOOT43"><sup>43</sup></a>.  (There may be
rare false positives.)

</li><li> The <samp>DESCRIPTION</samp> file is checked for completeness, and some of its
entries for correctness.  Unless installation tests are skipped,
checking is aborted if the package dependencies cannot be resolved at
run time.  (You may need to set <code>R_LIBS</code> in the environment if
dependent packages are in a separate library tree.) One check is that
the package name is not that of a standard package, nor one of the
defunct standard packages (&lsquo;<samp>ctest</samp>&rsquo;, &lsquo;<samp>eda</samp>&rsquo;, &lsquo;<samp>lqs</samp>&rsquo;,
&lsquo;<samp>mle</samp>&rsquo;, &lsquo;<samp>modreg</samp>&rsquo;, &lsquo;<samp>mva</samp>&rsquo;, &lsquo;<samp>nls</samp>&rsquo;, &lsquo;<samp>stepfun</samp>&rsquo; and
&lsquo;<samp>ts</samp>&rsquo;).  Another check is that all packages mentioned in
<code>library</code> or <code>require</code>s or from which the <samp>NAMESPACE</samp>
file imports or are called <em>via</em> <code>::</code> or <code>:::</code> are listed
(in &lsquo;<samp>Depends</samp>&rsquo;, &lsquo;<samp>Imports</samp>&rsquo;, &lsquo;<samp>Suggests</samp>&rsquo;): this is not an
exhaustive check of the actual imports.

</li><li> Available index information (in particular, for demos and vignettes) is
checked for completeness.

</li><li> The package subdirectories are checked for suitable file names and for
not being empty.  The checks on file names are controlled by the option
<samp>--check-subdirs=<var>value</var></samp>.  This defaults to &lsquo;<samp>default</samp>&rsquo;,
which runs the checks only if checking a tarball: the default can be
overridden by specifying the value as &lsquo;<samp>yes</samp>&rsquo; or &lsquo;<samp>no</samp>&rsquo;.  Further,
the check on the <samp>src</samp> directory is only run if the package
does not contain a <samp>configure</samp> script (which corresponds to the
value &lsquo;<samp>yes-maybe</samp>&rsquo;) and there is no <samp>src/Makefile</samp> or
<samp>src/Makefile.in</samp>.

<p>To allow a <samp>configure</samp> script to generate suitable files, files
ending in &lsquo;<samp>.in</samp>&rsquo; will be allowed in the <samp>R</samp> directory.
</p>
<p>A warning is given for directory names that look like R package check
directories &ndash; many packages have been submitted to <acronym>CRAN</acronym>
containing these.
</p>
</li><li> The R files are checked for syntax errors.  Bytes which are
non-<acronym>ASCII</acronym> are reported as warnings, but these should be
regarded as errors unless it is known that the package will always be
used in the same locale.

</li><li> It is checked that the package can be loaded, first with the usual
default packages and then only with package <strong>base</strong> already
loaded. It is checked that the namespace this can be loaded in an empty
session with only the <strong>base</strong> namespace loaded.  (Namespaces and
packages can be loaded very early in the session, before the default
packages are available, so packages should work then.)

</li><li> The R files are checked for correct calls to <code>library.dynam</code>.
Package startup functions are checked for correct argument lists and
(incorrect) calls to functions which modify the search path or
inappropriately generate messages.  The R code is checked for
possible problems using <a href="https://CRAN.R-project.org/package=codetools"><strong>codetools</strong></a>.  In addition, it is checked
whether S3 methods have all arguments of the corresponding generic, and
whether the final argument of replacement functions is called
&lsquo;<samp>value</samp>&rsquo;.  All foreign function calls (<code>.C</code>, <code>.Fortran</code>,
<code>.Call</code> and <code>.External</code> calls) are tested to see if they have
a <code>PACKAGE</code> argument, and if not, whether the appropriate DLL might
be deduced from the namespace of the package.  Any other calls are
reported.  (The check is generous, and users may want to supplement this
by examining the output of <code>tools::checkFF(&quot;mypkg&quot;, verbose=TRUE)</code>,
especially if the intention were to always use a <code>PACKAGE</code>
argument)

</li><li> The <samp>Rd</samp> files are checked for correct syntax and metadata,
including the presence of the mandatory fields (<code>\name</code>, <code>\alias</code>,
<code>\title</code> and <code>\description</code>).  The <samp>Rd</samp> name and
title are checked for being non-empty, and there is a check for missing
cross-references (links).

</li><li> A check is made for missing documentation entries, such as undocumented
user-level objects in the package.

</li><li> Documentation for functions, data sets, and S4 classes is checked for
consistency with the corresponding code.

</li><li> It is checked whether all function arguments given in <code>\usage</code>
sections of <samp>Rd</samp> files are documented in the corresponding
<code>\arguments</code> section.

</li><li> The <samp>data</samp> directory is checked for non-<acronym>ASCII</acronym> characters
and for the use of reasonable levels of compression.

</li><li> C, C++ and FORTRAN source and header files<a name="DOCF44" href="#FOOT44"><sup>44</sup></a> are
tested for portable (LF-only) line endings.  If there is a
<samp>Makefile</samp> or <samp>Makefile.in</samp> or <samp>Makevars</samp> or
<samp>Makevars.in</samp> file under the <samp>src</samp> directory, it is checked
for portable line endings and the correct use of &lsquo;<samp>$(BLAS_LIBS)</samp>&rsquo; and
&lsquo;<samp>$(LAPACK_LIBS)</samp>&rsquo;

<p>Compiled code is checked for symbols corresponding to functions which
might terminate R or write to <samp>stdout</samp>/<samp>stderr</samp> instead of
the console.  Note that the latter might give false positives in that
the symbols might be pulled in with external libraries and could never
be called.  Windows<a name="DOCF45" href="#FOOT45"><sup>45</sup></a> users
should note that the Fortran and C++ runtime libraries are examples of
such external libraries.
</p>
</li><li> Some checks are made of the contents of the <samp>inst/doc</samp> directory.
These always include checking for files that look like leftovers, and if
suitable tools (such as <code>qpdf</code>) are available, checking that the
PDF documentation is of minimal size.

</li><li> The examples provided by the package&rsquo;s documentation are run.
(see <a href="#Writing-R-documentation-files">Writing R documentation files</a>, for information on using
<code>\examples</code> to create executable example code.)  If there is a file
<samp>tests/Examples/<var>pkg</var>-Ex.Rout.save</samp>, the output of running the
examples is compared to that file.

<p>Of course, released packages should be able to run at least their own
examples.  Each example is run in a &lsquo;clean&rsquo; environment (so earlier
examples cannot be assumed to have been run), and with the variables
<code>T</code> and <code>F</code> redefined to generate an error unless they are set
in the example: See <a href="http://cran.r-project.org/doc/manuals/R-intro.html#Logical-vectors">Logical vectors</a> in <cite>An
Introduction to R</cite>.
</p>
</li><li> If the package sources contain a <samp>tests</samp> directory then the tests
specified in that directory are run.  (Typically they will consist of a
set of <samp>.R</samp> source files and target output files
<samp>.Rout.save</samp>.)  Please note that the comparison will be done in the
end user&rsquo;s locale, so the target output files should be <acronym>ASCII</acronym>
if at all possible.  (The command line option <code>--test-dir=foo</code> may
be used to specify tests in a non-standard location.  For example,
unusually slow tests could be placed in <samp>inst/slowTests</samp> and then
<code>R CMD check --test-dir=inst/slowTests</code> would be used to run them.
Other names that have been suggested are, for example,
<samp>inst/testWithOracle</samp> for tests that require Oracle to be installed,
<samp>inst/randomTests</samp> for tests which use random values and may
occasionally fail by chance, etc.)

</li><li> The code in package vignettes (see <a href="#Writing-package-vignettes">Writing package vignettes</a>) is
executed, and the vignette PDFs re-made from their sources as a check of
completeness of the sources (unless there is a &lsquo;<samp>BuildVignettes</samp>&rsquo;
field in the package&rsquo;s <samp>DESCRIPTION</samp> file with a false value).  If
there is a target output file <samp>.Rout.save</samp> in the vignette source
directory, the output from running the code in that vignette is compared
with the target output file and any differences are reported (but not
recorded in the log file).  (If the vignette sources are in the
deprecated location <samp>inst/doc</samp>, do mark such target output files to
not be installed in <samp>.Rinstignore</samp>.)

<p>If there is an error<a name="DOCF46" href="#FOOT46"><sup>46</sup></a> in executing the R code in vignette <samp><var>foo.ext</var></samp>, a log
file <samp><var>foo.ext</var>.log</samp> is created in the check directory.  The
vignette PDFs are re-made in a copy of the package sources in the
<samp>vign_test</samp> subdirectory of the check directory, so for further
information on errors look in directory
<samp><var>pkgname</var>/vign_test/vignettes</samp>.  (It is only retained if there
are errors or if environment variable <code>_R_CHECK_CLEAN_VIGN_TEST_</code> is
set to a false value.)
</p>
</li><li> The PDF version of the package&rsquo;s manual is created (to check that the
<samp>Rd</samp> files can be converted successfully).  This needs LaTeX and
suitable fonts and LaTeX packages to be installed.
See <a href="http://cran.r-project.org/doc/manuals/R-admin.html#Making-the-manuals">Making the manuals</a> in <cite>R Installation and Administration</cite>.

</li></ol>

<p>All these tests are run with collation set to the <code>C</code> locale, and
for the examples and tests with environment variable <code>LANGUAGE=en</code>:
this is to minimize differences between platforms.
</p>
<p>Use <kbd>R CMD check --help</kbd> to obtain more information about the usage
of the R package checker.  A subset of the checking steps can be
selected by adding command-line options. It also allows customization by
setting environment variables <code><span class="nolinebreak">_R_CHECK_*_</span></code><!-- /@w --> as described in
<a href="http://cran.r-project.org/doc/manuals/R-ints.html#Tools">Tools</a> in <cite>R Internals</cite>:
a set of these customizations similar to those used by <acronym>CRAN</acronym>
can be selected by the option <samp>--as-cran</samp> (which works best if
Internet access is available).  Some Windows users may
need to set environment variable <code>R_WIN_NO_JUNCTIONS</code> to a non-empty
value.  The test of cyclic declarations<a name="DOCF47" href="#FOOT47"><sup>47</sup></a>in <samp>DESCRIPTION</samp> files needs
repositories (including <acronym>CRAN</acronym>) set: do this in
<samp>~/.Rprofile</samp>, by e.g.
</p><div class="example">
<pre class="example">options(repos = c(CRAN=&quot;https://cran.r-project.org&quot;))
</pre></div>

<p>One check customization which can be revealing is
</p><div class="example">
<pre class="example">_R_CHECK_CODETOOLS_PROFILE_=&quot;suppressLocalUnused=FALSE&quot;
</pre></div>
<p>which reports unused local assignments.  Not only does this point out
computations which are unnecessary because their results are unused, it
also can uncover errors.  (Two such are to intend to update an object by
assigning a value but mistype its name or assign in the wrong scope,
for example using <code>&lt;-</code> where <code>&lt;&lt;-</code> was intended.)  This can
give false positives, most commonly because of non-standard evaluation
for formulae and because the intention is to return objects in the
environment of a function for later use.
</p>
<p>Complete checking of a package which contains a file <samp>README.md</samp>
needs <code>pandoc</code> installed: see
<a href="http://johnmacfarlane.net/pandoc/installing.html">http://johnmacfarlane.net/pandoc/installing.html</a>.  This
should be reasonably current: at the time of writing <acronym>CRAN</acronym> used
version 1.12.4.2 to process these files.
</p>
<p>You do need to ensure that the package is checked in a suitable locale
if it contains non-<acronym>ASCII</acronym> characters.  Such packages are likely
to fail some of the checks in a <code>C</code> locale, and <code>R CMD
check</code> will warn if it spots the problem.  You should be able to check
any package in a UTF-8 locale (if one is available).  Beware that
although a <code>C</code> locale is rarely used at a console, it may be the
default if logging in remotely or for batch jobs.
</p>
<blockquote>
<p><b>Multiple sub-architectures:</b> On systems which support multiple sub-architectures (principally
Windows), <code>R CMD check</code> will install and check a package which
contains compiled code under all available sub-architectures. (Use
option <samp>--force-multiarch</samp> to force this for packages without
compiled code, which are otherwise only checked under the main
sub-architecture.)  This will run the loading tests, examples and
<samp>tests</samp> directory under each installed sub-architecture in turn,
and give an error if any fail.  Where environment variables (including
perhaps <code>PATH</code>) need to be set differently for each
sub-architecture, these can be set in architecture-specific files such
as <samp><var>R_HOME</var>/etc/i386/Renviron.site</samp>.
</p>
<p>An alternative approach is to use <code>R CMD check --no-multiarch</code>
to check the primary sub-architecture, and then to use something like
<code>R --arch=x86_64 CMD check --extra-arch</code> or (Windows)
<code>/path/to/R/bin/x64/Rcmd check --extra-arch</code> to run for each
additional sub-architecture just the checks<a name="DOCF48" href="#FOOT48"><sup>48</sup></a> which differ by sub-architecture.  (This
approach is required for packages which are installed by <code>R CMD
INSTALL --merge-multiarch</code>.)
</p>
<p>Where packages need additional commands to install all the
sub-architectures these can be supplied by e.g.
<samp>--install-args=--force-biarch</samp>.
</p>
</blockquote>


<hr>
<a name="Building-package-tarballs"></a>
<div class="header">
<p>
Next: <a href="#Building-binary-packages" accesskey="n" rel="next">Building binary packages</a>, Previous: <a href="#Checking-packages" accesskey="p" rel="prev">Checking packages</a>, Up: <a href="#Checking-and-building-packages" accesskey="u" rel="up">Checking and building packages</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Building-package-tarballs-1"></a>
<h4 class="subsection">1.3.2 Building package tarballs</h4>
<a name="index-Building-source-packages"></a>

<a name="index-R-CMD-build"></a>
<a name="index-Package-builder"></a>
<a name="index-tarballs"></a>
<p>Packages may be distributed in source form as &ldquo;tarballs&rdquo;
(<samp>.tar.gz</samp> files) or in binary form. The source form can be
installed on all platforms with suitable tools and is the usual form for
Unix-like systems; the binary form is platform-specific, and is the more
common distribution form for the Windows and macOS platforms.
</p>
<p>Using <code>R CMD build</code>, the R package builder, one can build R
package tarballs from their sources (for example, for subsequent release).
</p>
<p>Prior to actually building the package in the standard gzipped tar file
format, a few diagnostic checks and cleanups are performed.  In
particular, it is tested whether object indices exist and can be assumed
to be up-to-date, and C, C++ and FORTRAN source files and relevant
makefiles in a <samp>src</samp> directory are tested and converted to LF
line-endings if necessary.
</p>
<p>Run-time checks whether the package works correctly should be performed
using <code>R CMD check</code> prior to invoking the final build procedure.
</p>
<a name="index-_002eRbuildignore-file"></a>
<p>To exclude files from being put into the package, one can specify a list
of exclude patterns in file <samp>.Rbuildignore</samp> in the top-level source
directory.  These patterns should be Perl-like regular expressions (see
the help for <code>regexp</code> in R for the precise details), one per
line, to be matched case-insensitively<a name="DOCF49" href="#FOOT49"><sup>49</sup></a> against the file and directory names relative to the
top-level package source directory.  In addition, directories from
source control systems<a name="DOCF50" href="#FOOT50"><sup>50</sup></a> or from <code>eclipse</code><a name="DOCF51" href="#FOOT51"><sup>51</sup></a>, directories with names ending <samp>.Rcheck</samp> or
<samp>Old</samp> or <samp>old</samp> and files <samp>GNUMakefile</samp><a name="DOCF52" href="#FOOT52"><sup>52</sup></a>, <samp>Read-and-delete-me</samp>
or with base names starting with &lsquo;<samp>.#</samp>&rsquo;, or starting and ending with
&lsquo;<samp>#</samp>&rsquo;, or ending in &lsquo;<samp>~</samp>&rsquo;, &lsquo;<samp>.bak</samp>&rsquo; or &lsquo;<samp>.swp</samp>&rsquo;, are
excluded by default.  In addition, those files in the <samp>R</samp>,
<samp>demo</samp> and <samp>man</samp> directories which are flagged by <code>R
CMD check</code> as having invalid names will be excluded.
</p>
<p>Use <kbd>R CMD build --help</kbd> to obtain more information about the usage
of the R package builder.
</p>
<p>Unless <kbd>R CMD build</kbd> is invoked with the
<samp>--no-build-vignettes</samp> option (or the package&rsquo;s
<samp>DESCRIPTION</samp> contains &lsquo;<samp>BuildVignettes: no</samp>&rsquo; or similar), it
will attempt to (re)build the vignettes (see <a href="#Writing-package-vignettes">Writing package vignettes</a>) in the package.  To do so it installs the current package
into a temporary library tree, but any dependent packages need to be
installed in an available library tree (see the Note: at the top of this
section).
</p>
<p>Similarly, if the <samp>.Rd</samp> documentation files contain any
<code>\Sexpr</code> macros (see <a href="#Dynamic-pages">Dynamic pages</a>), the package will be
temporarily installed to execute them.  Post-execution binary copies of
those pages containing build-time macros will be saved in
<samp>build/partial.rdb</samp>.  If there are any install-time or render-time
macros, a <samp>.pdf</samp> version of the package manual will be built and
installed in the <samp>build</samp> subdirectory.  (This allows
<acronym>CRAN</acronym> or other repositories to display the manual even if they
are unable to install the package.)  This can be suppressed by the
option <samp>--no-manual</samp> or if package&rsquo;s <samp>DESCRIPTION</samp> contains
&lsquo;<samp>BuildManual: no</samp>&rsquo; or similar.
</p>
<p>One of the checks that <code>R CMD build</code> runs is for empty source
directories.  These are in most (but not all) cases unintentional, if
they are intentional use the option <samp>--keep-empty-dirs</samp> (or set
the environment variable <code>_R_BUILD_KEEP_EMPTY_DIRS_</code> to &lsquo;<samp>TRUE</samp>&rsquo;,
or have a &lsquo;<samp>BuildKeepEmpty</samp>&rsquo; field with a true value in the
<samp>DESCRIPTION</samp> file).
</p>
<p>The <samp>--resave-data</samp> option allows saved images (<samp>.rda</samp> and
<samp>.RData</samp> files) in the <samp>data</samp> directory to be optimized for
size.  It will also compress tabular files and convert <samp>.R</samp> files
to saved images.  It can take values <code>no</code>, <code>gzip</code> (the default
if this option is not supplied, which can be changed by setting the
environment variable <code>_R_BUILD_RESAVE_DATA_</code>) and <code>best</code>
(equivalent to giving it without a value), which chooses the most
effective compression.  Using <code>best</code> adds a dependence on <code>R
(&gt;= 2.10)</code> to the <samp>DESCRIPTION</samp> file if <code>bzip2</code> or
<code>xz</code> compression is selected for any of the files.  If this is
thought undesirable, <samp>--resave-data=gzip</samp> (which is the default
if that option is not supplied) will do what compression it can with
<code>gzip</code>.  A package can control how its data is resaved by
supplying a &lsquo;<samp>BuildResaveData</samp>&rsquo; field (with one of the values given
earlier in this paragraph) in its <samp>DESCRIPTION</samp> file.
</p>
<p>The <samp>--compact-vignettes</samp> option will run
<code>tools::compactPDF</code> over the PDF files in <samp>inst/doc</samp> (and its
subdirectories) to losslessly compress them.  This is not enabled by
default (it can be selected by environment variable
<code>_R_BUILD_COMPACT_VIGNETTES_</code>) and needs <code>qpdf</code>
(<a href="http://qpdf.sourceforge.net/">http://qpdf.sourceforge.net/</a>) to be available.
</p>
<p>It can be useful to run <code>R CMD check --check-subdirs=yes</code> on the
built tarball as a final check on the contents.
</p>
<p>Where a non-POSIX file system is in use which does not utilize execute
permissions, some care is needed with permissions.  This applies on
Windows and to e.g. FAT-formatted drives and SMB-mounted file systems
on other OSes.  The &lsquo;mode&rsquo; of the file recorded in the tarball will be
whatever <code>file.info()</code> returns.  On Windows this will record only
directories as having execute permission and on other OSes it is likely
that all files have reported &lsquo;mode&rsquo; <code>0777</code>.  A particular issue is
packages being built on Windows which are intended to contain executable
scripts such as <samp>configure</samp> and <samp>cleanup</samp>: <code>R CMD
build</code> ensures those two are recorded with execute permission.
</p>
<p>Directory <samp>build</samp> of the package sources is reserved for use by
<code>R CMD build</code>: it contains information which may not easily be
created when the package is installed, including index information on
the vignettes and, rarely, information on the help pages and perhaps a
copy of the PDF reference manual (see above).
</p>
<hr>
<a name="Building-binary-packages"></a>
<div class="header">
<p>
Previous: <a href="#Building-package-tarballs" accesskey="p" rel="prev">Building package tarballs</a>, Up: <a href="#Checking-and-building-packages" accesskey="u" rel="up">Checking and building packages</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Building-binary-packages-1"></a>
<h4 class="subsection">1.3.3 Building binary packages</h4>
<a name="index-Building-binary-packages"></a>

<p>Binary packages are compressed copies of installed versions of
packages.  They contain compiled shared libraries rather than C, C++ or
Fortran source code, and the R functions are included in their installed
form.  The format and filename are platform-specific; for example, a
binary package for Windows is usually supplied as a <samp>.zip</samp> file,
and for the macOS platform the default binary package file extension is
<samp>.tgz</samp>.
</p>
<p>The recommended method of building binary packages is to use
</p>
<p><code>R CMD INSTALL --build pkg</code>
where <samp>pkg</samp> is either the name of a source tarball (in the usual
<samp>.tar.gz</samp> format) or the location of the directory of the package
source to be built.  This operates by first installing the package and
then packing the installed binaries into the appropriate binary package
file for the particular platform.
</p>
<p>By default, <code>R CMD INSTALL --build</code> will attempt to install the
package into the default library tree for the local installation of
R. This has two implications:
</p>
<ul>
<li> If the installation is successful, it will overwrite any existing installation
of the same package.

</li><li> The default library tree must have write permission; if not, the package will
not install and the binary will not be created.

</li></ul>

<p>To prevent changes to the present working installation or to provide an
install location with write access, create a suitably located directory
with write access and use the <code>-l</code> option to build the package
in the chosen location.  The usage is then
</p>
<p><code>R CMD INSTALL -l location --build pkg</code>
</p>
<p>where <samp>location</samp> is the chosen directory with write access. The package
will be installed as a subdirectory of <samp>location</samp>, and the package binary
will be created in the current directory.
</p>
<p>Other options for <code>R CMD INSTALL</code> can be found using <code>R
CMD INSTALL --help</code>, and platform-specific details for special cases are
discussed in the platform-specific FAQs.
</p>



<p>Finally, at least one web-based service is available for building binary
packages from (checked) source code: WinBuilder (see
<a href="https://win-builder.R-project.org/">https://win-builder.R-project.org/</a>) is able to build Windows
binaries. Note that this is intended for developers on other platforms
who do not have access to Windows but wish to provide binaries for the
Windows platform.
</p>
<hr>
<a name="Writing-package-vignettes"></a>
<div class="header">
<p>
Next: <a href="#Package-namespaces" accesskey="n" rel="next">Package namespaces</a>, Previous: <a href="#Checking-and-building-packages" accesskey="p" rel="prev">Checking and building packages</a>, Up: <a href="#Creating-R-packages" accesskey="u" rel="up">Creating R packages</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Writing-package-vignettes-1"></a>
<h3 class="section">1.4 Writing package vignettes</h3>
<a name="index-vignettes"></a>
<a name="index-Sweave"></a>

<table summary="" class="menu" border="0" cellspacing="0">
<tr><td align="left" valign="top">&bull; <a href="#Encodings-and-vignettes" accesskey="1">Encodings and vignettes</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Non_002dSweave-vignettes" accesskey="2">Non-Sweave vignettes</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
</table>

<p>In addition to the help files in <samp>Rd</samp> format, R packages allow
the inclusion of documents in arbitrary other formats.  The standard
location for these is subdirectory <samp>inst/doc</samp> of a source package,
the contents will be copied to subdirectory <samp>doc</samp> when the package
is installed.  Pointers from package help indices to the installed
documents are automatically created.  Documents in <samp>inst/doc</samp> can
be in arbitrary format, however we strongly recommend providing them in
PDF format, so users on almost all platforms can easily read them.  To
ensure that they can be accessed from a browser (as an <acronym>HTML</acronym> index is
provided), the file names should start with an <acronym>ASCII</acronym> letter
and be comprised entirely of <acronym>ASCII</acronym> letters or digits or hyphen
or underscore.
</p>
<p>A special case is <em>package vignettes</em>.  Vignettes are documents in
PDF or <acronym>HTML</acronym> format obtained from plain text literate source files
from which R knows how to extract R code and create output (in
PDF/<acronym>HTML</acronym> or intermediate LaTeX).  Vignette engines do this work,
using &ldquo;tangle&rdquo; and &ldquo;weave&rdquo; functions respectively.  Sweave, provided
by the R distribution, is the default engine.  Since R version 3.0.0,
other vignette engines besides Sweave are supported; see <a href="#Non_002dSweave-vignettes">Non-Sweave vignettes</a>.
</p>
<p>Package vignettes have their sources in subdirectory <samp>vignettes</samp> of
the package sources.  Note that the location of the vignette sources
only affects <code>R CMD build</code> and <code>R CMD check</code>: the
tarball built by <code>R CMD build</code> includes in <samp>inst/doc</samp> the
components intended to be installed.
</p>
<p>Sweave vignette sources are normally given the file extension
<samp>.Rnw</samp> or <samp>.Rtex</samp>, but for historical reasons
extensions<a name="DOCF53" href="#FOOT53"><sup>53</sup></a> <samp>.Snw</samp> and
<samp>.Stex</samp> are also recognized.  Sweave allows the integration of
LaTeX documents: see the <code>Sweave</code> help page in R and the
<code>Sweave</code> vignette in package <strong>utils</strong> for details on the
source document format.
</p>
<p>Package vignettes are tested by <code>R CMD check</code> by executing all R
code chunks they contain (except those marked for non-evaluation, e.g.,
with option <code>eval=FALSE</code> for Sweave).  The R working directory
for all vignette tests in <code>R CMD check</code> is a <em>copy</em> of the
vignette source directory.  Make sure all files needed to run the R
code in the vignette (data sets, &hellip;) are accessible by either
placing them in the <samp>inst/doc</samp> hierarchy of the source package or
by using calls to <code>system.file()</code>.  All other files needed to
re-make the vignettes (such as LaTeX style files, BibTeX input
files and files for any figures not created by running the code in the
vignette) must be in the vignette source directory.
</p>
<p><code>R CMD build</code> will automatically<a name="DOCF54" href="#FOOT54"><sup>54</sup></a> create the
(PDF or <acronym>HTML</acronym> versions of the) vignettes in <samp>inst/doc</samp> for
distribution with the package sources.  By including the vignette
outputs in the package sources it is not necessary that these can be
re-built at install time, i.e., the package author can use private R
packages, screen snapshots and LaTeX extensions which are only
available on his machine.<a name="DOCF55" href="#FOOT55"><sup>55</sup></a>
</p>
<p>By default <code>R CMD build</code> will run <code>Sweave</code> on all Sweave
vignette source files in <samp>vignettes</samp>.  If <samp>Makefile</samp> is found
in the vignette source directory, then <code>R CMD build</code> will try to
run <code>make</code> after the <code>Sweave</code> runs, otherwise
<code>texi2pdf</code> is run on each <samp>.tex</samp> file produced.
</p>
<p>The first target in the <samp>Makefile</samp> should take care of both
creation of PDF/<acronym>HTML</acronym> files and cleaning up afterwards (including
after <code>Sweave</code>), i.e., delete all files that shall not appear in
the final package archive.  Note that if the <code>make</code> step runs R
it needs to be careful to respect the environment values of <code>R_LIBS</code>
and <code>R_HOME</code><a name="DOCF56" href="#FOOT56"><sup>56</sup></a>.
Finally, if there is a <samp>Makefile</samp> and it has a &lsquo;<samp>clean:</samp>&rsquo;
target, <code>make clean</code> is run.
</p>
<p>All the usual <em>caveats</em> about including a <samp>Makefile</samp> apply.
It must be portable (no <acronym>GNU</acronym> extensions), use LF line endings
and must work correctly with a parallel <code>make</code>: too many authors
have written things like
</p>
<div class="example">
<pre class="example">## BAD EXAMPLE
all: pdf clean

pdf: ABC-intro.pdf ABC-details.pdf

%.pdf:  %.tex
	texi2dvi --pdf $*

clean:
	rm *.tex ABC-details-*.pdf
</pre></div>

<p>which will start removing the source files whilst <code>pdflatex</code> is
working.
</p>
<p>Metadata lines can be placed in the source file, preferably in LaTeX
comments in the preamble.  One such is a <code>\VignetteIndexEntry</code> of
the form
</p><div class="example">
<pre class="example">%\VignetteIndexEntry{Using Animal}
</pre></div>
<p>Others you may see are <code>\VignettePackage</code> (currently ignored),
<code>\VignetteDepends</code> and <code>\VignetteKeyword</code> (which replaced
<code>\VignetteKeywords</code>).  These are processed at package installation
time to create the saved data frame <samp>Meta/vignette.rds</samp>, but only
the <code>\VignetteIndexEntry</code> and <code>\VignetteKeyword</code> statements
are currently used.  The <code>\VignetteEngine</code> statement
is described in <a href="#Non_002dSweave-vignettes">Non-Sweave vignettes</a>.
</p>
<p>At install time an <acronym>HTML</acronym> index for all vignettes in the package is
automatically created from the <code>\VignetteIndexEntry</code> statements
unless a file <samp>index.html</samp> exists in directory
<samp>inst/doc</samp>. This index is linked from the <acronym>HTML</acronym> help index for
the package.  If you do supply a <samp>inst/doc/index.html</samp> file it
should contain relative links only to files under the installed
<samp>doc</samp> directory, or perhaps (not really an index) to <acronym>HTML</acronym> help
files or to the <samp>DESCRIPTION</samp> file, and be valid <acronym>HTML</acronym> as
confirmed via the <a href="https://validator.w3.org">W3C Markup Validation
Service</a> or <a href="https://validator.nu/">Validator.nu</a>.
</p>
<p>Sweave/Stangle allows the document to specify the <code>split=TRUE</code>
option to create a single R file for each code chunk: this will not
work for vignettes where it is assumed that each vignette source
generates a single file with the vignette extension replaced by
<samp>.R</samp>.
</p>
<p>Do watch that PDFs are not too large &ndash; one in a <acronym>CRAN</acronym> package
was 72MB!  This is usually caused by the inclusion of overly detailed
figures, which will not render well in PDF viewers.  Sometimes it is
much better to generate fairly high resolution bitmap (PNG, JPEG)
figures and include those in the PDF document.
</p>
<a name="index-_002einstall_005fextras-file"></a>
<p>When <code>R CMD build</code> builds the vignettes, it copies these and
the vignette sources from directory <samp>vignettes</samp> to <samp>inst/doc</samp>.
To install any other files from the <samp>vignettes</samp> directory, include
a file <samp>vignettes/.install_extras</samp> which specifies these as
Perl-like regular expressions on one or more lines.  (See the
description of the <samp>.Rinstignore</samp> file for full details.)
</p>

<hr>
<a name="Encodings-and-vignettes"></a>
<div class="header">
<p>
Next: <a href="#Non_002dSweave-vignettes" accesskey="n" rel="next">Non-Sweave vignettes</a>, Previous: <a href="#Writing-package-vignettes" accesskey="p" rel="prev">Writing package vignettes</a>, Up: <a href="#Writing-package-vignettes" accesskey="u" rel="up">Writing package vignettes</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Encodings-and-vignettes-1"></a>
<h4 class="subsection">1.4.1 Encodings and vignettes</h4>

<p>Vignettes will in general include descriptive text, R input, R
output and figures, LaTeX include files and bibliographic references.
As any of these may contain non-<acronym>ASCII</acronym> characters, the handling
of encodings can become very complicated.
</p>
<p>The vignette source file should be written in <acronym>ASCII</acronym> or contain
a declaration of the encoding (see below).  This applies even to
comments within the source file, since vignette engines process comments
to look for options and metadata lines.  When an engine&rsquo;s weave and
tangle functions are called on the vignette source, it will be converted
to the encoding of the current R session.
</p>
<p><code>Stangle()</code> will produce an R code file in the current locale&rsquo;s
encoding: for a non-<acronym>ASCII</acronym> vignette what that is is recorded in a
comment at the top of the file.
</p>
<p><code>Sweave()</code> will produce a <samp>.tex</samp> file in the current
encoding, or in UTF-8 if that is declared.  Non-<acronym>ASCII</acronym> encodings
need to be declared to LaTeX via a line like
</p><div class="example">
<pre class="example">\usepackage[utf8]{inputenc}
</pre></div>
<p>(It is also possible to use the more recent &lsquo;<samp>inputenx</samp>&rsquo; LaTeX
package.) For files where this line is not needed (e.g. chapters
included within the body of a larger document, or non-Sweave
vignettes), the encoding may be declared using a comment like
</p><div class="example">
<pre class="example">%\VignetteEncoding{UTF-8}
</pre></div>
<p>If the encoding is UTF-8, this can also be declared using
the declaration
</p><div class="example">
<pre class="example">%\SweaveUTF8
</pre></div>
<p>If no declaration is given in the vignette, it will be assumed to be
in the encoding declared for the package.  If there is no encoding
declared in either place, then it is an error to use non-<acronym>ASCII</acronym>
characters in the vignette.
</p>
<p>In any case, be aware that LaTeX may require the &lsquo;<samp>usepackage</samp>&rsquo;
declaration.
</p>
<p><code>Sweave()</code> will also parse and evaluate the R code in each
chunk.  The R output will also be in the current locale (or <acronym>UTF-8</acronym>
if so declared), and should
be covered by the &lsquo;<samp>inputenc</samp>&rsquo; declaration.  One thing people often
forget is that the R output may not be <acronym>ASCII</acronym> even for
<acronym>ASCII</acronym> R sources, for many possible reasons.  One common one
is the use of &lsquo;fancy&rsquo; quotes: see the R help on <code>sQuote</code>: note
carefully that it is not portable to declare UTF-8 or CP1252 to cover
such quotes, as their encoding will depend on the locale used to run
<code>Sweave()</code>:  this can be circumvented by setting
<code>options(useFancyQuotes=&quot;UTF-8&quot;)</code> in the vignette.
</p>
<p>The final issue is the encoding of figures &ndash; this applies only to PDF
figures and not PNG etc.  The PDF figures will contain declarations for
their encoding, but the Sweave option <code>pdf.encoding</code> may need to be
set appropriately: see the help for the <code>pdf()</code> graphics device.
</p>
<p>As a real example of the complexities, consider the <a href="https://CRAN.R-project.org/package=fortunes"><strong>fortunes</strong></a>
package version &lsquo;<samp>1.4-0</samp>&rsquo;.  That package did not have a declared
encoding, and its vignette was in <acronym>ASCII</acronym>.  However, the data it
displays are read from a UTF-8 CSV file and will be assumed to be in the
current encoding, so <samp>fortunes.tex</samp> will be in UTF-8 in any locale.
Had <code>read.table</code> been told the data were UTF-8, <samp>fortunes.tex</samp>
would have been in the locale&rsquo;s encoding.
</p>
<hr>
<a name="Non_002dSweave-vignettes"></a>
<div class="header">
<p>
Previous: <a href="#Encodings-and-vignettes" accesskey="p" rel="prev">Encodings and vignettes</a>, Up: <a href="#Writing-package-vignettes" accesskey="u" rel="up">Writing package vignettes</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Non_002dSweave-vignettes-1"></a>
<h4 class="subsection">1.4.2 Non-Sweave vignettes</h4>

<p>R 3.0.0 and later allow vignettes in formats other than Sweave by
means of &ldquo;vignette engines&rdquo;.  For example <a href="https://CRAN.R-project.org/package=knitr"><strong>knitr</strong></a> version 1.1
or later can create <samp>.tex</samp> files from a variation on Sweave format,
and <samp>.html</samp> files from a variation on &ldquo;markdown&rdquo; format. These
engines replace the <code>Sweave()</code> function with other functions to
convert vignette source files into LaTeX files for processing into
<samp>.pdf</samp>, or directly into <samp>.pdf</samp> or <samp>.html</samp> files. The
<code>Stangle()</code> function is replaced with a function that extracts the
R source from a vignette.
</p>
<p>R recognizes non-Sweave vignettes using filename extensions specified
by the engine.  For example, the <a href="https://CRAN.R-project.org/package=knitr"><strong>knitr</strong></a> package supports
the extension <samp>.Rmd</samp> (standing for
&ldquo;R markdown&rdquo;).  The user indicates the vignette engine
within the vignette source using a <code>\VignetteEngine</code> line, for example
</p><div class="example">
<pre class="example">%\VignetteEngine{knitr::knitr}
</pre></div>
<p>This specifies the name of a package and an engine to use in place of
Sweave in processing the vignette.  As <code>Sweave</code> is the only engine
supplied with the R distribution, the package providing any other
engine must be specified in the &lsquo;<samp>VignetteBuilder</samp>&rsquo; field of the
package <samp>DESCRIPTION</samp> file, and also specified in the
&lsquo;<samp>Suggests</samp>&rsquo;, &lsquo;<samp>Imports</samp>&rsquo; or &lsquo;<samp>Depends</samp>&rsquo; field (since its
namespace must be available to build or check your package).  If more
than one package is specified as a builder, they will be searched in the
order given there.  The <strong>utils</strong> package is always implicitly
appended to the list of builder packages, but may be included earlier
to change the search order.
</p>
<p>Note that a package with non-Sweave vignettes should always have a
&lsquo;<samp>VignetteBuilder</samp>&rsquo; field in the <samp>DESCRIPTION</samp> file, since this
is how <code>R CMD check</code> recognizes that there are vignettes to be
checked: packages listed there are required when the package is checked.
</p>
<p>The vignette engine can produce <samp>.tex</samp>, <samp>.pdf</samp>, or <samp>.html</samp>
files as output.  If it produces <samp>.tex</samp> files, R will
call <code>texi2pdf</code> to convert them to <samp>.pdf</samp> for display
to the user (unless there is a <samp>Makefile</samp> in the <samp>vignettes</samp>
directory).
</p>
<p>Package writers who would like to supply vignette engines need
to register those engines in the package <code>.onLoad</code> function.
For example, that function could make the call
</p><div class="example">
<pre class="example">tools::vignetteEngine(&quot;knitr&quot;, weave = vweave, tangle = vtangle,
		      pattern = &quot;[.]Rmd$&quot;, package = &quot;knitr&quot;)
</pre></div>
<p>(The actual registration in <a href="https://CRAN.R-project.org/package=knitr"><strong>knitr</strong></a> is more complicated, because
it supports other input formats.)  See the <code>?tools::vignetteEngine</code>
help topic for details on engine registration.
</p>

<hr>
<a name="Package-namespaces"></a>
<div class="header">
<p>
Next: <a href="#Writing-portable-packages" accesskey="n" rel="next">Writing portable packages</a>, Previous: <a href="#Writing-package-vignettes" accesskey="p" rel="prev">Writing package vignettes</a>, Up: <a href="#Creating-R-packages" accesskey="u" rel="up">Creating R packages</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Package-namespaces-1"></a>
<h3 class="section">1.5 Package namespaces</h3>
<a name="index-namespaces"></a>

<p>R has a namespace management system for code in packages.  This
system allows the package writer to specify which variables in the
package should be <em>exported</em> to make them available to package
users, and which variables should be <em>imported</em> from other
packages.
</p>
<p>The namespace for a package is specified by the
<samp>NAMESPACE</samp> file in the top level package directory.  This file
contains <em>namespace directives</em> describing the imports and exports
of the namespace.  Additional directives register any shared objects to
be loaded and any S3-style methods that are provided.  Note that
although the file looks like R code (and often has R-style
comments) it is not processed as R code.  Only very simple
conditional processing of <code>if</code> statements is implemented.
</p>
<p>Packages are loaded and attached to the search path by calling
<code>library</code> or <code>require</code>.  Only the exported variables are
placed in the attached frame.  Loading a package that imports variables
from other packages will cause these other packages to be loaded as well
(unless they have already been loaded), but they will <em>not</em> be
placed on the search path by these implicit loads.  Thus code in the
package can only depend on objects in its own namespace and its imports
(including the <strong>base</strong> namespace) being visible<a name="DOCF57" href="#FOOT57"><sup>57</sup></a>.
</p>
<p>Namespaces are <em>sealed</em> once they are loaded.  Sealing means that
imports and exports cannot be changed and that internal variable
bindings cannot be changed.  Sealing allows a simpler implementation
strategy for the namespace mechanism.  Sealing also allows code
analysis and compilation tools to accurately identify the definition
corresponding to a global variable reference in a function body.
</p>
<p>The namespace controls the search strategy for variables used by
functions in the package.  If not found locally, R searches the
package namespace first, then the imports, then the base namespace and
then the normal search path.
</p>

<table summary="" class="menu" border="0" cellspacing="0">
<tr><td align="left" valign="top">&bull; <a href="#Specifying-imports-and-exports" accesskey="1">Specifying imports and exports</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Registering-S3-methods" accesskey="2">Registering S3 methods</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Load-hooks" accesskey="3">Load hooks</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#useDynLib" accesskey="4">useDynLib</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#An-example" accesskey="5">An example</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Namespaces-with-S4-classes-and-methods" accesskey="6">Namespaces with S4 classes and methods</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
</table>


<hr>
<a name="Specifying-imports-and-exports"></a>
<div class="header">
<p>
Next: <a href="#Registering-S3-methods" accesskey="n" rel="next">Registering S3 methods</a>, Previous: <a href="#Package-namespaces" accesskey="p" rel="prev">Package namespaces</a>, Up: <a href="#Package-namespaces" accesskey="u" rel="up">Package namespaces</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Specifying-imports-and-exports-1"></a>
<h4 class="subsection">1.5.1 Specifying imports and exports</h4>

<p>Exports are specified using the <code>export</code> directive in the
<samp>NAMESPACE</samp> file.  A directive of the form
</p>
<a name="index-export"></a>
<div class="example">
<pre class="example">export(f, g)
</pre></div>

<p>specifies that the variables <code>f</code> and <code>g</code> are to be exported.
(Note that variable names may be quoted, and reserved words and
non-standard names such as <code>[&lt;-.fractions</code> must be.)
</p>
<p>For packages with many variables to export it may be more convenient to
specify the names to export with a regular expression using
<code>exportPattern</code>.  The directive
</p>
<a name="index-exportPattern"></a>
<div class="example">
<pre class="example">exportPattern(&quot;^[^\\.]&quot;)
</pre></div>

<p>exports all variables that do not start with a period.  However, such
broad patterns are not recommended for production code: it is better to
list all exports or use narrowly-defined groups.  (This pattern applies
to S4 classes.)  Beware of patterns which include names starting with a
period: some of these are internal-only variables and should never be
exported, e.g. &lsquo;<samp>.__S3MethodsTable__.</samp>&rsquo; (and the code nowadays
excludes known cases).
</p>
<p>Packages implicitly import the base namespace.
Variables exported from other packages with namespaces need to be
imported explicitly using the directives <code>import</code> and
<code>importFrom</code>.  The <code>import</code> directive imports all exported
variables from the specified package(s).  Thus the directives
</p>
<a name="index-import"></a>
<div class="example">
<pre class="example">import(foo, bar)
</pre></div>

<p>specifies that all exported variables in the packages <strong>foo</strong> and
<strong>bar</strong> are to be imported.  If only some of the exported variables
from a package are needed, then they can be imported using
<code>importFrom</code>.  The directive
</p>
<a name="index-importFrom"></a>
<div class="example">
<pre class="example">importFrom(foo, f, g)
</pre></div>

<p>specifies that the exported variables <code>f</code> and <code>g</code> of the
package <strong>foo</strong> are to be imported.  Using <code>importFrom</code>
selectively rather than <code>import</code> is good practice and recommended
notably when importing from packages with more than a dozen exports.
</p>
<p>To import every symbol from a package but for a few exceptions,
pass the <code>except</code> argument to <code>import</code>. The directive
</p>
<div class="example">
<pre class="example">import(foo, except=c(bar, baz))
</pre></div>

<p>imports every symbol from <strong>foo</strong> except <code>bar</code> and
<code>baz</code>. The value of <code>except</code> should evaluate to something
coercible to a character vector, after substituting each symbol for
its corresponding string.
</p>
<p>It is possible to export variables from a namespace which it has
imported from other namespaces: this has to be done explicitly and not
<em>via</em> <code>exportPattern</code>.
</p>
<p>If a package only needs a few objects from another package it can use a
fully qualified variable reference in the code instead of a formal
import.  A fully qualified reference to the function <code>f</code> in package
<strong>foo</strong> is of the form <code>foo::f</code>.  This is slightly less efficient
than a formal import and also loses the advantage of recording all
dependencies in the <samp>NAMESPACE</samp> file (but they still need to be
recorded in the <samp>DESCRIPTION</samp> file).  Evaluating <code>foo::f</code> will
cause package <strong>foo</strong> to be loaded, but not attached, if it was not
loaded already&mdash;this can be an advantage in delaying the loading of a
rarely used package.
</p>
<p>Using <code>foo:::f</code> instead of <code>foo::f</code> allows access to
unexported objects.  This is generally not recommended, as the
semantics of unexported objects may be changed by the package author
in routine maintenance.
</p>
<hr>
<a name="Registering-S3-methods"></a>
<div class="header">
<p>
Next: <a href="#Load-hooks" accesskey="n" rel="next">Load hooks</a>, Previous: <a href="#Specifying-imports-and-exports" accesskey="p" rel="prev">Specifying imports and exports</a>, Up: <a href="#Package-namespaces" accesskey="u" rel="up">Package namespaces</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Registering-S3-methods-1"></a>
<h4 class="subsection">1.5.2 Registering S3 methods</h4>

<p>The standard method for S3-style <code>UseMethod</code> dispatching might fail
to locate methods defined in a package that is imported but not attached
to the search path.  To ensure that these methods are available the
packages defining the methods should ensure that the generics are
imported and register the methods using <code>S3method</code> directives.  If
a package defines a function <code>print.foo</code> intended to be used as a
<code>print</code> method for class <code>foo</code>, then the directive
</p>
<a name="index-S3method"></a>
<div class="example">
<pre class="example">S3method(print, foo)
</pre></div>

<p>ensures that the method is registered and available for <code>UseMethod</code>
dispatch, and the function <code>print.foo</code> does not need to be exported.
Since the generic <code>print</code> is defined in <strong>base</strong> it does not need
to be imported explicitly.
</p>
<p>(Note that function and class names may be quoted, and reserved words
and non-standard names such as <code>[&lt;-</code> and <code>function</code> must
be.)
</p>
<p>It is possible to specify a third argument to S3method, the function to
be used as the method, for example
</p>
<div class="example">
<pre class="example">S3method(print, check_so_symbols, .print.via.format)
</pre></div>

<p>when <code>print.check_so_symbols</code> is not needed.
</p>
<p>There used to be a limit on the number of <code>S3method</code> directives: it
was <code>500</code> prior to R 3.0.2.
</p>
<hr>
<a name="Load-hooks"></a>
<div class="header">
<p>
Next: <a href="#useDynLib" accesskey="n" rel="next">useDynLib</a>, Previous: <a href="#Registering-S3-methods" accesskey="p" rel="prev">Registering S3 methods</a>, Up: <a href="#Package-namespaces" accesskey="u" rel="up">Package namespaces</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Load-hooks-1"></a>
<h4 class="subsection">1.5.3 Load hooks</h4>

<a name="index-_002eonLoad"></a>
<a name="index-_002eonAttach"></a>
<p>There are a number of hooks called as packages are loaded, attached,
detached, and unloaded.  See <code>help(&quot;.onLoad&quot;)</code> for more details.
</p>
<p>Since loading and attaching are distinct operations, separate hooks are
provided for each.  These hook functions are called <code>.onLoad</code> and
<code>.onAttach</code>.  They both take arguments<a name="DOCF58" href="#FOOT58"><sup>58</sup></a> <code>libname</code> and
<code>pkgname</code>; they should be defined in the namespace but not
exported.
</p>
<a name="index-_002eonUnload"></a>
<a name="index-_002eonDetach"></a>
<a name="index-_002eLast_002elib"></a>
<p>Packages can use a <code>.onDetach</code> (as from R 3.0.0) or <code>.Last.lib</code>
function (provided the latter is exported from the namespace) when
<code>detach</code> is called on the package.  It is called with a single
argument, the full path to the installed package.  There is also a hook
<code>.onUnload</code> which is called when the namespace is unloaded
(<em>via</em> a call to <code>unloadNamespace</code>, perhaps called by
<code>detach(unload = TRUE)</code>) with argument the full path to the installed
package&rsquo;s directory.  <code>.onUnload</code> and <code>.onDetach</code> should be
defined in the namespace and not exported, but <code>.Last.lib</code> does
need to be exported.
</p>
<p>Packages are not likely to need <code>.onAttach</code> (except perhaps for a
start-up banner); code to set options and load shared objects should be
placed in a <code>.onLoad</code> function, or use made of the <code>useDynLib</code>
directive described next.
</p>
<p>User-level hooks are also available: see the help on function
<code>setHook</code>.
</p>
<p>These hooks are often used incorrectly.  People forget to export
<code>.Last.lib</code>.   Compiled code should be loaded in <code>.onLoad</code> (or
<em>via</em> a <code>useDynLb</code> directive: see below) and unloaded in
<code>.onUnload</code>.  Do remember that a package&rsquo;s namespace can be loaded
without the namespace being attached (e.g. by <code>pkgname::fun</code>) and
that a package can be detached and re-attached whilst its namespace
remains loaded.
</p>
<hr>
<a name="useDynLib"></a>
<div class="header">
<p>
Next: <a href="#An-example" accesskey="n" rel="next">An example</a>, Previous: <a href="#Load-hooks" accesskey="p" rel="prev">Load hooks</a>, Up: <a href="#Package-namespaces" accesskey="u" rel="up">Package namespaces</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="useDynLib-1"></a>
<h4 class="subsection">1.5.4 useDynLib</h4>

<p>A <samp>NAMESPACE</samp> file can contain one or more <code>useDynLib</code>
directives which allows shared objects that need to be
loaded.<a name="DOCF59" href="#FOOT59"><sup>59</sup></a>  The directive
</p>
<a name="index-useDynLib"></a>
<div class="example">
<pre class="example">useDynLib(foo)
</pre></div>

<p>registers the shared object <code>foo</code><a name="DOCF60" href="#FOOT60"><sup>60</sup></a> for loading with <code>library.dynam</code>.
Loading of registered object(s) occurs after the package code has been
loaded and before running the load hook function.  Packages that would
only need a load hook function to load a shared object can use the
<code>useDynLib</code> directive instead.
</p>
<p>The <code>useDynLib</code> directive also accepts the names of the native
routines that are to be used in R <em>via</em> the <code>.C</code>, <code>.Call</code>,
<code>.Fortran</code> and <code>.External</code> interface functions.  These are given as
additional arguments to the directive, for example,
</p>
<div class="example">
<pre class="example">useDynLib(foo, myRoutine, myOtherRoutine)
</pre></div>

<p>By specifying these names in the <code>useDynLib</code> directive, the native
symbols are resolved when the package is loaded and R variables
identifying these symbols are added to the package&rsquo;s namespace with
these names.  These can be used in the <code>.C</code>, <code>.Call</code>,
<code>.Fortran</code> and <code>.External</code> calls in place of the name of the
routine and the <code>PACKAGE</code> argument.  For instance, we can call the
routine <code>myRoutine</code> from R with the code
</p>
<div class="example">
<pre class="example"> .Call(myRoutine, x, y)
</pre></div>

<p>rather than
</p>
<div class="example">
<pre class="example"> .Call(&quot;myRoutine&quot;, x, y, PACKAGE = &quot;foo&quot;)
</pre></div>

<p>There are at least two benefits to this approach.  Firstly, the symbol
lookup is done just once for each symbol rather than each time the
routine is invoked.  Secondly, this removes any ambiguity in resolving
symbols that might be present in several compiled DLLs.  However, this
approach is nowadays deprecated in favour of supplying registration
information (see below).
</p>
<p>In some circumstances, there will already be an R variable in the
package with the same name as a native symbol. For example, we may have
an R function in the package named <code>myRoutine</code>.  In this case,
it is necessary to map the native symbol to a different R variable
name. This can be done in the <code>useDynLib</code> directive by using named
arguments. For instance, to map the native symbol name <code>myRoutine</code>
to the R variable <code>myRoutine_sym</code>, we would use
</p>
<div class="example">
<pre class="example">useDynLib(foo, myRoutine_sym = myRoutine, myOtherRoutine)
</pre></div>

<p>We could then call that routine from R using the command
</p>
<div class="example">
<pre class="example"> .Call(myRoutine_sym, x, y)
</pre></div>

<p>Symbols without explicit names are assigned to the R variable with
that name.
</p>
<p>In some cases, it may be preferable not to create R variables in the
package&rsquo;s namespace that identify the native routines.  It may be too
costly to compute these for many routines when the package is loaded
if many of these routines are not likely to be used.  In this case,
one can still perform the symbol resolution correctly using the DLL,
but do this each time the routine is called.  Given a reference to the
DLL as an R variable, say <code>dll</code>, we can call the routine
<code>myRoutine</code> using the expression
</p>
<div class="example">
<pre class="example"> .Call(dll$myRoutine, x, y)
</pre></div>

<p>The <code>$</code> operator resolves the routine with the given name in the
DLL using a call to <code>getNativeSymbol</code>.  This is the same
computation as above where we resolve the symbol when the package is
loaded. The only difference is that this is done each time in the case
of <code>dll$myRoutine</code>.
</p>
<p>In order to use this dynamic approach (e.g., <code>dll$myRoutine</code>), one
needs the reference to the DLL as an R variable in the package.  The
DLL can be assigned to a variable by using the <code>variable =
dllName</code> format used above for mapping symbols to R variables.  For
example, if we wanted to assign the DLL reference for the DLL
<code>foo</code> in the example above to the variable <code>myDLL</code>, we would
use the following directive in the <samp>NAMESPACE</samp> file:
</p>
<div class="example">
<pre class="example">myDLL = useDynLib(foo, myRoutine_sym = myRoutine, myOtherRoutine)
</pre></div>

<p>Then, the R variable <code>myDLL</code> is in the package&rsquo;s namespace and
available for calls such as <code>myDLL$dynRoutine</code> to access routines
that are not explicitly resolved at load time.
</p>
<p>If the package has registration information (see <a href="#Registering-native-routines">Registering native routines</a>), then we can use that directly rather than specifying the
list of symbols again in the <code>useDynLib</code> directive in the
<samp>NAMESPACE</samp> file.  Each routine in the registration information is
specified by giving a name by which the routine is to be specified along
with the address of the routine and any information about the number and
type of the parameters.  Using the <code>.registration</code> argument of
<code>useDynLib</code>, we can instruct the namespace mechanism to create
R variables for these symbols.  For example, suppose we have the
following registration information for a DLL named <code>myDLL</code>:
</p>
<div class="example">
<pre class="example">static R_NativePrimitiveArgType foo_t[] = {
    REALSXP, INTSXP, STRSXP, LGLSXP
};

static R_CMethodDef cMethods[] = {
   {&quot;foo&quot;, (DL_FUNC) &amp;foo, 4, foo_t},
   {&quot;bar_sym&quot;, (DL_FUNC) &amp;bar, 0},
   {NULL, NULL, 0}
};

static R_CallMethodDef callMethods[] = {
   {&quot;R_call_sym&quot;, (DL_FUNC) &amp;R_call, 4},
   {&quot;R_version_sym&quot;, (DL_FUNC) &amp;R_version, 0},
   {NULL, NULL, 0}
};
</pre></div>

<p>Then, the directive in the <samp>NAMESPACE</samp> file
</p>
<div class="example">
<pre class="example">useDynLib(myDLL, .registration = TRUE)
</pre></div>

<p>causes the DLL to be loaded and also for the R variables <code>foo</code>,
<code>bar_sym</code>, <code>R_call_sym</code> and <code>R_version_sym</code> to be
defined in the package&rsquo;s namespace.
</p>
<p>Note that the names for the R variables are taken from the entry in
the registration information and do not need to be the same as the name
of the native routine.  This allows the creator of the registration
information to map the native symbols to non-conflicting variable names
in R, e.g. <code>R_version</code> to <code>R_version_sym</code> for use in an
R function such as
</p>
<div class="example">
<pre class="example">R_version &lt;- function()
{
  .Call(R_version_sym)
}
</pre></div>

<p>Using argument <code>.fixes</code> allows an automatic prefix to be added to
the registered symbols, which can be useful when working with an
existing package.  For example, package <a href="https://CRAN.R-project.org/package=KernSmooth"><strong>KernSmooth</strong></a> has
</p>
<div class="example">
<pre class="example">useDynLib(KernSmooth, .registration = TRUE, .fixes = &quot;F_&quot;)
</pre></div>

<p>which makes the R variables corresponding to the FORTRAN symbols
<code>F_bkde</code> and so on, and so avoid clashes with R code in the
namespace.
</p>
<p><strong>NB</strong>: Using these arguments for a package which does not register
native symbols merely slows down the package loading (although at the
time of writing 90 <acronym>CRAN</acronym> packages did so).  Once symbols are
registered, check that the corresponding R variables are not
accidentally exported by a pattern in the <samp>NAMESPACE</samp> file.
</p>

<hr>
<a name="An-example"></a>
<div class="header">
<p>
Next: <a href="#Namespaces-with-S4-classes-and-methods" accesskey="n" rel="next">Namespaces with S4 classes and methods</a>, Previous: <a href="#useDynLib" accesskey="p" rel="prev">useDynLib</a>, Up: <a href="#Package-namespaces" accesskey="u" rel="up">Package namespaces</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="An-example-1"></a>
<h4 class="subsection">1.5.5 An example</h4>

<p>As an example consider two packages named <strong>foo</strong> and <strong>bar</strong>.  The
R code for package <strong>foo</strong> in file <samp>foo.R</samp> is
</p>
<blockquote>
<table summary="" class="cartouche" border="1"><tr><td>
<div class="example">
<pre class="example">x &lt;- 1
f &lt;- function(y) c(x,y)
foo &lt;- function(x) .Call(&quot;foo&quot;, x, PACKAGE=&quot;foo&quot;)
print.foo &lt;- function(x, ...) cat(&quot;&lt;a foo&gt;\n&quot;)
</pre></div>
</td></tr></table>
</blockquote>

<p>Some C code defines a C function compiled into DLL <code>foo</code> (with an
appropriate extension).  The <samp>NAMESPACE</samp> file for this package is
</p>
<blockquote>
<table summary="" class="cartouche" border="1"><tr><td>
<div class="example">
<pre class="example">useDynLib(foo)
export(f, foo)
S3method(print, foo)
</pre></div>
</td></tr></table>
</blockquote>

<p>The second package <strong>bar</strong> has code file <samp>bar.R</samp>
</p>
<blockquote>
<table summary="" class="cartouche" border="1"><tr><td>
<div class="example">
<pre class="example">c &lt;- function(...) sum(...)
g &lt;- function(y) f(c(y, 7))
h &lt;- function(y) y+9
</pre></div>
</td></tr></table>
</blockquote>

<p>and <samp>NAMESPACE</samp> file
</p>
<blockquote>
<table summary="" class="cartouche" border="1"><tr><td>
<div class="example">
<pre class="example">import(foo)
export(g, h)
</pre></div>
</td></tr></table>
</blockquote>

<p>Calling <code>library(bar)</code> loads <strong>bar</strong> and attaches its exports to
the search path.  Package <strong>foo</strong> is also loaded but not attached to
the search path.  A call to <code>g</code> produces
</p>
<div class="example">
<pre class="example">&gt; g(6)
[1]  1 13
</pre></div>

<p>This is consistent with the definitions of <code>c</code> in the two settings:
in <strong>bar</strong> the function <code>c</code> is defined to be equivalent to
<code>sum</code>, but in <strong>foo</strong> the variable <code>c</code> refers to the
standard function <code>c</code> in <strong>base</strong>.
</p>
<hr>
<a name="Namespaces-with-S4-classes-and-methods"></a>
<div class="header">
<p>
Previous: <a href="#An-example" accesskey="p" rel="prev">An example</a>, Up: <a href="#Package-namespaces" accesskey="u" rel="up">Package namespaces</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Namespaces-with-S4-classes-and-methods-1"></a>
<h4 class="subsection">1.5.6 Namespaces with S4 classes and methods</h4>

<p>Some additional steps are needed for packages which make use of formal
(S4-style) classes and methods (unless these are purely used
internally).  The package should have <code>Depends: methods</code> in its
<samp>DESCRIPTION</samp> file<a name="DOCF61" href="#FOOT61"><sup>61</sup></a> and <code>import(methods)</code> or
<code>importFrom(methods, ...)</code> plus any classes and methods which are
to be exported need to be declared in the <samp>NAMESPACE</samp> file.  For
example, the <strong>stats4</strong> package has
</p>
<a name="index-exportClasses"></a>
<a name="index-exportMethods"></a>

<div class="example">
<pre class="example">export(mle) # exporting methods implicitly exports the generic
importFrom(&quot;graphics&quot;, plot)
importFrom(&quot;stats&quot;, optim, qchisq)
## For these, we define methods or (AIC, BIC, nobs) an implicit generic:
importFrom(&quot;stats&quot;, AIC, BIC, coef, confint, logLik, nobs, profile,
	   update, vcov)
exportClasses(mle, profile.mle, summary.mle)
## All methods for imported generics:
exportMethods(coef, confint, logLik, plot, profile, summary,
	      show, update, vcov)
## implicit generics which do not have any methods here
export(AIC, BIC, nobs)
</pre></div>

<a name="index-exportPattern-1"></a>
<a name="index-exportClassPattern"></a>
<p>All S4 classes to be used outside the package need to be listed in an
<code>exportClasses</code> directive. Alternatively, they can be specified
using <code>exportClassPattern</code><a name="DOCF62" href="#FOOT62"><sup>62</sup></a> in the same style as
for <code>exportPattern</code>.  To export methods for generics from other
packages an <code>exportMethods</code> directive can be used.
</p>
<p>Note that exporting methods on a generic in the namespace will also
export the generic, and exporting a generic in the namespace will also
export its methods.  If the generic function is not local to this
package, either because it was imported as a generic function or because
the non-generic version has been made generic solely to add S4 methods
to it (as for functions such as <code>plot</code> in the example above), it
can be declared <em>via</em> either or both of <code>export</code> or
<code>exportMethods</code>, but the latter is clearer (and is used in the
<strong>stats4</strong> example above).  In particular, for primitive functions
there is no generic function, so <code>export</code> would export the
primitive, which makes no sense.  On the other hand, if the generic is
local to this package, it is more natural to export the function itself
using <code>export()</code>, and this <em>must</em> be done if an implicit
generic is created without setting any methods for it (as is the case
for <code>AIC</code> in <strong>stats4</strong>).
</p>
<p>A non-local generic function is only exported to ensure that calls to
the function will dispatch the methods from this package (and that is
not done or required when the methods are for primitive functions).  For
this reason, you do not need to document such implicitly created generic
functions, and <code>undoc</code> in package <strong>tools</strong> will not report them.
</p>
<p>If a package uses S4 classes and methods exported from another package,
but does not import the entire namespace of the other
package<a name="DOCF63" href="#FOOT63"><sup>63</sup></a>, it needs
to import the classes and methods explicitly, with directives
</p>
<a name="index-importClassesFrom"></a>
<a name="index-importMethodsFrom"></a>

<div class="example">
<pre class="example">importClassesFrom(package, ...)
importMethodsFrom(package, ...)
</pre></div>

<p>listing the classes and functions with methods respectively.  Suppose we
had two small packages <strong>A</strong> and <strong>B</strong> with <strong>B</strong> using <strong>A</strong>.
Then they could have <code>NAMESPACE</code> files
</p>
<blockquote>
<table summary="" class="cartouche" border="1"><tr><td>
<div class="example">
<pre class="example">export(f1, ng1)
exportMethods(&quot;[&quot;)
exportClasses(c1)
</pre></div>
</td></tr></table>
</blockquote>

<p>and
</p>
<blockquote>
<table summary="" class="cartouche" border="1"><tr><td>
<div class="example">
<pre class="example">importFrom(A, ng1)
importClassesFrom(A, c1)
importMethodsFrom(A, f1)
export(f4, f5)
exportMethods(f6, &quot;[&quot;)
exportClasses(c1, c2)
</pre></div>
</td></tr></table>
</blockquote>

<p>respectively.
</p>
<p>Note that <code>importMethodsFrom</code> will also import any generics defined
in the namespace on those methods.
</p>
<p>It is important if you export S4 methods that the corresponding generics
are available.  You may for example need to import <code>plot</code> from
<strong>graphics</strong> to make visible a function to be converted into its
implicit generic.  But it is better practice to make use of the generics
exported by <strong>stats4</strong> as this enables multiple packages to
unambiguously set methods on those generics.
</p>
<hr>
<a name="Writing-portable-packages"></a>
<div class="header">
<p>
Next: <a href="#Diagnostic-messages" accesskey="n" rel="next">Diagnostic messages</a>, Previous: <a href="#Package-namespaces" accesskey="p" rel="prev">Package namespaces</a>, Up: <a href="#Creating-R-packages" accesskey="u" rel="up">Creating R packages</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Writing-portable-packages-1"></a>
<h3 class="section">1.6 Writing portable packages</h3>

<p>This section contains advice on writing packages to be used on multiple
platforms or for distribution (for example to be submitted to a package
repository such as <acronym>CRAN</acronym>).
</p>
<table summary="" class="menu" border="0" cellspacing="0">
<tr><td align="left" valign="top">&bull; <a href="#PDF-size" accesskey="1">PDF size</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Check-timing" accesskey="2">Check timing</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Encoding-issues" accesskey="3">Encoding issues</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Portable-C-and-C_002b_002b-code" accesskey="4">Portable C and C++ code</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Binary-distribution" accesskey="5">Binary distribution</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
</table>

<p>Portable packages should have simple file names: use only alphanumeric
<acronym>ASCII</acronym> characters and period (<code>.</code>), and avoid those names
not allowed under Windows which are mentioned above.
</p>
<p>Many of the graphics devices are platform-specific: even <code>X11()</code>
(aka <code>x11()</code>) which although emulated on Windows may not be
available on a Unix-alike (and is not the preferred screen device on OS
X).  It is rarely necessary for package code or examples to open a new
device, but if essential,<a name="DOCF64" href="#FOOT64"><sup>64</sup></a> use <code>dev.new()</code>.
</p>
<p>Use <code>R CMD build</code> to make the release <samp>.tar.gz</samp> file.
</p>
<p><code>R CMD check</code> provides a basic set of checks, but often further
problems emerge when people try to install and use packages submitted to
<acronym>CRAN</acronym> &ndash; many of these involve compiled code.  Here are some
further checks that you can do to make your package more portable.
</p>
<ul>
<li> If your package has a <samp>configure</samp> script, provide a
<samp>configure.win</samp> script to be used on Windows (an empty file if no
actions are needed).

</li><li> If your package has a <samp>Makevars</samp> or <samp>Makefile</samp> file, make sure
that you use only portable make features.  Such files should be
LF-terminated<a name="DOCF65" href="#FOOT65"><sup>65</sup></a> (including the final
line of the file) and not make use of GNU extensions.  (The POSIX
specification is available at
<a href="http://pubs.opengroup.org/onlinepubs/9699919799/utilities/make.html">http://pubs.opengroup.org/onlinepubs/9699919799/utilities/make.html</a>;
anything not documented there should be regarded as an extension to be
avoided.)  Commonly misused GNU extensions are conditional inclusions
(<code>ifeq</code> and the like), <code>${shell ...}</code>, <code>${wildcard
...}</code> and similar, and the use of <code>+=</code><a name="DOCF66" href="#FOOT66"><sup>66</sup></a> and <code>:=</code>.  Also, the
use of <code>$&lt;</code> other than in implicit rules is a GNU extension, as is
the <code>$^</code> macro.  Unfortunately makefiles which use GNU extensions
often run on other platforms but do not have the intended results.

<p>The use of <code>${shell ...}</code> can be avoided by using backticks, e.g.
</p>
<div class="example">
<pre class="example">PKG_CPPFLAGS = `gsl-config --cflags`
</pre></div>

<p>which works in all versions of <code>make</code> known<a name="DOCF67" href="#FOOT67"><sup>67</sup></a> to be used with
R.
</p>
<p>If you really must require GNU make, declare it in the <samp>DESCRIPTION</samp>
file by
</p>
<div class="example">
<pre class="example">SystemRequirements: GNU make
</pre></div>

<p>and ensure that you use the value of environment variable <code>MAKE</code>
(and not just <code>make</code>) in your scripts.  (On some platforms GNU
make is available under a name such as <code>gmake</code>, and there
<code>SystemRequirements</code> is used to set <code>MAKE</code>.)
</p>
<p>If you only need GNU make for parts of the package which are rarely
needed (for example to create bibliography files under
<samp>vignettes</samp>), use a file called <samp>GNUmakefile</samp> rather than
<samp>Makefile</samp> as GNU make (only) will use the former.
</p>
<p>Since the only viable make for Windows is GNU make, it is permissible to
use GNU extensions in files <samp>Makevars.win</samp> or <samp>Makefile.win</samp>.
</p>
</li><li> Bash extensions also need to be avoided in shell scripts, including
expressions in Makefiles (which are passed to the shell for processing).
Some R platforms use strict<a name="DOCF68" href="#FOOT68"><sup>68</sup></a>
Bourne shells: the R toolset on Windows and some Unix-alike OSes use
<code>ash</code> (<a href="https://en.wikipedia.org/wiki/Almquist_shell">https://en.wikipedia.org/wiki/Almquist_shell</a>),
a rather minimal shell with few builtins.  Beware of assuming that all
the POSIX command-line utilities are available, especially on Windows
where only a minimal set is provided for use with R.
(See <a href="http://cran.r-project.org/doc/manuals/R-admin.html#The-command-line-tools">The command line tools</a> in <cite>R Installation and Administration</cite>.)
One particular issue is the use of <code>echo</code>, for which two
behaviours are allowed
(<a href="http://pubs.opengroup.org/onlinepubs/9699919799/utilities/echo.html">http://pubs.opengroup.org/onlinepubs/9699919799/utilities/echo.html</a>)
and both occur as defaults on R platforms: portable applications
should not use <samp>-n</samp> (as the first argument) nor escape
sequences. Another common issue is the construction
<div class="example">
<pre class="example">export FOO=value
</pre></div>
<p>which is bash-specific (first set the variable then export it by name).
</p>
</li><li> Make use of the abilities of your compilers to check the
standards-conformance of your code.  For example, <code>gcc</code> and
<code>gfortran</code><a name="DOCF69" href="#FOOT69"><sup>69</sup></a>  can be used with options <samp>-Wall -pedantic</samp> to alert
you to potential problems.  This is particularly important for C++,
where <code>g++ -Wall -pedantic</code> will alert you to the use of some of
the GNU extensions which fail to compile on most other C++ compilers. If
R was not configured accordingly, one can achieve this <em>via</em>
personal <samp>Makevars</samp> files.
See <a href="http://cran.r-project.org/doc/manuals/R-admin.html#Customizing-package-compilation">Customizing package compilation</a> in <cite>R Installation and Administration</cite>,

<p>Portable C++ code needs to follow the 1998 standard (and not use
features from C99), or to specify a C++11 compiler (see <a href="#Using-C_002b_002b11-code">Using C++11 code</a>) where available (which is not the case on all R platforms).
</p>
<p>If you use FORTRAN 77, <code>ftnchek</code>
(<a href="http://www.dsm.fordham.edu/~ftnchek/">http://www.dsm.fordham.edu/~ftnchek/</a>) provides thorough testing
of conformance to the standard.
</p>
<p>If using Fortran 9x with the GNU compiler, use the flags
<samp>-std=f95 -Wall -pedantic</samp> which reject most GNU extensions and
features from later standards.
</p>
<p>R has tested that <code>DOUBLE COMPLEX</code> works (although an extension
to the Fortran standards) and so is preferred to <code>COMPLEX*16</code>.
(Fortran 9x code can use something like
<code>COMPLEX(KIND=KIND(0.0D0))</code><a name="DOCF70" href="#FOOT70"><sup>70</sup></a>.)
</p>

<p>Not all common R platforms conform to the expected standards, e.g.
C99 for C code.  One common area of problems is the <code>*printf</code>
functions where Windows does not support <code>%lld</code>, <code>%Lf</code> and
similar formats (and has its own formats such as <code>%I64d</code> for 64-bit
integers).  It is very rare to need to output such types, and 64-bit
integers can usually be converted to doubles for output.
</p>
</li><li> <code>R CMD check</code> performs some checks for non-portable
compiler/linker flags in <samp>src/Makevars</samp>.  However, it cannot check
the meaning of such flags, and some are commonly accepted but with
compiler-specific meanings.  There are other non-portable flags which
are not checked, nor are <samp>src/Makefile</samp> files and makefiles in
sub-directories.  As a comment in the code says

<blockquote>
<p>It is hard to think of anything apart from <samp>-I*</samp> and <samp>-D*</samp>
that is safe for general use &hellip;
</p></blockquote>

<p>although <samp>-pthread</samp> is pretty close to portable.  (Option
<samp>-U</samp> is portable but little use on the command line as it will
only cancel built-in defines (not portable) and those defined earlier on
the command line (R does not use any).)
</p>
</li><li> Do be very careful with passing arguments between R, C and
<acronym>FORTRAN</acronym> code.  In particular, <code>long</code> in C will be 32-bit
on some R platforms (including 64-bit Windows), but 64-bit on most
modern Unix and Linux platforms.  It is rather unlikely that the use of
<code>long</code> in C code has been thought through: if you need a longer
type than <code>int</code> you should use a configure test for a C99/C++11
type such as <code>int_fast64_t</code> (and failing that, <code>long long</code>
<a name="DOCF71" href="#FOOT71"><sup>71</sup></a>) and
typedef your own type to be <code>long</code> or <code>long long</code>, or use
another suitable type (such as <code>size_t</code>).

<p>It is not safe to assume that <code>long</code> and pointer types are the same
size, and they are not on 64-bit Windows.  If you need to convert
pointers to and from integers use the C99/C++11 integer types
<code>intptr_t</code> and <code>uintptr_t</code> (which are defined in the header
<code>&lt;stdint.h&gt;</code> and are not required to be implemented by the C99
standard but are used in C code by R itself).
</p>
<p>Note that <code>integer</code> in <acronym>FORTRAN</acronym> corresponds to <code>int</code>
in C on all R platforms.
</p>
</li><li> Under no circumstances should your compiled code ever call <code>abort</code>
or <code>exit</code><a name="DOCF72" href="#FOOT72"><sup>72</sup></a>: these terminate the user&rsquo;s R process, quite possibly
including all his unsaved work.  One usage that could call <code>abort</code>
is the <code>assert</code> macro in C or C++ functions, which should never be
active in production code.  The normal way to ensure that is to define
the macro <code>NDEBUG</code>, and <code>R CMD INSTALL</code> does so as part of
the compilation flags.  If you wish to use <code>assert</code> during
development. you can include <code>-UNDEBUG</code> in <code>PKG_CPPFLAGS</code>.
Note that your own <samp>src/Makefile</samp> or makefiles in sub-directories
may also need to define <code>NDEBUG</code>.

<p>This applies not only to your own code but to any external software you
compile in or link to.
</p>
</li><li> Compiled code should not write to <samp>stdout</samp> or <samp>stderr</samp> and C++
and Fortran I/O should not be used.  As with the previous item such
calls may come from external software and may never be called, but
package authors are often mistaken about that.

</li><li> Compiled code should not call the system random number generators such
as <code>rand</code>, <code>drand48</code> and <code>random</code><a name="DOCF73" href="#FOOT73"><sup>73</sup></a>, but rather use the
interfaces to R&rsquo;s RNGs described in <a href="#Random-numbers">Random numbers</a>.  In
particular, if more than one package initializes the system RNG (e.g.
<em>via</em> <code>srand</code>), they will interfere with each other.

<p>Nor should the C++11 random number library be used.
</p>
</li><li> Errors in memory allocation and reading/writing outside arrays are very
common causes of crashes (e.g., segfaults) on some machines.
See <a href="#Checking-memory-access">Checking memory access</a> for tools which can be used to look for this.

</li><li> Many platforms will allow unsatisfied entry points in compiled code, but
will crash the application (here R) if they are ever used.  Some
(notably Windows) will not.  Looking at the output of

<div class="example">
<pre class="example">nm -pg mypkg.so
</pre></div>

<p>and checking if any of the symbols marked <code>U</code> is unexpected is a
good way to avoid this.
</p>
</li><li> Linkers have a lot of freedom in how to resolve entry points in
dynamically-loaded code, so the results may differ by platform.  One
area that has caused grief is packages including copies of standard
system software such as <code>libz</code> (especially those already linked
into R).  In the case in point, entry point <code>gzgets</code> was
sometimes resolved against the old version compiled into the package,
sometimes against the copy compiled into R and sometimes against the
system dynamic library.  The only safe solution is to rename the entry
points in the copy in the package.  We have even seen problems with
entry point name <code>myprintf</code>, which is a system entry
point<a name="DOCF74" href="#FOOT74"><sup>74</sup></a> on some Linux systems.

</li><li> Conflicts between symbols in DLLs are handled in very platform-specific
ways.  Good ways to avoid trouble are to make as many symbols as
possible static (check with <code>nm -pg</code>), and to use names which are
clearly tied to your package (which also helps users if anything does go
wrong).  Note that symbol names starting with <code>R_</code> are regarded as
part of R&rsquo;s namespace and should not be used in packages.

</li><li> It is good practice for DLLs to register their symbols
(see <a href="#Registering-native-routines">Registering native routines</a>), restrict visibility
(see <a href="#Controlling-visibility">Controlling visibility</a>) and not allow symbol search
(see <a href="#Registering-native-routines">Registering native routines</a>).  It should be possible for a DLL
to have only one visible symbol, <code>R_init_<var>pkgname</var></code>, on
suitable platforms<a name="DOCF75" href="#FOOT75"><sup>75</sup></a>,
which would completely avoid symbol conflicts.

</li><li> It is not portable to call compiled code in R or other packages
<em>via</em> <code>.Internal</code>, <code>.C</code>, <code>.Fortran</code>, <code>.Call</code> or
<code>.External</code>, since such interfaces are subject to change without
notice and will probably result in your code terminating the R
process.

</li><li> Do not use (hard or symbolic) file links in your package sources.
Where possible <code>R CMD build</code> will replace them by copies.

</li><li> If you do not yourself have a Windows system, consider submitting your
source package to WinBuilder (<a href="https://win-builder.r-project.org/">https://win-builder.r-project.org/</a>)
before distribution.

</li><li> It is bad practice for package code to alter the search path using
<code>library</code>, <code>require</code> or <code>attach</code> and this often does not
work as intended.  For alternatives, see <a href="#Suggested-packages">Suggested packages</a> and
<code>with</code>.

</li><li> Examples can be run interactively <em>via</em> <code>example</code> as well as
in batch mode when checking.  So they should behave appropriately in
both scenarios, conditioning by <code>interactive()</code> the parts which
need an operator or observer.  For instance, progress
bars<a name="DOCF76" href="#FOOT76"><sup>76</sup></a> are only appropriate in
interactive use, as is displaying help pages or calling <code>View()</code>
(see below).

</li><li> Be careful with the order of entries in macros such as <code>PKG_LIBS</code>.
Some linkers will re-order the entries, and behaviour can differ between
dynamic and static libraries.  Generally <samp>-L</samp> options should
precede<a name="DOCF77" href="#FOOT77"><sup>77</sup></a> the libraries (typically
specified by <samp>-l</samp> options) to be found from those directories,
and libraries are searched once in the order they are specified.  Not
all linkers allow a space after <samp>-L</samp> .

</li><li> Great care is needed with the use of <code>LinkingTo</code>.  This puts one or
more directories on the include search path ahead of system headers but
currently<a name="DOCF78" href="#FOOT78"><sup>78</sup></a> after those
specified in the <code>CPPFLAGS</code> macro of the R build (which normally
includes <code>-I/usr/local/include</code>, but most platforms ignore that and
include it with the system headers).

<p>Any confusion would be avoided by having <code>LinkingTo</code> headers in a
directory named after the package.  In any case, name conflicts of
headers and directories under package <samp>include</samp> directories should
be avoided, both between packages and between a package and system and
third-party software.
</p>
</li><li> The <code>ar</code> utility is often used in makefiles to make static
libraries.  Its modifier <code>u</code> is defined by POSIX but is disabled in
GNU <code>ar</code> on some recent Linux distributions which use
&lsquo;deterministic mode&rsquo;.  The safest way to make a static library is to first
remove any existing file of that name then use <code>ar -cr</code> and then
<code>ranlib</code> if needed (which is system-dependent: on most
systems<a name="DOCF79" href="#FOOT79"><sup>79</sup></a> <code>ar</code> always
maintains a symbol table).  The POSIX standard says options should be
preceded by a hyphen (as in <samp>-cr</samp>), although most OSes accept
them without.
Note that on some systems <code>ar -cr</code> must have at least one file
specified.

</li><li> Some people have a need to set a locale.  Locale names are not portable,
and e.g. &lsquo;<samp>fr_FR.utf8</samp>&rsquo; is commonly used on Linux but not accepted on
either Solaris or macOS.  &lsquo;<samp>fr_FR.UTF-8</samp>&rsquo; is more portable, being
accepted on recent Linux, AIX, FreeBSD, macOS and Solaris (at least).
However, some Linux distributions micro-package, so locales defined by
<strong>glibc</strong> (including these examples) may not be installed.

</li><li> Avoid spaces in file names, not least as they can cause difficulties for
external tools.  A recent example was a package with a <a href="https://CRAN.R-project.org/package=knitr"><strong>knitr</strong></a>
vignette that used spaces in plot names: this caused some versions of
<code>pandoc</code> to fail with a baffling error message.

<p>Non-ASCII filenames can also cause problems (particularly in non-UTF-8
locales).
</p>
</li><li> Make sure that any version requirement for Java code is both declared in
the &lsquo;<samp>SystemRequirements</samp>&rsquo; field and tested at runtime (not least as
the Java installation when the package is installed might not be the
same as when the package is run and will not be for binary packages).
Java 8 (aka 1.8) is available for fewer platforms than Java 7.  A
suitable test for packages using <a href="https://CRAN.R-project.org/package=rJava"><strong>rJava</strong></a> would be
<div class="example">
<pre class="example">.jinit()
jv &lt;- .jcall(&quot;java/lang/System&quot;, &quot;S&quot;, &quot;getProperty&quot;, &quot;java.runtime.version&quot;)
jvn &lt;- as.numeric(paste0(strsplit(jv, &quot;[.]&quot;)[[1L]][1:2], collapse = &quot;.&quot;))
if(jvn &lt; 1.8) stop(&quot;Java 8 is needed for this package but not available&quot;)
</pre></div>

<p>Some packages have stated a requirement on a particular JDK, but a
package should only be requiring a JRE unless providing its own Java
interface.
</p>
</li><li> A package with a hard-to-satisfy system requirement is by definition not
portable, annoyingly so if this is not declared in the
&lsquo;<samp>SystemRequirements</samp>&rsquo; field.  The most common example is the use of
<code>pandoc</code>, which is only available for a very limited range of
platforms (and has onerous requirements to install from source) and has
capabilities<a name="DOCF80" href="#FOOT80"><sup>80</sup></a> that vary by build but are not documented.

<p>An external command can be an optional requirement for an imported
package but needed for examples or tests in the package itself.  Such
usage should always be conditional on a test for existence (perhaps
using <code>Sys.which</code>), as well as declared in the
&lsquo;<samp>SystemRequirements</samp>&rsquo; field.
</p>
</li><li> Be sure to use portable encoding names: none of <code>utf8</code>, <code>mac</code>
and <code>macroman</code> are.  See the help for <code>file</code> for more details.
</li></ul>

<p>Do be careful in what your tests (and examples) actually test.  Bad
practice seen in distributed packages include:
</p>
<ul>
<li> It is not reasonable to test the time taken by a command: you cannot
know how fast or how heavily loaded an R platform might be.  At best
you can test a ratio of times, and even that is fraught with
difficulties.

</li><li> Do not test the exact format of R error messages: They change, and
they can be translated.

<p>Worse, packages have tested the exact format of system error messages,
which are platform-dependent and perhaps locale-dependent.
</p>
</li><li> If you use functions such as <code>View</code>, remember that in testing there
is no one to look at the output.  It is better to use something like one of
<div class="example">
<pre class="example">if(interactive()) View(obj) else print(head(obj))
if(interactive()) View(obj) else str(obj)
</pre></div>

</li><li> Only test the accuracy of results if you have done a formal error
analysis.  Things such as checking that probabilities numerically sum to
one are silly: numerical tests should always have a tolerance.  That the
tests on your platform achieve a particular tolerance says little about
other platforms.  R is configured by default to make use of long
doubles where available, but they may not be available or be too slow
for routine use.  Most R platforms use &lsquo;<samp>ix86</samp>&rsquo; or
&lsquo;<samp>x86_64</samp>&rsquo; CPUs: these use extended precision registers on some
but not all of their FPU instructions.  Thus the achieved precision can
depend on the compiler version and optimization flags&mdash;our experience
is that 32-bit builds tend to be less precise than 64-bit ones.  But not
all platforms use those CPUs, and not all<a name="DOCF81" href="#FOOT81"><sup>81</sup></a> which use them configure them to
allow the use of extended precision.  In particular, ARM CPUs do not
(currently) have extended precision nor long doubles, and long double
was 64-bit on HP/PA Linux.

<p>If you must try to establish a tolerance empirically, configure and
build R with <samp>--disable-long-double</samp> and use appropriate
compiler flags (such as <samp>-ffloat-store</samp> and
<samp>-fexcess-precision=standard</samp> for <code>gcc</code>, depending on the
CPU type<a name="DOCF82" href="#FOOT82"><sup>82</sup></a>) to
mitigate the effects of extended-precision calculations.
</p>
<p>Tests which involve random inputs or non-deterministic algorithms should
normally set a seed or be tested for many seeds.
</p>
</li></ul>

<hr>
<a name="PDF-size"></a>
<div class="header">
<p>
Next: <a href="#Check-timing" accesskey="n" rel="next">Check timing</a>, Previous: <a href="#Writing-portable-packages" accesskey="p" rel="prev">Writing portable packages</a>, Up: <a href="#Writing-portable-packages" accesskey="u" rel="up">Writing portable packages</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="PDF-size-1"></a>
<h4 class="subsection">1.6.1 PDF size</h4>

<p>There are a several tools available to reduce the size of PDF files:
often the size can be reduced substantially with no or minimal loss in
quality. Not only do large files take up space: they can stress the PDF
viewer and take many minutes to print (if they can be printed at all).
</p>
<p><code>qpdf</code> (<a href="http://qpdf.sourceforge.net/">http://qpdf.sourceforge.net/</a>) can compress
losslessly.  It is fairly readily available (e.g. it has binaries for
Windows and packages in Debian/Ubuntu/Fedora, and is installed as part
of the <acronym>CRAN</acronym> macOS distribution of R).  <code>R CMD build</code>
has an option to run <code>qpdf</code> over PDF files under <samp>inst/doc</samp>
and replace them if at least 10Kb and 10% is saved.  The full path to
the <code>qpdf</code> command can be supplied as environment variable
<code>R_QPDF</code> (and is on the <acronym>CRAN</acronym> binary of R for macOS).  It seems
MiKTeX does not use PDF object compression and so <code>qpdf</code> can
reduce considerably the files it outputs: MiKTeX can be overridden by
code in the preamble of an Sweave or LaTeX file &mdash; see how this is
done for the R reference manual at
<a href="https://svn.r-project.org/R/trunk/doc/manual/refman.top">https://svn.r-project.org/R/trunk/doc/manual/refman.top</a>.
</p>
<p>Other tools can reduce the size of PDFs containing bitmap images at
excessively high resolution.  These are often best re-generated (for
example <code>Sweave</code> defaults to 300 ppi, and 100&ndash;150 is more
appropriate for a package manual).  These tools include Adobe Acrobat
(not Reader), Apple&rsquo;s Preview<a name="DOCF83" href="#FOOT83"><sup>83</sup></a>  and Ghostscript (which
converts PDF to PDF by
</p>
<div class="example">
<pre class="example">ps2pdf <var>options</var> -dAutoRotatePages=/None <var>in</var>.pdf <var>out</var>.pdf
</pre></div>

<p>and suitable options might be
</p>
<div class="example">
<pre class="example">-dPDFSETTINGS=/ebook
-dPDFSETTINGS=/screen
</pre></div>

<p>; see <a href="http://www.ghostscript.com/doc/current/Ps2pdf.htm">http://www.ghostscript.com/doc/current/Ps2pdf.htm</a> for
more such and consider all the options for image downsampling).  There
have been examples in <acronym>CRAN</acronym> packages for which Ghostscript 9.06
and later produced much better reductions than 9.05 or earlier.
</p>
<p>We come across occasionally large PDF files containing excessively
complicated figures using PDF vector graphics: such figures are often
best redesigned or failing that, output as PNG files.
</p>
<p>Option <samp>--compact-vignettes</samp> to <code>R CMD build</code> defaults to
value &lsquo;<samp>qpdf</samp>&rsquo;: use &lsquo;<samp>both</samp>&rsquo; to try harder to reduce the size,
provided you have Ghostscript available (see the help for
<code>tools::compactPDF</code>).
</p>
<hr>
<a name="Check-timing"></a>
<div class="header">
<p>
Next: <a href="#Encoding-issues" accesskey="n" rel="next">Encoding issues</a>, Previous: <a href="#PDF-size" accesskey="p" rel="prev">PDF size</a>, Up: <a href="#Writing-portable-packages" accesskey="u" rel="up">Writing portable packages</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Check-timing-1"></a>
<h4 class="subsection">1.6.2 Check timing</h4>

<p>There are several ways to find out where time is being spent in the
check process.  Start by setting the environment variable
<code>_R_CHECK_TIMINGS_</code> to &lsquo;<samp>0</samp>&rsquo;.  This will report the total CPU
times (not Windows) and elapsed times for installation and running
examples, tests and vignettes, under each sub-architecture if
appropriate.  For tests and vignettes, it reports the time for each as
well as the total.
</p>
<p>Setting <code>_R_CHECK_TIMINGS_</code> to a positive value sets a threshold (in
seconds elapsed time) for reporting timings.
</p>
<p>If you need to look in more detail at the timings for examples, use
option <samp>--timings</samp> to <code>R CMD check</code> (this is set by
<samp>--as-cran</samp>).  This adds a summary to the check output for all
the examples with CPU or elapsed time of more than 5 seconds.  It
produces a file <samp><var>mypkg</var>.Rcheck/<var>mypkg</var>-Ex.timings</samp>
containing timings for each help file: it is a tab-delimited file which
can be read into R for further analysis.
</p>
<p>Timings for the tests and vignette runs are given at the bottom of the
corresponding log file: note that log files for successful vignette runs
are only retained if environment variable
<code>_R_CHECK_ALWAYS_LOG_VIGNETTE_OUTPUT_</code> is set to a true value.
</p>

<hr>
<a name="Encoding-issues"></a>
<div class="header">
<p>
Next: <a href="#Portable-C-and-C_002b_002b-code" accesskey="n" rel="next">Portable C and C++ code</a>, Previous: <a href="#Check-timing" accesskey="p" rel="prev">Check timing</a>, Up: <a href="#Writing-portable-packages" accesskey="u" rel="up">Writing portable packages</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Encoding-issues-1"></a>
<h4 class="subsection">1.6.3 Encoding issues</h4>

<p>Care is needed if your package contains non-<acronym>ASCII</acronym> text, and in
particular if it is intended to be used in more than one locale.  It is
possible to mark the encoding used in the <samp>DESCRIPTION</samp> file and in
<samp>.Rd</samp> files, as discussed elsewhere in this manual.
</p>
<p>First, consider carefully if you really need non-<acronym>ASCII</acronym> text.
Many users of R will only be able to view correctly text in their
native language group (e.g. Western European, Eastern European,
Simplified Chinese) and <acronym>ASCII</acronym>.<a name="DOCF84" href="#FOOT84"><sup>84</sup></a>.  Other characters may not be rendered at all,
rendered incorrectly, or cause your R code to give an error.  For
<samp>.Rd</samp> documentation, marking the encoding and including
<acronym>ASCII</acronym> transliterations is likely to do a reasonable job.  The
set of characters which is commonly supported is wider than it used to
be around 2000, but non-Latin alphabets (Greek, Russian, Georgian,
&hellip;) are still often problematic and those with double-width
characters (Chinese, Japanese, Korean) often need specialist fonts to
render correctly.
</p>
<p>Several <acronym>CRAN</acronym> packages have messages in their R code in French (and a
few in German).  A better way to tackle this is to use the
internationalization facilities discussed elsewhere in this manual.
</p>
<p>Function <code>showNonASCIIfile</code> in package <strong>tools</strong> can help in
finding non-<acronym>ASCII</acronym> bytes in files.
</p>
<p>There is a portable way to have arbitrary text in character strings
(only) in your R code, which is to supply them in Unicode as
<code>\uxxxx</code> escapes.  If there are any characters not in the current
encoding the parser will encode the character string as UTF-8 and mark
it as such.  This applies also to character strings in datasets: they
can be prepared using <code>\uxxxx</code> escapes or encoded in UTF-8 in a
UTF-8 locale, or even converted to UTF-8 via &lsquo;<samp>iconv()</samp>&rsquo;.  If you do
this, make sure you have &lsquo;<samp>R (&gt;= 2.10)</samp>&rsquo; (or later) in the
&lsquo;<samp>Depends</samp>&rsquo; field of the <samp>DESCRIPTION</samp> file.
</p>
<p>R sessions running in non-UTF-8 locales will if possible re-encode
such strings for display (and this is done by <code>RGui</code> on Windows,
for example).  Suitable fonts will need to be selected or made
available<a name="DOCF85" href="#FOOT85"><sup>85</sup></a>  both for the console/terminal and graphics devices such as
&lsquo;<samp>X11()</samp>&rsquo; and &lsquo;<samp>windows()</samp>&rsquo;.  Using &lsquo;<samp>postscript</samp>&rsquo; or
&lsquo;<samp>pdf</samp>&rsquo; will choose a default 8-bit encoding depending on the
language of the UTF-8 locale, and your users would need to be told how
to select the &lsquo;<samp>encoding</samp>&rsquo; argument.
</p>
<p>If you want to run <code>R CMD check</code> on a Unix-alike over a package
that sets a package encoding in its <samp>DESCRIPTION</samp> file <em>and do
not use a UTF-8 locale</em> you may need to specify a suitable locale
<em>via</em> environment variable <code>R_ENCODING_LOCALES</code>.  The default
is equivalent to the value
</p>
<div class="example">
<pre class="example">&quot;latin1=en_US:latin2=pl_PL:UTF-8=en_US.UTF-8:latin9=fr_FR.iso885915@euro&quot;
</pre></div>

<p>(which is appropriate for a system based on <code>glibc</code>: macOS requires
<code>latin9=fr_FR.ISO8859-15</code>) except that if the current locale is
UTF-8 then the package code is translated to UTF-8 for syntax checking,
so it is strongly recommended to check in a UTF-8 locale.
</p>
<hr>
<a name="Portable-C-and-C_002b_002b-code"></a>
<div class="header">
<p>
Next: <a href="#Binary-distribution" accesskey="n" rel="next">Binary distribution</a>, Previous: <a href="#Encoding-issues" accesskey="p" rel="prev">Encoding issues</a>, Up: <a href="#Writing-portable-packages" accesskey="u" rel="up">Writing portable packages</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Portable-C-and-C_002b_002b-code-1"></a>
<h4 class="subsection">1.6.4 Portable C and C++ code</h4>

<p>Writing portable C and C++ code is mainly a matter of observing the
standards (C99, C++98 or where declared C++11) and testing that
extensions (such as POSIX functions) are supported.
</p>
<p>Note that the &lsquo;TR1&rsquo; C++ extensions are not part of any of these
standards and the <code>&lt;tr1/<var>name</var>&gt;</code> headers are not supplied by some of
the compilers used for R, including on macOS.  (Use the C++11
versions instead.)
</p>
<p>Note too that the POSIX standards only require recently-defined
functions to be declared if certain macros are defined with large enough
values, and on some compiler/OS combinations<a name="DOCF86" href="#FOOT86"><sup>86</sup></a> they are not declared otherwise.  So you may
need to include something like one of <a name="DOCF87" href="#FOOT87"><sup>87</sup></a>
</p><div class="example">
<pre class="example">#define _XOPEN_SOURCE 500
</pre></div>
<p>or
</p><div class="example">
<pre class="example">#ifdef __GLIBC__
# define _POSIX_C_SOURCE 200809L
#endif
</pre></div>
<p>before <em>any</em> headers.  (<code>strdup</code> and <code>strncasecmp</code> are
two such functions.)
</p>
<p>However, some common errors are worth pointing out here.  It can be
helpful to look up functions at
<a href="http://www.cplusplus.com/reference/">http://www.cplusplus.com/reference/</a> or
<a href="http://en.cppreference.com/w/">http://en.cppreference.com/w/</a> and compare what is defined in the
various standards.
</p>
<p>Both the compiler and OS (<em>via</em> system header files, which may
differ by architecture even for nominally the same OS) affect the
compilability of C/C++ code.  Compilers from the GCC, <code>clang</code>,
Intel and Oracle Studio suites are routinely used with R, and both
<code>clang</code> and Oracle have more than one implementation of C++
headers and library.  The range of possibilities makes comprehensive
empirical checking impossible, and regrettably compilers are patchy at
best on warning about non-standard code.
</p>
<ul>
<li> Mathematical functions such as <code>sqrt</code> are defined in C++ for
floating-point arguments.  It is legitimate in C++ to overload these
with versions for types <code>float</code>, <code>double</code>, <code>long double</code>
and possibly more.  This means that calling <code>sqrt</code> on an integer
type may have &lsquo;overloading ambiguity&rsquo; as it could be promoted to any of
the supported floating-point types: this is commonly seen on Solaris,
but for <code>pow</code> also seen on macOS.  (C++98 has an overload for
<code>std::pow(&lt;double&gt;, &lt;int&gt;)</code>, but this may not be visible from the
main namespace.  C++11 requires additional overloads for integer types,
and ambiguous overloads are more common in C++11 (and later) compiler
modes.)

<p>A not-uncommonly-seen problem is to mistakenly call <code>floor(x/y)</code> or
<code>ceil(x/y)</code> for <code>int</code> arguments <code>x</code> and <code>y</code>.  Since
<code>x/y</code> does integer division, the result is an <code>int</code> and
&lsquo;overloading ambiguity&rsquo; may be reported.  Some people have (pointlessly)
called <code>floor</code> and <code>ceil</code> on integer arguments, which may have
an &lsquo;overloading ambiguity&rsquo;.
</p>
<p>A surprising common misuse is things like <code>pow(10, -3)</code>: this
should be the constant <code>1e-3</code>.
</p>
</li><li> Function <code>fabs</code> is defined only for floating-point types, except in
C++11 which has overloads for <code>std::fabs</code> in <samp>&lt;cmath&gt;</samp> for
integer types.  Function <code>abs</code> is defined in C99&rsquo;s
<samp>&lt;stdlib.h&gt;</samp> for <code>int</code> and in C++98&rsquo;s <samp>&lt;cstdlib&gt;</samp> for
integer types, overloaded in <samp>&lt;cmath&gt;</samp> for floating-point types.
C++11 has additional overloads for <code>std::abs</code> in <samp>&lt;cmath&gt;</samp> for
integer types.  The effect of calling <code>abs</code> with a floating-point
type is implementation-specific: it may truncate to an integer.

</li><li> Functions/macros such as <code>isnan</code>, <code>isinf</code> and <code>isfinite</code>
are not required by C++98: where compilers support them they may be only
in the <code>std</code> namespace or only in the main namespace.  There is no
way to make use of these functions which works with all C++ compilers
currently in use on R platforms: use R&rsquo;s versions such as
<code>ISNAN</code> and <code>R_FINITE</code> instead.

<p>If you must use them in C++11, beware that some
compilers<a name="DOCF88" href="#FOOT88"><sup>88</sup></a> provide both
<code>std::isnan</code> and <code>::isnan</code>, so using
</p>
<div class="example">
<pre class="example">using namespace std;
</pre></div>

<p>may cause &lsquo;overloading ambiguity&rsquo; and you must use <code>std::isnan</code>
<em>etc</em> explicitly.
</p>
<p>It is an error (and make little sense, although has been seen) to call
these functions for integer arguments: a few compilers give a compilation
error.
</p>
</li><li> The GNU C/C++ compilers support a large number of non-portable
extensions.  For example, <code>INFINITY</code> (which is in C99 but not
C++98), for which R provides the portable <code>R_PosInf</code> (and
<code>R_NegInf</code> for <code>-INFINITY</code>).  And <code>NAN</code> is just one NaN
value: in R code <code>NA_REAL</code> is usually what is intended, but
<code>R_NaN</code> is also available.

<p>Some (but not all) extensions are listed at
<a href="https://gcc.gnu.org/onlinedocs/gcc/C-Extensions.html">https://gcc.gnu.org/onlinedocs/gcc/C-Extensions.html</a> and
<a href="https://gcc.gnu.org/onlinedocs/gcc/C_002b_002b-Extensions.html">https://gcc.gnu.org/onlinedocs/gcc/C_002b_002b-Extensions.html</a>.
</p>
<p>Other GNU extensions which have bitten package writers is the use of
non-portable characters such as &lsquo;<samp>$</samp>&rsquo; in identifiers and use of C++
headers under <samp>ext</samp>.
</p>
<p>The GNU Fortran compiler also supports a large number of non-portable
extensions, the most commonly encountered one being
<code>ISNAN</code><a name="DOCF89" href="#FOOT89"><sup>89</sup></a>.
Some are listed at
<a href="https://gcc.gnu.org/onlinedocs/gfortran/Extensions-implemented-in-GNU-Fortran.html">https://gcc.gnu.org/onlinedocs/gfortran/Extensions-implemented-in-GNU-Fortran.html</a>.
One that frequently catches package writers is that it allows
out-of-order declarations: in standard-conformant Fortran variables must
be declared (explicitly or implicitly) before use in other declarations
such as dimensions.
</p>
</li><li> Including C-style headers in C++ code is not portable.  Including the
legacy header<a name="DOCF90" href="#FOOT90"><sup>90</sup></a> <samp>math.h</samp> in C++ code may conflict with <samp>cmath</samp> which
may be included by other headers.  This is particularly problematic with
C++11 compilers, as functions like <code>sqrt</code> and <code>isnan</code> are
defined for <code>double</code> arguments in <samp>math.h</samp> and for a range of
types including <code>double</code> in <samp>cmath</samp>.  Similar issues have been
seen for <samp>stdlib.h</samp> and <samp>cstdlib</samp>.  Including the C++ version
first used to be a sufficient workaround but for some 2016 compilers
only one could be included.

</li><li> Variable-length arrays are C99, not supported by C++98 nor by the C++
compilers in use with R on some platforms.

</li><li> The <code>restrict</code> qualifier is C99/C11 but not part of C++11 and not
supported by some C++ compilers used with R.

</li><li> Be careful to include the headers which define the functions you use.
Some compilers/OSes include other system headers in their headers which
are not required by the standards, and so code may compile on such
systems and not on others.  (A prominent example is the C++11 header
<code>&lt;random&gt;</code> which is indirectly included by <code>&lt;algorithm&gt;</code> by
<code>g++</code>.  Another frequent issue is the C header <code>&lt;time.h&gt;</code>
which is included by other headers on Linux and Windows but not macOS nor
Solaris.)  Another common issue is that <code>malloc</code>, <code>calloc</code>,
<code>realloc</code> and <code>free</code> are defined by C99 in the header
<samp>stdlib.h</samp> and (in the <code>std::</code> namespace) by C++ header
<samp>cstdlib</samp>.  Some earlier implementations used a header
<samp>malloc.h</samp>, but that is not portable and does not exist on macOS.

</li><li> For C++ code, be careful to specify namespaces where needed.  Many
functions are defined by the standards to be in the <code>std</code>
namespace, but <code>g++</code> puts many such also in the C++ main
namespace.  One way to do so is to use declarations such as
<div class="example">
<pre class="example">using std::floor;
</pre></div>
<p>but it is usually preferable to use explicit namespace prefixes in the code.
</p>
<p>Examples seen in <acronym>CRAN</acronym> packages include
</p><div class="example">
<pre class="example">abs acos atan calloc ceil div exp fabs floor fmod free log malloc memcpy
memset pow printf qsort round sin sprintf sqrt strcmp strcpy strerror
strlen strncmp strtol tan trunc
</pre></div>

</li><li> Some C++ compilers refuse to compile constructs such as
<div class="example">
<pre class="example">      if(ptr &gt; 0) { ....}
</pre></div>
<p>which compares a pointer to the integer <code>0</code>.  This could just use
<code>if(ptr)</code> (pointer addresses cannot be negative) but if needed
pointers can be tested against <code>nullptr</code> (C++11 and later) or
<code>NULL</code>.
</p>
<p>Note that although <code>nullptr</code> was only introduced in C++11, some
compilers accept it in C++98 mode (but most do not).
</p>
</li><li> Macros defined by the compiler/OS can cause problems.  Identifiers
starting with an underscore followed by an upper-case letter or another
underscore are reserved for system macros and should not be used in
portable code (including not as guards in C/C++ headers).  Other macros,
typically upper-case, may be defined by the compiler or system headers
and can cause problems.
The most common issue involves the names of the Intel CPU registers such
as <code>CS</code>, <code>DS</code>, <code>ES</code>, <code>FS</code>, <code>GS</code> and <code>SS</code>
(and more with longer abbreviations) defined on i586/x64 Solaris in
<samp>&lt;sys/regset.h&gt;</samp> and often included indirectly by <samp>&lt;stdlib.h&gt;</samp>
and other core headers.  Further examples are <code>ERR</code>,
<code>LITTLE_ENDIAN</code>, <code>zero</code> and <code>I</code> (which is defined in
Solaris&rsquo; <samp>&lt;complex.h&gt;</samp> as a compiler intrinsic for the imaginary
unit).  Some of these can be avoided by defining <code>_POSIX_C_SOURCE</code>
before including any system headers, but it is better to only use
all-upper-case names which have a unique prefix such as the package name.

</li><li> <code>typedef</code>s in OS headers can conflict with those in the package:
two examples are <code>ulong</code> on several OSes and <code>index_t</code> on
Solaris.

</li><li> If you use OpenMP, check carefully that you have followed the advice in
the subsection on <a href="#OpenMP-support">OpenMP support</a>.  In particular, any use of
OpenMP in C/C++ code will need to use
<div class="example">
<pre class="example">#ifdef _OPENMP
# include &lt;omp.h&gt;
#endif
</pre></div>
<p>Any use of OpenMP functions, e.g. <code>omp_set_num_threads</code> also
needs to be conditioned.
</p>
<p>And do not hardcode <samp>-lgomp</samp>: not only is that specific to the
GCC family of compilers, using the correct linker flag sets up the
run-time path to the library.
</p>
</li><li> Package authors commonly assume things are part of C99 when they are
not: the most common example is POSIX function <code>strdup</code>.  The most
common C library on Linux, <code>glibc</code>, will hide the declarations of
such extensions unless a &lsquo;feature-test macro&rsquo; is defined <strong>before</strong>
(almost) any system header is included.  So for <code>strdup</code> you need
<div class="example">
<pre class="example">#define _POSIX_C_SOURCE 200809L
...
#include &lt;string.h&gt;
...
strdup call(s)
</pre></div>
<p>where the appropriate value can be found by <code>man strdup</code> on
Linux.  (Use of <code>strncasecmp</code> is similar.)
</p>
<p>However, modes of <code>gcc</code> with &lsquo;GNU EXTENSIONS&rsquo; (which are the
default, either <samp>-std=gnu99</samp> or <samp>-std=gnu11</samp>) declare
enough macros to ensure that missing declarations are rarely seen.
</p>
<p>This applies also to constants such as <code>M_PI</code>, which are part of
the X/Open standard: to use these define <code>_XOPEN_SOURCE</code> before
including any headers.
</p>
</li><li> Similarly, package authors commonly assume things are part of C++ when
they were introduced in C++11 if at all.  Recent examples from
<acronym>CRAN</acronym> packages include the C99/C++11 functions
<div class="example">
<pre class="example">erf expm1 fmin fmax lgamma lround loglp round snprintf strcasecmp trunc
</pre></div>
<p>and the POSIX function <code>strdup</code>.
</p>
</li></ul>

<p>Some additional information for C++ is available at
<a href="http://journal.r-project.org/archive/2011-2/RJournal_2011-2_Plummer.pdf">http://journal.r-project.org/archive/2011-2/RJournal_2011-2_Plummer.pdf</a>
by Martyn Plummer.
</p>
<hr>
<a name="Binary-distribution"></a>
<div class="header">
<p>
Previous: <a href="#Portable-C-and-C_002b_002b-code" accesskey="p" rel="prev">Portable C and C++ code</a>, Up: <a href="#Writing-portable-packages" accesskey="u" rel="up">Writing portable packages</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Binary-distribution-1"></a>
<h4 class="subsection">1.6.5 Binary distribution</h4>

<p>If you want to distribute a binary version of a package on Windows or OS
X, there are further checks you need to do to check it is portable: it
is all too easy to depend on external software on your own machine that
other users will not have.
</p>
<p>For Windows, check what other DLLs your package&rsquo;s DLL depends on
(&lsquo;imports&rsquo; from in the DLL tools&rsquo; parlance).  A convenient GUI-based
tool to do so is &lsquo;Dependency Walker&rsquo;
(<a href="http://www.dependencywalker.com/">http://www.dependencywalker.com/</a>) for both 32-bit and 64-bit
DLLs &ndash; note that this will report as missing links to R&rsquo;s own DLLs
such as <samp>R.dll</samp> and <samp>Rblas.dll</samp>.  For 32-bit DLLs only, the
command-line tool <code>pedump.exe -i</code> (in <samp>Rtools*.exe</samp>) can be
used, and for the brave, the <code>objdump</code> tool in the appropriate
toolchain will also reveal what DLLs are imported from.  If you use a
toolchain other than one provided by the R developers or use your own
makefiles, watch out in particular for dependencies on the toolchain&rsquo;s
runtime DLLs such as <samp>libgfortran</samp>, <samp>libstdc++</samp> and
<samp>libgcc_s</samp>.
</p>
<p>For macOS, using <code>R CMD otool -L</code> on the package&rsquo;s shared object(s)
in the <samp>libs</samp> directory will show what they depend on: watch for
any dependencies in <samp>/usr/local/lib</samp> or
<samp>/usr/local/gfortran/lib</samp>, notably <samp>libgfortran.?.dylib</samp> and
<samp>libquadmath.0.dylib</samp>.
</p>
<p>Many people (including the <acronym>CRAN</acronym> package repository) will not
accept source packages containing binary files as the latter are a
security risk.  If you want to distribute a source package which needs
external software on Windows or macOS, options include
</p><ul>
<li> To arrange for installation of the package to download the
additional software from a URL, as e.g. package <a href="https://CRAN.R-project.org/package=Cairo"><strong>Cairo</strong></a> does.

</li><li> (For <acronym>CRAN</acronym>.)
To negotiate with Uwe Ligges to host the additional components on
WinBuilder, and write a <samp>configure.win</samp> file to install them.

</li></ul>

<p>Be aware that license requirements will need to be met so you may need
to supply the sources for the additional components (and will if your
package has a GPL-like license).
</p>

<hr>
<a name="Diagnostic-messages"></a>
<div class="header">
<p>
Next: <a href="#Internationalization" accesskey="n" rel="next">Internationalization</a>, Previous: <a href="#Writing-portable-packages" accesskey="p" rel="prev">Writing portable packages</a>, Up: <a href="#Creating-R-packages" accesskey="u" rel="up">Creating R packages</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Diagnostic-messages-1"></a>
<h3 class="section">1.7 Diagnostic messages</h3>

<p>Diagnostic messages can be made available for translation, so it is
important to write them in a consistent style.  Using the tools
described in the next section to extract all the messages can give a
useful overview of your consistency (or lack of it).
Some guidelines follow.
</p>
<ul>
<li> Messages are sentence fragments, and not viewed in isolation.  So it is
conventional not to capitalize the first word and not to end with a
period (or other punctuation).

</li><li> Try not to split up messages into small pieces.  In C error messages use
a single format string containing all English words in the messages.

<p>In R error messages do not construct a message with <code>paste</code> (such
messages will not be translated) but <em>via</em> multiple arguments to
<code>stop</code> or <code>warning</code>, or <em>via</em> <code>gettextf</code>.
</p>
</li><li> Do not use colloquialisms such as &ldquo;can&rsquo;t&rdquo; and &ldquo;don&rsquo;t&rdquo;.

</li><li> Conventionally single quotation marks are used for quotations such as

<div class="example">
<pre class="example">'ord' must be a positive integer, at most the number of knots
</pre></div>

<p>and double quotation marks when referring to an R character string or
a class, such as
</p>
<div class="example">
<pre class="example">'format' must be &quot;normal&quot; or &quot;short&quot; - using &quot;normal&quot;
</pre></div>

<p>Since <acronym>ASCII</acronym> does not contain directional quotation marks, it
is best to use &lsquo;<samp>'</samp>&rsquo; and let the translator (including automatic
translation) use directional quotations where available.  The range of
quotation styles is immense: unfortunately we cannot reproduce them in a
portable <code>texinfo</code> document.  But as a taster, some languages use
&lsquo;up&rsquo; and &lsquo;down&rsquo; (comma) quotes rather than left or right quotes, and
some use guillemets (and some use what Adobe calls &lsquo;guillemotleft&rsquo; to
start and others use it to end).
</p>
<p>In R messages it is also possible to use <code>sQuote</code> or <code>dQuote</code> as in
</p>
<div class="example">
<pre class="example">	stop(gettextf(&quot;object must be of class %s or %s&quot;,
		      dQuote(&quot;manova&quot;), dQuote(&quot;maov&quot;)),
	     domain = NA)
</pre></div>

</li><li> Occasionally messages need to be singular or plural (and in other
languages there may be no such concept or several plural forms &ndash;
Slovenian has four).  So avoid constructions such as was once used in
<code>library</code>

<div class="example">
<pre class="example">if((length(nopkgs) &gt; 0) &amp;&amp; !missing(lib.loc)) {
    if(length(nopkgs) &gt; 1)
	warning(&quot;libraries &quot;,
		paste(sQuote(nopkgs), collapse = &quot;, &quot;),
		&quot; contain no packages&quot;)
    else
	warning(&quot;library &quot;, paste(sQuote(nopkgs)),
		&quot; contains no package&quot;)
}
</pre></div>

<p>and was replaced by
</p>
<div class="example">
<pre class="example">if((length(nopkgs) &gt; 0) &amp;&amp; !missing(lib.loc)) {
    pkglist &lt;- paste(sQuote(nopkgs), collapse = &quot;, &quot;)
    msg &lt;- sprintf(ngettext(length(nopkgs),
			    &quot;library %s contains no packages&quot;,
			    &quot;libraries %s contain no packages&quot;,
			    domain = &quot;R-base&quot;),
		   pkglist)
    warning(msg, domain=NA)
}
</pre></div>

<p>Note that it is much better to have complete clauses as here, since
in another language one might need to say
&lsquo;There is no package in library %s&rsquo; or
&lsquo;There are no packages in libraries %s&rsquo;.
</p>
</li></ul>

<hr>
<a name="Internationalization"></a>
<div class="header">
<p>
Next: <a href="#CITATION-files" accesskey="n" rel="next">CITATION files</a>, Previous: <a href="#Diagnostic-messages" accesskey="p" rel="prev">Diagnostic messages</a>, Up: <a href="#Creating-R-packages" accesskey="u" rel="up">Creating R packages</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Internationalization-1"></a>
<h3 class="section">1.8 Internationalization</h3>

<p>There are mechanisms to translate the R- and C-level error and warning
messages.  There are only available if R is compiled with NLS support
(which is requested by <code>configure</code> option <samp>--enable-nls</samp>,
the default).
</p>
<p>The procedures make use of <code>msgfmt</code> and <code>xgettext</code> which are
part of <acronym>GNU</acronym> <code>gettext</code> and this will need to be installed:
Windows users can find pre-compiled binaries at
<a href="https://www.stats.ox.ac.uk/pub/Rtools/goodies/gettext-tools.zip">https://www.stats.ox.ac.uk/pub/Rtools/goodies/gettext-tools.zip</a>.
</p>
<table summary="" class="menu" border="0" cellspacing="0">
<tr><td align="left" valign="top">&bull; <a href="#C_002dlevel-messages" accesskey="1">C-level messages</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#R-messages" accesskey="2">R messages</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Preparing-translations" accesskey="3">Preparing translations</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
</table>

<hr>
<a name="C_002dlevel-messages"></a>
<div class="header">
<p>
Next: <a href="#R-messages" accesskey="n" rel="next">R messages</a>, Previous: <a href="#Internationalization" accesskey="p" rel="prev">Internationalization</a>, Up: <a href="#Internationalization" accesskey="u" rel="up">Internationalization</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="C_002dlevel-messages-1"></a>
<h4 class="subsection">1.8.1 C-level messages</h4>

<p>The process of enabling translations is
</p>
<ul>
<li> In a header file that will be included in all the C (or C++ or Objective
C/C++) files containing messages that should be translated, declare

<div class="example">
<pre class="example">#include &lt;R.h&gt;  /* to include Rconfig.h */

#ifdef ENABLE_NLS
#include &lt;libintl.h&gt;
#define _(String) dgettext (&quot;<var>pkg</var>&quot;, String)
/* replace <var>pkg</var> as appropriate */
#else
#define _(String) (String)
#endif
</pre></div>

</li><li> For each message that should be translated, wrap it in <code>_(...)</code>,
for example

<div class="example">
<pre class="example">error(_(&quot;'ord' must be a positive integer&quot;));
</pre></div>

<p>If you want to use different messages for singular and plural forms, you
need to add
</p>
<div class="example">
<pre class="example">#ifndef ENABLE_NLS
#define dngettext(pkg, String, StringP, N) (N &gt; 1 ? StringP : String)
#endif
</pre></div>

<p>and mark strings by
</p>
<div class="example">
<pre class="example">dngettext(&quot;<var>pkg</var>&quot;, <var>&lt;singular string&gt;</var>, <var>&lt;plural string&gt;</var>, n)
</pre></div>

</li><li> In the package&rsquo;s <samp>src</samp> directory run

<div class="example">
<pre class="example">xgettext --keyword=_ -o <var>pkg</var>.pot *.c
</pre></div>

</li></ul>

<p>The file <samp>src/<var>pkg</var>.pot</samp> is the template file, and
conventionally this is shipped as <samp>po/<var>pkg</var>.pot</samp>.
</p>
<hr>
<a name="R-messages"></a>
<div class="header">
<p>
Next: <a href="#Preparing-translations" accesskey="n" rel="next">Preparing translations</a>, Previous: <a href="#C_002dlevel-messages" accesskey="p" rel="prev">C-level messages</a>, Up: <a href="#Internationalization" accesskey="u" rel="up">Internationalization</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="R-messages-1"></a>
<h4 class="subsection">1.8.2 R messages</h4>

<p>Mechanisms are also available to support the automatic translation of
R <code>stop</code>, <code>warning</code> and <code>message</code> messages.  They make
use of message catalogs in the same way as C-level messages, but using
domain <code>R-<var>pkg</var></code> rather than <code><var>pkg</var></code>.  Translation of
character strings inside <code>stop</code>, <code>warning</code> and <code>message</code>
calls is automatically enabled, as well as other messages enclosed in
calls to <code>gettext</code> or <code>gettextf</code>.  (To suppress this, use
argument <code>domain=NA</code>.)
</p>
<p>Tools to prepare the <samp>R-<var>pkg</var>.pot</samp> file are provided in package
<strong>tools</strong>: <code>xgettext2pot</code> will prepare a file from all strings
occurring inside <code>gettext</code>/<code>gettextf</code>, <code>stop</code>,
<code>warning</code> and <code>message</code> calls.  Some of these are likely to be
spurious and so the file is likely to need manual editing.
<code>xgettext</code> extracts the actual calls and so is more useful when
tidying up error messages.
</p>
<p>The R function <code>ngettext</code> provides an interface to the C
function of the same name: see example in the previous section.  It is
safest to use <code>domain=&quot;R-<var>pkg</var>&quot;</code> explicitly in calls to
<code>ngettext</code>, and necessary for earlier versions of R unless they
are calls directly from a function in the package.
</p>

<hr>
<a name="Preparing-translations"></a>
<div class="header">
<p>
Previous: <a href="#R-messages" accesskey="p" rel="prev">R messages</a>, Up: <a href="#Internationalization" accesskey="u" rel="up">Internationalization</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Preparing-translations-1"></a>
<h4 class="subsection">1.8.3 Preparing translations</h4>

<p>Once the template files have been created, translations can be made.
Conventional translations have file extension <samp>.po</samp> and are placed
in the <samp>po</samp> subdirectory of the package with a name that is either
&lsquo;<samp><var>ll</var>.po</samp>&rsquo; or &lsquo;<samp>R-<var>ll</var>.po</samp>&rsquo; for translations of the C and R
messages respectively to language with code &lsquo;<samp><var>ll</var></samp>&rsquo;.
</p>
<p>See <a href="http://cran.r-project.org/doc/manuals/R-admin.html#Localization-of-messages">Localization of messages</a> in <cite>R Installation and Administration</cite>, for details of language codes.
</p>
<p>There is an R function, <code>update_pkg_po</code> in package <strong>tools</strong>,
to automate much of the maintenance of message translations.  See its
help for what it does in detail.
</p>
<p>If this is called on a package with no existing translations, it creates
the directory <samp><var>pkgdir</var>/po</samp>, creates a template file of R
messages, <samp><var>pkgdir</var>/po/R-<var>pkg</var>.pot</samp>, within it, creates the
&lsquo;<samp>en@quot</samp>&rsquo; translation and installs that.  (The &lsquo;<samp>en@quot</samp>&rsquo;
pseudo-language interprets quotes in their directional forms in suitable
(e.g. UTF-8) locales.)
</p>
<p>If the package has C source files in its <samp>src</samp> directory
that are marked for translation, use
</p>
<div class="example">
<pre class="example">touch <var>pkgdir</var>/po/<var>pkg</var>.pot
</pre></div>

<p>to create a dummy template file, then call <code>update_pkg_po</code> again
(this can also be done before it is called for the first time).
</p>
<p>When translations to new languages are added in the <samp><var>pkgdir</var>/po</samp>
directory, running the same command will check and then
install the translations.
</p>
<p>If the package sources are updated, the same command will update the
template files, merge the changes into the translation <samp>.po</samp> files
and then installed the updated translations.  You will often see that
merging marks translations as &lsquo;fuzzy&rsquo; and this is reported in the
coverage statistics.  As fuzzy translations are <em>not</em> used, this is
an indication that the translation files need human attention.
</p>
<p>The merged translations are run through <code>tools::checkPofile</code> to
check that C-style formats are used correctly: if not the mismatches are
reported and the broken translations are not installed.
</p>
<p>This function needs the GNU <code>gettext-tools</code> installed and on the
path: see its help page.
</p>

<a name="index-CITATION-1"></a>
<a name="index-citation-1"></a>
<hr>
<a name="CITATION-files"></a>
<div class="header">
<p>
Next: <a href="#Package-types" accesskey="n" rel="next">Package types</a>, Previous: <a href="#Internationalization" accesskey="p" rel="prev">Internationalization</a>, Up: <a href="#Creating-R-packages" accesskey="u" rel="up">Creating R packages</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="CITATION-files-1"></a>
<h3 class="section">1.9 CITATION files</h3>

<p>An installed file named <samp>CITATION</samp> will be used by the
<code>citation()</code> function.  (It should be in the <samp>inst</samp>
subdirectory of the package sources.)
</p>
<p>The <samp>CITATION</samp> file is parsed as R code (in the package&rsquo;s
declared encoding, or in <acronym>ASCII</acronym> if none is declared).  If no
such file is present, <code>citation</code> auto-generates citation
information from the package <samp>DESCRIPTION</samp> metadata, and an example
of what that would look like as a <samp>CITATION</samp> file can be seen in
recommended package <a href="https://CRAN.R-project.org/package=nlme"><strong>nlme</strong></a> (see below): recommended packages
<a href="https://CRAN.R-project.org/package=boot"><strong>boot</strong></a>, <a href="https://CRAN.R-project.org/package=cluster"><strong>cluster</strong></a> and <a href="https://CRAN.R-project.org/package=mgcv"><strong>mgcv</strong></a> have further
examples.
</p>
<p>A <samp>CITATION</samp> file will contain calls to function <code>bibentry</code>.
</p>
<p>Here is that for <a href="https://CRAN.R-project.org/package=nlme"><strong>nlme</strong></a>:
</p>
<div class="example">
<pre class="example">year &lt;- sub(&quot;-.*&quot;, &quot;&quot;, meta$Date)
note &lt;- sprintf(&quot;R package version %s&quot;, meta$Version)

bibentry(bibtype = &quot;Manual&quot;,
         title = &quot;{nlme}: Linear and Nonlinear Mixed Effects Models&quot;,
         author = c(person(&quot;Jose&quot;, &quot;Pinheiro&quot;),
                    person(&quot;Douglas&quot;, &quot;Bates&quot;),
                    person(&quot;Saikat&quot;, &quot;DebRoy&quot;),
                    person(&quot;Deepayan&quot;, &quot;Sarkar&quot;),
                    person(&quot;R Core Team&quot;)),
         year = year,
         note = note,
         url = &quot;https://CRAN.R-project.org/package=nlme&quot;)
</pre></div>

<p>Note the way that information that may need to be updated is picked up
from object <code>meta</code>, a parsed version of the <samp>DESCRIPTION</samp> file
&ndash; it is tempting to hardcode such information, but it normally then
gets outdated.  See <code>?bibentry</code> for further details of the
information which can be provided.
</p>
<p>In case a bibentry contains LaTeX markup (e.g., for accented
characters or mathematical symbols), it may be necessary to provide a
text representation to be used for printing via the <code>textVersion</code>
argument to <code>bibentry</code>.  E.g., earlier versions of
<a href="https://CRAN.R-project.org/package=nlme"><strong>nlme</strong></a> additionally used
</p>
<div class="example">
<pre class="example">         textVersion =
         paste0(&quot;Jose Pinheiro, Douglas Bates, Saikat DebRoy,&quot;,
                &quot;Deepayan Sarkar and the R Core Team (&quot;,
                year,
                &quot;). nlme: Linear and Nonlinear Mixed Effects Models. &quot;,
                note, &quot;.&quot;)
</pre></div>

<p>The <samp>CITATION</samp> file should itself produce no output when
<code>source</code>-d.
</p>
<p>It is desirable (and essential for <acronym>CRAN</acronym>) that the
<samp>CITATION</samp> file does not contain calls to functions such as
<code>packageDescription</code> which assume the package is installed in a
library tree on the package search path.
</p>
<hr>
<a name="Package-types"></a>
<div class="header">
<p>
Next: <a href="#Services" accesskey="n" rel="next">Services</a>, Previous: <a href="#CITATION-files" accesskey="p" rel="prev">CITATION files</a>, Up: <a href="#Creating-R-packages" accesskey="u" rel="up">Creating R packages</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Package-types-1"></a>
<h3 class="section">1.10 Package types</h3>

<p>The <samp>DESCRIPTION</samp> file has an optional field <code>Type</code> which if
missing is assumed to be &lsquo;<samp>Package</samp>&rsquo;, the sort of extension discussed
so far in this chapter.  Currently one other type is recognized; there
used also to be a &lsquo;<samp>Translation</samp>&rsquo; type.
</p>
<table summary="" class="menu" border="0" cellspacing="0">
<tr><td align="left" valign="top">&bull; <a href="#Frontend" accesskey="1">Frontend</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
</table>

<hr>
<a name="Frontend"></a>
<div class="header">
<p>
Previous: <a href="#Package-types" accesskey="p" rel="prev">Package types</a>, Up: <a href="#Package-types" accesskey="u" rel="up">Package types</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Frontend-1"></a>
<h4 class="subsection">1.10.1 Frontend</h4>

<p>This is a rather general mechanism, designed for adding new front-ends
such as the former <strong>gnomeGUI</strong> package (see the <samp>Archive</samp> area on
<acronym>CRAN</acronym>).  If a <samp>configure</samp> file is found in the top-level
directory of the package it is executed, and then if a <samp>Makefile</samp>
is found (often generated by <samp>configure</samp>), <code>make</code> is called.
If <code>R CMD INSTALL --clean</code> is used <code>make clean</code> is called.  No
other action is taken.
</p>
<p><code>R CMD build</code> can package up this type of extension, but <code>R
CMD check</code> will check the type and skip it.
</p>
<p>Many packages of this type need write permission for the R
installation directory.
</p>
<hr>
<a name="Services"></a>
<div class="header">
<p>
Previous: <a href="#Package-types" accesskey="p" rel="prev">Package types</a>, Up: <a href="#Creating-R-packages" accesskey="u" rel="up">Creating R packages</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Services-1"></a>
<h3 class="section">1.11 Services</h3>

<p>Several members of the R project have set up services to assist those
writing R packages, particularly those intended for public
distribution.
</p>
<p><a href="https://win-builder.r-project.org">win-builder.r-project.org</a>
offers the automated preparation of (32/64-bit) Windows binaries from
well-tested source packages.
</p>
<p>R-Forge (<a href="https://R-Forge.r-project.org">R-Forge.r-project.org</a>) and
RForge (<a href="https://www.rforge.net">www.rforge.net</a>) are similar
services with similar names.  Both provide source-code management
through SVN, daily building and checking, mailing lists and a repository
that can be accessed <em>via</em> <code>install.packages</code> (they can be
selected by <code>setRepositories</code> and the GUI menus that use it).
Package developers have the opportunity to present their work on the
basis of project websites or news announcements.  Mailing lists, forums
or wikis provide useRs with convenient instruments for discussions and
for exchanging information between developers and/or interested useRs.
</p>

<hr>
<a name="Writing-R-documentation-files"></a>
<div class="header">
<p>
Next: <a href="#Tidying-and-profiling-R-code" accesskey="n" rel="next">Tidying and profiling R code</a>, Previous: <a href="#Creating-R-packages" accesskey="p" rel="prev">Creating R packages</a>, Up: <a href="#Top" accesskey="u" rel="up">Top</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Writing-R-documentation-files-1"></a>
<h2 class="chapter">2 Writing R documentation files</h2>
<a name="index-Documentation_002c-writing"></a>

<table summary="" class="menu" border="0" cellspacing="0">
<tr><td align="left" valign="top">&bull; <a href="#Rd-format" accesskey="1">Rd format</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Sectioning" accesskey="2">Sectioning</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Marking-text" accesskey="3">Marking text</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Lists-and-tables" accesskey="4">Lists and tables</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Cross_002dreferences" accesskey="5">Cross-references</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Mathematics" accesskey="6">Mathematics</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Figures" accesskey="7">Figures</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Insertions" accesskey="8">Insertions</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Indices" accesskey="9">Indices</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Platform_002dspecific-sections">Platform-specific sections</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Conditional-text">Conditional text</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Dynamic-pages">Dynamic pages</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#User_002ddefined-macros">User-defined macros</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Encoding">Encoding</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Processing-documentation-files">Processing documentation files</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Editing-Rd-files">Editing Rd files</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
</table>

<hr>
<a name="Rd-format"></a>
<div class="header">
<p>
Next: <a href="#Sectioning" accesskey="n" rel="next">Sectioning</a>, Previous: <a href="#Writing-R-documentation-files" accesskey="p" rel="prev">Writing R documentation files</a>, Up: <a href="#Writing-R-documentation-files" accesskey="u" rel="up">Writing R documentation files</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Rd-format-1"></a>
<h3 class="section">2.1 Rd format</h3>

<p>R objects are documented in files written in &ldquo;R documentation&rdquo;
(Rd) format, a simple markup language much of which closely resembles
(La)TeX, which can be processed into a variety of formats,
including LaTeX, <acronym>HTML</acronym> and plain text.  The translation is
carried out by functions in the <strong>tools</strong> package called by the
script <code>Rdconv</code> in <samp><var>R_HOME</var>/bin</samp> and by the
installation scripts for packages.
</p>
<p>The R distribution contains more than 1300 such files which can be
found in the <samp>src/library/<var>pkg</var>/man</samp> directories of the R
source tree, where <var>pkg</var> stands for one of the standard packages
which are included in the R distribution.
</p>
<p>As an example, let us look at a simplified version of
<samp>src/library/base/man/load.Rd</samp> which documents the R function
<code>load</code>.
</p>
<blockquote>
<table summary="" class="cartouche" border="1"><tr><td>
<div class="smallexample">
<pre class="smallexample">% File src/library/base/man/load.Rd
\name{load}
\alias{load}
\title{Reload Saved Datasets}
\description{
  Reload the datasets written to a file with the function
  \code{save}.
}
\usage{
load(file, envir = parent.frame())
}
\arguments{
  \item{file}{a connection or a character string giving the
    name of the file to load.}
  \item{envir}{the environment where the data should be
    loaded.}
}
\seealso{
  \code{\link{save}}.
}
\examples{
## save all data
save(list = ls(), file= &quot;all.RData&quot;)

## restore the saved values to the current environment
load(&quot;all.RData&quot;)

## restore the saved values to the workspace
load(&quot;all.RData&quot;, .GlobalEnv)
}
\keyword{file}
</pre></div>
</td></tr></table>
</blockquote>

<p>An <samp>Rd</samp> file consists of three parts.  The header gives basic
information about the name of the file, the topics documented, a title,
a short textual description and R usage information for the objects
documented.  The body gives further information (for example, on the
function&rsquo;s arguments and return value, as in the above example).
Finally, there is an optional footer with keyword information.  The
header is mandatory.
</p>
<p>Information is given within a series of <em>sections</em> with standard
names (and user-defined sections are also allowed).  Unless otherwise
specified<a name="DOCF91" href="#FOOT91"><sup>91</sup></a> these should occur only once in an <samp>Rd</samp>
file (in any order), and the processing software will retain only the
first occurrence of a standard section in the file, with a warning.
</p>
<p>See <a href="https://developer.r-project.org/Rds.html">&ldquo;Guidelines for Rd
files&rdquo;</a> for guidelines for writing documentation in <samp>Rd</samp> format
which should be useful for package writers.
<a name="index-prompt"></a>
The R
generic function <code>prompt</code> is used to construct a bare-bones <samp>Rd</samp>
file ready for manual editing.  Methods are defined for documenting
functions (which fill in the proper function and argument names) and
data frames.  There are also functions <code>promptData</code>,
<code>promptPackage</code>, <code>promptClass</code>, and <code>promptMethods</code> for
other types of <samp>Rd</samp> file.
</p>
<p>The general syntax of <samp>Rd</samp> files is summarized below.  For a detailed
technical discussion of current <samp>Rd</samp> syntax, see
<a href="https://developer.r-project.org/parseRd.pdf">&ldquo;Parsing Rd files&rdquo;</a>.
</p>
<p><samp>Rd</samp> files consist of four types of text input.  The most common
is LaTeX-like, with the backslash used as a prefix on markup
(e.g. <code>\alias</code>), and braces used to indicate arguments
(e.g. <code>{load}</code>).  The least common type of text is &lsquo;verbatim&rsquo;
text, where no markup other than the comment marker (<code>%</code>) is
processed.  There is also a rare variant of &lsquo;verbatim&rsquo; text
(used in <code>\eqn</code>, <code>\deqn</code>, <code>\figure</code>,
and <code>\newcommand</code>) where comment markers need not be escaped.
The final type is R-like, intended for R code, but allowing some
embedded macros.  Quoted strings within R-like text are handled
specially: regular character escapes such as <code>\n</code> may be entered
as-is.  Only markup starting with <code>\l</code> (e.g.  <code>\link</code>) or
<code>\v</code> (e.g. <code>\var</code>) will be recognized within quoted strings.
The rarely used vertical tab <code>\v</code> must be entered as <code>\\v</code>.
</p>
<p>Each macro defines the input type for its argument.  For example, the
file initially uses LaTeX-like syntax, and this is also used in the
<code>\description</code> section, but the <code>\usage</code> section uses
R-like syntax, and the <code>\alias</code> macro uses &lsquo;verbatim&rsquo; syntax.
Comments run from a percent symbol <code>%</code> to the end of the line in
all types of text except the rare &lsquo;verbatim&rsquo; variant
(as on the first line of the <code>load</code> example).
</p>
<p>Because backslashes, braces and percent symbols have special meaning, to
enter them into text sometimes requires escapes using a backslash.  In
general balanced braces do not need to be escaped, but percent symbols
always do, except in the &lsquo;verbatim&rsquo; variant.
For the complete list of macros and rules for escapes, see
<a href="https://developer.r-project.org/parseRd.pdf">&ldquo;Parsing Rd files&rdquo;</a>.
</p>
<table summary="" class="menu" border="0" cellspacing="0">
<tr><td align="left" valign="top">&bull; <a href="#Documenting-functions" accesskey="1">Documenting functions</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Documenting-data-sets" accesskey="2">Documenting data sets</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Documenting-S4-classes-and-methods" accesskey="3">Documenting S4 classes and methods</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Documenting-packages" accesskey="4">Documenting packages</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
</table>

<hr>
<a name="Documenting-functions"></a>
<div class="header">
<p>
Next: <a href="#Documenting-data-sets" accesskey="n" rel="next">Documenting data sets</a>, Previous: <a href="#Rd-format" accesskey="p" rel="prev">Rd format</a>, Up: <a href="#Rd-format" accesskey="u" rel="up">Rd format</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Documenting-functions-1"></a>
<h4 class="subsection">2.1.1 Documenting functions</h4>

<p>The basic markup commands used for documenting R objects (in
particular, functions) are given in this subsection.
</p>
<dl compact="compact">
<dt><code>\name{<var>name</var>}</code></dt>
<dd><a name="index-_005cname"></a>
<p><var>name</var> typically<a name="DOCF92" href="#FOOT92"><sup>92</sup></a> is the basename of
the <samp>Rd</samp> file containing the documentation.  It is the &ldquo;name&rdquo; of
the <samp>Rd</samp> object represented by the file and has to be unique in a
package.  To avoid problems with indexing the package manual, it may not
contain &lsquo;<samp>!</samp>&rsquo;  &lsquo;<samp>|</samp>&rsquo; nor &lsquo;<samp>@</samp>&rsquo;, and to avoid possible problems
with the <acronym>HTML</acronym> help system it should not contain &lsquo;<samp>/</samp>&rsquo; nor a space.
(LaTeX special characters are allowed, but may not be collated
correctly in the index.)  There can only be one <code>\name</code> entry in a
file, and it must not contain any markup.  Entries in the package manual
will be in alphabetic<a name="DOCF93" href="#FOOT93"><sup>93</sup></a> order
of the <code>\name</code> entries.
</p>
</dd>
<dt><code>\alias{<var>topic</var>}</code></dt>
<dd><a name="index-_005calias"></a>
<p>The <code>\alias</code> sections specify all &ldquo;topics&rdquo; the file documents.
This information is collected into index data bases for lookup by the
on-line (plain text and <acronym>HTML</acronym>) help systems.  The <var>topic</var> can
contain spaces, but (for historical reasons) leading and trailing spaces
will be stripped.  Percent and left brace need to be escaped by
a backslash.
</p>
<p>There may be several <code>\alias</code> entries.  Quite often it is
convenient to document several R objects in one file.  For example,
file <samp>Normal.Rd</samp> documents the density, distribution function,
quantile function and generation of random variates for the normal
distribution, and hence starts with
</p>
<div class="example">
<pre class="example">\name{Normal}
\alias{Normal}
\alias{dnorm}
\alias{pnorm}
\alias{qnorm}
\alias{rnorm}
</pre></div>

<p>Also, it is often convenient to have several different ways to refer to
an R object, and an <code>\alias</code> does not need to be the name of an
object.
</p>
<p>Note that the <code>\name</code> is not necessarily a topic documented, and if
so desired it needs to have an explicit <code>\alias</code> entry (as in this
example).
</p>
</dd>
<dt><code>\title{<var>Title</var>}</code></dt>
<dd><a name="index-_005ctitle"></a>
<p>Title information for the <samp>Rd</samp> file.  This should be capitalized
and not end in a period; try to limit its length to at most 65
characters for widest compatibility.
</p>
<p>Markup is supported in the text, but use of characters other than
English text and punctuation (e.g., &lsquo;<samp>&lt;</samp>&rsquo;) may limit portability.
</p>
<p>There must be one (and only one) <code>\title</code> section in a help file.
</p>
</dd>
<dt><code>\description{&hellip;}</code></dt>
<dd><a name="index-_005cdescription"></a>
<p>A short description of what the function(s) do(es) (one paragraph, a few
lines only).  (If a description is too long and cannot easily be
shortened, the file probably tries to document too much at once.)
This is mandatory except for package-overview files.
</p>
</dd>
<dt><code>\usage{<var>fun</var>(<var>arg1</var>, <var>arg2</var>, &hellip;)}</code></dt>
<dd><a name="index-_005cusage"></a>
<p>One or more lines showing the synopsis of the function(s) and variables
documented in the file.  These are set in typewriter font.  This is an
R-like command.
</p>
<p>The usage information specified should match the function definition
<em>exactly</em> (such that automatic checking for consistency between
code and documentation is possible).
</p>
<p>It is no longer advisable to use <code>\synopsis</code> for the actual
synopsis and show modified synopses in the <code>\usage</code>.  Support for
<code>\synopsis</code> will be removed in \R 3.1.0.  To indicate that a
function can be used in several different ways, depending on the named
arguments specified, use section <code>\details</code>.  E.g.,
<samp>abline.Rd</samp> contains
</p>
<div class="example">
<pre class="example">\details{
  Typical usages are
\preformatted{abline(a, b, untf = FALSE, \dots)
......
}
</pre></div>

<a name="index-_005cmethod"></a>
<p>Use <code>\method{<var>generic</var>}{<var>class</var>}</code> to indicate the name
of an S3 method for the generic function <var>generic</var> for objects
inheriting from class <code>&quot;<var>class</var>&quot;</code>.  In the printed versions,
this will come out as <var>generic</var> (reflecting the understanding that
methods should not be invoked directly but <em>via</em> method dispatch), but
<code>codoc()</code> and other QC tools always have access to the full name.
</p>
<p>For example, <samp>print.ts.Rd</samp> contains
</p>
<div class="example">
<pre class="example">\usage{
\method{print}{ts}(x, calendar, \dots)
}
</pre></div>

<p>which will print as
</p>
<div class="example">
<pre class="example">Usage:

     ## S3 method for class ‘ts’:
     print(x, calendar, ...)
</pre></div>

<p>Usage for replacement functions should be given in the style of
<code>dim(x) &lt;- value</code> rather than explicitly indicating the name of the
replacement function (<code><span class="nolinebreak">&quot;dim&lt;-&quot;</span></code><!-- /@w --> in the above).  Similarly, one
can use <code>\method{<var>generic</var>}{<var>class</var>}(<var>arglist</var>) &lt;-
value</code> to indicate the usage of an S3 replacement method for the generic
replacement function <code>&quot;<var>generic</var>&lt;-&quot;</code> for objects inheriting
from class <code>&quot;<var>class</var>&quot;</code>.
</p>
<p>Usage for S3 methods for extracting or replacing parts of an object, S3
methods for members of the Ops group, and S3 methods for user-defined
(binary) infix operators (&lsquo;<samp>%<var>xxx</var>%</samp>&rsquo;) follows the above rules,
using the appropriate function names.  E.g., <samp>Extract.factor.Rd</samp>
contains
</p>
<div class="example">
<pre class="example">\usage{
\method{[}{factor}(x, \dots, drop = FALSE)
\method{[[}{factor}(x, \dots)
\method{[}{factor}(x, \dots) &lt;- value
}
</pre></div>

<p>which will print as
</p>
<div class="example">
<pre class="example">Usage:

     ## S3 method for class ‘factor’:
     x[..., drop = FALSE]
     ## S3 method for class ‘factor’:
     x[[...]]
     ## S3 replacement method for class ‘factor’:
     x[...] &lt;- value
</pre></div>

<a name="index-_005cS3method"></a>
<p><code>\S3method</code> is accepted as an alternative to <code>\method</code>.
</p>
</dd>
<dt><code>\arguments{&hellip;}</code></dt>
<dd><a name="index-_005carguments"></a>
<p>Description of the function&rsquo;s arguments, using an entry of the form
</p>
<div class="example">
<pre class="example">\item{<var>arg_i</var>}{<var>Description of arg_i</var>.}
</pre></div>

<p>for each element of the argument list.  (Note that there is
no whitespace between the three parts of the entry.) There may be
optional text outside the <code>\item</code> entries, for example to give
general information about groups of parameters.
</p>

</dd>
<dt><code>\details{&hellip;}</code></dt>
<dd><a name="index-_005cdetails"></a>
<p>A detailed if possible precise description of the functionality
provided, extending the basic information in the <code>\description</code>
slot.
</p>
</dd>
<dt><code>\value{&hellip;}</code></dt>
<dd><a name="index-_005cvalue"></a>
<p>Description of the function&rsquo;s return value.
</p>
<p>If a list with multiple values is returned, you can use entries of the
form
</p>
<div class="example">
<pre class="example">\item{<var>comp_i</var>}{<var>Description of comp_i</var>.}
</pre></div>

<p>for each component of the list returned.  Optional text may
precede<a name="DOCF94" href="#FOOT94"><sup>94</sup></a> this
list (see for example the help for <code>rle</code>).  Note that <code>\value</code>
is implicitly a <code>\describe</code> environment, so that environment should
not be used for listing components, just individual <code>\item{}{}</code>
entries.
</p>
</dd>
<dt><code>\references{&hellip;}</code></dt>
<dd><a name="index-_005creferences"></a>
<p>A section with references to the literature.  Use <code>\url{}</code> or
<code>\href{}{}</code> for web pointers.
</p>
</dd>
<dt><code>\note{...}</code></dt>
<dd><a name="index-_005cnote"></a>
<p>Use this for a special note you want to have pointed out.  Multiple
<code>\note</code> sections are allowed, but might be confusing to the end users.
</p>
<p>For example, <samp>pie.Rd</samp> contains
</p>
<div class="example">
<pre class="example">\note{
  Pie charts are a very bad way of displaying information.
  The eye is good at judging linear measures and bad at
  judging relative areas.
  ......
}
</pre></div>

</dd>
<dt><code>\author{&hellip;}</code></dt>
<dd><a name="index-_005cauthor"></a>
<p>Information about the author(s) of the <samp>Rd</samp> file.  Use
<code>\email{}</code> without extra delimiters (such as &lsquo;<samp>( )</samp>&rsquo; or
&lsquo;<samp>&lt; &gt;</samp>&rsquo;) to specify email addresses, or <code>\url{}</code> or
<code>\href{}{}</code> for web pointers.
</p>
</dd>
<dt><code>\seealso{&hellip;}</code></dt>
<dd><a name="index-_005cseealso"></a>
<p>Pointers to related R objects, using <code>\code{\link{...}}</code> to
refer to them (<code>\code</code> is the correct markup for R object names,
and <code>\link</code> produces hyperlinks in output formats which support
this.  See <a href="#Marking-text">Marking text</a>, and <a href="#Cross_002dreferences">Cross-references</a>).
</p>
<a name="index-_005cexamples"></a>
</dd>
<dt><code>\examples{&hellip;}</code></dt>
<dd><p>Examples of how to use the function.  Code in this section is set
in typewriter font without reformatting and is run by
<code>example()</code> unless marked otherwise (see below).
</p>
<p>Examples are not only useful for documentation purposes, but also
provide test code used for diagnostic checking of R code.  By
default, text inside <code>\examples{}</code> will be displayed in the
output of the help page and run by <code>example()</code> and by <code>R CMD
check</code>.  You can use <code>\dontrun{}</code>
<a name="index-_005cdontrun"></a>
for text that should only be shown, but not run, and
<code>\dontshow{}</code>
<a name="index-_005cdontshow"></a>
for extra commands for testing that should not be shown to users, but
will be run by <code>example()</code>.  (Previously this was called
<code>\testonly</code>, and that is still accepted.)
</p>
<p>Text inside <code>\dontrun{}</code> is &lsquo;verbatim&rsquo;, but the other parts
of the <code>\examples</code> section are R-like text.
</p>
<p>For example,
</p>
<div class="example">
<pre class="example">x &lt;- runif(10)       # <span class="roman">Shown and run.</span>
\dontrun{plot(x)}    # <span class="roman">Only shown.</span>
\dontshow{log(x)}    # <span class="roman">Only run.</span>
</pre></div>

<p>Thus, example code not included in <code>\dontrun</code> must be executable!
In addition, it should not use any system-specific features or require
special facilities (such as Internet access or write permission to
specific directories).  Text included in <code>\dontrun</code> is indicated by
comments in the processed help files: it need not be valid R code but
the escapes must still be used for <code>%</code>, <code>\</code> and unpaired
braces as in other &lsquo;verbatim&rsquo; text.
</p>
<p>Example code must be capable of being run by <code>example</code>, which uses
<code>source</code>.  This means that it should not access <samp>stdin</samp>,
e.g. to <code>scan()</code> data from the example file.
</p>
<p>Data needed for making the examples executable can be obtained by random
number generation (for example, <code>x &lt;- rnorm(100)</code>), or by using
standard data sets listed by <code>data()</code> (see <code>?data</code> for more
info).
</p>
<p>Finally, there is <code>\donttest</code>, used (at the beginning of a separate
line) to mark code that should be run by <code>example()</code> but not by
<code>R CMD check</code> (by default: the option <samp>--run-donttest</samp> can
be used).  This should be needed only occasionally but can be used for
code which might fail in circumstances that are hard to test for, for
example in some locales.  (Use e.g. <code>capabilities()</code> or
<code>nzchar(Sys.which(&quot;someprogram&quot;))</code> to test for features needed in
the examples wherever possible, and you can also use <code>try()</code> or
<code>tryCatch()</code>.  Use <code>interactive()</code> to condition examples which
need someone to interact with.)  Note that code included in
<code>\donttest</code> must be correct R code, and any packages used should
be declared in the <samp>DESCRIPTION</samp> file.  It is good practice to
include a comment in the <code>\donttest</code> section explaining why it is
needed.
</p>
<a name="index-_005ckeyword"></a>
</dd>
<dt><code>\keyword{<var>key</var>}</code></dt>
<dd><p>There can be zero or more <code>\keyword</code> sections per file.
Each <code>\keyword</code> section should specify a single keyword, preferably
one of the standard keywords as listed in file <samp>KEYWORDS</samp> in the
R documentation directory (default <samp><var>R_HOME</var>/doc</samp>).  Use
e.g. <code>RShowDoc(&quot;KEYWORDS&quot;)</code> to inspect the standard keywords from
within R.  There can be more than one <code>\keyword</code> entry if the R
object being documented falls into more than one category, or none.
</p>
<p>Do strongly consider using <code>\concept</code> (see <a href="#Indices">Indices</a>) instead of
<code>\keyword</code> if you are about to use more than very few non-standard
keywords.
</p>
<p>The special keyword &lsquo;<samp>internal</samp>&rsquo; marks a page of internal objects
that are not part of the package&rsquo;s API.  If the help page for object
<code>foo</code> has keyword &lsquo;<samp>internal</samp>&rsquo;, then <code>help(foo)</code> gives this
help page, but <code>foo</code> is excluded from several object indices,
including the alphabetical list of objects in the <acronym>HTML</acronym> help system.
</p>
<p><code>help.search()</code> can search by keyword, including user-defined
values: however the &lsquo;Search Engine &amp; Keywords&rsquo; <acronym>HTML</acronym> page accessed
<em>via</em> <code>help.start()</code> provides single-click access only to a
pre-defined list of keywords.
</p></dd>
</dl>


<hr>
<a name="Documenting-data-sets"></a>
<div class="header">
<p>
Next: <a href="#Documenting-S4-classes-and-methods" accesskey="n" rel="next">Documenting S4 classes and methods</a>, Previous: <a href="#Documenting-functions" accesskey="p" rel="prev">Documenting functions</a>, Up: <a href="#Rd-format" accesskey="u" rel="up">Rd format</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Documenting-data-sets-1"></a>
<h4 class="subsection">2.1.2 Documenting data sets</h4>

<p>The structure of <samp>Rd</samp> files which document R data sets is slightly
different.  Sections such as <code>\arguments</code> and <code>\value</code> are not
needed but the format and source of the data should be explained.
</p>
<p>As an example, let us look at <samp>src/library/datasets/man/rivers.Rd</samp>
which documents the standard R data set <code>rivers</code>.
</p>
<blockquote>
<table summary="" class="cartouche" border="1"><tr><td>
<div class="smallexample">
<pre class="smallexample">\name{rivers}
\docType{data}
\alias{rivers}
\title{Lengths of Major North American Rivers}
\description{
  This data set gives the lengths (in miles) of 141 \dQuote{major}
  rivers in North America, as compiled by the US Geological
  Survey.
}
\usage{rivers}
\format{A vector containing 141 observations.}
\source{World Almanac and Book of Facts, 1975, page 406.}
\references{
  McNeil, D. R. (1977) \emph{Interactive Data Analysis}.
  New York: Wiley.
}
\keyword{datasets}
</pre></div>
</td></tr></table>
</blockquote>

<p>This uses the following additional markup commands.
</p>
<dl compact="compact">
<dt><code>\docType{&hellip;}</code></dt>
<dd><p>Indicates the &ldquo;type&rdquo; of the documentation object.  Always &lsquo;<samp>data</samp>&rsquo;
for data sets, and &lsquo;<samp>package</samp>&rsquo; for <samp><var>pkg</var>-package.Rd</samp>
overview files.  Documentation for S4 methods and classes uses
&lsquo;<samp>methods</samp>&rsquo; (from <code>promptMethods()</code>) and &lsquo;<samp>class</samp>&rsquo; (from
<code>promptClass()</code>).
</p>
</dd>
<dt><code>\format{&hellip;}</code></dt>
<dd><a name="index-_005cformat"></a>
<p>A description of the format of the data set (as a vector, matrix, data
frame, time series, &hellip;).  For matrices and data frames this should
give a description of each column, preferably as a list or table.
See <a href="#Lists-and-tables">Lists and tables</a>, for more information.
</p>
</dd>
<dt><code>\source{&hellip;}</code></dt>
<dd><a name="index-_005csource"></a>
<p>Details of the original source (a reference or <acronym>URL</acronym>,
see <a href="#Specifying-URLs">Specifying URLs</a>).  In addition, section <code>\references</code> could
give secondary sources and usages.
</p></dd>
</dl>

<p>Note also that when documenting data set <var>bar</var>,
</p>
<ul>
<li> The <code>\usage</code> entry is always <code><var>bar</var></code> or (for packages
which do not use lazy-loading of data) <code>data(<var>bar</var>)</code>.  (In
particular, only document a <em>single</em> data object per <samp>Rd</samp> file.)
</li><li> The <code>\keyword</code> entry should always be &lsquo;<samp>datasets</samp>&rsquo;.
</li></ul>

<p>If <code><var>bar</var></code> is a data frame, documenting it as a data set can
be initiated <em>via</em> <code>prompt(<var>bar</var>)</code>.  Otherwise, the <code>promptData</code>
function may be used.
</p>
<hr>
<a name="Documenting-S4-classes-and-methods"></a>
<div class="header">
<p>
Next: <a href="#Documenting-packages" accesskey="n" rel="next">Documenting packages</a>, Previous: <a href="#Documenting-data-sets" accesskey="p" rel="prev">Documenting data sets</a>, Up: <a href="#Rd-format" accesskey="u" rel="up">Rd format</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Documenting-S4-classes-and-methods-1"></a>
<h4 class="subsection">2.1.3 Documenting S4 classes and methods</h4>

<p>There are special ways to use the &lsquo;<samp>?</samp>&rsquo;  operator, namely
&lsquo;<samp>class?<var>topic</var></samp>&rsquo; and &lsquo;<samp>methods?<var>topic</var></samp>&rsquo;, to access
documentation for S4 classes and methods, respectively.  This mechanism
depends on conventions for the topic names used in <code>\alias</code>
entries.  The topic names for S4 classes and methods respectively are of
the form
</p>
<div class="example">
<pre class="example"><var>class</var>-class
<var>generic</var>,<var>signature_list</var>-method
</pre></div>

<p>where <var>signature_list</var> contains the names of the classes in the
signature of the method (without quotes) separated by &lsquo;<samp>,</samp>&rsquo; (without
whitespace), with &lsquo;<samp>ANY</samp>&rsquo; used for arguments without an explicit
specification.  E.g., &lsquo;<samp>genericFunction-class</samp>&rsquo; is the topic name for
documentation for the S4 class <code>&quot;genericFunction&quot;</code>, and
&lsquo;<samp>coerce,ANY,NULL-method</samp>&rsquo; is the topic name for documentation for
the S4 method for <code>coerce</code> for signature <code>c(&quot;ANY&quot;, &quot;NULL&quot;)</code>.
</p>
<p>Skeletons of documentation for S4 classes and methods can be generated
by using the functions <code>promptClass()</code> and <code>promptMethods()</code>
from package <strong>methods</strong>.  If it is necessary or desired to provide an
explicit function declaration (in a <code>\usage</code> section) for an S4
method (e.g., if it has &ldquo;surprising arguments&rdquo; to be mentioned
explicitly), one can use the special markup
</p>
<div class="example">
<pre class="example">\S4method{<var>generic</var>}{<var>signature_list</var>}(<var>argument_list</var>)
</pre></div>

<p>(e.g., &lsquo;<samp>\S4method{coerce}{ANY,NULL}(from, to)</samp>&rsquo;).
</p>
<p>To make full use of the potential of the on-line documentation system,
all user-visible S4 classes and methods in a package should at least
have a suitable <code>\alias</code> entry in one of the package&rsquo;s <samp>Rd</samp> files.
If a package has methods for a function defined originally somewhere
else, and does not change the underlying default method for the
function, the package is responsible for documenting the methods it
creates, but not for the function itself or the default method.
</p>
<p>An S4 replacement method is documented in the same way as an S3 one: see
the description of  <code>\method</code> in <a href="#Documenting-functions">Documenting functions</a>.
</p>

<p>See <kbd>help(&quot;Documentation&quot;, package = &quot;methods&quot;)</kbd> for more
information on using and creating on-line documentation for S4 classes and
methods.
</p>
<hr>
<a name="Documenting-packages"></a>
<div class="header">
<p>
Previous: <a href="#Documenting-S4-classes-and-methods" accesskey="p" rel="prev">Documenting S4 classes and methods</a>, Up: <a href="#Rd-format" accesskey="u" rel="up">Rd format</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Documenting-packages-1"></a>
<h4 class="subsection">2.1.4 Documenting packages</h4>

<p>Packages may have an overview help page with an <code>\alias</code>
<code><var>pkgname</var>-package</code>, e.g. &lsquo;<samp>utils-package</samp>&rsquo; for the
<strong>utils</strong> package, when <code>package?<var>pkgname</var></code> will open that
help page.  If a topic named <code><var>pkgname</var></code> does not exist in
another <samp>Rd</samp> file, it is helpful to use this as an additional
<code>\alias</code>.
</p>
<p>Skeletons of documentation for a package can be generated using the
function <code>promptPackage()</code>.  If the <code>final = LIBS</code> argument
is used, then the <samp>Rd</samp> file will be generated in final form, containing
the information that would be produced up to
<code>library(help = <var>pkgname</var>)</code>.  Otherwise (the default) comments
will be inserted giving suggestions for content.
</p>
<p>Apart from the mandatory <code>\name</code> and <code>\title</code> and the
<code><var>pkgname</var>-package</code> alias, the only requirement for the package
overview page is that it include a <code>\docType{package}</code> statement.
All other content is optional.  We suggest that it should be a short
overview, to give a reader unfamiliar with the package enough
information to get started.  More extensive documentation is better
placed into a package vignette (see <a href="#Writing-package-vignettes">Writing package vignettes</a>) and
referenced from this page, or into individual man pages for the
functions, datasets, or classes.
</p>
<hr>
<a name="Sectioning"></a>
<div class="header">
<p>
Next: <a href="#Marking-text" accesskey="n" rel="next">Marking text</a>, Previous: <a href="#Rd-format" accesskey="p" rel="prev">Rd format</a>, Up: <a href="#Writing-R-documentation-files" accesskey="u" rel="up">Writing R documentation files</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Sectioning-1"></a>
<h3 class="section">2.2 Sectioning</h3>

<p>To begin a new paragraph or leave a blank line in an example, just
insert an empty line (as in (La)TeX).  To break a line, use
<code>\cr</code>.
<a name="index-_005ccr"></a>
</p>
<p>In addition to the predefined sections (such as <code>\description{}</code>,
<code>\value{}</code>, etc.), you can &ldquo;define&rdquo; arbitrary ones by
<code>\section{<var>section_title</var>}{&hellip;}</code>.
<a name="index-_005csection"></a>
For example
</p>
<div class="example">
<pre class="example">\section{Warning}{
  You must not call this function unless &hellip;
}
</pre></div>

<p>For consistency with the pre-assigned sections, the section name (the
first argument to <code>\section</code>) should be capitalized (but not all
upper case).  Whitespace between the first and second braced expressions
is not allowed.  Markup (e.g. <code>\code</code>) within the section title
may cause problems with the latex conversion (depending on the version
of macro packages such as &lsquo;<samp>hyperref</samp>&rsquo;) and so should be avoided.
</p>
<p>The <code>\subsection</code> macro takes arguments in the same format as
<code>\section</code>, but is used within a section, so it may be used to
nest subsections within sections or other subsections.  There is no
predefined limit on the nesting level, but formatting is not designed
for more than 3 levels (i.e. subsections within subsections within
sections).
</p>
<p>Note that additional named sections are always inserted at a fixed
position in the output (before <code>\note</code>, <code>\seealso</code> and the
examples), no matter where they appear in the input (but in the same
order amongst themselves as in the input).
</p>

<hr>
<a name="Marking-text"></a>
<div class="header">
<p>
Next: <a href="#Lists-and-tables" accesskey="n" rel="next">Lists and tables</a>, Previous: <a href="#Sectioning" accesskey="p" rel="prev">Sectioning</a>, Up: <a href="#Writing-R-documentation-files" accesskey="u" rel="up">Writing R documentation files</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Marking-text-1"></a>
<h3 class="section">2.3 Marking text</h3>
<a name="index-Marking-text-in-documentation"></a>

<p>The following logical markup commands are available for emphasizing or
quoting text.
</p>
<dl compact="compact">
<dt><code>\emph{<var>text</var>}</code></dt>
<dd><a name="index-_005cemph"></a>
</dd>
<dt><code>\strong{<var>text</var>}</code></dt>
<dd><a name="index-_005cstrong"></a>
<p>Emphasize <var>text</var> using <em>italic</em> and <strong>bold</strong> font if
possible; <code>\strong</code> is regarded as stronger (more emphatic).
</p>
</dd>
<dt><code>\bold{<var>text</var>}</code></dt>
<dd><a name="index-_005cbold"></a>
<p>Set <var>text</var> in <b>bold</b> font where possible.
</p>
</dd>
<dt><code>\sQuote{<var>text</var>}</code></dt>
<dd><a name="index-_005csQuote"></a>
</dd>
<dt><code>\dQuote{<var>text</var>}</code></dt>
<dd><a name="index-_005cdQuote"></a>
<p>Portably single or double quote <var>text</var> (without hard-wiring the
characters used for quotation marks).
</p></dd>
</dl>

<p>Each of the above commands takes LaTeX-like input, so other macros
may be used within <var>text</var>.
</p>
<p>The following logical markup commands are available for indicating
specific kinds of text.  Except as noted, these take &lsquo;verbatim&rsquo; text
input, and so other macros may not be used within them.  Some characters
will need to be escaped (see <a href="#Insertions">Insertions</a>).
</p>
<dl compact="compact">
<dt><code>\code{<var>text</var>}</code></dt>
<dd><a name="index-_005ccode"></a>
<p>Indicate text that is a literal example of a piece of an R program,
e.g., a fragment of R code or the name of an R object.  Text is
entered in R-like syntax, and displayed using <code>typewriter</code> font
where possible.  Macros <code>\var</code> and <code>\link</code> are interpreted within
<var>text</var>.
</p>
</dd>
<dt><code>\preformatted{<var>text</var>}</code></dt>
<dd><a name="index-_005cpreformatted"></a>
<p>Indicate text that is a literal example of a piece of a program.  Text
is displayed using <code>typewriter</code> font where possible.  Formatting,
e.g. line breaks, is preserved.  (Note that this includes a line break
after the initial {, so typically text should start on the same line as
the command.)
</p>
<p>Due to limitations in LaTeX as of this writing, this macro may not be
nested within other markup macros other than <code>\dQuote</code> and
<code>\sQuote</code>, as errors or bad formatting may result.
</p>
</dd>
<dt><code>\kbd{<var>keyboard-characters</var>}</code></dt>
<dd><a name="index-_005ckbd"></a>
<p>Indicate keyboard input, using <kbd>slanted typewriter</kbd> font if
possible, so users can distinguish the characters they are supposed to
type from computer output.  Text is entered &lsquo;verbatim&rsquo;.
</p>
</dd>
<dt><code>\samp{<var>text</var>}</code></dt>
<dd><a name="index-_005csamp"></a>
<p>Indicate text that is a literal example of a sequence of characters,
entered &lsquo;verbatim&rsquo;.  No wrapping or reformatting will occur.  Displayed
using <code>typewriter</code> font where possible.
</p>

</dd>
<dt><code>\verb{<var>text</var>}</code></dt>
<dd><a name="index-_005cverb"></a>
<p>Indicate text that is a literal example of a sequence of characters,
with no interpretation of e.g. <code>\var</code>, but which will be included
within word-wrapped text.  Displayed using <code>typewriter</code> font if
possible.
</p>
</dd>
<dt><code>\pkg{<var>package_name</var>}</code></dt>
<dd><a name="index-_005cpkg"></a>
<p>Indicate the name of an R package.  LaTeX-like.
</p>
</dd>
<dt><code>\file{<var>file_name</var>}</code></dt>
<dd><a name="index-_005cfile"></a>
<p>Indicate the name of a file.  Text is LaTeX-like, so backslash needs
to be escaped.  Displayed using a distinct font where possible.
</p>
</dd>
<dt><code>\email{<var>email_address</var>}</code></dt>
<dd><a name="index-_005cemail"></a>
<p>Indicate an electronic mail address.  LaTeX-like, will be rendered as
a hyperlink in <acronym>HTML</acronym> and PDF conversion.  Displayed using
<code>typewriter</code> font where possible.
</p>
</dd>
<dt><code>\url{<var>uniform_resource_locator</var>}</code></dt>
<dd><a name="index-_005curl"></a>
<p>Indicate a uniform resource locator (<acronym>URL</acronym>) for the World Wide
Web.  The argument is handled as &lsquo;verbatim&rsquo; text (with percent and
braces escaped by backslash), and rendered as a hyperlink in <acronym>HTML</acronym> and
PDF conversion.  Linefeeds are removed, and as from R 3.2.0 leading
and trailing whitespace<a name="DOCF95" href="#FOOT95"><sup>95</sup></a> is removed. See <a href="#Specifying-URLs">Specifying URLs</a>.
</p>
<p>Displayed using <code>typewriter</code> font where possible.
</p>
</dd>
<dt><code>\href{<var>uniform_resource_locator</var>}{<var>text</var>}</code></dt>
<dd><a name="index-_005chref"></a>
<p>Indicate a hyperlink to the World Wide Web. The first argument is
handled as &lsquo;verbatim&rsquo; text (with percent and braces escaped by
backslash) and is used as the <acronym>URL</acronym> in the hyperlink, with the
second argument of LaTeX-like text displayed to the user.  Linefeeds
are removed from the first argument, and as from R 3.2.0 leading and
trailing whitespace is removed.
</p>
<p>Note that RFC3986-encoded URLs (e.g. using &lsquo;<samp>\%28VS.85\%29</samp>&rsquo; in
place of &lsquo;<samp>(VS.85)</samp>&rsquo;) may not work correctly in versions of R
before 3.1.3 and are best avoided&mdash;use <code>URLdecode()</code> to decode
them.
</p>
</dd>
<dt><code>\var{<var>metasyntactic_variable</var>}</code></dt>
<dd><a name="index-_005cvar"></a>
<p>Indicate a metasyntactic variable.  In some cases this will be rendered
distinctly, e.g. in italic, but not in all<a name="DOCF96" href="#FOOT96"><sup>96</sup></a>. LaTeX-like.
</p></dd>
<dt><code>\env{<var>environment_variable</var>}</code></dt>
<dd><a name="index-_005cenv"></a>
<p>Indicate an environment variable. &lsquo;Verbatim&rsquo;.
Displayed using <code>typewriter</code> font where possible
</p></dd>
<dt><code>\option{<var>option</var>}</code></dt>
<dd><a name="index-_005coption"></a>
<p>Indicate a command-line option.  &lsquo;Verbatim&rsquo;.
Displayed using <code>typewriter</code> font where possible.
</p></dd>
<dt><code>\command{<var>command_name</var>}</code></dt>
<dd><a name="index-_005ccommand"></a>
<p>Indicate the name of a command. LaTeX-like, so <code>\var</code> is
interpreted.  Displayed using <code>typewriter</code> font where possible.
</p></dd>
<dt><code>\dfn{<var>term</var>}</code></dt>
<dd><a name="index-_005cdfn"></a>
<p>Indicate the introductory or defining use of a term. LaTeX-like.
</p></dd>
<dt><code>\cite{<var>reference</var>}</code></dt>
<dd><a name="index-_005ccite"></a>
<p>Indicate a reference without a direct cross-reference <em>via</em> <code>\link</code>
(see <a href="#Cross_002dreferences">Cross-references</a>), such as the name of a book. LaTeX-like.
</p></dd>
<dt><code>\acronym{<var>acronym</var>}</code></dt>
<dd><a name="index-_005cacronym"></a>
<p>Indicate an acronym (an abbreviation written in all capital letters),
such as <acronym>GNU</acronym>. LaTeX-like.
</p></dd>
</dl>


<hr>
<a name="Lists-and-tables"></a>
<div class="header">
<p>
Next: <a href="#Cross_002dreferences" accesskey="n" rel="next">Cross-references</a>, Previous: <a href="#Marking-text" accesskey="p" rel="prev">Marking text</a>, Up: <a href="#Writing-R-documentation-files" accesskey="u" rel="up">Writing R documentation files</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Lists-and-tables-1"></a>
<h3 class="section">2.4 Lists and tables</h3>
<a name="index-Lists-and-tables-in-documentation"></a>

<a name="index-_005citemize"></a>
<a name="index-_005cenumerate"></a>
<p>The <code>\itemize</code> and <code>\enumerate</code> commands take a single
argument, within which there may be one or more <code>\item</code> commands.
The text following each <code>\item</code> is formatted as one or more
paragraphs, suitably indented and with the first paragraph marked with a
bullet point (<code>\itemize</code>) or a number (<code>\enumerate</code>).
</p>
<p>Note that unlike argument lists, <code>\item</code> in these formats is
followed by a space and the text (not enclosed in braces).  For example
</p>
<div class="example">
<pre class="example">  \enumerate{
    \item A database consists of one or more records, each with one or
    more named fields.
    \item Regular lines start with a non-whitespace character.
    \item Records are separated by one or more empty lines.
  }
</pre></div>

<p><code>\itemize</code> and <code>\enumerate</code> commands may be nested.
</p>
<a name="index-_005cdescribe"></a>
<p>The <code>\describe</code> command is similar to <code>\itemize</code> but allows
initial labels to be specified.  Each <code>\item</code> takes two arguments,
the label and the body of the item, in exactly the same way as an
argument or value <code>\item</code>.  <code>\describe</code> commands are mapped to
<code>&lt;DL&gt;</code> lists in <acronym>HTML</acronym> and <code>\description</code> lists in LaTeX.
</p>
<a name="index-_005ctabular"></a>
<p>The <code>\tabular</code> command takes two arguments.  The first gives for
each of the columns the required alignment (&lsquo;<samp>l</samp>&rsquo; for
left-justification, &lsquo;<samp>r</samp>&rsquo; for right-justification or &lsquo;<samp>c</samp>&rsquo; for
centring.)  The second argument consists of an arbitrary number of
lines separated by <code>\cr</code>, and with fields separated by <code>\tab</code>.
For example:
</p>
<div class="example">
<pre class="example">  \tabular{rlll}{
    [,1] \tab Ozone   \tab numeric \tab Ozone (ppb)\cr
    [,2] \tab Solar.R \tab numeric \tab Solar R (lang)\cr
    [,3] \tab Wind    \tab numeric \tab Wind (mph)\cr
    [,4] \tab Temp    \tab numeric \tab Temperature (degrees F)\cr
    [,5] \tab Month   \tab numeric \tab Month (1--12)\cr
    [,6] \tab Day     \tab numeric \tab Day of month (1--31)
  }
</pre></div>

<p>There must be the same number of fields on each line as there are
alignments in the first argument, and they must be non-empty (but can
contain only spaces).  (There is no whitespace between <code>\tabular</code>
and the first argument, nor between the two arguments.)
</p>
<hr>
<a name="Cross_002dreferences"></a>
<div class="header">
<p>
Next: <a href="#Mathematics" accesskey="n" rel="next">Mathematics</a>, Previous: <a href="#Lists-and-tables" accesskey="p" rel="prev">Lists and tables</a>, Up: <a href="#Writing-R-documentation-files" accesskey="u" rel="up">Writing R documentation files</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Cross_002dreferences-1"></a>
<h3 class="section">2.5 Cross-references</h3>
<a name="index-Cross_002dreferences-in-documentation"></a>

<a name="index-_005clink"></a>
<p>The markup <code>\link{<var>foo</var>}</code> (usually in the combination
<code>\code{\link{<var>foo</var>}}</code>) produces a hyperlink to the help for
<var>foo</var>.  Here <var>foo</var> is a <em>topic</em>, that is the argument of
<code>\alias</code> markup in another <samp>Rd</samp> file (possibly in another package).
Hyperlinks are supported in some of the formats to which <samp>Rd</samp> files are
converted, for example <acronym>HTML</acronym> and PDF, but ignored in others, e.g.
the text format.
</p>
<p>One main usage of <code>\link</code> is in the <code>\seealso</code> section of the
help page, see <a href="#Rd-format">Rd format</a>.
</p>
<p>Note that whereas leading and trailing spaces are stripped when
extracting a topic from a <code>\alias</code>, they are not stripped when
looking up the topic of a <code>\link</code>.
</p>
<a name="index-_005clinkS4class"></a>
<p>You can specify a link to a different topic than its name by
<code>\link[=<var>dest</var>]{<var>name</var>}</code> which links to topic <var>dest</var>
with name <var>name</var>.  This can be used to refer to the documentation
for S3/4 classes, for example <code>\code{&quot;\link[=abc-class]{abc}&quot;}</code>
would be a way to refer to the documentation of an S4 class <code>&quot;abc&quot;</code>
defined in your package, and
<code>\code{&quot;\link[=terms.object]{terms}&quot;}</code> to the S3 <code>&quot;terms&quot;</code>
class (in package <strong>stats</strong>).  To make these easy to read in the
source file, <code>\code{&quot;\linkS4class{abc}&quot;}</code> expands to the form
given above.
</p>
<p>There are two other forms of optional argument specified as
<code>\link[<var>pkg</var>]{<var>foo</var>}</code> and
<code>\link[<var>pkg:bar</var>]{<var>foo</var>}</code> to link to the package
<strong><var>pkg</var></strong>, to <em>files</em> <samp><var>foo</var>.html</samp> and
<samp><var>bar</var>.html</samp> respectively.  These are rarely needed, perhaps to
refer to not-yet-installed packages (but there the <acronym>HTML</acronym> help system
will resolve the link at run time) or in the normally undesirable event
that more than one package offers help on a topic<a name="DOCF97" href="#FOOT97"><sup>97</sup></a> (in
which case the present package has precedence so this is only needed to
refer to other packages).  They are currently only used in <acronym>HTML</acronym> help
(and ignored for hyperlinks in LaTeX conversions of help pages), and
link to the file rather than the topic (since there is no way to know
which topics are in which files in an uninstalled package).  The
<strong>only</strong> reason to use these forms for base and recommended
packages is to force a reference to a package that might be further down
the search path.  Because they have been frequently misused, the <acronym>HTML</acronym>
help system looks for topic <code><var>foo</var></code> in package <strong><var>pkg</var></strong>
if it does not find file <samp><var>foo</var>.html</samp>.
</p>
<hr>
<a name="Mathematics"></a>
<div class="header">
<p>
Next: <a href="#Figures" accesskey="n" rel="next">Figures</a>, Previous: <a href="#Cross_002dreferences" accesskey="p" rel="prev">Cross-references</a>, Up: <a href="#Writing-R-documentation-files" accesskey="u" rel="up">Writing R documentation files</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Mathematics-1"></a>
<h3 class="section">2.6 Mathematics</h3>
<a name="index-Mathematics-in-documentation"></a>
<a name="index-_005ceqn"></a>
<a name="index-_005cdeqn"></a>

<p>Mathematical formulae should be set beautifully for printed
documentation yet we still want something useful for text and <acronym>HTML</acronym>
online help.  To this end, the two commands
<code>\eqn{<var>latex</var>}{<var>ascii</var>}</code> and
<code>\deqn{<var>latex</var>}{<var>ascii</var>}</code> are used.  Whereas <code>\eqn</code>
is used for &ldquo;inline&rdquo; formulae (corresponding to TeX&rsquo;s
<code>$&hellip;$</code>), <code>\deqn</code> gives &ldquo;displayed equations&rdquo; (as in
LaTeX&rsquo;s <code>displaymath</code> environment, or TeX&rsquo;s
<code>$$&hellip;$$</code>).  Both arguments are treated as &lsquo;verbatim&rsquo; text.
</p>
<p>Both commands can also be used as <code>\eqn{<var>latexascii</var>}</code> (only
<em>one</em> argument) which then is used for both <var>latex</var> and
<var>ascii</var>.  No whitespace is allowed between command and the first
argument, nor between the first and second arguments.
</p>
<p>The following example is from <samp>Poisson.Rd</samp>:
</p>
<div class="example">
<pre class="example">  \deqn{p(x) = \frac{\lambda^x e^{-\lambda}}{x!}}{%
	p(x) = \lambda^x exp(-\lambda)/x!}
  for \eqn{x = 0, 1, 2, \ldots}.
</pre></div>


<p>For text on-line help we get
</p>
<blockquote>
<table summary="" class="cartouche" border="1"><tr><td>
<div class="example">
<pre class="example">    p(x) = lambda^x exp(-lambda)/x!

for x = 0, 1, 2, ....
</pre></div>
</td></tr></table>
</blockquote>

<p>Greek letters (both cases) will be rendered in <acronym>HTML</acronym> if preceded by a
backslash, <code>\dots</code> and <code>\ldots</code> will be rendered as ellipses
and <code>\sqrt</code>, <code>\ge</code> and <code>\le</code> as mathematical symbols.
</p>
<p>Note that only basic LaTeX can be used, there being no provision to
specify LaTeX style files such as the AMS extensions.
</p>
<hr>
<a name="Figures"></a>
<div class="header">
<p>
Next: <a href="#Insertions" accesskey="n" rel="next">Insertions</a>, Previous: <a href="#Mathematics" accesskey="p" rel="prev">Mathematics</a>, Up: <a href="#Writing-R-documentation-files" accesskey="u" rel="up">Writing R documentation files</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Figures-1"></a>
<h3 class="section">2.7 Figures</h3>
<a name="index-Figures-in-documentation"></a>
<a name="index-_005cfigure"></a>

<p>To include figures in help pages, use the <code>\figure</code> markup.  There
are three forms.
</p>
<p>The two commonly used simple forms are <code>\figure{<var>filename</var>}</code>
and <code>\figure{<var>filename</var>}{<var>alternate text</var>}</code>.  This will
include a copy of the figure in either <acronym>HTML</acronym> or LaTeX output.  In text
output, the alternate text will be displayed instead.  (When the second
argument is omitted, the filename will be used.)  Both the filename and
the alternate text will be parsed verbatim, and should not include
special characters that are significant in <acronym>HTML</acronym> or LaTeX.
</p>
<p>The expert form is <code>\figure{<var>filename</var>}{options:
<var>string</var>}</code>.  (The word &lsquo;<samp>options:</samp>&rsquo; must be typed exactly as
shown and followed by at least one space.)  In this form, the
<var>string</var> is copied into the <acronym>HTML</acronym> <code>img</code> tag as attributes
following the <code>src</code> attribute, or into the second argument of the
<code>\Figure</code> macro in LaTeX, which by default is used as options to
an <code>\includegraphics</code> call.  As it is unlikely that any single
string would suffice for both display modes, the expert form would
normally be wrapped in conditionals.  It is up to the author to make
sure that legal <acronym>HTML</acronym>/LaTeX is used.  For example, to include a
logo in both <acronym>HTML</acronym> (using the simple form) and LaTeX (using the
expert form), the following could be used:
</p>
<div class="example">
<pre class="example">\if{html}{\figure{Rlogo.svg}{options: width=100 alt=&quot;R logo&quot;}}
\if{latex}{\figure{Rlogo.pdf}{options: width=0.5in}}
</pre></div>

<p>The files containing the figures should be stored in the directory
<samp>man/figures</samp>.  Files with extensions <samp>.jpg</samp>, <samp>.jpeg</samp>,
<samp>.pdf</samp>, <samp>.png</samp> and <samp>.svg</samp> from that directory will be
copied to the <samp>help/figures</samp> directory at install time. (Figures in
PDF format will not display in most <acronym>HTML</acronym> browsers, but might be the
best choice in reference manuals.)  Specify the filename relative to
<samp>man/figures</samp> in the <code>\figure</code> directive.
</p>
<hr>
<a name="Insertions"></a>
<div class="header">
<p>
Next: <a href="#Indices" accesskey="n" rel="next">Indices</a>, Previous: <a href="#Figures" accesskey="p" rel="prev">Figures</a>, Up: <a href="#Writing-R-documentation-files" accesskey="u" rel="up">Writing R documentation files</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Insertions-1"></a>
<h3 class="section">2.8 Insertions</h3>

<a name="index-_005cR"></a>
<p>Use <code>\R</code> for the R system itself.  Use <code>\dots</code>
<a name="index-_005cdots"></a>
for the dots in function argument lists &lsquo;<samp>&hellip;</samp>&rsquo;, and
<code>\ldots</code>
<a name="index-_005cldots"></a>
for ellipsis dots in ordinary text.<a name="DOCF98" href="#FOOT98"><sup>98</sup></a>  These can be followed by
<code>{}</code>, and should be unless followed by whitespace.
</p>
<p>After an unescaped &lsquo;<samp>%</samp>&rsquo;, you can put your own comments regarding the
help text.  The rest of the line (but not the newline at the end) will
be completely disregarded.  Therefore, you can also use it to make part
of the &ldquo;help&rdquo; invisible.
</p>
<p>You can produce a backslash (&lsquo;<samp>\</samp>&rsquo;) by escaping it by another
backslash.  (Note that <code>\cr</code> is used for generating line breaks.)
</p>
<p>The &ldquo;comment&rdquo; character &lsquo;<samp>%</samp>&rsquo; and unpaired braces<a name="DOCF99" href="#FOOT99"><sup>99</sup></a>
<em>almost always</em> need to be escaped by &lsquo;<samp>\</samp>&rsquo;, and &lsquo;<samp>\\</samp>&rsquo; can
be used for backslash and needs to be when there are two or more adjacent
backslashes.  In R-like code quoted strings are handled slightly
differently; see <a href="https://developer.r-project.org/parseRd.pdf">&ldquo;Parsing Rd files&rdquo;</a> for details &ndash; in particular braces should not be
escaped in quoted strings.
</p>
<p>All of &lsquo;<samp>% { } \</samp>&rsquo; should be escaped in LaTeX-like text.
</p>
<a name="index-_005cenc"></a>
<p>Text which might need to be represented differently in different
encodings should be marked by <code>\enc</code>, e.g.
<code>\enc{J&ouml;reskog}{Joreskog}</code> (with no whitespace between the
braces) where the first argument will be used where encodings are
allowed and the second should be <acronym>ASCII</acronym> (and is used for e.g.
the text conversion in locales that cannot represent the encoded form).
(This is intended to be used for individual words, not whole sentences
or paragraphs.)
</p>
<hr>
<a name="Indices"></a>
<div class="header">
<p>
Next: <a href="#Platform_002dspecific-sections" accesskey="n" rel="next">Platform-specific sections</a>, Previous: <a href="#Insertions" accesskey="p" rel="prev">Insertions</a>, Up: <a href="#Writing-R-documentation-files" accesskey="u" rel="up">Writing R documentation files</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Indices-1"></a>
<h3 class="section">2.9 Indices</h3>
<a name="index-Indices"></a>

<p>The <code>\alias</code> command (see <a href="#Documenting-functions">Documenting functions</a>) is used to
specify the &ldquo;topics&rdquo; documented, which should include <em>all</em> R
objects in a package such as functions and variables, data sets, and S4
classes and methods (see <a href="#Documenting-S4-classes-and-methods">Documenting S4 classes and methods</a>).  The
on-line help system searches the index data base consisting of all
alias topics.
</p>
<a name="index-_005cconcept"></a>
<p>In addition, it is possible to provide &ldquo;concept index entries&rdquo; using
<code>\concept</code>, which can be used for <code>help.search()</code> lookups.
E.g., file <samp>cor.test.Rd</samp> in the standard package <strong>stats</strong>
contains
</p>
<div class="example">
<pre class="example">\concept{Kendall correlation coefficient}
\concept{Pearson correlation coefficient}
\concept{Spearman correlation coefficient}
</pre></div>

<p>so that e.g. <kbd>??Spearman</kbd> will succeed in finding the
help page for the test for association between paired samples using
Spearman&rsquo;s rho.
</p>

<p>(Note that <code>help.search()</code> only uses &ldquo;sections&rdquo; of documentation
objects with no additional markup.)
</p>
<p>If you want to cross reference such items from other help files <em>via</em>
<code>\link</code>, you need to use <code>\alias</code> and not <code>\concept</code>.
</p>

<hr>
<a name="Platform_002dspecific-sections"></a>
<div class="header">
<p>
Next: <a href="#Conditional-text" accesskey="n" rel="next">Conditional text</a>, Previous: <a href="#Indices" accesskey="p" rel="prev">Indices</a>, Up: <a href="#Writing-R-documentation-files" accesskey="u" rel="up">Writing R documentation files</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Platform_002dspecific-documentation"></a>
<h3 class="section">2.10 Platform-specific documentation</h3>
<a name="index-Platform_002dspecific-documentation"></a>

<p>Sometimes the documentation needs to differ by platform.  Currently two
OS-specific options are available, &lsquo;<samp>unix</samp>&rsquo; and &lsquo;<samp>windows</samp>&rsquo;, and
lines in the help source file can be enclosed in
</p>
<div class="example">
<pre class="example">#ifdef <var>OS</var>
   ...
#endif
</pre></div>

<p>or
</p>
<div class="example">
<pre class="example">#ifndef <var>OS</var>
   ...
#endif
</pre></div>

<p>for OS-specific inclusion or exclusion.  Such blocks should not be
nested, and should be entirely within a block (that, is between the
opening and closing brace of a section or item), or at top-level contain
one or more complete sections.
</p>
<p>If the differences between platforms are extensive or the R objects
documented are only relevant to one platform, platform-specific <samp>Rd</samp> files
can be put in a <samp>unix</samp> or <samp>windows</samp> subdirectory.
</p>
<hr>
<a name="Conditional-text"></a>
<div class="header">
<p>
Next: <a href="#Dynamic-pages" accesskey="n" rel="next">Dynamic pages</a>, Previous: <a href="#Platform_002dspecific-sections" accesskey="p" rel="prev">Platform-specific sections</a>, Up: <a href="#Writing-R-documentation-files" accesskey="u" rel="up">Writing R documentation files</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Conditional-text-1"></a>
<h3 class="section">2.11 Conditional text</h3>
<a name="index-conditionals"></a>
<a name="index-_005cif"></a>
<a name="index-_005cifelse"></a>
<a name="index-_005cout"></a>

<p>Occasionally the best content for one output format is different from
the best content for another.  For this situation, the
<code>\if{<var>format</var>}{<var>text</var>}</code> or
<code>\ifelse{<var>format</var>}{<var>text</var>}{<var>alternate</var>}</code> markup
is used.  Here <var>format</var> is a comma separated list of formats in
which the <var>text</var> should be rendered.  The <var>alternate</var> will be
rendered if the format does not match.  Both <var>text</var> and
<var>alternate</var> may be any sequence of text and markup.
</p>
<p>Currently the following formats are recognized:  <code>example</code>,
<code>html</code>, <code>latex</code> and <code>text</code>.  These select output for
the corresponding targets. (Note that <code>example</code> refers to
extracted example code rather than the displayed example in some other
format.)  Also accepted are <code>TRUE</code> (matching all formats) and
<code>FALSE</code> (matching no formats).  These could be the output
of the <code>\Sexpr</code> macro (see <a href="#Dynamic-pages">Dynamic pages</a>).
</p>
<p>The <code>\out{<var>literal</var>}</code> macro would usually be used within
the <var>text</var> part of <code>\if{<var>format</var>}{<var>text</var>}</code>.  It
causes the renderer to output the literal text exactly, with no
attempt to escape special characters.  For example, use
the following to output the markup necessary to display the Greek letter in
LaTeX or <acronym>HTML</acronym>, and the text string <code>alpha</code> in other formats:
</p><div class="example">
<pre class="example">\ifelse{latex}{\out{$\alpha$}}{\ifelse{html}{\out{&amp;alpha;}}{alpha}}
</pre></div>

<hr>
<a name="Dynamic-pages"></a>
<div class="header">
<p>
Next: <a href="#User_002ddefined-macros" accesskey="n" rel="next">User-defined macros</a>, Previous: <a href="#Conditional-text" accesskey="p" rel="prev">Conditional text</a>, Up: <a href="#Writing-R-documentation-files" accesskey="u" rel="up">Writing R documentation files</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Dynamic-pages-1"></a>
<h3 class="section">2.12 Dynamic pages</h3>
<a name="index-dynamic-pages"></a>
<a name="index-_005cSexpr"></a>
<a name="index-_005cRdOpts"></a>

<p>Two macros supporting dynamically generated man pages are <code>\Sexpr</code>
and <code>\RdOpts</code>.  These are modelled after Sweave, and are intended
to contain executable R expressions in the <samp>Rd</samp> file.
</p>
<p>The main argument to <code>\Sexpr</code> must be valid R code that can be
executed. It may also take options in square brackets before the main
argument. Depending on the options, the code may be executed at
package build time, package install time, or man page rendering time.
</p>
<p>The options follow the same format as in Sweave, but different options
are supported.  Currently the allowed options and their defaults are:
</p>
<ul>
<li> <code>eval=TRUE</code>
Whether the R code should be evaluated.

</li><li> <code>echo=FALSE</code>
Whether the R code should be echoed.  If <code>TRUE</code>, a display will
be given in a preformatted block.  For example,
<code>\Sexpr[echo=TRUE]{ x &lt;- 1 }</code> will be displayed as
<div class="example">
<pre class="example">&gt; x &lt;- 1
</pre></div>

</li><li> <code>keep.source=TRUE</code>
Whether to keep the author&rsquo;s formatting when displaying the
code, or throw it away and use a deparsed version.

</li><li> <code>results=text</code>
How should the results be displayed?  The possibilities
are:

<ul class="no-bullet">
<li>- <code>results=text</code>
Apply <code>as.character()</code> to the result of the code, and insert it
as a text element.

</li><li>- <code>results=verbatim</code>
Print the results of the code just as if it was executed at the console,
and include the printed results verbatim.  (Invisible results will not print.)

</li><li>- <code>results=rd</code>
The result is assumed to be a character vector containing markup to be
passed to <code>parse_Rd()</code>, with the result inserted in place.  This
could be used to insert computed aliases, for instance.
<code>parse_Rd()</code> is called first with <code>fragment = FALSE</code> to allow
a single Rd section macro to be inserted.  If that fails, it is called
again with <code>fragment = TRUE</code>, the older behavior.

</li><li>- <code>results=hide</code>
Insert no output.
</li></ul>

</li><li> <code>strip.white=TRUE</code>
Remove leading and trailing white space from each line of
output if <code>strip.white=TRUE</code>.  With
<code>strip.white=all</code>, also remove blank lines.

</li><li> <code>stage=install</code>
Control when this macro is run.  Possible values are
<ul class="no-bullet">
<li>- <code>stage=build</code>
The macro is run when building a source tarball.

</li><li>- <code>stage=install</code>
The macro is run when installing from source.

</li><li>- <code>stage=render</code>
The macro is run when displaying the help page.
</li></ul>

<p>Conditionals such as <code>#ifdef</code>
(see <a href="#Platform_002dspecific-sections">Platform-specific sections</a>) are applied after the
<code>build</code> macros but before the <code>install</code> macros.  In some
situations (e.g. installing directly from a source directory without a
tarball, or building a binary package) the above description is not
literally accurate, but authors can rely on the sequence being
<code>build</code>, <code>#ifdef</code>, <code>install</code>, <code>render</code>, with all
stages executed.
</p>
<p>Code is only run once in each stage, so a <code>\Sexpr[results=rd]</code>
macro can output an <code>\Sexpr</code> macro designed for a later stage,
but not for the current one or any earlier stage.
</p>
</li><li> <code>width, height, fig</code>
These options are currently allowed but ignored.
</li></ul>

<p>The <code>\RdOpts</code> macro is used to set new defaults for options to apply
to following uses of <code>\Sexpr</code>.
</p>
<p>For more details, see the online document
<a href="https://developer.r-project.org/parseRd.pdf">&ldquo;Parsing Rd files&rdquo;</a>.
</p>
<hr>
<a name="User_002ddefined-macros"></a>
<div class="header">
<p>
Next: <a href="#Encoding" accesskey="n" rel="next">Encoding</a>, Previous: <a href="#Dynamic-pages" accesskey="p" rel="prev">Dynamic pages</a>, Up: <a href="#Writing-R-documentation-files" accesskey="u" rel="up">Writing R documentation files</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="User_002ddefined-macros-1"></a>
<h3 class="section">2.13 User-defined macros</h3>
<a name="index-user_002ddefined-macros"></a>
<a name="index-_005cnewcommand"></a>
<a name="index-_005crenewcommand"></a>

<p>The <code>\newcommand</code> and <code>\renewcommand</code> macros allow new macros
to be defined within an Rd file.  These are similar but not identical to
the same-named LaTeX macros.
</p>
<p>They each take two arguments which are parsed verbatim.  The first is
the name of the new macro including the initial backslash, and the second
is the macro definition.  As in LaTeX, <code>\newcommand</code> requires that the
new macro not have been previously defined, whereas <code>\renewcommand</code>
allows existing macros (including all built-in ones) to be replaced.
(As from version 3.2.0, this test is disabled by default, but may
be enabled by setting the environment variable <code>_WARN_DUPLICATE_RD_MACROS_</code>
to a true value.)
</p>
<p>Also as in LaTeX, the new macro may be defined to take arguments,
and numeric placeholders such as <code>#1</code> are used in the macro
definition. However, unlike LaTeX, the number of arguments is
determined automatically from the highest placeholder number seen in
the macro definition.  For example, a macro definition containing
<code>#1</code> and <code>#3</code> (but no other placeholders) will define a
three argument macro (whose second argument will be ignored). As in
LaTeX, at most 9 arguments may be defined. If the <code>#</code>
character is followed by a non-digit it will have no special
significance.  All arguments to user-defined macros will be parsed as
verbatim text, and simple text-substitution will be used to replace
the place-holders, after which the replacement text will be parsed.
</p>
<p>As of R version 3.2.0, a number of macros are defined in the file
<samp>share/Rd/macros/system.Rd</samp> of the R source or home
directory, and these will normally be available in all <samp>.Rd</samp> files.
For example, that file contains the definition
</p><div class="example">
<pre class="example">\newcommand{\PR}{\Sexpr[results=rd]{tools:::Rd_expr_PR(#1)}}
</pre></div>
<p>which defines <code>\PR</code> to be a single argument macro; then code
(typically used in the <samp>NEWS.Rd</samp> file) like
</p><div class="example">
<pre class="example">\PR{1234}
</pre></div>
<p>will expand to
</p><div class="example">
<pre class="example">\Sexpr[results=rd]{tools:::Rd_expr_PR(1234)}
</pre></div>
<p>when parsed.  
</p>
<p>Some macros that might be of general use are:
</p><dl compact="compact">
<dt><code>\CRANpkg{<var>pkg</var>}</code>
<a name="index-_005cCRANpkg_007bpkg_007d"></a>
</dt>
<dd><p>A package on CRAN
</p>
</dd>
<dt><code>\sspace</code>
<a name="index-_005csspace"></a>
</dt>
<dd><p>A single space (used after a period that does not end a sentence).
</p>
</dd>
<dt><code>\doi{<var>numbers</var>}</code>
<a name="index-_005cdoi_007bnumbers_007d"></a>
</dt>
<dd><p>A digital object identifier (DOI).
</p></dd>
</dl>
<p>See the <samp>system.Rd</samp> file in <samp>share/Rd/macros</samp> for more details
and macro definitions, including macros <code>\packageTitle</code>,
<code>\packageDescription</code>, <code>\packageAuthor</code>, <code>\packageMaintainer</code>,
<code>\packageDESCRIPTION</code> and <code>\packageIndices</code>.
<a name="index-_005cpackageTitle"></a>
<a name="index-_005cpackageDescription"></a>
<a name="index-_005cpackageAuthor"></a>
<a name="index-_005cpackageMaintainer"></a>
<a name="index-_005cpackageDESCRIPTION"></a>
<a name="index-_005cpackageIndices"></a>
</p>

<p>Packages may also define their own common macros; these would be stored
in an <samp>.Rd</samp> file in <samp>man/macros</samp> in the package source and
will be installed into <samp>help/macros</samp> when the package is installed.
A package may also use the macros from a different package by listing
the other package in the &lsquo;<samp>RdMacros</samp>&rsquo; field in the <samp>DESCRIPTION</samp>
file.
</p>


<hr>
<a name="Encoding"></a>
<div class="header">
<p>
Next: <a href="#Processing-documentation-files" accesskey="n" rel="next">Processing documentation files</a>, Previous: <a href="#User_002ddefined-macros" accesskey="p" rel="prev">User-defined macros</a>, Up: <a href="#Writing-R-documentation-files" accesskey="u" rel="up">Writing R documentation files</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Encoding-1"></a>
<h3 class="section">2.14 Encoding</h3>
<a name="index-encoding"></a>

<p>Rd files are text files and so it is impossible to deduce the encoding
they are written in unless <acronym>ASCII</acronym>: files with 8-bit characters
could be UTF-8, Latin-1, Latin-9, KOI8-R, EUC-JP, <em>etc</em>.  So an
<code>\encoding{}</code> section must be used to specify the encoding if it
is not <acronym>ASCII</acronym>.  (The <code>\encoding{}</code> section must be on a
line by itself, and in particular one containing no non-<acronym>ASCII</acronym>
characters.  The encoding declared in the <samp>DESCRIPTION</samp> file will
be used if none is declared in the file.)  The <samp>Rd</samp> files are
converted to UTF-8 before parsing and so the preferred encoding for the
files themselves is now UTF-8.
</p>
<p>Wherever possible, avoid non-<acronym>ASCII</acronym> chars in <samp>Rd</samp> files, and
even symbols such as &lsquo;<samp>&lt;</samp>&rsquo;, &lsquo;<samp>&gt;</samp>&rsquo;, &lsquo;<samp>$</samp>&rsquo;, &lsquo;<samp>^</samp>&rsquo;, &lsquo;<samp>&amp;</samp>&rsquo;,
&lsquo;<samp>|</samp>&rsquo;, &lsquo;<samp>@</samp>&rsquo;, &lsquo;<samp>~</samp>&rsquo;, and &lsquo;<samp>*</samp>&rsquo; outside &lsquo;verbatim&rsquo;
environments (since they may disappear in fonts designed to render
text).  (Function <code>showNonASCIIfile</code> in package <strong>tools</strong> can help
in finding non-<acronym>ASCII</acronym> bytes in the files.)
</p>
<p>For convenience, encoding names &lsquo;<samp>latin1</samp>&rsquo; and &lsquo;<samp>latin2</samp>&rsquo; are
always recognized: these and &lsquo;<samp>UTF-8</samp>&rsquo; are likely to work fairly
widely.  However, this does not mean that all characters in UTF-8 will
be recognized, and the coverage of non-Latin characters<a name="DOCF100" href="#FOOT100"><sup>100</sup></a> is fairly low.  Using LaTeX
<code>inputenx</code> (see <code>?Rd2pdf</code> in R) will give greater coverage
of UTF-8.
</p>
<p>The <code>\enc</code> command (see <a href="#Insertions">Insertions</a>) can be used to provide
transliterations which will be used in conversions that do not support
the declared encoding.
</p>
<p>The LaTeX conversion converts the file to UTF-8 from the declared
encoding, and includes a
</p>
<div class="example">
<pre class="example">\inputencoding{utf8}
</pre></div>

<p>command, and this needs to be matched by a suitable invocation of the
<code>\usepackage{inputenc}</code> command.  The R utility <code>R
CMD Rd2pdf</code> looks at the converted code and includes the encodings used:
it might for example use
</p>
<div class="example">
<pre class="example">\usepackage[utf8]{inputenc}
</pre></div>

<p>(Use of <code>utf8</code> as an encoding requires LaTeX dated 2003/12/01 or
later.  Also, the use of Cyrillic characters in &lsquo;<samp>UTF-8</samp>&rsquo; appears to
also need &lsquo;<samp>\usepackage[T2A]{fontenc}</samp>&rsquo;, and <code>R CMD Rd2pdf</code>
includes this conditionally on the file <samp>t2aenc.def</samp> being present
and environment variable <code>_R_CYRILLIC_TEX_</code> being set.)
</p>
<p>Note that this mechanism works best with Latin letters: the coverage of
UTF-8 in LaTeX is quite low.
</p>


<hr>
<a name="Processing-documentation-files"></a>
<div class="header">
<p>
Next: <a href="#Editing-Rd-files" accesskey="n" rel="next">Editing Rd files</a>, Previous: <a href="#Encoding" accesskey="p" rel="prev">Encoding</a>, Up: <a href="#Writing-R-documentation-files" accesskey="u" rel="up">Writing R documentation files</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Processing-documentation-files-1"></a>
<h3 class="section">2.15 Processing documentation files</h3>
<a name="index-Processing-Rd-format"></a>

<p>There are several commands to process Rd files from the system command
line.
</p>
<a name="index-R-CMD-Rdconv"></a>
<p>Using <code>R CMD Rdconv</code> one can convert R documentation format to
other formats, or extract the executable examples for run-time testing.
The currently supported conversions are to plain text, <acronym>HTML</acronym> and
LaTeX as well as extraction of the examples.
</p>
<a name="index-R-CMD-Rd2pdf"></a>
<p><code>R CMD Rd2pdf</code> generates PDF output from documentation in <samp>Rd</samp>
files, which can be specified either explicitly or by the path to a
directory with the sources of a package.  In the latter case, a
reference manual for all documented objects in the package is created,
including the information in the <samp>DESCRIPTION</samp> files.
</p>
<a name="index-R-CMD-Sweave"></a>
<a name="index-R-CMD-Stangle"></a>
<p><code>R CMD Sweave</code> and <code>R CMD Stangle</code> process vignette-like
documentation files (e.g. Sweave vignettes with extension
&lsquo;<samp>.Snw</samp>&rsquo; or &lsquo;<samp>.Rnw</samp>&rsquo;, or other non-Sweave vignettes).
<code>R CMD Stangle</code> is used to extract the R code fragments.
</p>
<p>The exact usage and a detailed list of available options for all of
these commands can be obtained by running <code>R CMD <var>command</var>
--help</code>, e.g., <kbd>R CMD Rdconv --help</kbd>.  All available commands can be
listed using <kbd>R --help</kbd> (or <kbd>Rcmd --help</kbd> under Windows).
</p>
<p>All of these work under Windows.  You may need to have installed the
the tools to build packages from source as described in the &ldquo;R
Installation and Administration&rdquo; manual, although typically all that is
needed is a LaTeX installation.
</p>
<hr>
<a name="Editing-Rd-files"></a>
<div class="header">
<p>
Previous: <a href="#Processing-documentation-files" accesskey="p" rel="prev">Processing documentation files</a>, Up: <a href="#Writing-R-documentation-files" accesskey="u" rel="up">Writing R documentation files</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Editing-Rd-files-1"></a>
<h3 class="section">2.16 Editing Rd files</h3>
<a name="index-Editing-Rd-files"></a>

<p>It can be very helpful to prepare <samp>.Rd</samp> files using a editor which
knows about their syntax and will highlight commands, indent to show the
structure and detect mis-matched braces, and so on.
</p>
<p>The system most commonly used for this is some version of
<code>Emacs</code> (including <code>XEmacs</code>) with the <acronym>ESS</acronym>
package (<a href="https://ESS.R-project.org/">https://ESS.R-project.org/</a>: it is often is installed with
<code>Emacs</code> but may need to be loaded, or even installed,
separately).
</p>
<p>Another is the Eclipse IDE with the Stat-ET plugin
(<a href="http://www.walware.de/goto/statet">http://www.walware.de/goto/statet</a>), and (on Windows only)
Tinn-R (<a href="http://sourceforge.net/projects/tinn-r/">http://sourceforge.net/projects/tinn-r/</a>).
</p>
<p>People have also used LaTeX mode in a editor, as <samp>.Rd</samp> files are
rather similar to LaTeX files.
</p>
<p>Some R front-ends provide editing support for  <samp>.Rd</samp> files, for
example RStudio (<a href="https://rstudio.org/">https://rstudio.org/</a>).
</p>
<hr>
<a name="Tidying-and-profiling-R-code"></a>
<div class="header">
<p>
Next: <a href="#Debugging" accesskey="n" rel="next">Debugging</a>, Previous: <a href="#Writing-R-documentation-files" accesskey="p" rel="prev">Writing R documentation files</a>, Up: <a href="#Top" accesskey="u" rel="up">Top</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Tidying-and-profiling-R-code-1"></a>
<h2 class="chapter">3 Tidying and profiling R code</h2>

<table summary="" class="menu" border="0" cellspacing="0">
<tr><td align="left" valign="top">&bull; <a href="#Tidying-R-code" accesskey="1">Tidying R code</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Profiling-R-code-for-speed" accesskey="2">Profiling R code for speed</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Profiling-R-code-for-memory-use" accesskey="3">Profiling R code for memory use</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Profiling-compiled-code" accesskey="4">Profiling compiled code</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
</table>

<p>R code which is worth preserving in a package and perhaps making
available for others to use is worth documenting, tidying up and perhaps
optimizing. The last two of these activities are the subject of this
chapter.
</p>
<hr>
<a name="Tidying-R-code"></a>
<div class="header">
<p>
Next: <a href="#Profiling-R-code-for-speed" accesskey="n" rel="next">Profiling R code for speed</a>, Previous: <a href="#Tidying-and-profiling-R-code" accesskey="p" rel="prev">Tidying and profiling R code</a>, Up: <a href="#Tidying-and-profiling-R-code" accesskey="u" rel="up">Tidying and profiling R code</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Tidying-R-code-1"></a>
<h3 class="section">3.1 Tidying R code</h3>
<a name="index-Tidying-R-code"></a>

<p>R treats function code loaded from packages and code entered by users
differently.  By default code entered by users has the source code stored
internally, and when the function is listed, the original source is
reproduced.  Loading code from a package (by default) discards the
source code, and the function listing is re-created from the parse tree
of the function.
</p>
<p>Normally keeping the source code is a good idea, and in particular it
avoids comments being removed from the source.  However, we can make
use of the ability to re-create a function listing from its parse tree
to produce a tidy version of the function, for example with consistent
indentation and spaces around operators.  If the original source
does not follow the standard format this tidied version can be much
easier to read.
</p>
<p>We can subvert the keeping of source in two ways.
</p>
<ol>
<li> The option <code>keep.source</code> can be set to <code>FALSE</code> before the code
is loaded into R.
</li><li> The stored source code can be removed by calling the <code>removeSource()</code>
function, for example by

<div class="example">
<pre class="example">myfun &lt;- removeSource(myfun)
</pre></div>

</li></ol>

<p>In each case if we then list the function we will get the standard
layout.
</p>
<p>Suppose we have a file of functions <samp>myfuns.R</samp> that we want to
tidy up.  Create a file <samp>tidy.R</samp> containing
</p>
<div class="example">
<pre class="example">source(&quot;myfuns.R&quot;, keep.source = FALSE)
dump(ls(all = TRUE), file = &quot;new.myfuns.R&quot;)
</pre></div>

<p>and run R with this as the source file, for example by <kbd>R
--vanilla &lt; tidy.R</kbd> or by pasting into an R session.  Then the file
<samp>new.myfuns.R</samp> will contain the functions in alphabetical order in
the standard layout.  Warning:  comments in your functions will be lost.
</p>
<p>The standard format provides a good starting point for further tidying.
Although the deparsing cannot do so, we recommend the consistent use of
the preferred assignment operator &lsquo;<samp>&lt;-</samp>&rsquo; (rather than &lsquo;<samp>=</samp>&rsquo;) for
assignment.  Many package authors use a version of Emacs (on a
Unix-alike or Windows) to edit R code, using the ESS[S] mode of the
<acronym>ESS</acronym> Emacs package.  See <a href="http://cran.r-project.org/doc/manuals/R-ints.html#R-coding-standards">R coding
standards</a> in <cite>R Internals</cite> for style options within the ESS[S] mode
recommended for the source code of R itself.
</p>


<hr>
<a name="Profiling-R-code-for-speed"></a>
<div class="header">
<p>
Next: <a href="#Profiling-R-code-for-memory-use" accesskey="n" rel="next">Profiling R code for memory use</a>, Previous: <a href="#Tidying-R-code" accesskey="p" rel="prev">Tidying R code</a>, Up: <a href="#Tidying-and-profiling-R-code" accesskey="u" rel="up">Tidying and profiling R code</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Profiling-R-code-for-speed-1"></a>
<h3 class="section">3.2 Profiling R code for speed</h3>
<a name="index-Profiling"></a>
<a name="index-Rprof"></a>

<p>It is possible to profile R code on Windows and most<a name="DOCF101" href="#FOOT101"><sup>101</sup></a> Unix-alike versions of
R.
</p>
<p>The command <code>Rprof</code> is used to control profiling, and its help
page can be consulted for full details.  Profiling works by recording
at fixed intervals<a name="DOCF102" href="#FOOT102"><sup>102</sup></a> (by default every 20 msecs)
which line in which R function is being used, and recording the
results in a file (default <samp>Rprof.out</samp> in the working directory).
Then the function <code>summaryRprof</code> or the command-line utility
<code>R CMD Rprof <var>Rprof.out</var></code> can be used to summarize the
activity.
</p>
<p>As an example, consider the following code (from Venables &amp; Ripley,
2002, pp. 225&ndash;6).
</p>
<div class="smallexample">
<pre class="smallexample">library(MASS); library(boot)
storm.fm &lt;- nls(Time ~ b*Viscosity/(Wt - c), stormer,
		start = c(b=30.401, c=2.2183))
st &lt;- cbind(stormer, fit=fitted(storm.fm))
storm.bf &lt;- function(rs, i) {
    st$Time &lt;-  st$fit + rs[i]
    tmp &lt;- nls(Time ~ (b * Viscosity)/(Wt - c), st,
	       start = coef(storm.fm))
    tmp$m$getAllPars()
}
rs &lt;- scale(resid(storm.fm), scale = FALSE) # remove the mean
Rprof(&quot;boot.out&quot;)
storm.boot &lt;- boot(rs, storm.bf, R = 4999) # slow enough to profile
Rprof(NULL)
</pre></div>

<p>Having run this we can summarize the results by
</p>
<div class="smallexample">
<pre class="smallexample">R CMD Rprof boot.out

Each sample represents 0.02 seconds.
Total run time: 22.52 seconds.

Total seconds: time spent in function and callees.
Self seconds: time spent in function alone.
</pre><pre class="smallexample">

</pre><pre class="smallexample">   %       total       %        self
 total    seconds     self    seconds    name
 100.0     25.22       0.2      0.04     &quot;boot&quot;
  99.8     25.18       0.6      0.16     &quot;statistic&quot;
  96.3     24.30       4.0      1.02     &quot;nls&quot;
  33.9      8.56       2.2      0.56     &quot;&lt;Anonymous&gt;&quot;
  32.4      8.18       1.4      0.36     &quot;eval&quot;
  31.8      8.02       1.4      0.34     &quot;.Call&quot;
  28.6      7.22       0.0      0.00     &quot;eval.parent&quot;
  28.5      7.18       0.3      0.08     &quot;model.frame&quot;
  28.1      7.10       3.5      0.88     &quot;model.frame.default&quot;
  17.4      4.38       0.7      0.18     &quot;sapply&quot;
  15.0      3.78       3.2      0.80     &quot;nlsModel&quot;
  12.5      3.16       1.8      0.46     &quot;lapply&quot;
  12.3      3.10       2.7      0.68     &quot;assign&quot;
 ...
</pre><pre class="smallexample">

</pre><pre class="smallexample">   %        self        %      total
  self    seconds     total   seconds    name
   5.7      1.44       7.5      1.88     &quot;inherits&quot;
   4.0      1.02      96.3     24.30     &quot;nls&quot;
   3.6      0.92       3.6      0.92     &quot;$&quot;
   3.5      0.88      28.1      7.10     &quot;model.frame.default&quot;
   3.2      0.80      15.0      3.78     &quot;nlsModel&quot;
   2.8      0.70       9.8      2.46     &quot;qr.coef&quot;
   2.7      0.68      12.3      3.10     &quot;assign&quot;
   2.5      0.64       2.5      0.64     &quot;.Fortran&quot;
   2.5      0.62       7.1      1.80     &quot;qr.default&quot;
   2.2      0.56      33.9      8.56     &quot;&lt;Anonymous&gt;&quot;
   2.1      0.54       5.9      1.48     &quot;unlist&quot;
   2.1      0.52       7.9      2.00     &quot;FUN&quot;
  ...
</pre></div>

<p>This often produces
surprising results and can be used to identify bottlenecks or pieces of
R code that could benefit from being replaced by compiled code.
</p>
<p>Two warnings: profiling does impose a small performance penalty, and the
output files can be very large if long runs are profiled at the default
sampling interval.
</p>
<p>Profiling short runs can sometimes give misleading results.  R from
time to time performs <em>garbage collection</em> to reclaim unused
memory, and this takes an appreciable amount of time which profiling
will charge to whichever function happens to provoke it.  It may be
useful to compare profiling code immediately after a call to <code>gc()</code>
with a profiling run without a preceding call to <code>gc</code>.
</p>
<p>More detailed analysis of the output can be achieved by the tools in the
<acronym>CRAN</acronym> packages <a href="https://CRAN.R-project.org/package=proftools"><strong>proftools</strong></a> and <a href="https://CRAN.R-project.org/package=profr"><strong>profr</strong></a>: in
particular these allow call graphs to be studied.
</p>
<hr>
<a name="Profiling-R-code-for-memory-use"></a>
<div class="header">
<p>
Next: <a href="#Profiling-compiled-code" accesskey="n" rel="next">Profiling compiled code</a>, Previous: <a href="#Profiling-R-code-for-speed" accesskey="p" rel="prev">Profiling R code for speed</a>, Up: <a href="#Tidying-and-profiling-R-code" accesskey="u" rel="up">Tidying and profiling R code</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Profiling-R-code-for-memory-use-1"></a>
<h3 class="section">3.3 Profiling R code for memory use</h3>
<a name="index-Profiling-1"></a>
<a name="index-Memory-use"></a>

<p>Measuring memory use in R code is useful either when the code takes
more memory than is conveniently available or when memory allocation and
copying of objects is responsible for slow code. There are three ways to
profile memory use over time in R code. All three require R to
have been compiled with <samp>--enable-memory-profiling</samp>, which is not
the default, but is currently used for the macOS and Windows binary
distributions. All can be misleading, for different reasons.
</p>
<p>In understanding the memory profiles it is useful to know a little more
about R&rsquo;s memory allocation. Looking at the results of <code>gc()</code>
shows a division of memory into <code>Vcells</code> used to store the contents
of vectors and <code>Ncells</code> used to store everything else, including
all the administrative overhead for vectors such as type and length
information.  In fact the vector contents are divided into two
pools. Memory for small vectors (by default 128 bytes or less) is
obtained in large chunks and then parcelled out by R; memory for
larger vectors is obtained directly from the operating system.
</p>
<p>Some memory allocation is obvious in interpreted code, for example,
</p>
<div class="smallexample">
<pre class="smallexample">y &lt;- x + 1
</pre></div>

<p>allocates memory for a new vector <code>y</code>. Other memory allocation is
less obvious and occurs because <code>R</code> is forced to make good on its
promise of &lsquo;call-by-value&rsquo; argument passing.  When an argument is
passed to a function it is not immediately copied. Copying occurs (if
necessary) only when the argument is modified.  This can lead to
surprising memory use. For example, in the &lsquo;survey&rsquo; package we have
</p>
<div class="smallexample">
<pre class="smallexample">print.svycoxph &lt;- function (x, ...)
{
    print(x$survey.design, varnames = FALSE, design.summaries = FALSE, ...)
    x$call &lt;- x$printcall
    NextMethod()
}
</pre></div>

<p>It may not be obvious that the assignment to <code>x$call</code> will cause
the entire object <code>x</code> to be copied.  This copying to preserve the
call-by-value illusion is usually done by the internal C function
<code>duplicate</code>.
</p>
<p>The main reason that memory-use profiling is difficult is garbage
collection. Memory is allocated at well-defined times in an R
program, but is freed whenever the garbage collector happens to run.
</p>
<table summary="" class="menu" border="0" cellspacing="0">
<tr><td align="left" valign="top">&bull; <a href="#Memory-statistics-from-Rprof" accesskey="1">Memory statistics from Rprof</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Tracking-memory-allocations" accesskey="2">Tracking memory allocations</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Tracing-copies-of-an-object" accesskey="3">Tracing copies of an object</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
</table>

<hr>
<a name="Memory-statistics-from-Rprof"></a>
<div class="header">
<p>
Next: <a href="#Tracking-memory-allocations" accesskey="n" rel="next">Tracking memory allocations</a>, Previous: <a href="#Profiling-R-code-for-memory-use" accesskey="p" rel="prev">Profiling R code for memory use</a>, Up: <a href="#Profiling-R-code-for-memory-use" accesskey="u" rel="up">Profiling R code for memory use</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Memory-statistics-from-Rprof-1"></a>
<h4 class="subsection">3.3.1 Memory statistics from <code>Rprof</code></h4>
<a name="index-Rprof-1"></a>
<a name="index-summaryRprof"></a>

<p>The sampling profiler <code>Rprof</code> described in the previous section can
be given the option <code>memory.profiling=TRUE</code>. It then writes out the
total R memory allocation in small vectors, large vectors, and cons
cells or nodes at each sampling interval. It also writes out the number
of calls to the internal function <code>duplicate</code>, which is called to
copy R objects. <code>summaryRprof</code> provides summaries of this
information.  The main reason that this can be misleading is that the
memory use is attributed to the function running at the end of the
sampling interval. A second reason is that garbage collection can make
the amount of memory in use decrease, so a function appears to use
little memory.  Running under <code>gctorture</code> helps with both problems:
it slows down the code to effectively increase the sampling frequency
and it makes each garbage collection release a smaller amount of memory.
Changing the memory limits with <code>mem.limits()</code> may also be useful,
to see how the code would run under different memory conditions.
</p>
<hr>
<a name="Tracking-memory-allocations"></a>
<div class="header">
<p>
Next: <a href="#Tracing-copies-of-an-object" accesskey="n" rel="next">Tracing copies of an object</a>, Previous: <a href="#Memory-statistics-from-Rprof" accesskey="p" rel="prev">Memory statistics from Rprof</a>, Up: <a href="#Profiling-R-code-for-memory-use" accesskey="u" rel="up">Profiling R code for memory use</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Tracking-memory-allocations-1"></a>
<h4 class="subsection">3.3.2 Tracking memory allocations</h4>
<a name="index-Rprofmem"></a>

<p>The second method of memory profiling uses a memory-allocation
profiler, <code>Rprofmem()</code>, which writes out a stack trace to an
output file every time a large vector is allocated (with a
user-specified threshold for &lsquo;large&rsquo;) or a new page of memory is
allocated for the R heap. Summary functions for this output are still
being designed.
</p>
<p>Running the example from the previous section with
</p>
<div class="smallexample">
<pre class="smallexample">&gt; Rprofmem(&quot;boot.memprof&quot;,threshold=1000)
&gt; storm.boot &lt;- boot(rs, storm.bf, R = 4999)
&gt; Rprofmem(NULL)
</pre></div>

<p>shows that apart from some initial and final work in <code>boot</code> there
are no vector allocations over 1000 bytes.
</p>
<hr>
<a name="Tracing-copies-of-an-object"></a>
<div class="header">
<p>
Previous: <a href="#Tracking-memory-allocations" accesskey="p" rel="prev">Tracking memory allocations</a>, Up: <a href="#Profiling-R-code-for-memory-use" accesskey="u" rel="up">Profiling R code for memory use</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Tracing-copies-of-an-object-1"></a>
<h4 class="subsection">3.3.3 Tracing copies of an object</h4>
<a name="index-tracemem"></a>
<a name="index-untracemem"></a>

<p>The third method of memory profiling involves tracing copies made of a
specific (presumably large) R object. Calling <code>tracemem</code> on an
object marks it so that a message is printed to standard output when
the object is copied <em>via</em> <code>duplicate</code> or coercion to another type,
or when a new object of the same size is created in arithmetic
operations. The main reason that this can be misleading is that
copying of subsets or components of an object is not tracked. It may
be helpful to use <code>tracemem</code> on these components.
</p>

<p>In the example above we can run <code>tracemem</code> on the data frame
<code>st</code>
</p>
<div class="smallexample">
<pre class="smallexample">&gt; tracemem(st)
[1] &quot;&lt;0x9abd5e0&gt;&quot;
&gt; storm.boot &lt;- boot(rs, storm.bf, R = 4)
memtrace[0x9abd5e0-&gt;0x92a6d08]: statistic boot
memtrace[0x92a6d08-&gt;0x92a6d80]: $&lt;-.data.frame $&lt;- statistic boot
memtrace[0x92a6d80-&gt;0x92a6df8]: $&lt;-.data.frame $&lt;- statistic boot
memtrace[0x9abd5e0-&gt;0x9271318]: statistic boot
memtrace[0x9271318-&gt;0x9271390]: $&lt;-.data.frame $&lt;- statistic boot
memtrace[0x9271390-&gt;0x9271408]: $&lt;-.data.frame $&lt;- statistic boot
memtrace[0x9abd5e0-&gt;0x914f558]: statistic boot
memtrace[0x914f558-&gt;0x914f5f8]: $&lt;-.data.frame $&lt;- statistic boot
memtrace[0x914f5f8-&gt;0x914f670]: $&lt;-.data.frame $&lt;- statistic boot
memtrace[0x9abd5e0-&gt;0x972cbf0]: statistic boot
memtrace[0x972cbf0-&gt;0x972cc68]: $&lt;-.data.frame $&lt;- statistic boot
memtrace[0x972cc68-&gt;0x972cd08]: $&lt;-.data.frame $&lt;- statistic boot
memtrace[0x9abd5e0-&gt;0x98ead98]: statistic boot
memtrace[0x98ead98-&gt;0x98eae10]: $&lt;-.data.frame $&lt;- statistic boot
memtrace[0x98eae10-&gt;0x98eae88]: $&lt;-.data.frame $&lt;- statistic boot
</pre></div>

<p>The object is duplicated fifteen times, three times for each of the
<code>R+1</code> calls to <code>storm.bf</code>.  This is surprising, since none of the duplications happen inside <code>nls</code>. Stepping through <code>storm.bf</code> in the debugger shows that all three happen in the line
</p>
<div class="smallexample">
<pre class="smallexample">st$Time &lt;- st$fit + rs[i]
</pre></div>

<p>Data frames are slower than matrices and this is an example of why.
Using <code>tracemem(st$Viscosity)</code> does not reveal any additional
copying.
</p>
<hr>
<a name="Profiling-compiled-code"></a>
<div class="header">
<p>
Previous: <a href="#Profiling-R-code-for-memory-use" accesskey="p" rel="prev">Profiling R code for memory use</a>, Up: <a href="#Tidying-and-profiling-R-code" accesskey="u" rel="up">Tidying and profiling R code</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Profiling-compiled-code-1"></a>
<h3 class="section">3.4 Profiling compiled code</h3>
<a name="index-Profiling-2"></a>

<p>Profiling compiled code is highly system-specific, but this section
contains some hints gleaned from various R users.  Some methods need
to be different for a compiled executable and for dynamic/shared
libraries/objects as used by R packages.  We know of no good way to
profile DLLs on Windows.
</p>
<table summary="" class="menu" border="0" cellspacing="0">
<tr><td align="left" valign="top">&bull; <a href="#Linux" accesskey="1">Linux</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Solaris" accesskey="2">Solaris</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#macOS" accesskey="3">macOS</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
</table>

<hr>
<a name="Linux"></a>
<div class="header">
<p>
Next: <a href="#Solaris" accesskey="n" rel="next">Solaris</a>, Previous: <a href="#Profiling-compiled-code" accesskey="p" rel="prev">Profiling compiled code</a>, Up: <a href="#Profiling-compiled-code" accesskey="u" rel="up">Profiling compiled code</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Linux-1"></a>
<h4 class="subsection">3.4.1 Linux</h4>

<p>Options include using <code>sprof</code> for a shared object, and
<code>oprofile</code> (see <a href="http://oprofile.sourceforge.net/">http://oprofile.sourceforge.net/</a>) and
<code>perf</code> (see
<a href="https://perf.wiki.kernel.org/index.php/Tutorial">https://perf.wiki.kernel.org/index.php/Tutorial</a>) for any
executable or shared object.
</p>
<a name="sprof"></a>
<h4 class="subsubsection">3.4.1.1 sprof</h4>

<p>You can select shared objects to be profiled with <code>sprof</code> by
setting the environment variable <code>LD_PROFILE</code>.  For example
</p>
<div class="example">
<pre class="example">% setenv LD_PROFILE /path/to/R_HOME/library/stats/libs/stats.so
R
... run the boot example
% sprof /path/to/R_HOME/library/stats/libs/stats.so \
  /var/tmp/path/to/R_HOME/library/stats/libs/stats.so.profile

Flat profile:

Each sample counts as 0.01 seconds.
  %   cumulative   self              self     total
 time   seconds   seconds    calls  us/call  us/call  name
 76.19      0.32     0.32        0     0.00           numeric_deriv
 16.67      0.39     0.07        0     0.00           nls_iter
  7.14      0.42     0.03        0     0.00           getListElement

rm /var/tmp/path/to/R_HOME/library/stats/libs/stats.so.profile
... to clean up ...
</pre></div>

<p>It is possible that root access is needed to create the directories used
for the profile data.
</p>
<a name="oprofile-and-operf"></a>
<h4 class="subsubsection">3.4.1.2 oprofile and operf</h4>

<p>The <code>oprofile</code> project has two modes of operation.  In what is
now called &lsquo;legacy&rsquo; mode, it is uses a daemon to collect information on
a process (see below).  Since version 0.9.8 (August 2012), the preferred
mode is to use <code>operf</code>, so we discuss that first.  The modes
differ in how the profiling data is collected: it is analysed by tools
such as <code>opreport</code> and <code>oppannote</code> in both.
</p>
<p>Here is an example on <code>x86_64</code> Linux using R 3.0.2.  File
<samp>pvec.R</samp> contains the part of the examples from <code>pvec</code> in
package <strong>parallel</strong>:
</p><div class="example">
<pre class="example">library(parallel)
N &lt;- 1e6
dates &lt;- sprintf('%04d-%02d-%02d', as.integer(2000+rnorm(N)),
                 as.integer(runif(N, 1, 12)), as.integer(runif(N, 1, 28)))
system.time(a &lt;- as.POSIXct(dates, format = &quot;%Y-%m-%d&quot;))
</pre></div>
<p>with timings from the final step
</p><div class="example">
<pre class="example">   user  system elapsed
  0.371   0.237   0.612
</pre></div>

<p>R-level profiling by <code>Rprof</code> shows
</p><div class="example">
<pre class="example">                     self.time self.pct total.time total.pct
&quot;strptime&quot;                1.70    41.06       1.70     41.06
&quot;as.POSIXct.POSIXlt&quot;      1.40    33.82       1.42     34.30
&quot;sprintf&quot;                 0.74    17.87       0.98     23.67
...
</pre></div>
<p>so the conversion from character to <code>POSIXlt</code> takes most of the
time.
</p>
<p>This can be run under <code>operf</code> and analysed by
</p><div class="example">
<pre class="example">operf R -f pvec.R
opreport
opreport -l /path/to/R_HOME/bin/exec/R
opannotate --source /path/to/R_HOME/bin/exec/R
## And for the system time
opreport -l /lib64/libc.so.6
</pre></div>
<p>The first report shows where (which library etc) the time was spent:
</p><div class="example">
<pre class="example">CPU_CLK_UNHALT...|
  samples|      %|
------------------
   166761 99.9161 Rdev
	CPU_CLK_UNHALT...|
	  samples|      %|
	------------------
	    70586 42.3276 no-vmlinux
	    56963 34.1585 libc-2.16.so
	    36922 22.1407 R
	     1584  0.9499 stats.so
	      624  0.3742 libm-2.16.so
...
</pre></div>

<p>The rest of the output is voluminous, and only extracts are shown below.
</p>
<p>Most of the time within R is spent in
</p><div class="example">
<pre class="example">samples  %        image name symbol name
10397    28.5123  R           R_gc_internal
5683     15.5848  R           do_sprintf
3036      8.3258  R           do_asPOSIXct
2427      6.6557  R           do_strptime
2421      6.6392  R           Rf_mkCharLenCE
1480      4.0587  R           w_strptime_internal
1202      3.2963  R           Rf_qnorm5
1165      3.1948  R           unif_rand
675       1.8511  R           mktime0
617       1.6920  R           makelt
617       1.6920  R           validate_tm
584       1.6015  R           day_of_the_week
...
</pre></div>
<p><code>opannotate</code> shows that 31% of the time in R is spent in
<samp>memory.c</samp>, 21% in <samp>datetime.c</samp> and 7% in <samp>Rstrptime.h</samp>.
The analysis for <samp>libc</samp> showed that calls to <code>wcsftime</code>
dominated, so those calls were cached for R 3.0.3: the time spent in
<code>no-vmlinux</code> (the kernel) was reduced dramatically.
</p>
<p>On platforms which support it, call graphs can be produced by
<code>opcontrol --callgraph</code> if collected via <code>operf
--callgraph</code>.
</p>
<p>The profiling data is by default stored in sub-directory
<samp>oprofile_data</samp> of the current directory, which can be removed at
the end of the session.
</p>
<p>Another example, from <a href="https://CRAN.R-project.org/package=sm"><strong>sm</strong></a> version 2.2-5.4.  The example for
<code>sm.variogram</code> took a long time:
</p><div class="example">
<pre class="example">system.time(example(sm.variogram))
...
   user  system elapsed
  5.543   3.202   8.785
</pre></div>

<p>including a lot of system time.  Profiling just the slow part, the
second plot, showed
</p>
<div class="example">
<pre class="example">  samples|      %|
------------------
   381845 99.9885 R
	CPU_CLK_UNHALT...|
	  samples|      %|
	------------------
	   187484 49.0995 sm.so
	   169627 44.4230 no-vmlinux
	    12636  3.3092 libgfortran.so.3.0.0
	     6455  1.6905 R
</pre></div>

<p>so the system time was almost all in the Linux kernel.  It is possible
to dig deeper if you have a matching uncompressed kernel with debug
symbols to specify <em>via</em> <samp>--vmlinux</samp>: we did not.
</p>
<p>In &lsquo;legacy&rsquo; mode <code>oprofile</code> works by running a daemon which
collects information.  The daemon must be started as root, e.g.
</p>
<div class="example">
<pre class="example">% su
% opcontrol --no-vmlinux
% (optional, some platforms) opcontrol --callgraph=5
% opcontrol --start
% exit
</pre></div>

<p>Then as a user
</p>
<div class="example">
<pre class="example">% R
... run the boot example
% opcontrol --dump
% opreport -l /path/to/R_HOME/library/stats/libs/stats.so
...
samples  %        symbol name
1623     75.5939  anonymous symbol from section .plt
349      16.2552  numeric_deriv
113       5.2632  nls_iter
62        2.8878  getListElement
% opreport -l /path/to/R_HOME/bin/exec/R
...
samples  %        symbol name
76052    11.9912  Rf_eval
54670     8.6198  Rf_findVarInFrame3
37814     5.9622  Rf_allocVector
31489     4.9649  Rf_duplicate
28221     4.4496  Rf_protect
26485     4.1759  Rf_cons
23650     3.7289  Rf_matchArgs
21088     3.3250  Rf_findFun
19995     3.1526  findVarLocInFrame
14871     2.3447  Rf_evalList
13794     2.1749  R_Newhashpjw
13522     2.1320  R_gc_internal
...
</pre></div>

<p>Shutting down the profiler and clearing the records needs to be done as
root.
</p>

<hr>
<a name="Solaris"></a>
<div class="header">
<p>
Next: <a href="#macOS" accesskey="n" rel="next">macOS</a>, Previous: <a href="#Linux" accesskey="p" rel="prev">Linux</a>, Up: <a href="#Profiling-compiled-code" accesskey="u" rel="up">Profiling compiled code</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Solaris-1"></a>
<h4 class="subsection">3.4.2 Solaris</h4>

<p>On 64-bit (only) Solaris, the standard profiling tool <code>gprof</code>
collects information from shared objects compiled with <samp>-pg</samp>.
</p>
<hr>
<a name="macOS"></a>
<div class="header">
<p>
Previous: <a href="#Solaris" accesskey="p" rel="prev">Solaris</a>, Up: <a href="#Profiling-compiled-code" accesskey="u" rel="up">Profiling compiled code</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="macOS-1"></a>
<h4 class="subsection">3.4.3 macOS</h4>

<p>Developers have recommended <code>sample</code> (or <code>Sampler.app</code>,
which is a GUI version), <code>Shark</code> (in version of <code>Xcode</code>
up to those for Snow Leopard), and <code>Instruments</code> (part of
<code>Xcode</code>, see
<a href="https://developer.apple.com/library/mac/#documentation/DeveloperTools/Conceptual/InstrumentsUserGuide/Introduction/Introduction.html">https://developer.apple.com/library/mac/#documentation/DeveloperTools/Conceptual/InstrumentsUserGuide/Introduction/Introduction.html</a>).
</p>

<hr>
<a name="Debugging"></a>
<div class="header">
<p>
Next: <a href="#System-and-foreign-language-interfaces" accesskey="n" rel="next">System and foreign language interfaces</a>, Previous: <a href="#Tidying-and-profiling-R-code" accesskey="p" rel="prev">Tidying and profiling R code</a>, Up: <a href="#Top" accesskey="u" rel="up">Top</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Debugging-1"></a>
<h2 class="chapter">4 Debugging</h2>

<p>This chapter covers the debugging of R extensions, starting with the
ways to get useful error information and moving on to how to deal with
errors that crash R.  For those who prefer other styles there are
contributed packages such as <a href="https://CRAN.R-project.org/package=debug"><strong>debug</strong></a> on <acronym>CRAN</acronym>
(described in an article in
<a href="https://CRAN.R-project.org/doc/Rnews/Rnews_2003-3.pdf">R-News
3/3</a>).  (There are notes from 2002 provided by Roger Peng at
<a href="http://www.biostat.jhsph.edu/~rpeng/docs/R-debug-tools.pdf">http://www.biostat.jhsph.edu/~rpeng/docs/R-debug-tools.pdf</a>
which provide complementary examples to those given here.)
</p>
<table summary="" class="menu" border="0" cellspacing="0">
<tr><td align="left" valign="top">&bull; <a href="#Browsing" accesskey="1">Browsing</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Debugging-R-code" accesskey="2">Debugging R code</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Checking-memory-access" accesskey="3">Checking memory access</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Debugging-compiled-code" accesskey="4">Debugging compiled code</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
</table>


<hr>
<a name="Browsing"></a>
<div class="header">
<p>
Next: <a href="#Debugging-R-code" accesskey="n" rel="next">Debugging R code</a>, Previous: <a href="#Debugging" accesskey="p" rel="prev">Debugging</a>, Up: <a href="#Debugging" accesskey="u" rel="up">Debugging</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Browsing-1"></a>
<h3 class="section">4.1 Browsing</h3>

<a name="index-browser"></a>
<p>Most of the R-level debugging facilities are based around the
built-in browser.  This can be used directly by inserting a call to
<code>browser()</code> into the code of a function (for example, using
<code>fix(my_function)</code> ).  When code execution reaches that point in
the function, control returns to the R console with a special prompt.
For example
</p>
<div class="example">
<pre class="example">&gt; fix(summary.data.frame) ## insert browser() call after for() loop
&gt; summary(women)
Called from: summary.data.frame(women)
Browse[1]&gt; ls()
 [1] &quot;digits&quot; &quot;i&quot;      &quot;lbs&quot;    &quot;lw&quot;     &quot;maxsum&quot; &quot;nm&quot;     &quot;nr&quot;     &quot;nv&quot;
 [9] &quot;object&quot; &quot;sms&quot;    &quot;z&quot;
Browse[1]&gt; maxsum
[1] 7
Browse[1]&gt;
     height         weight
 Min.   :58.0   Min.   :115.0
 1st Qu.:61.5   1st Qu.:124.5
 Median :65.0   Median :135.0
 Mean   :65.0   Mean   :136.7
 3rd Qu.:68.5   3rd Qu.:148.0
 Max.   :72.0   Max.   :164.0
&gt; rm(summary.data.frame)
</pre></div>

<p>At the browser prompt one can enter any R expression, so for example
<code>ls()</code> lists the objects in the current frame, and entering the
name of an object will<a name="DOCF103" href="#FOOT103"><sup>103</sup></a> print it.  The following commands are
also accepted
</p>
<ul>
<li> <code>n</code>

<p>Enter &lsquo;step-through&rsquo; mode.  In this mode, hitting return executes the
next line of code (more precisely one line and any continuation lines).
Typing <code>c</code> will continue to the end of the current context, e.g.
to the end of the current loop or function.
</p>
</li><li> <code>c</code>

<p>In normal mode, this quits the browser and continues execution, and just
return works in the same way.  <code>cont</code> is a synonym.
</p>
</li><li> <code>where</code>

<p>This prints the call stack.  For example
</p>
<div class="example">
<pre class="example">&gt; summary(women)
Called from: summary.data.frame(women)
Browse[1]&gt; where
where 1: summary.data.frame(women)
where 2: summary(women)

Browse[1]&gt;
</pre></div>

</li><li> <code>Q</code>

<p>Quit both the browser and the current expression, and return to the
top-level prompt.
</p></li></ul>

<p>Errors in code executed at the browser prompt will normally return
control to the browser prompt.  Objects can be altered by assignment,
and will keep their changed values when the browser is exited.  If
really necessary, objects can be assigned to the workspace from the
browser prompt (by using <code>&lt;&lt;-</code> if the name is not already in
scope).
</p>
<hr>
<a name="Debugging-R-code"></a>
<div class="header">
<p>
Next: <a href="#Checking-memory-access" accesskey="n" rel="next">Checking memory access</a>, Previous: <a href="#Browsing" accesskey="p" rel="prev">Browsing</a>, Up: <a href="#Debugging" accesskey="u" rel="up">Debugging</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Debugging-R-code-1"></a>
<h3 class="section">4.2 Debugging R code</h3>

<a name="index-traceback"></a>
<p>Suppose your R program gives an error message.  The first thing to
find out is what R was doing at the time of the error, and the most
useful tool is <code>traceback()</code>.  We suggest that this is run whenever
the cause of the error is not immediately obvious.  Daily, errors are
reported to the R mailing lists as being in some package when
<code>traceback()</code> would show that the error was being reported by some
other package or base R.  Here is an example from the regression
suite.
</p>
<div class="smallexample">
<pre class="smallexample">&gt; success &lt;- c(13,12,11,14,14,11,13,11,12)
&gt; failure &lt;- c(0,0,0,0,0,0,0,2,2)
&gt; resp &lt;- cbind(success, failure)
&gt; predictor &lt;- c(0, 5^(0:7))
&gt; glm(resp ~ 0+predictor, family = binomial(link=&quot;log&quot;))
Error: no valid set of coefficients has been found: please supply starting values
&gt; traceback()
3: stop(&quot;no valid set of coefficients has been found: please supply
	 starting values&quot;, call. = FALSE)
2: glm.fit(x = X, y = Y, weights = weights, start = start, etastart = etastart,
       mustart = mustart, offset = offset, family = family, control = control,
       intercept = attr(mt, &quot;intercept&quot;) &gt; 0)
1: glm(resp ~ 0 + predictor, family = binomial(link =&quot;log&quot;))
</pre></div>

<p>The calls to the active frames are given in reverse order (starting with
the innermost).  So we see the error message comes from an explicit
check in <code>glm.fit</code>.  (<code>traceback()</code> shows you all the lines of
the function calls, which can be limited by setting <code>option</code>
<samp>&quot;deparse.max.lines&quot;</samp>.)
</p>
<p>Sometimes the traceback will indicate that the error was detected inside
compiled code, for example (from <code>?nls</code>)
</p>
<div class="smallexample">
<pre class="smallexample">Error in nls(y ~ a + b * x, start = list(a = 0.12345, b = 0.54321), trace = TRUE) :
	step factor 0.000488281 reduced below ‘minFactor’ of 0.000976563
&gt;  traceback()
2: .Call(R_nls_iter, m, ctrl, trace)
1: nls(y ~ a + b * x, start = list(a = 0.12345, b = 0.54321), trace = TRUE)
</pre></div>

<p>This will be the case if the innermost call is to <code>.C</code>,
<code>.Fortran</code>, <code>.Call</code>, <code>.External</code> or <code>.Internal</code>, but
as it is also possible for such code to evaluate R expressions, this
need not be the innermost call, as in
</p>
<div class="smallexample">
<pre class="smallexample">&gt; traceback()
9: gm(a, b, x)
8: .Call(R_numeric_deriv, expr, theta, rho, dir)
7: numericDeriv(form[[3]], names(ind), env)
6: getRHS()
5: assign(&quot;rhs&quot;, getRHS(), envir = thisEnv)
4: assign(&quot;resid&quot;, .swts * (lhs - assign(&quot;rhs&quot;, getRHS(), envir = thisEnv)),
       envir = thisEnv)
3: function (newPars)
   {
       setPars(newPars)
       assign(&quot;resid&quot;, .swts * (lhs - assign(&quot;rhs&quot;, getRHS(), envir = thisEnv)),
	   envir = thisEnv)
       assign(&quot;dev&quot;, sum(resid^2), envir = thisEnv)
       assign(&quot;QR&quot;, qr(.swts * attr(rhs, &quot;gradient&quot;)), envir = thisEnv)
       return(QR$rank &lt; min(dim(QR$qr)))
   }(c(-0.00760232418963883, 1.00119632515036))
2: .Call(R_nls_iter, m, ctrl, trace)
1: nls(yeps ~ gm(a, b, x), start = list(a = 0.12345, b = 0.54321))
</pre></div>

<p>Occasionally <code>traceback()</code> does not help, and this can be the case
if S4 method dispatch is involved.  Consider the following example
</p>
<div class="example">
<pre class="example">&gt; xyd &lt;- new(&quot;xyloc&quot;, x=runif(20), y=runif(20))
Error in as.environment(pkg) : no item called &quot;package:S4nswv&quot;
on the search list
Error in initialize(value, ...) : S language method selection got
an error when called from internal dispatch for function ‘initialize’
&gt; traceback()
2: initialize(value, ...)
1: new(&quot;xyloc&quot;, x = runif(20), y = runif(20))
</pre></div>

<p>which does not help much, as there is no call to <code>as.environment</code>
in <code>initialize</code> (and the note &ldquo;called from internal dispatch&rdquo;
tells us so).  In this case we searched the R sources for the quoted
call, which occurred in only one place,
<code>methods:::.asEnvironmentPackage</code>.  So now we knew where the
error was occurring.  (This was an unusually opaque example.)
</p>
<p>The error message
</p>
<div class="example">
<pre class="example">evaluation nested too deeply: infinite recursion / options(expressions=)?
</pre></div>

<p>can be hard to handle with the default value (5000).  Unless you know
that there actually is deep recursion going on, it can help to set
something like
</p>
<div class="example">
<pre class="example">options(expressions=500)
</pre></div>

<p>and re-run the example showing the error.
</p>
<p>Sometimes there is warning that clearly is the precursor to some later
error, but it is not obvious where it is coming from.  Setting
<code>options(warn = 2)</code> (which turns warnings into errors) can help here.
</p>
<p>Once we have located the error, we have some choices.  One way to proceed
is to find out more about what was happening at the time of the crash by
looking a <em>post-mortem</em> dump.  To do so, set
<a name="index-dump_002eframes"></a>
<code>options(error=dump.frames)</code> and run the code again.  Then invoke
<code>debugger()</code> and explore the dump.  Continuing our example:
</p>
<div class="smallexample">
<pre class="smallexample">&gt; options(error = dump.frames)
&gt; glm(resp ~ 0 + predictor, family = binomial(link =&quot;log&quot;))
Error: no valid set of coefficients has been found: please supply starting values
</pre></div>

<p>which is the same as before, but an object called <code>last.dump</code> has
appeared in the workspace.  (Such objects can be large, so remove it
when it is no longer needed.)  We can examine this at a later time by
calling the function <code>debugger</code>.
<a name="index-debugger"></a>
</p>
<div class="smallexample">
<pre class="smallexample">&gt; debugger()
Message:  Error: no valid set of coefficients has been found: please supply starting values
Available environments had calls:
1: glm(resp ~ 0 + predictor, family = binomial(link = &quot;log&quot;))
2: glm.fit(x = X, y = Y, weights = weights, start = start, etastart = etastart, mus
3: stop(&quot;no valid set of coefficients has been found: please supply starting values
Enter an environment number, or 0 to exit  Selection:
</pre></div>

<p>which gives the same sequence of calls as <code>traceback</code>, but in
outer-first order and with only the first line of the call, truncated to
the current width.  However, we can now examine in more detail what was
happening at the time of the error.  Selecting an environment opens the
browser in that frame.  So we select the function call which spawned the
error message, and explore some of the variables (and execute two
function calls).
</p>
<div class="smallexample">
<pre class="smallexample">Enter an environment number, or 0 to exit  Selection: 2
Browsing in the environment with call:
   glm.fit(x = X, y = Y, weights = weights, start = start, etas
Called from: debugger.look(ind)
Browse[1]&gt; ls()
 [1] &quot;aic&quot;        &quot;boundary&quot;   &quot;coefold&quot;    &quot;control&quot;    &quot;conv&quot;
 [6] &quot;dev&quot;        &quot;dev.resids&quot; &quot;devold&quot;     &quot;EMPTY&quot;      &quot;eta&quot;
[11] &quot;etastart&quot;   &quot;family&quot;     &quot;fit&quot;        &quot;good&quot;       &quot;intercept&quot;
[16] &quot;iter&quot;       &quot;linkinv&quot;    &quot;mu&quot;         &quot;mu.eta&quot;     &quot;mu.eta.val&quot;
[21] &quot;mustart&quot;    &quot;n&quot;          &quot;ngoodobs&quot;   &quot;nobs&quot;       &quot;nvars&quot;
[26] &quot;offset&quot;     &quot;start&quot;      &quot;valideta&quot;   &quot;validmu&quot;    &quot;variance&quot;
[31] &quot;varmu&quot;      &quot;w&quot;          &quot;weights&quot;    &quot;x&quot;          &quot;xnames&quot;
[36] &quot;y&quot;          &quot;ynames&quot;     &quot;z&quot;
Browse[1]&gt; eta
	    1             2             3             4             5
 0.000000e+00 -2.235357e-06 -1.117679e-05 -5.588393e-05 -2.794197e-04
	    6             7             8             9
-1.397098e-03 -6.985492e-03 -3.492746e-02 -1.746373e-01
Browse[1]&gt; valideta(eta)
[1] TRUE
Browse[1]&gt; mu
	1         2         3         4         5         6         7         8
1.0000000 0.9999978 0.9999888 0.9999441 0.9997206 0.9986039 0.9930389 0.9656755
	9
0.8397616
Browse[1]&gt; validmu(mu)
[1] FALSE
Browse[1]&gt; c
Available environments had calls:
1: glm(resp ~ 0 + predictor, family = binomial(link = &quot;log&quot;))
2: glm.fit(x = X, y = Y, weights = weights, start = start, etastart = etastart
3: stop(&quot;no valid set of coefficients has been found: please supply starting v

Enter an environment number, or 0 to exit  Selection: 0
&gt; rm(last.dump)
</pre></div>

<p>Because <code>last.dump</code> can be looked at later or even in another R
session, post-mortem debugging is possible even for batch usage of R.
We do need to arrange for the dump to be saved: this can be done either
using the command-line flag <samp>--save</samp> to save the workspace at the
end of the run, or <em>via</em> a setting such as
</p>
<div class="example">
<pre class="example">&gt; options(error = quote({dump.frames(to.file=TRUE); q()}))
</pre></div>

<p>See the help on <code>dump.frames</code> for further options and a worked
example.
</p>
<a name="index-recover"></a>
<p>An alternative error action is to use the function <code>recover()</code>:
</p>
<div class="smallexample">
<pre class="smallexample">&gt; options(error = recover)
&gt; glm(resp ~ 0 + predictor, family = binomial(link = &quot;log&quot;))
Error: no valid set of coefficients has been found: please supply starting values

Enter a frame number, or 0 to exit

1: glm(resp ~ 0 + predictor, family = binomial(link = &quot;log&quot;))
2: glm.fit(x = X, y = Y, weights = weights, start = start, etastart = etastart

Selection:
</pre></div>

<p>which is very similar to <code>dump.frames</code>.  However, we can examine
the state of the program directly, without dumping and re-loading the
dump.  As its help page says, <code>recover</code> can be routinely used as
the error action in place of <code>dump.calls</code> and <code>dump.frames</code>,
since it behaves like <code>dump.frames</code> in non-interactive use.
</p>

<a name="index-debug"></a>
<p>Post-mortem debugging is good for finding out exactly what went wrong,
but not necessarily why.  An alternative approach is to take a closer
look at what was happening just before the error, and a good way to do
that is to use <code>debug</code>.  This inserts a call to the browser
at the beginning of the function, starting in step-through mode.  So in
our example we could use
</p>
<div class="smallexample">
<pre class="smallexample">&gt; debug(glm.fit)
&gt; glm(resp ~ 0 + predictor, family = binomial(link =&quot;log&quot;))
debugging in: glm.fit(x = X, y = Y, weights = weights, start = start, etastart = etastart,
    mustart = mustart, offset = offset, family = family, control = control,
    intercept = attr(mt, &quot;intercept&quot;) &gt; 0)
debug: {
## lists the whole function
Browse[1]&gt;
debug: x &lt;- as.matrix(x)
...
Browse[1]&gt; start
[1] -2.235357e-06
debug: eta &lt;- drop(x %*% start)
Browse[1]&gt; eta
	    1             2             3             4             5
 0.000000e+00 -2.235357e-06 -1.117679e-05 -5.588393e-05 -2.794197e-04
	    6             7             8             9
-1.397098e-03 -6.985492e-03 -3.492746e-02 -1.746373e-01
Browse[1]&gt;
debug: mu &lt;- linkinv(eta &lt;- eta + offset)
Browse[1]&gt; mu
	1         2         3         4         5         6         7         8
1.0000000 0.9999978 0.9999888 0.9999441 0.9997206 0.9986039 0.9930389 0.9656755
	9
0.8397616
</pre></div>

<p>(The prompt <code>Browse[1]&gt;</code> indicates that this is the first level of
browsing: it is possible to step into another function that is itself
being debugged or contains a call to <code>browser()</code>.)
</p>
<p><code>debug</code> can be used for hidden functions and S3 methods by
e.g. <code>debug(stats:::predict.Arima)</code>.  (It cannot be used for S4
methods, but an alternative is given on the help page for <code>debug</code>.)
Sometimes you want to debug a function defined inside another function,
e.g. the function <code>arimafn</code> defined inside <code>arima</code>.  To do so,
set <code>debug</code> on the outer function (here <code>arima</code>) and
step through it until the inner function has been defined.  Then
call <code>debug</code> on the inner function (and use <code>c</code> to get out of
step-through mode in the outer function).
</p>
<a name="index-undebug"></a>
<p>To remove debugging of a function, call <code>undebug</code> with the argument
previously given to <code>debug</code>; debugging otherwise lasts for the rest
of the R session (or until the function is edited or otherwise
replaced).
</p>
<a name="index-trace"></a>
<p><code>trace</code> can be used to temporarily insert debugging code into a
function, for example to insert a call to <code>browser()</code> just before
the point of the error.  To return to our running example
</p>
<div class="example">
<pre class="example">## first get a numbered listing of the expressions of the function
&gt; page(as.list(body(glm.fit)), method=&quot;print&quot;)
&gt; trace(glm.fit, browser, at=22)
Tracing function &quot;glm.fit&quot; in package &quot;stats&quot;
[1] &quot;glm.fit&quot;
&gt; glm(resp ~ 0 + predictor, family = binomial(link =&quot;log&quot;))
Tracing glm.fit(x = X, y = Y, weights = weights, start = start,
   etastart = etastart,  .... step 22
Called from: eval(expr, envir, enclos)
Browse[1]&gt; n
## and single-step from here.
&gt; untrace(glm.fit)
</pre></div>
<p>For your own functions, it may be as easy to use <code>fix</code> to insert
temporary code, but <code>trace</code> can help with functions in a namespace
(as can <code>fixInNamespace</code>).  Alternatively, use
<code>trace(,edit=TRUE)</code> to insert code visually.
</p>

<hr>
<a name="Checking-memory-access"></a>
<div class="header">
<p>
Next: <a href="#Debugging-compiled-code" accesskey="n" rel="next">Debugging compiled code</a>, Previous: <a href="#Debugging-R-code" accesskey="p" rel="prev">Debugging R code</a>, Up: <a href="#Debugging" accesskey="u" rel="up">Debugging</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Checking-memory-access-1"></a>
<h3 class="section">4.3 Checking memory access</h3>

<p>Errors in memory allocation and reading/writing outside arrays are very
common causes of crashes (e.g., segfaults) on some machines.  Often
the crash appears long after the invalid memory access: in particular
damage to the structures which R itself has allocated may only become
apparent at the next garbage collection (or even at later garbage
collections after objects have been deleted).
</p>
<p>Note that memory access errors may be seen with LAPACK, BLAS, OpenMP and
Java-using packages: some at least of these seem to be intentional, and
some are related to passing characters to Fortran.
</p>
<p>Some of these tools can detect mismatched allocation and deallocation.
C++ programmers should note that memory allocated by <code>new []</code> must
be freed by <code>delete []</code>, other uses of <code>new</code> by <code>delete</code>,
and memory allocated by <code>malloc</code>, <code>calloc</code> and <code>realloc</code>
by <code>free</code>.  Some platforms will tolerate mismatches (perhaps with
memory leaks) but others will segfault.
</p>
<table summary="" class="menu" border="0" cellspacing="0">
<tr><td align="left" valign="top">&bull; <a href="#Using-gctorture" accesskey="1">Using gctorture</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Using-valgrind" accesskey="2">Using valgrind</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Using-Address-Sanitizer" accesskey="3">Using Address Sanitizer</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Using-Undefined-Behaviour-Sanitizer" accesskey="4">Using Undefined Behaviour Sanitizer</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Other-analyses-with-_0060clang_0027" accesskey="5">Other analyses with ‘clang’</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Using-_0060Dr_002e-Memory_0027" accesskey="6">Using ‘Dr. Memory’</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Fortran-array-bounds-checking" accesskey="7">Fortran array bounds checking</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
</table>

<hr>
<a name="Using-gctorture"></a>
<div class="header">
<p>
Next: <a href="#Using-valgrind" accesskey="n" rel="next">Using valgrind</a>, Previous: <a href="#Checking-memory-access" accesskey="p" rel="prev">Checking memory access</a>, Up: <a href="#Checking-memory-access" accesskey="u" rel="up">Checking memory access</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Using-gctorture-1"></a>
<h4 class="subsection">4.3.1 Using gctorture</h4>

<a name="index-gctorture"></a>
<p>We can help to detect memory problems in R objects earlier by running
garbage collection as often as possible.  This is achieved by
<code>gctorture(TRUE)</code>, which as described on its help page
</p>
<blockquote>
<p>Provokes garbage collection on (nearly) every memory allocation.
Intended to ferret out memory protection bugs.  Also makes R run
<em>very</em> slowly, unfortunately.
</p></blockquote>

<p>The reference to &lsquo;memory protection&rsquo; is to missing C-level calls to
<code>PROTECT</code>/<code>UNPROTECT</code> (see <a href="#Garbage-Collection">Garbage Collection</a>) which if
missing allow R objects to be garbage-collected when they are still
in use.  But it can also help with other memory-related errors.
</p>
<p>Normally running under <code>gctorture(TRUE)</code> will just produce a crash
earlier in the R program, hopefully close to the actual cause. See
the next section for how to decipher such crashes.
</p>
<p>It is possible to run all the examples, tests and vignettes covered by
<code>R CMD check</code> under <code>gctorture(TRUE)</code> by using the option
<samp>--use-gct</samp>.
</p>
<p>The function <code>gctorture2</code> provides more refined control over the GC
torture process.  Its arguments <code>step</code>, <code>wait</code> and
<code>inhibit_release</code> are documented on its help page.  Environment
variables can also be used at the start of the R session to turn on
GC torture: <code>R_GCTORTURE</code> corresponds to the <code>step</code> argument to
<code>gctorture2</code>, <code>R_GCTORTURE_WAIT</code> to <code>wait</code>, and
<code>R_GCTORTURE_INHIBIT_RELEASE</code> to <code>inhibit_release</code>.
</p>
<p>If R is configured with <samp>--enable-strict-barrier</samp> then a
variety of tests for the integrity of the write barrier are enabled.  In
addition tests to help detect protect issues are enabled:
</p>
<ul>
<li> All GCs are full GCs.

</li><li> New nodes in small node pages are marked as <code>NEWSXP</code> on creation.

</li><li> After a GC all free nodes that are not of type <code>NEWSXP</code> are marked
as type <code>FREESXP</code> and their previous type is recorded.

</li><li> Most calls to accessor functions check their <code>SEXP</code> inputs and
<code>SEXP</code> outputs and signal an error if a <code>FREESXP</code> is found.
The address of the node and the old type are included in the error
message.

</li></ul>

<p><code>R CMD check --use-gct</code> can be set to use
<code>gctorture2(<var>n</var>)</code> rather than <code>gctorture(TRUE)</code> by setting
environment variable <code>_R_CHECK_GCT_N_</code> to a positive integer value
to be used as <code><var>n</var></code>.
</p>
<p>Used with a debugger and with <code>gctorture</code> or <code>gctorture2</code> this
mechanism can be helpful in isolating memory protect problems.
</p>

<hr>
<a name="Using-valgrind"></a>
<div class="header">
<p>
Next: <a href="#Using-Address-Sanitizer" accesskey="n" rel="next">Using Address Sanitizer</a>, Previous: <a href="#Using-gctorture" accesskey="p" rel="prev">Using gctorture</a>, Up: <a href="#Checking-memory-access" accesskey="u" rel="up">Checking memory access</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Using-valgrind-1"></a>
<h4 class="subsection">4.3.2 Using valgrind</h4>

<p>If you have access to Linux on a common CPU type or supported versions
of macOS<a name="DOCF104" href="#FOOT104"><sup>104</sup></a> you can use
<code>valgrind</code> (<a href="http://www.valgrind.org/">http://www.valgrind.org/</a>, pronounced to rhyme
with &lsquo;tinned&rsquo;) to check for possible problems.  To run some examples
under <code>valgrind</code> use something like
</p>
<div class="example">
<pre class="example">R -d valgrind --vanilla &lt; mypkg-Ex.R
R -d &quot;valgrind --tool=memcheck --leak-check=full&quot; --vanilla &lt; mypkg-Ex.R
</pre></div>

<p>where <samp>mypkg-Ex.R</samp> is a set of examples, e.g. the file created in
<samp>mypkg.Rcheck</samp> by <code>R CMD check</code>.  Occasionally this reports
memory reads of &lsquo;uninitialised values&rsquo; that are the result of compiler
optimization, so can be worth checking under an unoptimized compile: for
maximal information use a build with debugging symbols.  We know there
will be some small memory leaks from <code>readline</code> and R itself &mdash;
these are memory areas that are in use right up to the end of the R
session.  Expect this to run around 20x slower than without
<code>valgrind</code>, and in some cases much slower than that.  Several
versions of <code>valgrind</code> were not happy with some optimized BLASes
that use <acronym>CPU</acronym>-specific instructions so you may need to build a
version of R specifically to use with <code>valgrind</code>.
</p>
<p>On platforms where <code>valgrind</code> is installed you can build a version
of R with extra instrumentation to help <code>valgrind</code> detect errors
in the use of memory allocated from the R heap.  The
<code>configure</code> option is
<samp>--with-valgrind-instrumentation=<var>level</var></samp>, where <var>level</var>
is 0, 1 or 2.  Level 0 is the default and does not add any anything.
Level 1 will detect some uses<a name="DOCF105" href="#FOOT105"><sup>105</sup></a> of uninitialised memory and has little impact on speed
(compared to level 0). Level 2 will detect many other memory-use
bugs<a name="DOCF106" href="#FOOT106"><sup>106</sup></a>  but make R much slower when running under
<code>valgrind</code>.  Using this in conjunction with <code>gctorture</code> can be
even more effective (and even slower).
</p>
<p>An example of <code>valgrind</code> output is
</p><div class="smallexample">
<pre class="smallexample">==12539== Invalid read of size 4
==12539==    at 0x1CDF6CBE: csc_compTr (Mutils.c:273)
==12539==    by 0x1CE07E1E: tsc_transpose (dtCMatrix.c:25)
==12539==    by 0x80A67A7: do_dotcall (dotcode.c:858)
==12539==    by 0x80CACE2: Rf_eval (eval.c:400)
==12539==    by 0x80CB5AF: R_execClosure (eval.c:658)
==12539==    by 0x80CB98E: R_execMethod (eval.c:760)
==12539==    by 0x1B93DEFA: R_standardGeneric (methods_list_dispatch.c:624)
==12539==    by 0x810262E: do_standardGeneric (objects.c:1012)
==12539==    by 0x80CAD23: Rf_eval (eval.c:403)
==12539==    by 0x80CB2F0: Rf_applyClosure (eval.c:573)
==12539==    by 0x80CADCC: Rf_eval (eval.c:414)
==12539==    by 0x80CAA03: Rf_eval (eval.c:362)
==12539==  Address 0x1C0D2EA8 is 280 bytes inside a block of size 1996 alloc'd
==12539==    at 0x1B9008D1: malloc (vg_replace_malloc.c:149)
==12539==    by 0x80F1B34: GetNewPage (memory.c:610)
==12539==    by 0x80F7515: Rf_allocVector (memory.c:1915)
...
</pre></div>
<p>This example is from an instrumented version of R, while tracking
down a bug in the <a href="https://CRAN.R-project.org/package=Matrix"><strong>Matrix</strong></a> package in 2006.  The first line
indicates that R has tried to read 4 bytes from a memory address that
it does not have access to. This is followed by a C stack trace showing
where the error occurred. Next is a description of the memory that was
accessed. It is inside a block allocated by <code>malloc</code>, called from
<code>GetNewPage</code>, that is, in the internal R heap.  Since this
memory all belongs to R, <code>valgrind</code> would not (and did not)
detect the problem in an uninstrumented build of R.  In this example
the stack trace was enough to isolate and fix the bug, which was in
<code>tsc_transpose</code>, and in this example running under
<code>gctorture()</code> did not provide any additional information.  When the
stack trace is not sufficiently informative the option
<samp>--db-attach=yes</samp> to <code>valgrind</code> may be helpful.  This starts
a post-mortem debugger (by default <code>gdb</code>) so that variables in the
C code can be inspected (see <a href="#Inspecting-R-objects">Inspecting R objects</a>).
</p>
<p><code>valgrind</code> is good at spotting the use of uninitialized values:
use option <samp>--track-origins=yes</samp> to show where these originated
from.  What it cannot detect is the misuse of arrays allocated on the
stack: this includes C automatic variables and some<a name="DOCF107" href="#FOOT107"><sup>107</sup></a>
Fortran arrays.
</p>
<p>It is possible to run all the examples, tests and vignettes covered by
<code>R CMD check</code> under <code>valgrind</code> by using the option
<samp>--use-valgrind</samp>.  If you do this you will need to select the
<code>valgrind</code> options some other way, for example by having a
<samp>~/.valgrindrc</samp> file containing
</p>
<div class="example">
<pre class="example">--leak-check=full
--track-origins=yes
</pre></div>

<p>or setting the environment variable <code>VALGRIND_OPTS</code>.
</p>
<p>On macOS you may need to ensure that debugging symbols are made available
(so <code>valgrind</code> reports line numbers in files).  This can usually
be done with the <code>valgrind</code> option <samp>--dsymutil=yes</samp> to
ask for the symbols to be dumped when the <samp>.so</samp> file is loaded.
This will not work where packages are installed into a system area (such
as the <samp>R.framework</samp>) and can be slow.  Installing packages with
<code>R CMD INSTALL --dsym</code> installs the dumped symbols.  (This can
also be done by setting environment variable <code>PKG_MAKE_DSYM</code> to a
non-empty value before the <code>INSTALL</code>.)
</p>
<p>This section has described the use of <code>memtest</code>, the default
(and most useful) of <code>valgrind</code>&rsquo;s tools.  There are others
described in its documentation: <code>helgrind</code> can be useful for
threaded programs.
</p>
<hr>
<a name="Using-Address-Sanitizer"></a>
<div class="header">
<p>
Next: <a href="#Using-Undefined-Behaviour-Sanitizer" accesskey="n" rel="next">Using Undefined Behaviour Sanitizer</a>, Previous: <a href="#Using-valgrind" accesskey="p" rel="prev">Using valgrind</a>, Up: <a href="#Checking-memory-access" accesskey="u" rel="up">Checking memory access</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Using-the-Address-Sanitizer"></a>
<h4 class="subsection">4.3.3 Using the Address Sanitizer</h4>

<p><code>AddressSanitizer</code> (&lsquo;ASan&rsquo;) is a tool with similar aims to the
memory checker in <code>valgrind</code>.  It is available with suitable
builds<a name="DOCF108" href="#FOOT108"><sup>108</sup></a> of <code>gcc</code> and <code>clang</code> on common Linux
and macOS platforms.  See
<a href="http://clang.llvm.org/docs/UsersManual.html#controlling-code-generation">http://clang.llvm.org/docs/UsersManual.html#controlling-code-generation</a>,
<a href="http://clang.llvm.org/docs/AddressSanitizer.html">http://clang.llvm.org/docs/AddressSanitizer.html</a> and
<a href="https://code.google.com/p/address-sanitizer/">https://code.google.com/p/address-sanitizer/</a>.
</p>
<p>More thorough checks of C++ code are done if the C++ library has been
&lsquo;annotated&rsquo;: at the time of writing this applied to <code>std::vector</code>
in <code>libc++</code> for use with <code>clang</code> and gives rise to
&lsquo;<samp>container-overflow</samp>&rsquo;<a name="DOCF109" href="#FOOT109"><sup>109</sup></a>
reports.
</p>
<p>It requires code to have been compiled <em>and linked</em> with
<samp>-fsanitize=address</samp> and compiling with <code>-fno-omit-frame-pointer</code>
will give more legible reports.  It has a runtime penalty of 2&ndash;3x,
extended compilation times and uses substantially more memory, often
1&ndash;2GB, at run time.  On 64-bit platforms it reserves (but does not
allocate) 16&ndash;20TB of virtual memory: restrictive shell settings can
cause problems.
</p>
<p>By comparison with <code>valgrind</code>, ASan can
detect misuse of stack and global variables but not the use of
uninitialized memory.
</p>
<p>Recent versions return symbolic addresses for the location of the error
provided <code>llvm-symbolizer</code><a name="DOCF110" href="#FOOT110"><sup>110</sup></a> is on the path: if it is available but not
on the path or has been renamed<a name="DOCF111" href="#FOOT111"><sup>111</sup></a>, one
can use an environment variable, e.g.
</p>
<div class="example">
<pre class="example">ASAN_SYMBOLIZER_PATH=/path/to/llvm-symbolizer
</pre></div>

<p>An alternative is to pipe the output through
<code>asan_symbolize.py</code><a name="DOCF112" href="#FOOT112"><sup>112</sup></a> and perhaps
then (for compiled C++ code) <code>c++filt</code>.  (On macOS, you may need
to run <code>dsymutil</code> to get line-number reports.)
</p>
<p>The simplest way to make use of this is to build a version of R with
something like
</p>
<div class="example">
<pre class="example">CC=&quot;gcc -std=gnu99 -fsanitize=address&quot;
CFLAGS=&quot;-fno-omit-frame-pointer -g -O2 -Wall -pedantic -mtune=native&quot;
</pre></div>

<p>which will ensure that the <code>libasan</code> run-time library is compiled
into the R executable.  However this check can be enabled on a
per-package basis by using a <samp>~/.R/Makevars</samp> file like
</p><div class="example">
<pre class="example">CC = gcc -std=gnu99 -fsanitize=address -fno-omit-frame-pointer
CXX = g++ -fsanitize=address -fno-omit-frame-pointer
F77 = gfortran -fsanitize=address
FC = gfortran -fsanitize=address
</pre></div>
<p>(Note that <code>-fsanitize=address</code> has to be part of the compiler
specification to ensure it is used for linking.  These settings will not
be honoured by packages which ignore <samp>~/.R/Makevars</samp>.)  It will
be necessary to build R with
</p>
<div class="example">
<pre class="example">MAIN_LDFLAGS = -fsanitize=address
</pre></div>

<p>to link the runtime libraries into the R executable if it was not
specified as part of &lsquo;<samp>CC</samp>&rsquo; when R was built.
</p>
<p>For options available <em>via</em> the environment variable
<code>ASAN_OPTIONS</code> see
<a href="https://code.google.com/p/address-sanitizer/wiki/Flags#Run-time_flags">https://code.google.com/p/address-sanitizer/wiki/Flags#Run-time_flags</a>.
With <code>gcc</code> additional control is available <em>via</em> the
<samp>--params</samp> flag: see its <code>man</code> page.
</p>
<p>For more detailed information on an error, R can be run under a
debugger with a breakpoint set before the address sanitizer report is
produced: for <code>gdb</code> or <code>lldb</code> you could use
</p><div class="example">
<pre class="example">break __asan_report_error
</pre></div>
<p>(See
<a href="https://code.google.com/p/address-sanitizer/wiki/AddressSanitizer#gdb">https://code.google.com/p/address-sanitizer/wiki/AddressSanitizer#gdb</a>.)
</p>
<p>Recent versions<a name="DOCF113" href="#FOOT113"><sup>113</sup></a> added the flag
<samp>-fsanitize-address-use-after-scope</samp>: see
<a href="https://github.com//google//sanitizers//wiki//AddressSanitizerUseAfterScope">https://github.com//google//sanitizers//wiki//AddressSanitizerUseAfterScope</a>.
</p>
<table summary="" class="menu" border="0" cellspacing="0">
<tr><td align="left" valign="top">&bull; <a href="#Using-Leak-Sanitizer" accesskey="1">Using Leak Sanitizer</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
</table>

<hr>
<a name="Using-Leak-Sanitizer"></a>
<div class="header">
<p>
Previous: <a href="#Using-Address-Sanitizer" accesskey="p" rel="prev">Using Address Sanitizer</a>, Up: <a href="#Using-Address-Sanitizer" accesskey="u" rel="up">Using Address Sanitizer</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Using-the-Leak-Sanitizer"></a>
<h4 class="subsubsection">4.3.3.1 Using the Leak Sanitizer</h4>

<p>For <code>x86_64</code> Linux there is a leak sanitizer, &lsquo;LSan&rsquo;: see
<a href="https://code.google.com/p/address-sanitizer/wiki/LeakSanitizer">https://code.google.com/p/address-sanitizer/wiki/LeakSanitizer</a>.
This is available on recent versions of <code>gcc</code> and <code>clang</code>, and
where available is compiled in as part of ASan.
</p>
<p>One way to invoke this from an ASan-enabled build is by the environment
variable
</p>
<div class="example">
<pre class="example">ASAN_OPTIONS='detect_leaks=1'
</pre></div>
<p>However, this was made the default for <code>clang</code> 3.5 and
<code>gcc</code> 5.1.0.
</p>
<p>When LSan is enabled, leaks give the process a failure error status (by
default <code>23</code>).  For an R package this means the R process,
and as the parser retains some memory to the end of the process, if R
itself was built against ASan, all runs will have a failure error status
(which may include running R as part of building R itself).
</p>
<p>To disable both this and some strict checking use
</p>
<div class="example">
<pre class="example">setenv ASAN_OPTIONS ‘alloc_dealloc_mismatch=0:detect_leaks=0:detect_odr_violation=0’
</pre></div>

<p>LSan also has a &lsquo;stand-alone&rsquo; mode where it is compiled in using
<samp>-fsanitize=leak</samp> and avoids the run-time overhead of ASan.
</p>
<hr>
<a name="Using-Undefined-Behaviour-Sanitizer"></a>
<div class="header">
<p>
Next: <a href="#Other-analyses-with-_0060clang_0027" accesskey="n" rel="next">Other analyses with ‘clang’</a>, Previous: <a href="#Using-Address-Sanitizer" accesskey="p" rel="prev">Using Address Sanitizer</a>, Up: <a href="#Checking-memory-access" accesskey="u" rel="up">Checking memory access</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Using-the-Undefined-Behaviour-Sanitizer"></a>
<h4 class="subsection">4.3.4 Using the Undefined Behaviour Sanitizer</h4>

<p>&lsquo;Undefined behaviour&rsquo; is where the language standard does not require
particular behaviour from the compiler.  Examples include division by
zero (where for doubles R requires the
<acronym>ISO</acronym>/<acronym>IEC</acronym>&nbsp;60559 behaviour but C/C++ do not), use
of zero-length arrays, shifts too far for signed types (e.g. <code>int
x, y; y = x &lt;&lt; 31;</code>), out-of-range coercion, invalid C++ casts and
mis-alignment.  Not uncommon examples of out-of-range coercion in R
packages are attempts to coerce a <code>NaN</code> or infinity to type
<code>int</code> or <code>NA_INTEGER</code> to an unsigned type such as
<code>size_t</code>.  Also common is <code>y[x - 1]</code> forgetting that <code>x</code>
might be <code>NA_INTEGER</code>.
</p>
<p>&lsquo;UBSanitizer&rsquo; is a tool for C/C++ source code selected by
<samp>-fsanitize=undefined</samp> in suitable builds<a name="DOCF114" href="#FOOT114"><sup>114</sup></a> of <code>clang</code> and GCC.  Its (main) runtime library is
linked into each package&rsquo;s DLL, so it is less often needed to be
included in <code>MAIN_LDFLAGS</code>.
</p>
<p>Some versions have greatly increased compilation times on a few
files<a name="DOCF115" href="#FOOT115"><sup>115</sup></a>.
</p>
<p>This sanitizer can be combined with the Address Sanitizer by
<samp>-fsanitize=undefined,address</samp> (where both are supported).
</p>
<p>Finer control of what is checked can be achieved by other options: for
<code>clang</code> see
<a href="http://clang.llvm.org/docs/UsersManual.html#controlling-code-generation">http://clang.llvm.org/docs/UsersManual.html#controlling-code-generation</a>.<a name="DOCF116" href="#FOOT116"><sup>116</sup></a>
The current set for <code>clang</code> is (on a single line):
</p><div class="example">
<pre class="example">-fsanitize=alignment,bool,bounds,enum,float-cast-overflow,
float-divide-by-zero,function,integer-divide-by-zero,nonnull-attribute,
null,object-size,return,returns-nonnull-attribute,shift,
signed-integer-overflow,unreachable,vla-bound,vptr
</pre></div>

<p>a subset of which could be combined with <code>address</code>, or use something
like
</p>
<div class="example">
<pre class="example">-fsanitize=undefined -fno-sanitize=float-divide-by-zero
</pre></div>

<p>(<code>function</code>, <code>return</code> and <code>vptr</code> apply only to C++).  In
addition,
</p><div class="example">
<pre class="example">-fsanitize=unsigned-integer-overflow
</pre></div>
<p>is available as a separate option in some versions of <code>clang</code>
(not enabled by <samp>-fsanitize=undefined</samp>).
</p>
<p><code>clang</code> 3.5 and later may need
</p>
<div class="example">
<pre class="example">-fsanitize=undefined -fno-sanitize=float-divide-by-zero,vptr
</pre></div>

<p>for C++ code (in <code>CXX</code> and <code>CXX1X</code>) as the run-time library
for <code>vptr</code> needs to be linked into the main R executable (and
that would need to be linked by <code>clang++</code>, not <code>clang</code>: you
could try building R with something like
</p><div class="example">
<pre class="example">MAIN_LD=&quot;clang++ -fsanitize=undefined&quot;
</pre></div>
<p>and perhaps for <code>clang</code> 3.7.x
</p><div class="example">
<pre class="example">R_OPENMP_CFLAGS=&quot;-fopenmp=libomp&quot;
</pre></div>
<p>or add <code>-lclang_rt.asan_cxx-x86_64</code><a name="DOCF117" href="#FOOT117"><sup>117</sup></a> or similar to <code>LD_FLAGS</code>).
</p>
<p>See <a href="https://gcc.gnu.org/onlinedocs/gcc/Debugging-Options.html">https://gcc.gnu.org/onlinedocs/gcc/Debugging-Options.html</a> (or
the manual for your version of GCC, installed or <em>via</em>
<a href="https://gcc.gnu.org/onlinedocs/">https://gcc.gnu.org/onlinedocs/</a>) for the options supported by
GCC: 5.3 supports
</p><div class="example">
<pre class="example">-fsanitize=alignment,bool,bounds,enum,float-cast-overflow,
integer-divide-by-zero,nonnull-attribute,null,object-size,
return,returns-nonnull-attribute,shift,signed-integer-overflow,
unreachable,vla-bound,vptr
</pre></div>
<p>with
</p><div class="example">
<pre class="example">-fsanitize=float-divide-by-zero
</pre></div>
<p>as a separate option not enabled by <code>-fsanitize=undefined</code> (and not
desirable for R uses).  At the time of writing the <code>object-size</code>
and <code>vptr</code> checks produced many warnings on GCC&rsquo;s own C++ headers,
so should be disabled.
</p>
<p>GCC 6 added
</p><div class="example">
<pre class="example">-fsanitize=bounds-strict
</pre></div>
<p>an extension of <code>bounds</code> to &lsquo;flexible array member-like arrays&rsquo;.
</p>

<p>Other useful flags include
</p><div class="example">
<pre class="example">-no-fsanitize-recover
</pre></div>

<p>which causes the first report to be fatal (it always is for the
<code>unreachable</code> and <code>return</code> suboptions).  For more detailed
information on where the runtime error occurs, R can be run under a
debugger with a breakpoint set before the sanitizer report is produced:
for <code>gdb</code> or <code>lldb</code> you could use
</p><div class="example">
<pre class="example">break __ubsan_handle_float_cast_overflow
break __ubsan_handle_float_cast_overflow_abort
</pre></div>
<p>or similar (there are handlers for each type of undefined behaviour).
</p>
<p>There are also the compiler flags <samp>-fcatch-undefined-behavior</samp>
and <samp>-ftrapv</samp>, said to be more reliable in <code>clang</code> than
<code>gcc</code>.
</p>
<p>For more details on the topic see
<a href="http://blog.regehr.org/archives/213">http://blog.regehr.org/archives/213</a> and
<a href="http://blog.llvm.org/2011/05/what-every-c-programmer-should-know.html">http://blog.llvm.org/2011/05/what-every-c-programmer-should-know.html</a>
(which has 3 parts).
</p>

<hr>
<a name="Other-analyses-with-_0060clang_0027"></a>
<div class="header">
<p>
Next: <a href="#Using-_0060Dr_002e-Memory_0027" accesskey="n" rel="next">Using ‘Dr. Memory’</a>, Previous: <a href="#Using-Undefined-Behaviour-Sanitizer" accesskey="p" rel="prev">Using Undefined Behaviour Sanitizer</a>, Up: <a href="#Checking-memory-access" accesskey="u" rel="up">Checking memory access</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Other-analyses-with-_0060clang_0027-1"></a>
<h4 class="subsection">4.3.5 Other analyses with &lsquo;clang&rsquo;</h4>

<p>Recent versions of <code>clang</code> on &lsquo;<samp>x86_64</samp>&rsquo; Linux have
&lsquo;ThreadSanitizer&rsquo; (<a href="https://code.google.com/p/thread-sanitizer/">https://code.google.com/p/thread-sanitizer/</a>),
a &lsquo;data race detector for C/C++ programs&rsquo;, and &lsquo;MemorySanitizer&rsquo;
(<a href="http://clang.llvm.org/docs/MemorySanitizer.html">http://clang.llvm.org/docs/MemorySanitizer.html</a>,
<a href="https://code.google.com/p/memory-sanitizer/wiki/MemorySanitizer">https://code.google.com/p/memory-sanitizer/wiki/MemorySanitizer</a>)
for the detection of uninitialized memory.  Both are based on and
provide similar functionality to tools in <code>valgrind</code>.
</p>
<p><code>clang</code> has a &lsquo;Static Analyser&rsquo; which can be run on the source
files during compilation: see <a href="http://clang-analyzer.llvm.org/">http://clang-analyzer.llvm.org/</a>.
</p>
<hr>
<a name="Using-_0060Dr_002e-Memory_0027"></a>
<div class="header">
<p>
Next: <a href="#Fortran-array-bounds-checking" accesskey="n" rel="next">Fortran array bounds checking</a>, Previous: <a href="#Other-analyses-with-_0060clang_0027" accesskey="p" rel="prev">Other analyses with ‘clang’</a>, Up: <a href="#Checking-memory-access" accesskey="u" rel="up">Checking memory access</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Using-_0060Dr_002e-Memory_0027-1"></a>
<h4 class="subsection">4.3.6 Using &lsquo;Dr. Memory&rsquo;</h4>

<p>&lsquo;Dr. Memory&rsquo; from <a href="http://www.drmemory.org/">http://www.drmemory.org/</a> is a memory checker
for (currently) 32-bit Windows, Linux and macOS with similar aims to
<code>valgrind</code>.  It works with unmodified executables<a name="DOCF118" href="#FOOT118"><sup>118</sup></a>
and detects memory access errors, uninitialized reads and memory leaks.
</p>
<hr>
<a name="Fortran-array-bounds-checking"></a>
<div class="header">
<p>
Previous: <a href="#Using-_0060Dr_002e-Memory_0027" accesskey="p" rel="prev">Using ‘Dr. Memory’</a>, Up: <a href="#Checking-memory-access" accesskey="u" rel="up">Checking memory access</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Fortran-array-bounds-checking-1"></a>
<h4 class="subsection">4.3.7 Fortran array bounds checking</h4>

<p>Most of the Fortran compilers used with R allow code to be compiled
with checking of array bounds: for example <code>gfortran</code> has option
<samp>-fbounds-check</samp> and Oracle Studio has <samp>-C</samp>.  This will
give an error when the upper or lower bound is exceeded, e.g.
</p><div class="example">
<pre class="example">At line 97 of file .../src/appl/dqrdc2.f
Fortran runtime error: Index ‘1’ of dimension 1 of array ‘x’ above upper bound of 0
</pre></div>

<p>One does need to be aware that lazy programmers often specify Fortran
dimensions as <code>1</code> rather than <code>*</code> or a real bound and these
will be reported.
</p>
<p>It is easy to arrange to use this check on just the code in your
package: add to <samp>~/.R/Makevars</samp> something like (for
<code>gfortran</code>)
</p><div class="example">
<pre class="example">FCFLAGS = -g -O2 -mtune=native -fbounds-check
FFLAGS = -g -O2 -mtune=native -fbounds-check
</pre></div>

<p>when you run <code>R CMD check</code>.
</p>
<p>This may report incorrectly errors with the way that Fortran character
variables are passed, particularly when Fortran subroutines are called
from C code.  This may include the use of BLAS and LAPACK subroutines in
R, so it is not advisable to build R itself with bounds checking
(and may not even be possible as these subroutines are called during the
R build).
</p>

<hr>
<a name="Debugging-compiled-code"></a>
<div class="header">
<p>
Previous: <a href="#Checking-memory-access" accesskey="p" rel="prev">Checking memory access</a>, Up: <a href="#Debugging" accesskey="u" rel="up">Debugging</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Debugging-compiled-code-1"></a>
<h3 class="section">4.4 Debugging compiled code</h3>
<a name="index-Debugging"></a>


<p>Sooner or later programmers will be faced with the need to debug
compiled code loaded into R.  This section is geared to platforms
using <code>gdb</code> with code compiled by <code>gcc</code>, but similar things
are possible with other debuggers such as <code>lldb</code>
(<a href="http://lldb.llvm.org/">http://lldb.llvm.org/</a>, used on macOS) and Sun&rsquo;s <code>dbx</code>:
some debuggers have graphical front-ends available.
</p>
<p>Consider first &lsquo;crashes&rsquo;, that is when R terminated unexpectedly with
an illegal memory access (a &lsquo;segfault&rsquo; or &lsquo;bus error&rsquo;), illegal
instruction or similar.  Unix-alike versions of R use a signal
handler which aims to give some basic information.  For example
</p>
<div class="example">
<pre class="example"> *** caught segfault ***
address 0x20000028, cause ‘memory not mapped’

Traceback:
 1: .identC(class1[[1]], class2)
 2: possibleExtends(class(sloti), classi, ClassDef2 = getClassDef(classi,
where = where))
 3: validObject(t(cu))
 4: stopifnot(validObject(cu &lt;- as(tu, &quot;dtCMatrix&quot;)), validObject(t(cu)),
validObject(t(tu)))

Possible actions:
1: abort (with core dump)
2: normal R exit
3: exit R without saving workspace
4: exit R saving workspace
Selection: 3
</pre></div>

<p>Since the R process may be damaged, the only really safe options are
the first or third.  (Note that a core dump is only produced where
enabled: a common default in a shell is to limit its size to 0, thereby
disabling it.)
</p>
<p>A fairly common cause of such crashes is a package which uses <code>.C</code>
or <code>.Fortran</code> and writes beyond (at either end) one of the
arguments it is passed.  As from R 3.0.0 there is a good way to
detect this: using <code>options(CBoundsCheck = TRUE)</code> (which can be
selected <em>via</em> the environment variable <code>R_C_BOUNDS_CHECK=yes)</code>
changes the way <code>.C</code> and <code>.Fortran</code> work to check if the
compiled code writes in the 64 bytes at either end of an argument.
</p>
<p>Another cause of a &lsquo;crash&rsquo; is to overrun the C stack.  R tries to
track that in its own code, but it may happen in third-party compiled
code.  For modern POSIX-compliant OSes R can safely catch that and
return to the top-level prompt, so one gets something like
</p>
<div class="example">
<pre class="example">&gt; .C(&quot;aaa&quot;)
Error: segfault from C stack overflow
&gt;
</pre></div>

<p>However, C stack overflows are fatal under Windows and normally defeat
attempts at debugging on that platform.  Further, the size of the stack
is set when R is compiled, whereas on POSIX OSes it can be set in the
shell from which R is launched.
</p>
<p>If you have a crash which gives a core dump you can use something like
</p>
<div class="example">
<pre class="example">gdb /path/to/R/bin/exec/R core.12345
</pre></div>

<p>to examine the core dump.  If core dumps are disabled or to catch errors
that do not generate a dump one can run R directly under a debugger
by for example
</p>
<div class="example">
<pre class="example">$ R -d gdb --vanilla
...
gdb&gt; run
</pre></div>

<p>at which point R will run normally, and hopefully the debugger will
catch the error and return to its prompt.  This can also be used to
catch infinite loops or interrupt very long-running code.  For a simple
example
</p>
<div class="example">
<pre class="example">&gt; for(i in 1:1e7) x &lt;- rnorm(100)
[hit Ctrl-C]
Program received signal SIGINT, Interrupt.
0x00397682 in _int_free () from /lib/tls/libc.so.6
(gdb) where
#0  0x00397682 in _int_free () from /lib/tls/libc.so.6
#1  0x00397eba in free () from /lib/tls/libc.so.6
#2  0xb7cf2551 in R_gc_internal (size_needed=313)
    at /users/ripley/R/svn/R-devel/src/main/memory.c:743
#3  0xb7cf3617 in Rf_allocVector (type=13, length=626)
    at /users/ripley/R/svn/R-devel/src/main/memory.c:1906
#4  0xb7c3f6d3 in PutRNGstate ()
    at /users/ripley/R/svn/R-devel/src/main/RNG.c:351
#5  0xb7d6c0a5 in do_random2 (call=0x94bf7d4, op=0x92580e8, args=0x9698f98,
    rho=0x9698f28) at /users/ripley/R/svn/R-devel/src/main/random.c:183
...
</pre></div>

<p>In many cases it is possible to attach a debugger to a running process:
this is helpful if an alternative front-end is in use or to investigate
a task that seems to be taking far too long.  This is done by something
like
</p>
<div class="example">
<pre class="example">gdb -p <var>pid</var>
</pre></div>

<p>where <code><var>pid</var></code> is the id of the R executable or front-end.
This stops the process so its state can be examined: use <code>continue</code>
to resume execution.
</p>
<p>Some &ldquo;tricks&rdquo; worth knowing follow:
</p>
<table summary="" class="menu" border="0" cellspacing="0">
<tr><td align="left" valign="top">&bull; <a href="#Finding-entry-points" accesskey="1">Finding entry points</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Inspecting-R-objects" accesskey="2">Inspecting R objects</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
</table>

<hr>
<a name="Finding-entry-points"></a>
<div class="header">
<p>
Next: <a href="#Inspecting-R-objects" accesskey="n" rel="next">Inspecting R objects</a>, Previous: <a href="#Debugging-compiled-code" accesskey="p" rel="prev">Debugging compiled code</a>, Up: <a href="#Debugging-compiled-code" accesskey="u" rel="up">Debugging compiled code</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Finding-entry-points-in-dynamically-loaded-code"></a>
<h4 class="subsection">4.4.1 Finding entry points in dynamically loaded code</h4>

<p>Under most compilation environments, compiled code dynamically loaded
into R cannot have breakpoints set within it until it is loaded.  To
use a symbolic debugger on such dynamically loaded code under
Unix-alikes use
</p>
<ul>
<li> Call the debugger on the R executable, for example by <kbd>R -d gdb</kbd>.
</li><li> Start R.
</li><li> At the R prompt, use <code>dyn.load</code> or <code>library</code> to load your
shared object.
</li><li> Send an interrupt signal.  This will put you back to the debugger
prompt.
</li><li> Set the breakpoints in your code.
</li><li> Continue execution of R by typing <kbd>signal 0<span class="key">RET</span></kbd>.
</li></ul>

<p>Under Windows signals may not be able to be used, and if so the procedure is
more complicated.  See the rw-FAQ and
<a href="http://www.stats.uwo.ca/faculty/murdoch/software/debuggingR/gdb.shtml"><code>www.stats.uwo.ca/faculty/murdoch/software/debuggingR/gdb.shtml</code></a>.
</p>

<hr>
<a name="Inspecting-R-objects"></a>
<div class="header">
<p>
Previous: <a href="#Finding-entry-points" accesskey="p" rel="prev">Finding entry points</a>, Up: <a href="#Debugging-compiled-code" accesskey="u" rel="up">Debugging compiled code</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Inspecting-R-objects-when-debugging"></a>
<h4 class="subsection">4.4.2 Inspecting R objects when debugging</h4>
<a name="index-Inspecting-R-objects-when-debugging"></a>

<p>The key to inspecting R objects from compiled code is the function
<code>PrintValue(SEXP <var>s</var>)</code> which uses the normal R printing
mechanisms to print the R object pointed to by <var>s</var>, or the safer
version <code>R_PV(SEXP <var>s</var>)</code> which will only print &lsquo;objects&rsquo;.
</p>
<p>One way to make use of <code>PrintValue</code> is to insert suitable calls
into the code to be debugged.
</p>
<p>Another way is to call <code>R_PV</code> from the symbolic debugger.
(<code>PrintValue</code> is hidden as <code>Rf_PrintValue</code>.)  For example,
from <code>gdb</code> we can use
</p>
<div class="example">
<pre class="example">(gdb) p R_PV(ab)
</pre></div>

<p>using the object <code>ab</code> from the convolution example, if we have
placed a suitable breakpoint in the convolution C code.
</p>
<p>To examine an arbitrary R object we need to work a little harder.
For example, let
</p>
<div class="example">
<pre class="example">R&gt; DF &lt;- data.frame(a = 1:3, b = 4:6)
</pre></div>

<p>By setting a breakpoint at <code>do_get</code> and typing <kbd>get(&quot;DF&quot;)</kbd> at
the R prompt, one can find out the address in memory of <code>DF</code>, for
example
</p>
<div class="example">
<pre class="example">Value returned is $1 = (SEXPREC *) 0x40583e1c
(gdb) p *$1
$2 = {
  sxpinfo = {type = 19, obj = 1, named = 1, gp = 0,
    mark = 0, debug = 0, trace = 0, = 0},
  attrib = 0x40583e80,
  u = {
    vecsxp = {
      length = 2,
      type = {c = 0x40634700 &quot;0&gt;X@D&gt;X@0&gt;X@&quot;, i = 0x40634700,
	f = 0x40634700, z = 0x40634700, s = 0x40634700},
      truelength = 1075851272,
    },
    primsxp = {offset = 2},
    symsxp = {pname = 0x2, value = 0x40634700, internal = 0x40203008},
    listsxp = {carval = 0x2, cdrval = 0x40634700, tagval = 0x40203008},
    envsxp = {frame = 0x2, enclos = 0x40634700},
    closxp = {formals = 0x2, body = 0x40634700, env = 0x40203008},
    promsxp = {value = 0x2, expr = 0x40634700, env = 0x40203008}
  }
}
</pre></div>

<p>(Debugger output reformatted for better legibility).
</p>
<p>Using <code>R_PV()</code> one can &ldquo;inspect&rdquo; the values of the various
elements of the SEXP, for example,
</p>
<div class="example">
<pre class="example">(gdb) p R_PV($1-&gt;attrib)
$names
[1] &quot;a&quot; &quot;b&quot;

$row.names
[1] &quot;1&quot; &quot;2&quot; &quot;3&quot;

$class
[1] &quot;data.frame&quot;

$3 = void
</pre></div>

<p>To find out where exactly the corresponding information is stored, one
needs to go &ldquo;deeper&rdquo;:
</p>
<div class="example">
<pre class="example">(gdb) set $a = $1-&gt;attrib
(gdb) p $a-&gt;u.listsxp.tagval-&gt;u.symsxp.pname-&gt;u.vecsxp.type.c
$4 = 0x405d40e8 &quot;names&quot;
(gdb) p $a-&gt;u.listsxp.carval-&gt;u.vecsxp.type.s[1]-&gt;u.vecsxp.type.c
$5 = 0x40634378 &quot;b&quot;
(gdb) p $1-&gt;u.vecsxp.type.s[0]-&gt;u.vecsxp.type.i[0]
$6 = 1
(gdb) p $1-&gt;u.vecsxp.type.s[1]-&gt;u.vecsxp.type.i[1]
$7 = 5
</pre></div>

<p>Another alternative is the <code>R_inspect</code> function which shows the
low-level structure of the objects recursively (addresses differ from
the above as this example is created on another machine):
</p>
<div class="example">
<pre class="example">(gdb) p R_inspect($1)
@100954d18 19 VECSXP g0c2 [OBJ,NAM(2),ATT] (len=2, tl=0)
  @100954d50 13 INTSXP g0c2 [NAM(2)] (len=3, tl=0) 1,2,3
  @100954d88 13 INTSXP g0c2 [NAM(2)] (len=3, tl=0) 4,5,6
ATTRIB:
  @102a70140 02 LISTSXP g0c0 []
    TAG: @10083c478 01 SYMSXP g0c0 [MARK,NAM(2),gp=0x4000] &quot;names&quot;
    @100954dc0 16 STRSXP g0c2 [NAM(2)] (len=2, tl=0)
      @10099df28 09 CHARSXP g0c1 [MARK,gp=0x21] &quot;a&quot;
      @10095e518 09 CHARSXP g0c1 [MARK,gp=0x21] &quot;b&quot;
    TAG: @100859e60 01 SYMSXP g0c0 [MARK,NAM(2),gp=0x4000] &quot;row.names&quot;
    @102a6f868 13 INTSXP g0c1 [NAM(2)] (len=2, tl=1) -2147483648,-3
    TAG: @10083c948 01 SYMSXP g0c0 [MARK,gp=0x4000] &quot;class&quot;
    @102a6f838 16 STRSXP g0c1 [NAM(2)] (len=1, tl=1)
      @1008c6d48 09 CHARSXP g0c2 [MARK,gp=0x21,ATT] &quot;data.frame&quot;
</pre></div>

<p>In general the representation of each object follows the format:
</p>
<div class="smallexample">
<pre class="smallexample">@&lt;address&gt; &lt;type-nr&gt; &lt;type-name&gt; &lt;gc-info&gt; [&lt;flags&gt;] ...
</pre></div>

<p>For a more fine-grained control over the depth of the recursion
and the output of vectors <code>R_inspect3</code> takes additional two character()
parameters: maximum depth and the maximal number of elements that will
be printed for scalar vectors. The defaults in <code>R_inspect</code> are
currently -1 (no limit) and 5 respectively.
</p>

<hr>
<a name="System-and-foreign-language-interfaces"></a>
<div class="header">
<p>
Next: <a href="#The-R-API" accesskey="n" rel="next">The R API</a>, Previous: <a href="#Debugging" accesskey="p" rel="prev">Debugging</a>, Up: <a href="#Top" accesskey="u" rel="up">Top</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="System-and-foreign-language-interfaces-1"></a>
<h2 class="chapter">5 System and foreign language interfaces</h2>

<table summary="" class="menu" border="0" cellspacing="0">
<tr><td align="left" valign="top">&bull; <a href="#Operating-system-access" accesskey="1">Operating system access</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Interface-functions-_002eC-and-_002eFortran" accesskey="2">Interface functions .C and .Fortran</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#dyn_002eload-and-dyn_002eunload" accesskey="3">dyn.load and dyn.unload</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Registering-native-routines" accesskey="4">Registering native routines</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Creating-shared-objects" accesskey="5">Creating shared objects</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Interfacing-C_002b_002b-code" accesskey="6">Interfacing C++ code</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Fortran-I_002fO" accesskey="7">Fortran I/O</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Linking-to-other-packages" accesskey="8">Linking to other packages</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Handling-R-objects-in-C" accesskey="9">Handling R objects in C</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Interface-functions-_002eCall-and-_002eExternal">Interface functions .Call and .External</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Evaluating-R-expressions-from-C">Evaluating R expressions from C</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Parsing-R-code-from-C">Parsing R code from C</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#External-pointers-and-weak-references">External pointers and weak references</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Vector-accessor-functions">Vector accessor functions</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Character-encoding-issues">Character encoding issues</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
</table>

<hr>
<a name="Operating-system-access"></a>
<div class="header">
<p>
Next: <a href="#Interface-functions-_002eC-and-_002eFortran" accesskey="n" rel="next">Interface functions .C and .Fortran</a>, Previous: <a href="#System-and-foreign-language-interfaces" accesskey="p" rel="prev">System and foreign language interfaces</a>, Up: <a href="#System-and-foreign-language-interfaces" accesskey="u" rel="up">System and foreign language interfaces</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Operating-system-access-1"></a>
<h3 class="section">5.1 Operating system access</h3>
<a name="index-Operating-system-access"></a>

<p>Access to operating system functions is <em>via</em> the R functions
<code>system</code> and <code>system2</code>.
<a name="index-system"></a>
<a name="index-system2"></a>
The details will differ by platform (see the on-line help), and about
all that can safely be assumed is that the first argument will be a
string <code>command</code> that will be passed for execution (not necessarily
by a shell) and the second argument to <code>system</code> will be
<code>internal</code> which if true will collect the output of the command
into an R character vector.
</p>
<p>On POSIX-compliant OSes these commands pass a command-line to a shell:
Windows is not POSIX-compliant and there is a separate function
<code>shell</code> to do so.
</p>
<p>The function <code>system.time</code>
<a name="index-system_002etime"></a>
is available for timing.  Timing on child processes is only available on
Unix-alikes, and may not be reliable there.
</p>
<hr>
<a name="Interface-functions-_002eC-and-_002eFortran"></a>
<div class="header">
<p>
Next: <a href="#dyn_002eload-and-dyn_002eunload" accesskey="n" rel="next">dyn.load and dyn.unload</a>, Previous: <a href="#Operating-system-access" accesskey="p" rel="prev">Operating system access</a>, Up: <a href="#System-and-foreign-language-interfaces" accesskey="u" rel="up">System and foreign language interfaces</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Interface-functions-_002eC-and-_002eFortran-1"></a>
<h3 class="section">5.2 Interface functions <code>.C</code> and <code>.Fortran</code></h3>
<a name="index-Interfaces-to-compiled-code"></a>

<a name="index-_002eC"></a>
<a name="index-_002eFortran"></a>

<p>These two functions provide an interface to compiled code that has been
linked into R, either at build time or <em>via</em> <code>dyn.load</code>
(see <a href="#dyn_002eload-and-dyn_002eunload">dyn.load and dyn.unload</a>).  They are primarily intended for
compiled C and FORTRAN 77 code respectively, but the <code>.C</code> function
can be used with other languages which can generate C interfaces, for
example C++ (see <a href="#Interfacing-C_002b_002b-code">Interfacing C++ code</a>).
</p>
<p>The first argument to each function is a character string specifying the
symbol name as known<a name="DOCF119" href="#FOOT119"><sup>119</sup></a> to C or
FORTRAN, that is the function or subroutine name.  (That the symbol is
loaded can be tested by, for example, <code>is.loaded(&quot;cg&quot;)</code>.  Use the
name you pass to <code>.C</code> or <code>.Fortran</code> rather than the translated
symbol name.)
</p>
<p>There can be up to 65 further arguments giving R objects to be passed
to compiled code.  Normally these are copied before being passed in, and
copied again to an R list object when the compiled code returns.  If
the arguments are given names, these are used as names for the
components in the returned list object (but not passed to the compiled
code).
</p>
<p>The following table gives the mapping between the modes of R atomic
vectors and the types of arguments to a C function or FORTRAN
subroutine.
</p>
<blockquote>
<table summary="">
<thead><tr><th>R storage mode</th><th>C type</th><th>FORTRAN type</th></tr></thead>
<tr><td><code>logical</code></td><td><code>int *</code></td><td><code>INTEGER</code></td></tr>
<tr><td><code>integer</code></td><td><code>int *</code></td><td><code>INTEGER</code></td></tr>
<tr><td><code>double</code></td><td><code>double *</code></td><td><code>DOUBLE PRECISION</code></td></tr>
<tr><td><code>complex</code></td><td><code>Rcomplex *</code></td><td><code>DOUBLE COMPLEX</code></td></tr>
<tr><td><code>character</code></td><td><code>char **</code></td><td><code>CHARACTER*255</code></td></tr>
<tr><td><code>raw</code></td><td><code>unsigned char *</code></td><td>none</td></tr>
</table>
</blockquote>

<p>Do please note the first two.  On the 64-bit Unix/Linux/macOS platforms,
<code>long</code> is 64-bit whereas <code>int</code> and <code>INTEGER</code> are 32-bit.
Code ported from S-PLUS (which uses <code>long *</code> for <code>logical</code> and
<code>integer</code>) will not work on all 64-bit platforms (although it may
appear to work on some, including Windows).  Note also that if your
compiled code is a mixture of C functions and FORTRAN subprograms the
argument types must match as given in the table above.
</p>
<p>C type <code>Rcomplex</code> is a structure with <code>double</code> members
<code>r</code> and <code>i</code> defined in the header file <samp>R_ext/Complex.h</samp>
included by <samp>R.h</samp>.  (On most platforms this is stored in a way
compatible with the C99 <code>double complex</code> type: however, it may not
be possible to pass <code>Rcomplex</code> to a C99 function expecting a
<code>double complex</code> argument.  Nor need it be compatible with a C++
<code>complex</code> type.  Moreover, the compatibility can depends on the
optimization level set for the compiler.)
</p>
<p>Only a single character string can be passed to or from FORTRAN, and the
success of this is compiler-dependent.  Other R objects can be passed
to <code>.C</code>, but it is much better to use one of the other interfaces.
</p>
<p>It is possible to pass numeric vectors of storage mode <code>double</code> to
C as <code>float *</code> or to FORTRAN as <code>REAL</code> by setting the
attribute <code>Csingle</code>, most conveniently by using the R functions
<code>as.single</code>, <code>single</code> or <code>mode</code>.  This is intended only
to be used to aid interfacing existing C or FORTRAN code.
</p>
<p>Logical values are sent as <code>0</code> (<code>FALSE</code>), <code>1</code>
(<code>TRUE</code>) or <code>INT_MIN = -2147483648</code> (<code>NA</code>, but only if
<code>NAOK</code> is true), and the compiled code should return one of these
three values.  (Non-zero values other than <code>INT_MIN</code> are mapped to
<code>TRUE</code>.)
</p>
<p>Unless formal argument <code>NAOK</code> is true, all the other arguments are
checked for missing values <code>NA</code> and for the <acronym>IEEE</acronym> special
values <code>NaN</code>, <code>Inf</code> and <code>-Inf</code>, and the presence of any
of these generates an error.  If it is true, these values are passed
unchecked.
</p>
<p>Argument <code>PACKAGE</code> confines the search for the symbol name to a
specific shared object (or use <code>&quot;base&quot;</code> for code compiled into
R).  Its use is highly desirable, as there is no way to avoid two
package writers using the same symbol name, and such name clashes are
normally sufficient to cause R to crash.  (If it is not present and
the call is from the body of a function defined in a package namespace,
the shared object loaded by the first (if any) <code>useDynLib</code>
directive will be used.  However, prior to R 2.15.2 the detection of
the correct namespace is unreliable and you are strongly recommended to
use the <code>PACKAGE</code> argument for packages to be used with earlier
versions of R.
</p>
<p>Note that the compiled code should not return anything except through
its arguments: C functions should be of type <code>void</code> and FORTRAN
subprograms should be subroutines.
</p>
<p>To fix ideas, let us consider a very simple example which convolves two
finite sequences. (This is hard to do fast in interpreted R code, but
easy in C code.)  We could do this using <code>.C</code> by
</p>
<div class="example">
<pre class="example">void convolve(double *a, int *na, double *b, int *nb, double *ab)
{
    int nab = *na + *nb - 1;

    for(int i = 0; i &lt; nab; i++)
	ab[i] = 0.0;
    for(int i = 0; i &lt; *na; i++)
	for(int j = 0; j &lt; *nb; j++)
	    ab[i + j] += a[i] * b[j];
}
</pre></div>

<p>called from R by
</p>
<div class="example">
<pre class="example">conv &lt;- function(a, b)
    .C(&quot;convolve&quot;,
       as.double(a),
       as.integer(length(a)),
       as.double(b),
       as.integer(length(b)),
       ab = double(length(a) + length(b) - 1))$ab
</pre></div>

<p>Note that we take care to coerce all the arguments to the correct R
storage mode before calling <code>.C</code>; mistakes in matching the types
can lead to wrong results or hard-to-catch errors.
</p>
<p>Special care is needed in handling <code>character</code> vector arguments in
C (or C++).  On entry the contents of the elements are duplicated and
assigned to the elements of a <code>char **</code> array, and on exit the
elements of the C array are copied to create new elements of a character
vector.  This means that the contents of the character strings of the
<code>char **</code> array can be changed, including to <code>\0</code> to shorten
the string, but the strings cannot be lengthened.  It is
possible<a name="DOCF120" href="#FOOT120"><sup>120</sup></a> to allocate a new string <em>via</em>
<code>R_alloc</code> and replace an entry in the <code>char **</code> array by the
new string.  However, when character vectors are used other than in a
read-only way, the <code>.Call</code> interface is much to be preferred.
</p>
<p>Passing character strings to FORTRAN code needs even more care, and
should be avoided where possible.  Only the first element of the
character vector is passed in, as a fixed-length (255) character array.
Up to 255 characters are passed back to a length-one character vector.
How well this works (or even if it works at all) depends on the C and
FORTRAN compilers on each platform (including on their options).  Often
what is being passed to FORTRAN is one of a small set of possible values
(a factor in R terms) which could alternatively be passed as an
integer code: similarly FORTRAN code that wants to generate diagnostic
messages can pass an integer code to a C or R wrapper which will
convert it to a character string.
</p>
<p>It is possible to pass some R objects other than atomic vectors via
<code>.C</code>, but this is only supported for historical compatibility: use
the <code>.Call</code> or <code>.External</code> interfaces for such objects.  Any
C/C++ code that includes <samp>Rinternals.h</samp> should be called via
<code>.Call</code> or <code>.External</code>.
</p>
<hr>
<a name="dyn_002eload-and-dyn_002eunload"></a>
<div class="header">
<p>
Next: <a href="#Registering-native-routines" accesskey="n" rel="next">Registering native routines</a>, Previous: <a href="#Interface-functions-_002eC-and-_002eFortran" accesskey="p" rel="prev">Interface functions .C and .Fortran</a>, Up: <a href="#System-and-foreign-language-interfaces" accesskey="u" rel="up">System and foreign language interfaces</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="dyn_002eload-and-dyn_002eunload-1"></a>
<h3 class="section">5.3 <code>dyn.load</code> and <code>dyn.unload</code></h3>
<a name="index-Dynamic-loading"></a>

<a name="index-dyn_002eload"></a>
<a name="index-dyn_002eunload"></a>

<p>Compiled code to be used with R is loaded as a shared object
(Unix-alikes including macOS, see <a href="#Creating-shared-objects">Creating shared objects</a> for more
information) or DLL (Windows).
</p>
<p>The shared object/DLL is loaded by <code>dyn.load</code> and unloaded by
<code>dyn.unload</code>.  Unloading is not normally necessary, but it is
needed to allow the DLL to be re-built on some platforms, including
Windows.
</p>
<p>The first argument to both functions is a character string giving the
path to the object.  Programmers should not assume a specific file
extension for the object/DLL (such as <samp>.so</samp>) but use a construction
like
</p>
<div class="example">
<pre class="example">file.path(path1, path2, paste0(&quot;mylib&quot;, .Platform$dynlib.ext))
</pre></div>

<p>for platform independence.  On Unix-alike systems the path supplied to
<code>dyn.load</code> can be an absolute path, one relative to the current
directory or, if it starts with &lsquo;<samp>~</samp>&rsquo;, relative to the user&rsquo;s home
directory.
</p>
<p>Loading is most often done automatically based on the <code>useDynLib()</code>
declaration in the <samp>NAMESPACE</samp> file, but may be done
explicitly <em>via</em> a call to <code>library.dynam</code>.
<a name="index-library_002edynam-1"></a>
This has the form
</p>
<div class="example">
<pre class="example">library.dynam(&quot;libname&quot;, package, lib.loc)
</pre></div>

<p>where <code>libname</code> is the object/DLL name <em>with the extension
omitted</em>.  Note that the first argument, <code>chname</code>, should
<strong>not</strong> be <code>package</code> since this will not work if the package
is installed under another name.
</p>
<p>Under some Unix-alike systems there is a choice of how the symbols are
resolved when the object is loaded, governed by the arguments
<code>local</code> and <code>now</code>.  Only use these if really necessary: in
particular using <code>now=FALSE</code> and then calling an unresolved symbol
will terminate R unceremoniously.
</p>
<p>R provides a way of executing some code automatically when a object/DLL
is either loaded or unloaded.  This can be used, for example, to
register native routines with R&rsquo;s dynamic symbol mechanism, initialize
some data in the native code, or initialize a third party library.  On
loading a DLL, R will look for a routine within that DLL named
<code>R_init_<var>lib</var></code> where <var>lib</var> is the name of the DLL file with
the extension removed.  For example, in the command
</p>
<div class="example">
<pre class="example">library.dynam(&quot;mylib&quot;, package, lib.loc)
</pre></div>

<p>R looks for the symbol named <code>R_init_mylib</code>.  Similarly, when
unloading the object, R looks for a routine named
<code>R_unload_<var>lib</var></code>, e.g., <code>R_unload_mylib</code>.  In either case,
if the routine is present, R will invoke it and pass it a single
argument describing the DLL.  This is a value of type <code>DllInfo</code>
which is defined in the <samp>Rdynload.h</samp> file in the <samp>R_ext</samp>
directory.
</p>
<p>Note that there are some implicit restrictions on this mechanism as the
basename of the DLL needs to be both a valid file name and valid as part
of a C entry point (e.g. it cannot contain &lsquo;<samp>.</samp>&rsquo;): for portable
code it is best to confine DLL names to be <acronym>ASCII</acronym> alphanumeric
plus underscore.  If entry point <code>R_init_<var>lib</var></code> is not found it
is also looked for with &lsquo;<samp>.</samp>&rsquo; replaced by &lsquo;<samp>_</samp>&rsquo;.
</p>

<p>The following example shows templates for the initialization and
unload routines for the <code>mylib</code> DLL.
</p>
<blockquote>
<table summary="" class="cartouche" border="1"><tr><td>
<div class="example">
<pre class="example">#include &lt;R_ext/Rdynload.h&gt;

void
R_init_mylib(DllInfo *info)
{
  /* Register routines,
     allocate resources. */
}

void
R_unload_mylib(DllInfo *info)
{
  /* Release resources. */
}
</pre></div>
</td></tr></table>
</blockquote>

<p>If a shared object/DLL is loaded more than once the most recent version
is used.<a name="DOCF121" href="#FOOT121"><sup>121</sup></a>  More generally, if the same symbol name
appears in several shared objects, the most recently loaded occurrence
is used.  The <code>PACKAGE</code> argument and registration (see the next
section) provide good ways to avoid any ambiguity in which occurrence is
meant.
</p>
<p>On Unix-alikes the paths used to resolve dynamically linked dependent
libraries are fixed (for security reasons) when the process is launched,
so <code>dyn.load</code> will only look for such libraries in the locations
set by the <samp>R</samp> shell script (<em>via</em> <samp>etc/ldpaths</samp>) and in
the OS-specific defaults.
</p>
<p>Windows allows more control (and less security) over where dependent
DLLs are looked for.  On all versions this includes the <code>PATH</code>
environment variable, but with lowest priority: note that it does not
include the directory from which the DLL was loaded.  It is possible to
add a single path with quite high priority <em>via</em> the <code>DLLpath</code>
argument to <code>dyn.load</code>.  This is (by default) used by
<code>library.dynam</code> to include the package&rsquo;s <samp>libs/i386</samp> or
<samp>libs/x64</samp> directory in the DLL search path.
</p>

<hr>
<a name="Registering-native-routines"></a>
<div class="header">
<p>
Next: <a href="#Creating-shared-objects" accesskey="n" rel="next">Creating shared objects</a>, Previous: <a href="#dyn_002eload-and-dyn_002eunload" accesskey="p" rel="prev">dyn.load and dyn.unload</a>, Up: <a href="#System-and-foreign-language-interfaces" accesskey="u" rel="up">System and foreign language interfaces</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Registering-native-routines-1"></a>
<h3 class="section">5.4 Registering native routines</h3>
<a name="index-Registering-native-routines"></a>

<table summary="" class="menu" border="0" cellspacing="0">
<tr><td align="left" valign="top">&bull; <a href="#Speed-considerations" accesskey="1">Speed considerations</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Converting-a-package-to-use-registration" accesskey="2">Converting a package to use registration</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Linking-to-native-routines-in-other-packages" accesskey="3">Linking to native routines in other packages</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
</table>

<p>By &lsquo;native&rsquo; routine, we mean an entry point in compiled code.
</p>
<p>In calls to <code>.C</code>, <code>.Call</code>, <code>.Fortran</code> and
<code>.External</code>, R must locate the specified native routine by
looking in the appropriate shared object/DLL.  By default, R uses the
operating-system-specific dynamic loader to lookup the symbol in
all<a name="DOCF122" href="#FOOT122"><sup>122</sup></a> loaded DLLs and the R executable
or libraries it is linked to.  Alternatively, the author of the DLL can
explicitly register routines with R and use a single,
platform-independent mechanism for finding the routines in the DLL.  One
can use this registration mechanism to provide additional information
about a routine, including the number and type of the arguments, and
also make it available to R programmers under a different name.
</p>
<p>Registering routines has two main advantages: it provides a
faster<a name="DOCF123" href="#FOOT123"><sup>123</sup></a> way to
find the address of the entry point <em>via</em> tables stored in the DLL
at compilation time, and it provides a run-time check that the entry
point is called with the right number of arguments and, optionally, the
right argument types.
</p>
<a name="index-R_005fregisterRoutines"></a>
<p>To register routines with R, one calls the C routine
<code>R_registerRoutines</code>.  This is typically done when the DLL is first
loaded within the initialization routine <code>R_init_<var>dll name</var></code>
described in <a href="#dyn_002eload-and-dyn_002eunload">dyn.load and dyn.unload</a>.  <code>R_registerRoutines</code>
takes 5 arguments.  The first is the <code>DllInfo</code> object passed by
R to the initialization routine. This is where R stores the
information about the methods.  The remaining 4 arguments are arrays
describing the routines for each of the 4 different interfaces:
<code>.C</code>, <code>.Call</code>, <code>.Fortran</code> and <code>.External</code>.  Each
argument is a <code>NULL</code>-terminated array of the element types given in
the following table:
</p>
<blockquote>
<table summary="">
<tr><td><code>.C</code></td><td><code>R_CMethodDef</code></td></tr>
<tr><td><code>.Call</code></td><td><code>R_CallMethodDef</code></td></tr>
<tr><td><code>.Fortran</code></td><td><code>R_FortranMethodDef</code></td></tr>
<tr><td><code>.External</code></td><td><code>R_ExternalMethodDef</code></td></tr>
</table>
</blockquote>

<p>Currently, the <code>R_ExternalMethodDef</code> type is the same as
<code>R_CallMethodDef</code> type and contains fields for the name of the
routine by which it can be accessed in R, a pointer to the actual
native symbol (i.e., the routine itself), and the number of arguments
the routine expects to be passed from R. For example, if we had a
routine named <code>myCall</code> defined as
</p>
<div class="example">
<pre class="example">SEXP myCall(SEXP a, SEXP b, SEXP c);
</pre></div>

<p>we would describe this as
</p>
<div class="example">
<pre class="example">static R_CallMethodDef callMethods[]  = {
  {&quot;myCall&quot;, (DL_FUNC) &amp;myCall, 3},
  {NULL, NULL, 0}
};
</pre></div>

<p>along with any other routines for the <code>.Call</code> interface. For
routines with a variable number of arguments invoked <em>via</em> the
<code>.External</code> interface, one specifies <code>-1</code> for the number of
arguments which tells R not to check the actual number passed.
</p>
<p>Routines for use with the <code>.C</code> and <code>.Fortran</code> interfaces are
described with similar data structures, but which have two additional
fields for describing the type and &ldquo;style&rdquo; of each argument.  Each of
these can be omitted. However, if specified, each should be an array
with the same number of elements as the number of parameters for the
routine.  The types array should contain the <code>SEXP</code> types
describing the expected type of the argument. (Technically, the elements
of the types array are of type <code>R_NativePrimitiveArgType</code> which is
just an unsigned integer.)  The R types and corresponding type
identifiers are provided in the following table:
</p>
<blockquote>
<table summary="">
<tr><td><code>numeric</code></td><td><code>REALSXP</code></td></tr>
<tr><td><code>integer</code></td><td><code>INTSXP</code></td></tr>
<tr><td><code>logical</code></td><td><code>LGLSXP</code></td></tr>
<tr><td><code>single</code></td><td><code>SINGLESXP</code></td></tr>
<tr><td><code>character</code></td><td><code>STRSXP</code></td></tr>
<tr><td><code>list</code></td><td><code>VECSXP</code></td></tr>
</table>
</blockquote>

<p>Consider a C routine, <code>myC</code>, declared as
</p>
<div class="example">
<pre class="example">void myC(double *x, int *n, char **names, int *status);
</pre></div>

<p>We would register it as
</p>
<div class="example">
<pre class="example">static R_NativePrimitiveArgType myC_t[] = {
    REALSXP, INTSXP, STRSXP, LGLSXP
};

static R_CMethodDef cMethods[] = {
   {&quot;myC&quot;, (DL_FUNC) &amp;myC, 4, myC_t}
   {NULL, NULL, 0}
};
</pre></div>


<p>Note that <code>.Fortran</code> entry points are mapped to lowercase, so
registration should use lowercase only.
</p>
<p>Having created the arrays describing each routine, the last step is to
actually register them with R.  We do this by calling
<code>R_registerRoutines</code>.  For example, if we have the descriptions
above for the routines accessed by the <code>.C</code> and <code>.Call</code>
we would use the following code:
</p>
<div class="example">
<pre class="example">void
R_init_myLib(DllInfo *info)
{
   R_registerRoutines(info, cMethods, callMethods, NULL, NULL);
}
</pre></div>

<p>This routine will be invoked when R loads the shared object/DLL named
<code>myLib</code>.  The last two arguments in the call to
<code>R_registerRoutines</code> are for the routines accessed by
<code>.Fortran</code> and <code>.External</code> interfaces.  In our example, these
are given as <code>NULL</code> since we have no routines of these types.
</p>
<p>When R unloads a shared object/DLL, its registrations are removed.
There is no other facility for unregistering a symbol.
</p>
<p>Examples of registering routines can be found in the different packages
in the R source tree (e.g., <strong>stats</strong> and <strong>graphics</strong>).  Also,
there is a brief, high-level introduction in <em>R News</em> (volume 1/3,
September 2001, pages 20&ndash;23,
<a href="https://www.r-project.org/doc/Rnews/Rnews_2001-3.pdf">https://www.r-project.org/doc/Rnews/Rnews_2001-3.pdf</a>).
</p>
<p>Once routines are registered, they can be referred to as R objects if
they this is arranged in the <code>useDynLib</code> call in the package&rsquo;s
<samp>NAMESPACE</samp> file (see <a href="#useDynLib">useDynLib</a>).  So for example the
<strong>stats</strong> package has
</p><div class="example">
<pre class="example"># Refer to all C/Fortran routines by their name prefixed by C_
useDynLib(stats, .registration = TRUE, .fixes = &quot;C_&quot;)
</pre></div>

<p>in its <samp>NAMESPACE</samp> file, and then <code>ansari.test</code>&rsquo;s default
methods can contain
</p><div class="example">
<pre class="example">	pansari &lt;- function(q, m, n)
	    .C(C_pansari, as.integer(length(q)), p = as.double(q),
		as.integer(m), as.integer(n))$p
</pre></div>

<p>This avoids the overhead of looking up an entry point each time it is
used, and ensures that the entry point in the package is the one used
(without a <code>PACKAGE = &quot;pkg&quot;</code> argument).
</p>
<p><code>R_init_</code> routines are often of the form
</p><div class="example">
<pre class="example">void attribute_visible R_init_mypkg(DllInfo *dll)
{
    R_registerRoutines(dll, CEntries, CallEntries, FortEntries,
                       ExternalEntries);
    R_useDynamicSymbols(dll, FALSE);
    R_forceSymbols(dll, TRUE);
...
}
</pre></div>

<p><a name="index-R_005fuseDynamicSymbols"></a>
<a name="index-R_005fforceSymbols"></a>
The <code>R_useDynamicSymbols</code> call says the DLL is not to be searched
for entry points specified by character strings so <code>.C</code> etc calls
will only find registered symbols: the <code>R_forceSymbols</code> call only
allows <code>.C</code> etc calls which specify entry points by R objects
such as <code>C_pansari</code> (and not by character strings).  Each provides
some protection against accidentally finding your entry points when
people supply a character string without a package, and avoids slowing
down such searches.  (For the visibility attribute see <a href="#Controlling-visibility">Controlling visibility</a>.)
</p>
<p>In more detail, if a package <code>mypkg</code> contains entry points
<code>reg</code> and <code>unreg</code> and the first is registered as a 0-argument
<code>.Call</code> routine, we could use (from code in the package)
</p>
<div class="example">
<pre class="example">.Call(&quot;reg&quot;)
.Call(&quot;unreg&quot;)
</pre></div>

<p>Without or with registration, these will both work.  If
<code>R_init_mypkg</code> calls <code>R_useDynamicSymbols(dll, FALSE)</code>, only
the first will work.  If in addition to registration the
<samp>NAMESPACE</samp> file contains
</p>
<div class="example">
<pre class="example">useDynLib(mypkg, .registration = TRUE, .fixes = &quot;C_&quot;)
</pre></div>

<p>then we can call <code>.Call(C_reg)</code>.  Finally, if <code>R_init_mypkg</code>
also calls <code>R_forceSymbols(dll, TRUE)</code>, only <code>.Call(C_reg)</code>
will work (and not <code>.Call(&quot;reg&quot;)</code>).  This is usually what we want:
it ensures that all of our own <code>.Call</code> calls go directly to the
intended code in our package and that no one else accidentally finds our
entry points.  (Should someone need to call our code from outside the
package, for example for debugging, they can use
<code>.Call(mypkg:::C_reg)</code>.)
</p>

<hr>
<a name="Speed-considerations"></a>
<div class="header">
<p>
Next: <a href="#Converting-a-package-to-use-registration" accesskey="n" rel="next">Converting a package to use registration</a>, Previous: <a href="#Registering-native-routines" accesskey="p" rel="prev">Registering native routines</a>, Up: <a href="#Registering-native-routines" accesskey="u" rel="up">Registering native routines</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Speed-considerations-1"></a>
<h4 class="subsection">5.4.1 Speed considerations</h4>

<p>Sometimes registering native routines or using a <code>PACKAGE</code> argument
can make a large difference.  The results can depend quite markedly on
the OS (and even if it is 32- or 64-bit), on the version of R and
what else is loaded into R at the time.
</p>
<p>To fix ideas, first consider <code>x84_64</code> OS 10.7 and R 2.15.2.  A
simple <code>.Call</code> function might be
</p><div class="example">
<pre class="example">foo &lt;- function(x) .Call(&quot;foo&quot;, x)
</pre></div>
<p>with C code
</p><div class="example">
<pre class="example">#include &lt;Rinternals.h&gt;

SEXP foo(SEXP x)
{
    return x;
}
</pre></div>
<p>If we compile with by <code>R CMD SHLIB foo.c</code>, load the code by
<code>dyn.load(&quot;foo.so&quot;)</code> and run <code>foo(pi)</code> it took around 22
microseconds (us). Specifying the DLL by
</p><div class="example">
<pre class="example">foo2 &lt;- function(x) .Call(&quot;foo&quot;, x, PACKAGE = &quot;foo&quot;)
</pre></div>
<p>reduced the time to 1.7 us.
</p>
<p>Now consider making these functions part of a package whose
<samp>NAMESPACE</samp> file uses <code>useDynlib(foo)</code>.  This immediately
reduces the running time as <code>&quot;foo&quot;</code> will be preferentially looked
for <samp>foo.dll</samp>.  Without specifying <code>PACKAGE</code> it took about 5
us (it needs to fathom out the appropriate DLL each time it is invoked
but it does not need to search all DLLs), and with the <code>PACKAGE</code>
argument it is again about 1.7 us.
</p>
<p>Next suppose the package has registered the native routine <code>foo</code>.
Then <code>foo()</code> still has to find the appropriate DLL but can get to
the entry point in the DLL faster, in about 4.2 us.  And <code>foo2()</code>
now takes about 1 us.  If we register the symbols in the
<samp>NAMESPACE</samp> file and use
</p><div class="example">
<pre class="example">foo3 &lt;- function(x) .Call(C_foo, x)
</pre></div>
<p>then the address for the native routine is looked up just once when the
package is loaded, and <code>foo3(pi)</code> takes about 0.8 us.
</p>
<p>Versions using <code>.C()</code> rather than <code>.Call()</code> took about 0.2 us
longer.
</p>
<p>These are all quite small differences, but C routines are not uncommonly
invoked millions of times for run times of a few microseconds each, and
those doing such things may wish to be aware of the differences.
</p>
<p>On Linux and Solaris there is a smaller overhead in looking up
symbols.
</p>
<p>Symbol lookup on Windows used to be far slower, so R maintains a
small cache.  If the cache is currently empty enough that the symbol can
be stored in the cache then the performance is similar to Linux and
Solaris: if not it may be slower.  R&rsquo;s own code always uses
registered symbols and so these never contribute to the cache: however
many other packages do rely on symbol lookup.
</p>
<p>In more recent versions of R all the standard packages register
native symbols and do not allow symbol search, so in a new session
<code>foo()</code> can only look in <samp>foo.so</samp> and may be as fast as
<code>foo2()</code>.  This will no longer apply when many contributed packages
are loaded, and generally those last loaded are searched first.  For
example, consider R 3.3.2 on x86_64 Linux.  In an empty R session,
both <code>foo()</code> and <code>foo2()</code> took about 0.75 us; however after
packages <a href="https://CRAN.R-project.org/package=igraph"><strong>igraph</strong></a> and <a href="https://CRAN.R-project.org/package=spatstat"><strong>spatstat</strong></a> had been loaded (which
loaded another 12 DLLs), <code>foo()</code> took 3.6 us but <code>foo2()</code>
still took about 0.80 us.  Using registration in a package reduced this
to 0.55 us and <code>foo3()</code> took 0.40 us, times which were unchanged
when further packages were loaded.
</p>
<hr>
<a name="Converting-a-package-to-use-registration"></a>
<div class="header">
<p>
Next: <a href="#Linking-to-native-routines-in-other-packages" accesskey="n" rel="next">Linking to native routines in other packages</a>, Previous: <a href="#Speed-considerations" accesskey="p" rel="prev">Speed considerations</a>, Up: <a href="#Registering-native-routines" accesskey="u" rel="up">Registering native routines</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Example_003a-converting-a-package-to-use-registration"></a>
<h4 class="subsection">5.4.2 Example: converting a package to use registration</h4>

<p>The <strong>splines</strong> package was converted to use symbol registration in
2001, but we can use it as an example<a name="DOCF124" href="#FOOT124"><sup>124</sup></a> of what needs to be done for a small package.
</p>
<ul>
<li> Find the relevant entry points.
This is somewhat OS-specific, but something like the following should be
possible at the OS command-line

<div class="example">
<pre class="example">nm -g /path/to/splines.so | grep &quot; T &quot;
0000000000002670 T _spline_basis
0000000000001ec0 T _spline_value
</pre></div>

<p>This indicates that there are two relevant entry points. (They may or
may not have a leading underscore, as here. Fortran entry points will
have a trailing underscore.)  Check in the R code that they are
called by the package and how: in this case they are used by
<code>.Call</code>.
</p>
<p>Alternatively, examine the package&rsquo;s R code for all <code>.C</code>,
<code>.Fortran</code>, <code>.Call</code> and <code>.External</code> calls.
</p>
</li><li> Construct the registration table.
First write skeleton registration code, conventionally in file
<samp>src/init.c</samp> (or at the end of the only C source file in the
package):

<div class="example">
<pre class="example">#include &lt;stdlib.h&gt; // for NULL
#include &lt;R_ext/Rdynload.h&gt;

#define CALLDEF(name, n)  {#name, (DL_FUNC) &amp;name, n}

const static R_CallMethodDef R_CallDef[] = {
   CALLDEF(spline_basis, ?),
   CALLDEF(spline_value, ?),
   {NULL, NULL, 0}
};

void R_init_splines(DllInfo *dll)
{
    R_registerRoutines(dll, NULL, R_CallDef, NULL, NULL);
}
</pre></div>

<p>and then replace the <code>?</code> in the skeleton with the actual numbers of
arguments.  You will need to add declarations of the functions unless
appending to the only C source file.  Some packages will already have
these in a header file, or you could create one and include it in
<samp>init.c</samp>, for example <samp>splines.h</samp> containing
</p>
<div class="smallexample">
<pre class="smallexample">#include &lt;Rinternals.h&gt; // for SEXP
extern SEXP spline_basis(SEXP knots, SEXP order, SEXP xvals, SEXP derivs);
extern SEXP spline_value(SEXP knots, SEXP coeff, SEXP order, SEXP x, SEXP deriv);
</pre></div>

<p>For examples of registering other types of calls, see packages
<strong>graphics</strong> and <strong>stats</strong>.  In particular, when registering entry
points for <code>.Fortran</code> one needs declarations as if called from C,
such as
</p>
<div class="example">
<pre class="example">#include &lt;R_ext/RS.h&gt;
void F77_NAME(supsmu)(int *n, double *x, double *y,
                      double *w, int *iper, double *span, double *alpha,
                      double *smo, double *sc, double *edf);
</pre></div>

<p>One can get away with inaccurate argument lists in the declarations: it
is easy to specify the arguments for <code>.Call</code> (all <code>SEXP</code>) and
<code>.External</code> (one <code>SEXP</code>) and as the arguments for <code>.C</code>
and <code>.Fortran</code> are all pointers, specifying them as <code>void *</code>
suffices. (For most platforms one can omit all the arguments.)
</p>
</li><li> (Optional but highly recommended.)  Restrict <code>.Call</code> etc to using the
symbols you chose to register by editing <samp>src/init.c</samp> to contain

<div class="example">
<pre class="example">void R_init_splines(DllInfo *dll)
{
    R_registerRoutines(dll, NULL, R_CallDef, NULL, NULL);
    R_useDynamicSymbols(dll, FALSE);
}
</pre></div>

</li></ul>

<p>A skeleton for the steps so far can be made using tools in R 3.4.0
and later.
</p>
<p>The remaining steps are optional but recommended.
</p>
<ul>
<li> Edit the <samp>NAMESPACE</samp> file to create R objects for the registered
symbols:

<div class="example">
<pre class="example">useDynLib(splines, .registration = TRUE, .fixes = &quot;C_&quot;)
</pre></div>

</li><li> Find all the relevant calls in the R code and edit them to use the
R objects.  This entailed changing the lines

<div class="smallexample">
<pre class="smallexample">temp &lt;- .Call(&quot;spline_basis&quot;, knots, ord, x, derivs, PACKAGE = &quot;splines&quot;)
y[accept] &lt;- .Call(&quot;spline_value&quot;, knots, coeff, ord, x[accept], deriv, PACKAGE = &quot;splines&quot;)
y = .Call&quot;spline_value&quot;, knots, coef(object), ord, x, deriv, PACKAGE = &quot;splines&quot;))
</pre></div>
<p>to
</p>
<div class="smallexample">
<pre class="smallexample">temp &lt;- .Call(C_spline_basis, knots, ord, x, derivs)
y[accept] &lt;- .Call(C_spline_value, knots, coeff, ord, x[accept], deriv)
y = .Call(C_spline_value, knots, coef(object), ord, x, deriv))
</pre></div>

</li><li> Restrict <code>.Call</code> to using the R symbols by editing
<samp>src/init.c</samp> to contain

<div class="example">
<pre class="example">void R_init_splines(DllInfo *dll)
{
    R_registerRoutines(dll, NULL, R_CallDef, NULL, NULL);
    R_useDynamicSymbols(dll, FALSE);
    R_forceSymbols(dll, TRUE);
}
</pre></div>

</li><li> Consider visibility.  On some OSes we can hide entry points from the
loader, which precludes any possible name clashes and calling them
accidentally (usually with incorrect arguments and crashing the R
process). If we repeat the first step we now see

<div class="example">
<pre class="example">nm -g /path/to/splines.so | grep &quot; T &quot;
0000000000002e00 T _R_init_splines
00000000000025e0 T _spline_basis
0000000000001e20 T _spline_value
</pre></div>

<p>If there were any entry points not intended to be used by the package we
should try to avoid exporting them, for example by making them
<code>static</code>.  Now the two relevant entry points are only accessed
<em>via</em> the registration table, we can hide them.  There are two ways
to do so on some Unix-alikes.  We can hide individual entry points
<em>via</em>
</p>
<div class="example">
<pre class="example">#include &lt;R_ext/Visibility.h&gt;

SEXP attribute_hidden
spline_basis(SEXP knots, SEXP order, SEXP xvals, SEXP derivs)
&hellip;

SEXP attribute_hidden
spline_value(SEXP knots, SEXP coeff, SEXP order, SEXP x, SEXP deriv)
&hellip;
</pre></div>

<p>Alternatively, we can change the default visibility for all C symbols by
including
</p>
<div class="example">
<pre class="example">PKG_CFLAGS = $(C_VISIBILITY)
</pre></div>

<p>in <samp>src/Makevars</samp>, and then we need to allow registration by
declaring <code>R_init_splines</code> to be visible:
</p>
<div class="example">
<pre class="example">#include &lt;R_ext/Visibility.h&gt;

void attribute_visible
R_init_splines(DllInfo *dll)
&hellip;
</pre></div>

<p>See <a href="#Controlling-visibility">Controlling visibility</a> for more details, including using Fortran
code and ways to restrict visibility on Windows.
</p>
</li><li> We end up with a file <samp>src/init.c</samp> containing

<blockquote>
<table summary="" class="cartouche" border="1"><tr><td>
<div class="example">
<pre class="example">#include &lt;stdlib.h&gt;
#include &lt;R_ext/Rdynload.h&gt;
#include &lt;R_ext/Visibility.h&gt;  // optional

#include &quot;splines.h&quot;

#define CALLDEF(name, n)  {#name, (DL_FUNC) &amp;name, n}

const static R_CallMethodDef R_CallDef[] = {
    CALLDEF(spline_basis, 4),
    CALLDEF(spline_value, 5),
    {NULL, NULL, 0}
};

void
attribute_visible  // optional
R_init_splines(DllInfo *dll)
{
    R_registerRoutines(dll, NULL, R_CallDef, NULL, NULL);
    R_useDynamicSymbols(dll, FALSE);
    R_forceSymbols(dll, TRUE);
}
</pre></div>
</td></tr></table>
</blockquote>

</li></ul>

<hr>
<a name="Linking-to-native-routines-in-other-packages"></a>
<div class="header">
<p>
Previous: <a href="#Converting-a-package-to-use-registration" accesskey="p" rel="prev">Converting a package to use registration</a>, Up: <a href="#Registering-native-routines" accesskey="u" rel="up">Registering native routines</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Linking-to-native-routines-in-other-packages-1"></a>
<h4 class="subsection">5.4.3 Linking to native routines in other packages</h4>

<p>In addition to registering C routines to be called by R, it can at
times be useful for one package to make some of its C routines available
to be called by C code in another package.  The interface consists of
two routines declared in header <samp>R_ext/Rdynload.h</samp> as
</p>
<a name="index-R_005fRegisterCCallable"></a>
<a name="index-R_005fGetCCallable"></a>
<div class="example">
<pre class="example">void R_RegisterCCallable(const char *package, const char *name,
			 DL_FUNC fptr);
DL_FUNC R_GetCCallable(const char *package, const char *name);
</pre></div>

<p>A package <strong>packA</strong> that wants to make a C routine <code>myCfun</code>
available to C code in other packages would include the call
</p>
<div class="example">
<pre class="example">R_RegisterCCallable(&quot;packA&quot;, &quot;myCfun&quot;, myCfun);
</pre></div>
<p>in its initialization function <code>R_init_packA</code>.  A package
<strong>packB</strong> that wants to use this routine would retrieve the function
pointer with a call of the form
</p>
<div class="example">
<pre class="example">p_myCfun = R_GetCCallable(&quot;packA&quot;, &quot;myCfun&quot;);
</pre></div>

<p>The author of <strong>packB</strong> is responsible for ensuring that
<code>p_myCfun</code> has an appropriate declaration. In the future R may
provide some automated tools to simplify exporting larger numbers of
routines.
</p>
<p>A package that wishes to make use of header files in other packages
needs to declare them as a comma-separated list in the field
&lsquo;<samp>LinkingTo</samp>&rsquo; in the <samp>DESCRIPTION</samp> file.  This then arranges
that the <samp>include</samp> directories in the installed linked-to packages
are added to the include paths for C and C++ code.
</p>
<p>It must specify<a name="DOCF125" href="#FOOT125"><sup>125</sup></a>
&lsquo;<samp>Imports</samp>&rsquo; or &lsquo;<samp>Depends</samp>&rsquo; of those packages, for they have to be
loaded<a name="DOCF126" href="#FOOT126"><sup>126</sup></a> prior to this one
(so the path to their compiled code has been registered).
</p>

<p><acronym>CRAN</acronym> examples of the use of this mechanism include <a href="https://CRAN.R-project.org/package=coxme"><strong>coxme</strong></a>
linking to <a href="https://CRAN.R-project.org/package=bdsmatrix"><strong>bdsmatrix</strong></a> and <a href="https://CRAN.R-project.org/package=xts"><strong>xts</strong></a> linking to
<a href="https://CRAN.R-project.org/package=zoo"><strong>zoo</strong></a>
</p>
<hr>
<a name="Creating-shared-objects"></a>
<div class="header">
<p>
Next: <a href="#Interfacing-C_002b_002b-code" accesskey="n" rel="next">Interfacing C++ code</a>, Previous: <a href="#Registering-native-routines" accesskey="p" rel="prev">Registering native routines</a>, Up: <a href="#System-and-foreign-language-interfaces" accesskey="u" rel="up">System and foreign language interfaces</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Creating-shared-objects-1"></a>
<h3 class="section">5.5 Creating shared objects</h3>
<a name="index-Creating-shared-objects"></a>
<a name="index-R-CMD-SHLIB"></a>

<p>Shared objects for loading into R can be created using <code>R CMD
SHLIB</code>.  This accepts as arguments a list of files which must be object
files (with extension <samp>.o</samp>) or sources for C, C++, FORTRAN 77,
Fortran 9x, Objective C or Objective C++ (with extensions <samp>.c</samp>,
<samp>.cc</samp> or <samp>.cpp</samp>, <samp>.f</samp>, <samp>.f90</samp> or <samp>.f95</samp>,
<samp>.m</samp>, and <samp>.mm</samp> or <samp>.M</samp>, respectively), or commands to be
passed to the linker.  See <kbd>R CMD SHLIB --help</kbd> (or the R help
for <code>SHLIB</code>) for usage information.
</p>
<p>If compiling the source files does not work &ldquo;out of the box&rdquo;, you can
specify additional flags by setting some of the variables
<a name="index-PKG_005fCPPFLAGS"></a>
<code>PKG_CPPFLAGS</code> (for the C preprocessor, typically &lsquo;<samp>-I</samp>&rsquo; flags),
<a name="index-PKG_005fCFLAGS"></a>
<a name="index-PKG_005fCXXFLAGS"></a>
<a name="index-PKG_005fFFLAGS"></a>
<a name="index-PKG_005fFCFLAGS"></a>
<a name="index-PKG_005fOBJCFLAGS"></a>
<a name="index-PKG_005fOBJCXXFLAGS"></a>
<code>PKG_CFLAGS</code>, <code>PKG_CXXFLAGS</code>, <code>PKG_FFLAGS</code>,
<code>PKG_FCFLAGS</code>, <code>PKG_OBJCFLAGS</code>, and <code>PKG_OBJCXXFLAGS</code>
(for the C, C++, FORTRAN 77, Fortran 9x, Objective C, and Objective C++
compilers, respectively) in the file <samp>Makevars</samp> in the compilation
directory (or, of course, create the object files directly from the
command line).
<a name="index-PKG_005fLIBS"></a>
Similarly, variable <code>PKG_LIBS</code> in <samp>Makevars</samp> can be used for
additional &lsquo;<samp>-l</samp>&rsquo; and &lsquo;<samp>-L</samp>&rsquo; flags to be passed to the linker when
building the shared object. (Supplying linker commands as arguments to
<code>R CMD SHLIB</code> will take precedence over <code>PKG_LIBS</code> in
<samp>Makevars</samp>.)
</p>
<a name="index-OBJECTS-1"></a>
<p>It is possible to arrange to include compiled code from other languages
by setting the macro &lsquo;<samp>OBJECTS</samp>&rsquo; in file <samp>Makevars</samp>, together
with suitable rules to make the objects.
</p>
<p>Flags which are already set (for example in file
<samp>etc<var>R_ARCH</var>/Makeconf</samp>) can be overridden by the environment
variable <code>MAKEFLAGS</code> (at least for systems using a POSIX-compliant
<code>make</code>), as in (Bourne shell syntax)
</p>
<div class="example">
<pre class="example">MAKEFLAGS=&quot;CFLAGS=-O3&quot; R CMD SHLIB *.c
</pre></div>

<p>It is also possible to set such variables in personal <samp>Makevars</samp>
files, which are read after the local <samp>Makevars</samp> and the system
makefiles or in a site-wide <samp>Makevars.site</samp> file.
See <a href="http://cran.r-project.org/doc/manuals/R-admin.html#Customizing-package-compilation">Customizing package compilation</a> in <cite>R Installation and Administration</cite>,
</p>

<p>Note that as <code>R CMD SHLIB</code> uses Make, it will not remake a shared
object just because the flags have changed, and if <samp>test.c</samp> and
<samp>test.f</samp> both exist in the current directory
</p>
<div class="example">
<pre class="example">R CMD SHLIB test.f
</pre></div>

<p>will compile <samp>test.c</samp>!
</p>

<p>If the <samp>src</samp> subdirectory of an add-on package contains source code
with one of the extensions listed above or a file <samp>Makevars</samp> but
<strong>not</strong> a file <samp>Makefile</samp>, <code>R CMD INSTALL</code> creates a
shared object (for loading into R through <code>useDynlib</code> in the
<samp>NAMESPACE</samp>, or in the <code>.onLoad</code> function of the package)
using the <code>R CMD SHLIB</code> mechanism.  If file <samp>Makevars</samp>
exists it is read first, then the system makefile and then any personal
<samp>Makevars</samp> files.
</p>
<p>If the <samp>src</samp> subdirectory of package contains a file
<samp>Makefile</samp>, this is used by <code>R CMD INSTALL</code> in place of the
<code>R CMD SHLIB</code> mechanism.  <code>make</code> is called with makefiles
<samp><var>R_HOME</var>/etc<var>R_ARCH</var>/Makeconf</samp>, <samp>src/Makefile</samp> and
any personal <samp>Makevars</samp> files (in that order).  The first target
found in <samp>src/Makefile</samp> is used.
</p>
<p>It is better to make use of a <samp>Makevars</samp> file rather than a
<samp>Makefile</samp>: the latter should be needed only exceptionally.
</p>

<p>Under Windows the same commands work, but <samp>Makevars.win</samp> will be
used in preference to <samp>Makevars</samp>, and only <samp>src/Makefile.win</samp>
will be used by <code>R CMD INSTALL</code> with <samp>src/Makefile</samp> being
ignored.  For past experiences of building DLLs with a variety of
compilers, see file &lsquo;<samp>README.packages</samp>&rsquo; and
<a href="http://www.stats.uwo.ca/faculty/murdoch/software/compilingDLLs/">http://www.stats.uwo.ca/faculty/murdoch/software/compilingDLLs/</a>
.  Under Windows you can supply an exports definitions file called
<samp><var>dllname</var>-win.def</samp>: otherwise all entry points in objects (but
not libraries) supplied to <code>R CMD SHLIB</code> will be exported from the
DLL.  An example is <samp>stats-win.def</samp> for the <strong>stats</strong> package: a
<acronym>CRAN</acronym> example in package <a href="https://CRAN.R-project.org/package=fastICA"><strong>fastICA</strong></a>.
</p>
<p>If you feel tempted to read the source code and subvert these
mechanisms, please resist.  Far too much developer time has been wasted
in chasing down errors caused by failures to follow this documentation,
and even more by package authors demanding explanations as to why their
packages no longer work.
In particular, undocumented environment or <code>make</code> variables are
not for use by package writers and are subject to change without notice.
</p>
<hr>
<a name="Interfacing-C_002b_002b-code"></a>
<div class="header">
<p>
Next: <a href="#Fortran-I_002fO" accesskey="n" rel="next">Fortran I/O</a>, Previous: <a href="#Creating-shared-objects" accesskey="p" rel="prev">Creating shared objects</a>, Up: <a href="#System-and-foreign-language-interfaces" accesskey="u" rel="up">System and foreign language interfaces</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Interfacing-C_002b_002b-code-1"></a>
<h3 class="section">5.6 Interfacing C++ code</h3>
<a name="index-Interfacing-C_002b_002b-code"></a>
<a name="index-C_002b_002b-code_002c-interfacing"></a>

<p>Suppose we have the following hypothetical C++ library, consisting of
the two files <samp>X.h</samp> and <samp>X.cpp</samp>, and implementing the two
classes <code>X</code> and <code>Y</code> which we want to use in R.
</p>
<blockquote>
<table summary="" class="cartouche" border="1"><tr><td>
<div class="example">
<pre class="example">// X.h

class X {
public: X (); ~X ();
};

class Y {
public: Y (); ~Y ();
};
</pre></div>
</td></tr></table>
</blockquote>

<blockquote>
<table summary="" class="cartouche" border="1"><tr><td>
<div class="example">
<pre class="example">// X.cpp

#include &lt;R.h&gt;
#include &quot;X.h&quot;

static Y y;

X::X()  { REprintf(&quot;constructor X\n&quot;); }
X::~X() { REprintf(&quot;destructor X\n&quot;);  }
Y::Y()  { REprintf(&quot;constructor Y\n&quot;); }
Y::~Y() { REprintf(&quot;destructor Y\n&quot;);  }
</pre></div>
</td></tr></table>
</blockquote>

<p>To use with R, the only thing we have to do is writing a wrapper
function and ensuring that the function is enclosed in
</p>
<div class="example">
<pre class="example">extern &quot;C&quot; {

}
</pre></div>

<p>For example,
</p>
<blockquote>
<table summary="" class="cartouche" border="1"><tr><td>
<div class="example">
<pre class="example">// X_main.cpp:

#include &quot;X.h&quot;

extern &quot;C&quot; {

void X_main () {
  X x;
}

} // extern &quot;C&quot;
</pre></div>
</td></tr></table>
</blockquote>

<p>Compiling and linking should be done with the C++ compiler-linker
(rather than the C compiler-linker or the linker itself); otherwise, the
C++ initialization code (and hence the constructor of the static
variable <code>Y</code>) are not called.  On a properly configured system, one
can simply use
</p>
<div class="example">
<pre class="example">R CMD SHLIB X.cpp X_main.cpp
</pre></div>

<p>to create the shared object, typically <samp>X.so</samp> (the file name
extension may be different on your platform).  Now starting R yields
</p>
<div class="example">
<pre class="example">R version 2.14.1 Patched (2012-01-16 r58124)
Copyright (C) 2012 The R Foundation for Statistical Computing
...
Type    &quot;q()&quot; to quit R.
</pre><pre class="example">

</pre><pre class="example">R&gt; dyn.load(paste(&quot;X&quot;, .Platform$dynlib.ext, sep = &quot;&quot;))
constructor Y
R&gt; .C(&quot;X_main&quot;)
constructor X
destructor X
list()
R&gt; q()
Save workspace image? [y/n/c]: y
destructor Y
</pre></div>

<p>The R for Windows <acronym>FAQ</acronym> (<samp>rw-FAQ</samp>) contains details of how
to compile this example under Windows.
</p>
<p>Earlier versions of this example used C++ iostreams: this is best
avoided.  There is no guarantee that the output will appear in the R
console, and indeed it will not on the R for Windows console.  Use
R code or the C entry points (see <a href="#Printing">Printing</a>) for all I/O if at all
possible.  Examples have been seen where merely loading a DLL that
contained calls to C++ I/O upset R&rsquo;s own C I/O (for example by
resetting buffers on open files).
</p>
<p>Most R header files can be included within C++ programs but they
should <strong>not</strong> be included within an <code>extern &quot;C&quot;</code> block (as
they include system headers<a name="DOCF127" href="#FOOT127"><sup>127</sup></a>).  The inclusion of system
headers in C++ changed in R 3.3.0<a name="DOCF128" href="#FOOT128"><sup>128</sup></a>, so if you care about
earlier versions of R please check your package there.
</p>
<p>Legacy header <samp>S.h</samp> cannot be used with C++.
</p>
<a name="External-C_002b_002b-code"></a>
<h4 class="subsection">5.6.1 External C++ code</h4>

<p>Quite a lot of external C++ software is header-only (e.g. most of the
Boost &lsquo;libraries&rsquo; including all those supplied by package <a href="https://CRAN.R-project.org/package=BH"><strong>BH</strong></a>,
and most of Armadillo as supplied by package <a href="https://CRAN.R-project.org/package=RcppArmadillo"><strong>RcppArmadillo</strong></a>)
and so is compiled when an R package which uses it is installed.
This causes few problems.
</p>
<p>A small number of external libraries used in R packages have a C++
interface to a library of compiled code, e.g. packages <a href="https://CRAN.R-project.org/package=rgdal"><strong>rgdal</strong></a>
and <a href="https://CRAN.R-project.org/package=rjags"><strong>rjags</strong></a>.  This raises many more problems!  The C++ interface
uses name-mangling and the
ABI<a name="DOCF129" href="#FOOT129"><sup>129</sup></a>
may depend on the compiler, version and even C++ defines<a name="DOCF130" href="#FOOT130"><sup>130</sup></a>,
so requires the package C++ code to be compiled in exactly the same way
as the library (and what that was is often undocumented).  Examples
include use of <code>g++</code> <em>vs</em> <code>clang++</code> or Solaris&rsquo;
<code>CC</code>, and the two ABIs available for C++11 in <code>g++</code> with
different defaults for GCC 4.9 and 5.x in some Linux distributions.
</p>
<p>Even fewer external libraries use C++ internally but present a C
interface, such as <a href="https://CRAN.R-project.org/package=rgeos"><strong>rgeos</strong></a>.  These require the C++ runtime
library to be linked into the package&rsquo;s shared object/DLL, and this is
best done by including a dummy C++ file in the package sources.
</p>
<p>There is a recent trend to link to the C++ interfaces offered by C
software such as <strong>hdf5</strong>, <strong>pcre</strong> and <strong>ImageMagick</strong>.  Their C
interfaces are much preferred for portability (and can be used from C++
code).  Also, the C++ interfaces are often optional in the software
build or packaged separately and so users installing from package
sources are far less likely to already have them installed.
</p>

<hr>
<a name="Fortran-I_002fO"></a>
<div class="header">
<p>
Next: <a href="#Linking-to-other-packages" accesskey="n" rel="next">Linking to other packages</a>, Previous: <a href="#Interfacing-C_002b_002b-code" accesskey="p" rel="prev">Interfacing C++ code</a>, Up: <a href="#System-and-foreign-language-interfaces" accesskey="u" rel="up">System and foreign language interfaces</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Fortran-I_002fO-1"></a>
<h3 class="section">5.7 Fortran I/O</h3>

<p>We have already warned against the use of C++ iostreams not least
because output is not guaranteed to appear on the R console, and this
warning applies equally to Fortran (77 or 9x) output to units <code>*</code>
and <code>6</code>. See <a href="#Printing-from-FORTRAN">Printing from FORTRAN</a>, which describes workarounds.
</p>
<p>In the past most Fortran compilers implemented I/O on top of the C I/O
system and so the two interworked successfully.  This was true of
<code>g77</code>, but it is less true of <code>gfortran</code> as used in
<code>gcc</code> 4 and later.  In particular, any package that makes use of
Fortran I/O will when compiled on Windows interfere with C I/O: when the
Fortran I/O is initialized (typically when the package is loaded) the C
<code>stdout</code> and <code>stderr</code> are switched to LF line endings.
(Function <code>init</code> in file <samp>src/modules/lapack/init_win.c</samp> shows
how to mitigate this.)
</p>
<hr>
<a name="Linking-to-other-packages"></a>
<div class="header">
<p>
Next: <a href="#Handling-R-objects-in-C" accesskey="n" rel="next">Handling R objects in C</a>, Previous: <a href="#Fortran-I_002fO" accesskey="p" rel="prev">Fortran I/O</a>, Up: <a href="#System-and-foreign-language-interfaces" accesskey="u" rel="up">System and foreign language interfaces</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Linking-to-other-packages-1"></a>
<h3 class="section">5.8 Linking to other packages</h3>

<p>It is not in general possible to link a DLL in package <strong>packA</strong> to a
DLL provided by package <strong>packB</strong> (for the security reasons mentioned
in <a href="#dyn_002eload-and-dyn_002eunload">dyn.load and dyn.unload</a>, and also because some platforms
distinguish between shared objects and dynamic libraries), but it is on
Windows.
</p>
<p>Note that there can be tricky versioning issues here, as package
<strong>packB</strong> could be re-installed after package <strong>packA</strong> &mdash; it is
desirable that the API provided by package <strong>packB</strong> remains
backwards-compatible.
</p>
<p>Shipping a static library in package <strong>packB</strong> for other packages to
link to avoids most of the difficulties.
</p>
<table summary="" class="menu" border="0" cellspacing="0">
<tr><td align="left" valign="top">&bull; <a href="#Unix_002dalikes" accesskey="1">Unix-alikes</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Windows" accesskey="2">Windows</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
</table>

<hr>
<a name="Unix_002dalikes"></a>
<div class="header">
<p>
Next: <a href="#Windows" accesskey="n" rel="next">Windows</a>, Previous: <a href="#Linking-to-other-packages" accesskey="p" rel="prev">Linking to other packages</a>, Up: <a href="#Linking-to-other-packages" accesskey="u" rel="up">Linking to other packages</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Unix_002dalikes-1"></a>
<h4 class="subsection">5.8.1 Unix-alikes</h4>

<p>It is possible to link a shared object in package <strong>packA</strong> to a
library provided by package <strong>packB</strong> under limited circumstances
on a Unix-alike OS.  There are severe portability issues, so this is not
recommended for a distributed package.
</p>
<p>This is easiest if <strong>packB</strong> provides a static library
<samp>packB/lib/libpackB.a</samp>.  (Note using directory <samp>lib</samp> rather
than <samp>libs</samp> is conventional, and architecture-specific
sub-directories may be needed and are assumed in the sample code
below. The code in the static library will need to be compiled with
<code>PIC</code> flags on platforms where it matters.)  Then as the code from
package <strong>packB</strong> is incorporated when package <strong>packA</strong> is
installed, we only need to find the static library at install time for
package <strong>packA</strong>.  The only issue is to find package <strong>packB</strong>, and
for that we can ask R by something like (long lines broken for
display here)
</p>
<div class="example">
<pre class="example">PKGB_PATH=‘echo ’library(packB);
  cat(system.file(&quot;lib&quot;,  package=&quot;packB&quot;, mustWork=TRUE))' \
 | &quot;${R_HOME}/bin/R&quot; --vanilla --slave`
PKG_LIBS=&quot;$(PKGB_PATH)$(R_ARCH)/libpackB.a&quot;
</pre></div>

<p>For a dynamic library <samp>packB/lib/libpackB.so</samp>
(<samp>packB/lib/libpackB.dylib</samp> on macOS: note that you cannot link to
a shared object, <samp>.so</samp>, on that platform) we could use
</p>
<div class="example">
<pre class="example">PKGB_PATH=‘echo ’library(packB);
  cat(system.file(&quot;lib&quot;, package=&quot;packB&quot;, mustWork=TRUE))' \
 | &quot;${R_HOME}/bin/R&quot; --vanilla --slave`
PKG_LIBS=-L&quot;$(PKGB_PATH)$(R_ARCH)&quot; -lpackB
</pre></div>

<p>This will work for installation, but very likely not when package
<code>packB</code> is loaded, as the path to package <strong>packB</strong>&rsquo;s <samp>lib</samp>
directory is not in the <code>ld.so</code><a name="DOCF131" href="#FOOT131"><sup>131</sup></a>  search path.  You can arrange to
put it there <strong>before</strong> R is launched by setting (on some
platforms) <code>LD_RUN_PATH</code> or <code>LD_LIBRARY_PATH</code> or adding to the
<code>ld.so</code> cache (see <code>man ldconfig</code>).  On platforms that
support it, the path to the directory containing the dynamic library can
be hardcoded at install time (which assumes that the location of package
<strong>packB</strong> will not be changed nor the package updated to a changed
API).  On systems with the <code>gcc</code> or <code>clang</code> and the
<acronym>GNU</acronym> linker (e.g. Linux) and some others this can be done by
e.g.
</p>
<div class="example">
<pre class="example">PKGB_PATH=‘echo ’library(packB);
  cat(system.file(&quot;lib&quot;, package=&quot;packB&quot;, mustWork=TRUE)))' \
 | &quot;${R_HOME}/bin/R&quot; --vanilla --slave`
PKG_LIBS=-L&quot;$(PKGB_PATH)$(R_ARCH)&quot; -Wl,-rpath,&quot;$(PKGB_PATH)$(R_ARCH)&quot; -lpackB
</pre></div>

<p>Some other systems (e.g. Solaris with its native linker) use
<samp>-Rdir</samp> rather than <samp>-rpath,dir</samp> (and this is accepted by
the compiler as well as the linker).
</p>
<p>It may be possible to figure out what is required semi-automatically
from the result of <code>R CMD libtool --config</code> (look for
&lsquo;<samp>hardcode</samp>&rsquo;).
</p>
<p>Making headers provided by package <strong>packB</strong> available to the code to
be compiled in package <strong>packA</strong> can be done by the <code>LinkingTo</code>
mechanism (see <a href="#Registering-native-routines">Registering native routines</a>).
</p>

<hr>
<a name="Windows"></a>
<div class="header">
<p>
Previous: <a href="#Unix_002dalikes" accesskey="p" rel="prev">Unix-alikes</a>, Up: <a href="#Linking-to-other-packages" accesskey="u" rel="up">Linking to other packages</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Windows-1"></a>
<h4 class="subsection">5.8.2 Windows</h4>

<p>Suppose package <strong>packA</strong> wants to make use of compiled code provided
by <strong>packB</strong> in DLL <samp>packB/libs/exB.dll</samp>, possibly the package&rsquo;s
DLL <samp>packB/libs/packB.dll</samp>.  (This can be extended to linking to
more than one package in a similar way.)  There are three issues to be
addressed:
</p>
<ul>
<li> Making headers provided by package <strong>packB</strong> available to the code to
be compiled in package <strong>packA</strong>.

<p>This is done by the <code>LinkingTo</code> mechanism (see <a href="#Registering-native-routines">Registering native routines</a>).
</p>
</li><li> preparing <code>packA.dll</code> to link to <samp>packB/libs/exB.dll</samp>.

<p>This needs an entry in <samp>Makevars.win</samp> of the form
</p>
<div class="example">
<pre class="example">PKG_LIBS= -L&lt;something&gt; -lexB
</pre></div>
<p>and one possibility is that <code>&lt;something&gt;</code> is the path to the
installed <samp>pkgB/libs</samp> directory.  To find that we need to ask R
where it is by something like
</p>
<div class="example">
<pre class="example">PKGB_PATH=‘echo ’library(packB);
  cat(system.file(&quot;libs&quot;, package=&quot;packB&quot;, mustWork=TRUE))' \
 | rterm --vanilla --slave`
PKG_LIBS= -L&quot;$(PKGB_PATH)$(R_ARCH)&quot; -lexB
</pre></div>

<p>Another possibility is to use an import library, shipping with package
<strong>packA</strong> an exports file <samp>exB.def</samp>.  Then <samp>Makevars.win</samp>
could contain
</p>
<div class="example">
<pre class="example">PKG_LIBS= -L. -lexB

all: $(SHLIB) before

before: libexB.dll.a
libexB.dll.a: exB.def
</pre></div>
<p>and then installing package <strong>packA</strong> will make and use the import
library for <samp>exB.dll</samp>.  (One way to prepare the exports file is to
use <samp>pexports.exe</samp>.)
</p>
</li><li> loading <samp>packA.dll</samp> which depends on <samp>exB.dll</samp>.

<p>If <code>exB.dll</code> was used by package <strong>packB</strong> (because it is in fact
<samp>packB.dll</samp> or <samp>packB.dll</samp> depends on it) and <strong>packB</strong> has
been loaded before <strong>packA</strong>, then nothing more needs to be done as
<samp>exB.dll</samp> will already be loaded into the R executable.  (This
is the most common scenario.)
</p>
<p>More generally, we can use the <code>DLLpath</code> argument to
<code>library.dynam</code> to ensure that <code>exB.dll</code> is found, for example
by setting
</p>
<div class="example">
<pre class="example">library.dynam(&quot;packA&quot;, pkg, lib,
	      DLLpath = system.file(&quot;libs&quot;, package=&quot;packB&quot;))
</pre></div>

<p>Note that <code>DLLpath</code> can only set one path, and so for linking to
two or more packages you would need to resort to setting environment
variable <code>PATH</code>.
</p>
</li></ul>

<hr>
<a name="Handling-R-objects-in-C"></a>
<div class="header">
<p>
Next: <a href="#Interface-functions-_002eCall-and-_002eExternal" accesskey="n" rel="next">Interface functions .Call and .External</a>, Previous: <a href="#Linking-to-other-packages" accesskey="p" rel="prev">Linking to other packages</a>, Up: <a href="#System-and-foreign-language-interfaces" accesskey="u" rel="up">System and foreign language interfaces</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Handling-R-objects-in-C-1"></a>
<h3 class="section">5.9 Handling R objects in C</h3>
<a name="index-Handling-R-objects-in-C"></a>

<p>Using C code to speed up the execution of an R function is often very
fruitful.  Traditionally this has been done <em>via</em> the <code>.C</code>
function in R.  However, if a user wants to write C code using
internal R data structures, then that can be done using the
<code>.Call</code> and <code>.External</code> functions.  The syntax for the calling
function in R in each case is similar to that of <code>.C</code>, but the
two functions have different C interfaces.  Generally the <code>.Call</code>
interface is simpler to use, but <code>.External</code> is a little more
general.
<a name="index-_002eCall"></a>
<a name="index-_002eExternal"></a>
</p>
<p>A call to <code>.Call</code> is very similar to <code>.C</code>, for example
</p>
<div class="example">
<pre class="example">.Call(&quot;convolve2&quot;, a, b)
</pre></div>

<p>The first argument should be a character string giving a C symbol name
of code that has already been loaded into R.  Up to 65 R objects
can passed as arguments.  The C side of the interface is
</p>
<div class="example">
<pre class="example">#include &lt;R.h&gt;
#include &lt;Rinternals.h&gt;

SEXP convolve2(SEXP a, SEXP b)
 ...
</pre></div>

<p>A call to <code>.External</code> is almost identical
</p>
<div class="example">
<pre class="example">.External(&quot;convolveE&quot;, a, b)
</pre></div>

<p>but the C side of the interface is different, having only one argument
</p>
<div class="example">
<pre class="example">#include &lt;R.h&gt;
#include &lt;Rinternals.h&gt;

SEXP convolveE(SEXP args)
 ...
</pre></div>

<p>Here <code>args</code> is a <code>LISTSXP</code>, a Lisp-style pairlist from which
the arguments can be extracted.
</p>
<p>In each case the R objects are available for manipulation <em>via</em>
a set of functions and macros defined in the header file
<samp>Rinternals.h</samp> or some S-compatibility macros<a name="DOCF132" href="#FOOT132"><sup>132</sup></a> defined
in <samp>Rdefines.h</samp>.  See <a href="#Interface-functions-_002eCall-and-_002eExternal">Interface functions .Call and .External</a>
for details on <code>.Call</code> and <code>.External</code>.
</p>
<p>Before you decide to use <code>.Call</code> or <code>.External</code>, you should
look at other alternatives.  First, consider working in interpreted R
code; if this is fast enough, this is normally the best option.  You
should also see if using <code>.C</code> is enough.  If the task to be
performed in C is simple enough involving only atomic vectors and
requiring no call to R, <code>.C</code> suffices. A great deal of useful
code was written using just <code>.C</code> before <code>.Call</code> and
<code>.External</code> were available.  These interfaces allow much more
control, but they also impose much greater responsibilities so need to
be used with care.  Neither <code>.Call</code> nor <code>.External</code> copy their
arguments: you should treat arguments you receive through these
interfaces as read-only.
</p>
<p>To handle R objects from within C code we use the macros and functions
that have been used to implement the core parts of R.  A
public<a name="DOCF133" href="#FOOT133"><sup>133</sup></a>  subset of these is defined in the header file
<samp>Rinternals.h</samp> in the directory <samp><var>R_INCLUDE_DIR</var></samp> (default
<samp><var>R_HOME</var>/include</samp>) that should be available on any R
installation.
</p>
<p>A substantial amount of R, including the standard packages, is
implemented using the functions and macros described here, so the R
source code provides a rich source of examples and &ldquo;how to do it&rdquo;: do
make use of the source code for inspirational examples.
</p>
<p>It is necessary to know something about how R objects are handled in
C code.  All the R objects you will deal with will be handled with
the type <em>SEXP</em><a name="DOCF134" href="#FOOT134"><sup>134</sup></a>, which is a
pointer to a structure with typedef <code>SEXPREC</code>.  Think of this
structure as a <em>variant type</em> that can handle all the usual types
of R objects, that is vectors of various modes, functions,
environments, language objects and so on.  The details are given later
in this section and in <a href="http://cran.r-project.org/doc/manuals/R-ints.html#R-Internal-Structures">R Internal
Structures</a> in <cite>R Internals</cite>, but for most
purposes the programmer does not need to know them.  Think rather of a
model such as that used by Visual Basic, in which R objects are
handed around in C code (as they are in interpreted R code) as the
variant type, and the appropriate part is extracted for, for example,
numerical calculations, only when it is needed.  As in interpreted R
code, much use is made of coercion to force the variant object to the
right type.
</p>
<table summary="" class="menu" border="0" cellspacing="0">
<tr><td align="left" valign="top">&bull; <a href="#Garbage-Collection" accesskey="1">Garbage Collection</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Allocating-storage" accesskey="2">Allocating storage</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Details-of-R-types" accesskey="3">Details of R types</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Attributes" accesskey="4">Attributes</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Classes" accesskey="5">Classes</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Handling-lists" accesskey="6">Handling lists</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Handling-character-data" accesskey="7">Handling character data</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Finding-and-setting-variables" accesskey="8">Finding and setting variables</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Some-convenience-functions" accesskey="9">Some convenience functions</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Named-objects-and-copying">Named objects and copying</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
</table>

<hr>
<a name="Garbage-Collection"></a>
<div class="header">
<p>
Next: <a href="#Allocating-storage" accesskey="n" rel="next">Allocating storage</a>, Previous: <a href="#Handling-R-objects-in-C" accesskey="p" rel="prev">Handling R objects in C</a>, Up: <a href="#Handling-R-objects-in-C" accesskey="u" rel="up">Handling R objects in C</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Handling-the-effects-of-garbage-collection"></a>
<h4 class="subsection">5.9.1 Handling the effects of garbage collection</h4>
<a name="index-Garbage-collection"></a>

<a name="index-PROTECT"></a>
<a name="index-UNPROTECT"></a>

<p>We need to know a little about the way R handles memory allocation.
The memory allocated for R objects is not freed by the user; instead,
the memory is from time to time <em>garbage collected</em>.  That is, some
or all of the allocated memory not being used is freed or marked as
re-usable.
</p>
<p>The R object types are represented by a C structure defined by a
typedef <code>SEXPREC</code> in <samp>Rinternals.h</samp>.  It contains several
things among which are pointers to data blocks and to other
<code>SEXPREC</code>s.  A <code>SEXP</code> is simply a pointer to a <code>SEXPREC</code>.
</p>
<p>If you create an R object in your C code, you must tell R that you
are using the object by using the <code>PROTECT</code> macro on a pointer to
the object. This tells R that the object is in use so it is not
destroyed during garbage collection.  Notice that it is the object which
is protected, not the pointer variable.  It is a common mistake to
believe that if you invoked <code>PROTECT(<var>p</var>)</code> at some point then
<var>p</var> is protected from then on, but that is not true once a new
object is assigned to <var>p</var>.
</p>
<p>Protecting an R object automatically protects all the R objects
pointed to in the corresponding <code>SEXPREC</code>, for example all elements
of a protected list are automatically protected.
</p>
<p>The programmer is solely responsible for housekeeping the calls to
<code>PROTECT</code>.  There is a corresponding macro <code>UNPROTECT</code> that
takes as argument an <code>int</code> giving the number of objects to
unprotect when they are no longer needed.  The protection mechanism is
stack-based, so <code>UNPROTECT(<var>n</var>)</code> unprotects the last <var>n</var>
objects which were protected.  The calls to <code>PROTECT</code> and
<code>UNPROTECT</code> must balance when the user&rsquo;s code returns.  R will
warn about <code>&quot;stack imbalance in .Call&quot;</code> (or <code>.External</code>) if
the housekeeping is wrong.
</p>
<p>Here is a small example of creating an R numeric vector in C code:
</p>
<div class="example">
<pre class="example">#include &lt;R.h&gt;
#include &lt;Rinternals.h&gt;

    SEXP ab;
      ....
    ab = PROTECT(allocVector(REALSXP, 2));
    REAL(ab)[0] = 123.45;
    REAL(ab)[1] = 67.89;
    UNPROTECT(1);
</pre></div>

<p>Now, the reader may ask how the R object could possibly get removed
during those manipulations, as it is just our C code that is running.
As it happens, we can do without the protection in this example, but in
general we do not know (nor want to know) what is hiding behind the R
macros and functions we use, and any of them might cause memory to be
allocated, hence garbage collection and hence our object <code>ab</code> to be
removed. It is usually wise to err on the side of caution and assume
that any of the R macros and functions might remove the object.
</p>
<p>In some cases it is necessary to keep better track of whether protection
is really needed.  Be particularly aware of situations where a large
number of objects are generated.  The pointer protection stack has a
fixed size (default 10,000) and can become full.  It is not a good idea
then to just <code>PROTECT</code> everything in sight and <code>UNPROTECT</code>
several thousand objects at the end. It will almost invariably be
possible to either assign the objects as part of another object (which
automatically protects them) or unprotect them immediately after use.
</p>
<p>Protection is not needed for objects which R already knows are in
use.  In particular, this applies to function arguments.
</p>
<p>There is a less-used macro <code>UNPROTECT_PTR(<var>s</var>)</code> that unprotects
the object pointed to by the <code>SEXP</code> <var>s</var>, even if it is not the
top item on the pointer protection stack. This is rarely needed outside
the parser (the R sources currently have three examples, one in
<samp>src/main/plot3d.c</samp>).
<a name="index-UNPROTECT_005fPTR"></a>
</p>
<p>Sometimes an object is changed (for example duplicated, coerced or
grown) yet the current value needs to be protected.  For these cases
<code>PROTECT_WITH_INDEX</code> saves an index of the protection location that
can be used to replace the protected value using <code>REPROTECT</code>.
<a name="index-PROTECT_005fWITH_005fINDEX"></a>
<a name="index-REPROTECT"></a>
For example (from the internal code for <code>optim</code>)
</p>
<div class="example">
<pre class="example">    PROTECT_INDEX ipx;

    ....
    s = PROTECT_WITH_INDEX(eval(OS-&gt;R_fcall, OS-&gt;R_env), &amp;ipx);
    s = REPROTECT(coerceVector(s, REALSXP), ipx);
</pre></div>

<p>Note that it is dangerous to mix <code>UNPROTECT_PTR</code> with
<code>PROTECT_WITH_INDEX</code>, as the former changes the protection
locations of objects that were protected after the one being
unprotected.
</p>
<a name="index-R_005fPreserveObject"></a>
<a name="index-R_005fReleaseObject"></a>
<p>There is another way to avoid the affects of garbage collection: a call
to <code>R_PreserveObject</code> adds an object to an internal list of objects
not to be collects, and a subsequent call to <code>R_ReleaseObject</code>
removes it from that list.  This provides a way for objects which are
not returned as part of R objects to be protected across calls to
compiled code: on the other hand it becomes the user&rsquo;s responsibility to
release them when they are no longer needed (and this often requires the
use of a finalizer).  It is less efficient that the normal protection
mechanism, and should be used sparingly.
</p>
<hr>
<a name="Allocating-storage"></a>
<div class="header">
<p>
Next: <a href="#Details-of-R-types" accesskey="n" rel="next">Details of R types</a>, Previous: <a href="#Garbage-Collection" accesskey="p" rel="prev">Garbage Collection</a>, Up: <a href="#Handling-R-objects-in-C" accesskey="u" rel="up">Handling R objects in C</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Allocating-storage-1"></a>
<h4 class="subsection">5.9.2 Allocating storage</h4>
<a name="index-Allocating-storage"></a>

<p>For many purposes it is sufficient to allocate R objects and
manipulate those.  There are quite a few <code>alloc<var>Xxx</var></code> functions
defined in <samp>Rinternals.h</samp>&mdash;you may want to explore them.
</p>
<a name="index-allocVector"></a>
<p>One that is commonly used is <code>allocVector</code>, the C-level equivalent
of R-level <code>vector()</code> and its wrappers such as <code>integer()</code>
and <code>character()</code>.  One distinction is that whereas the R
functions always initialize the elements of the vector,
<code>allocVector</code> only does so for lists, expressions and character
vectors (the cases where the elements are themselves R objects).
</p>
<p>If storage is required for C objects during the calculations this is
best allocating by calling <code>R_alloc</code>; see <a href="#Memory-allocation">Memory allocation</a>.
All of these memory allocation routines do their own error-checking, so
the programmer may assume that they will raise an error and not return
if the memory cannot be allocated.
</p>
<hr>
<a name="Details-of-R-types"></a>
<div class="header">
<p>
Next: <a href="#Attributes" accesskey="n" rel="next">Attributes</a>, Previous: <a href="#Allocating-storage" accesskey="p" rel="prev">Allocating storage</a>, Up: <a href="#Handling-R-objects-in-C" accesskey="u" rel="up">Handling R objects in C</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Details-of-R-types-1"></a>
<h4 class="subsection">5.9.3 Details of R types</h4>
<a name="index-Details-of-R-types"></a>

<p>Users of the <samp>Rinternals.h</samp> macros will need to know how the R
types are known internally.  The different R data types are
represented in C by <em>SEXPTYPE</em>.  Some of these are familiar from
R and some are internal data types.  The usual R object modes are
given in the table.
</p>
<blockquote>
<table summary="">
<thead><tr><th>SEXPTYPE</th><th>R equivalent</th></tr></thead>
<tr><td><code>REALSXP</code></td><td>numeric with storage mode <code>double</code></td></tr>
<tr><td><code>INTSXP</code></td><td>integer</td></tr>
<tr><td><code>CPLXSXP</code></td><td>complex</td></tr>
<tr><td><code>LGLSXP</code></td><td>logical</td></tr>
<tr><td><code>STRSXP</code></td><td>character</td></tr>
<tr><td><code>VECSXP</code></td><td>list (generic vector)</td></tr>
<tr><td><code>LISTSXP</code></td><td>pairlist</td></tr>
<tr><td><code>DOTSXP</code></td><td>a &lsquo;<samp>&hellip;</samp>&rsquo; object</td></tr>
<tr><td><code>NILSXP</code></td><td>NULL</td></tr>
<tr><td><code>SYMSXP</code></td><td>name/symbol</td></tr>
<tr><td><code>CLOSXP</code></td><td>function or function closure</td></tr>
<tr><td><code>ENVSXP</code></td><td>environment</td></tr>
</table>
</blockquote>

<p>Among the important internal <code>SEXPTYPE</code>s are <code>LANGSXP</code>,
<code>CHARSXP</code>, <code>PROMSXP</code>, etc.  (<strong>N.B.</strong>: although it is
possible to return objects of internal types, it is unsafe to do so as
assumptions are made about how they are handled which may be violated at
user-level evaluation.)  More details are given in <a href="http://cran.r-project.org/doc/manuals/R-ints.html#R-Internal-Structures">R Internal Structures</a> in <cite>R Internals</cite>.
</p>
<p>Unless you are very sure about the type of the arguments, the code
should check the data types.  Sometimes it may also be necessary to
check data types of objects created by evaluating an R expression in
the C code.  You can use functions like <code>isReal</code>, <code>isInteger</code>
and <code>isString</code> to do type checking.  See the header file
<samp>Rinternals.h</samp> for definitions of other such functions.  All of
these take a <code>SEXP</code> as argument and return 1 or 0 to indicate
<var>TRUE</var> or <var>FALSE</var>.
</p>
<p>What happens if the <code>SEXP</code> is not of the correct type?  Sometimes
you have no other option except to generate an error.  You can use the
function <code>error</code> for this.  It is usually better to coerce the
object to the correct type.  For example, if you find that an
<code>SEXP</code> is of the type <code>INTEGER</code>, but you need a <code>REAL</code>
object, you can change the type by using
</p>
<div class="example">
<pre class="example"><var>newSexp</var> = PROTECT(coerceVector(<var>oldSexp</var>, REALSXP));
</pre></div>

<p>Protection is needed as a new <em>object</em> is created; the object
formerly pointed to by the <code>SEXP</code> is still protected but now
unused.<a name="DOCF135" href="#FOOT135"><sup>135</sup></a>
</p>
<p>All the coercion functions do their own error-checking, and generate
<code>NA</code>s with a warning or stop with an error as appropriate.
</p>
<p>Note that these coercion functions are <em>not</em> the same as calling
<code>as.numeric</code> (and so on) in R code, as they do not dispatch on
the class of the object.  Thus it is normally preferable to do the
coercion in the calling R code.
</p>
<p>So far we have only seen how to create and coerce R objects from C
code, and how to extract the numeric data from numeric R vectors.
These can suffice to take us a long way in interfacing R objects to
numerical algorithms, but we may need to know a little more to create
useful return objects.
</p>
<hr>
<a name="Attributes"></a>
<div class="header">
<p>
Next: <a href="#Classes" accesskey="n" rel="next">Classes</a>, Previous: <a href="#Details-of-R-types" accesskey="p" rel="prev">Details of R types</a>, Up: <a href="#Handling-R-objects-in-C" accesskey="u" rel="up">Handling R objects in C</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Attributes-1"></a>
<h4 class="subsection">5.9.4 Attributes</h4>
<a name="index-Attributes"></a>

<p>Many R objects have attributes: some of the most useful are classes
and the <code>dim</code> and <code>dimnames</code> that mark objects as matrices or
arrays.  It can also be helpful to work with the <code>names</code> attribute
of vectors.
</p>
<p>To illustrate this, let us write code to take the outer product of two
vectors (which <code>outer</code> and <code>%o%</code> already do).  As usual the
R code is simple
</p>
<div class="example">
<pre class="example">out &lt;- function(x, y)
{
    storage.mode(x) &lt;- storage.mode(y) &lt;- &quot;double&quot;
    .Call(&quot;out&quot;, x, y)
}
</pre></div>

<p>where we expect <code>x</code> and <code>y</code> to be numeric vectors (possibly
integer), possibly with names.  This time we do the coercion in the
calling R code.
</p>
<p>C code to do the computations is
</p>
<div class="example">
<pre class="example">#include &lt;R.h&gt;
#include &lt;Rinternals.h&gt;

SEXP out(SEXP x, SEXP y)
{
    int nx = length(x), ny = length(y);
    SEXP ans = PROTECT(allocMatrix(REALSXP, nx, ny));
    double *rx = REAL(x), *ry = REAL(y), *rans = REAL(ans);
    for(int i = 0; i &lt; nx; i++) {
	double tmp = rx[i];
	for(int j = 0; j &lt; ny; j++)
	    rans[i + nx*j] = tmp * ry[j];
    }
    UNPROTECT(1);
    return ans;
}
</pre></div>

<p>Note the way <code>REAL</code> is used: as it is a function call it can be
considerably faster to store the result and index that.
</p>
<p>However, we would like to set the <code>dimnames</code> of the result.  We can use
</p>
<div class="example">
<pre class="example">#include &lt;R.h&gt;
#include &lt;Rinternals.h&gt;

</pre><pre class="example">SEXP out(SEXP x, SEXP y)
{
    int nx = length(x), ny = length(y);
    SEXP ans = PROTECT(allocMatrix(REALSXP, nx, ny));
    double *rx = REAL(x), *ry = REAL(y), *rans = REAL(ans);

    for(int i = 0; i &lt; nx; i++) {
      double tmp = rx[i];
      for(int j = 0; j &lt; ny; j++)
	rans[i + nx*j] = tmp * ry[j];
    }

    SEXP dimnames = PROTECT(allocVector(VECSXP, 2));
    SET_VECTOR_ELT(dimnames, 0, getAttrib(x, R_NamesSymbol));
    SET_VECTOR_ELT(dimnames, 1, getAttrib(y, R_NamesSymbol));
    setAttrib(ans, R_DimNamesSymbol, dimnames);
</pre><pre class="example">

</pre><pre class="example">    UNPROTECT(3);
    return ans;
}
</pre></div>

<p>This example introduces several new features.  The <code>getAttrib</code> and
<code>setAttrib</code>
<a name="index-getAttrib"></a>
<a name="index-setAttrib"></a>
functions get and set individual attributes.  Their second argument is a
<code>SEXP</code> defining the name in the symbol table of the attribute we
want; these and many such symbols are defined in the header file
<samp>Rinternals.h</samp>.
</p>
<p>There are shortcuts here too: the functions <code>namesgets</code>,
<code>dimgets</code> and <code>dimnamesgets</code> are the internal versions of the
default methods of <code>names&lt;-</code>, <code>dim&lt;-</code> and <code>dimnames&lt;-</code>
(for vectors and arrays), and there are functions such as
<code>GetMatrixDimnames</code> and <code>GetArrayDimnames</code>.
</p>
<p>What happens if we want to add an attribute that is not pre-defined? We
need to add a symbol for it <em>via</em> a call to
<a name="index-install"></a>
<code>install</code>.  Suppose for illustration we wanted to add an attribute
<code>&quot;version&quot;</code> with value <code>3.0</code>.  We could use
</p>
<div class="example">
<pre class="example">    SEXP version;
    version = PROTECT(allocVector(REALSXP, 1));
    REAL(version)[0] = 3.0;
    setAttrib(ans, install(&quot;version&quot;), version);
    UNPROTECT(1);
</pre></div>

<p>Using <code>install</code> when it is not needed is harmless and provides a
simple way to retrieve the symbol from the symbol table if it is already
installed. However, the lookup takes a non-trivial amount of time, so
consider code such as
</p>
<div class="example">
<pre class="example">static SEXP VerSymbol = NULL;
...
    if (VerSymbol == NULL) VerSymbol = install(&quot;version&quot;);
</pre></div>

<p>if it is to be done frequently.
</p>
<p>This example can be simplified by another convenience function:
</p>
<div class="example">
<pre class="example">    SEXP version = PROTECT(ScalarReal(3.0));
    setAttrib(ans, install(&quot;version&quot;), version);
    UNPROTECT(1);
</pre></div>


<hr>
<a name="Classes"></a>
<div class="header">
<p>
Next: <a href="#Handling-lists" accesskey="n" rel="next">Handling lists</a>, Previous: <a href="#Attributes" accesskey="p" rel="prev">Attributes</a>, Up: <a href="#Handling-R-objects-in-C" accesskey="u" rel="up">Handling R objects in C</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Classes-1"></a>
<h4 class="subsection">5.9.5 Classes</h4>
<a name="index-Classes"></a>

<p>In R the class is just the attribute named <code>&quot;class&quot;</code> so it can
be handled as such, but there is a shortcut <code>classgets</code>.  Suppose
we want to give the return value in our example the class <code>&quot;mat&quot;</code>.
We can use
</p>
<div class="example">
<pre class="example">#include &lt;R.h&gt;
#include &lt;Rinternals.h&gt;
      ....
    SEXP ans, dim, dimnames, class;
      ....
    class = PROTECT(allocVector(STRSXP, 1));
    SET_STRING_ELT(class, 0, mkChar(&quot;mat&quot;));
    classgets(ans, class);
    UNPROTECT(4);
    return ans;
}
</pre></div>

<p>As the value is a character vector, we have to know how to create that
from a C character array, which we do using the function
<code>mkChar</code>.
</p>
<hr>
<a name="Handling-lists"></a>
<div class="header">
<p>
Next: <a href="#Handling-character-data" accesskey="n" rel="next">Handling character data</a>, Previous: <a href="#Classes" accesskey="p" rel="prev">Classes</a>, Up: <a href="#Handling-R-objects-in-C" accesskey="u" rel="up">Handling R objects in C</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Handling-lists-1"></a>
<h4 class="subsection">5.9.6 Handling lists</h4>
<a name="index-Handling-lists"></a>

<p>Some care is needed with lists, as R moved early on from using
LISP-like lists (now called &ldquo;pairlists&rdquo;) to S-like generic vectors.
As a result, the appropriate test for an object of mode <code>list</code> is
<code>isNewList</code>, and we need <code>allocVector(VECSXP, <var>n</var></code>) and
<em>not</em> <code>allocList(<var>n</var>)</code>.
</p>
<p>List elements can be retrieved or set by direct access to the elements
of the generic vector.  Suppose we have a list object
</p>
<div class="example">
<pre class="example">a &lt;- list(f = 1, g = 2, h = 3)
</pre></div>

<p>Then we can access <code>a$g</code> as <code>a[[2]]</code> by
</p>
<div class="example">
<pre class="example">    double g;
      ....
    g = REAL(VECTOR_ELT(a, 1))[0];
</pre></div>

<p>This can rapidly become tedious, and the following function (based on
one in package <strong>stats</strong>) is very useful:
</p>
<div class="example">
<pre class="example">/* get the list element named str, or return NULL */

SEXP getListElement(SEXP list, const char *str)
{
    SEXP elmt = R_NilValue, names = getAttrib(list, R_NamesSymbol);
</pre><pre class="example">

</pre><pre class="example">    for (int i = 0; i &lt; length(list); i++)
	if(strcmp(CHAR(STRING_ELT(names, i)), str) == 0) {
	   elmt = VECTOR_ELT(list, i);
	   break;
	}
    return elmt;
}
</pre></div>

<p>and enables us to say
</p>
<div class="example">
<pre class="example">  double g;
  g = REAL(getListElement(a, &quot;g&quot;))[0];
</pre></div>

<hr>
<a name="Handling-character-data"></a>
<div class="header">
<p>
Next: <a href="#Finding-and-setting-variables" accesskey="n" rel="next">Finding and setting variables</a>, Previous: <a href="#Handling-lists" accesskey="p" rel="prev">Handling lists</a>, Up: <a href="#Handling-R-objects-in-C" accesskey="u" rel="up">Handling R objects in C</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Handling-character-data-1"></a>
<h4 class="subsection">5.9.7 Handling character data</h4>
<a name="index-handling-character-data"></a>

<p>R character vectors are stored as <code>STRSXP</code>s, a vector type like
<code>VECSXP</code> where every element is of type <code>CHARSXP</code>.  The
<code>CHARSXP</code> elements of <code>STRSXP</code>s are accessed using
<code>STRING_ELT</code> and <code>SET_STRING_ELT</code>.
</p>
<p><code>CHARSXP</code>s are read-only objects and must never be modified.  In
particular, the C-style string contained in a <code>CHARSXP</code> should be
treated as read-only and for this reason the <code>CHAR</code> function used
to access the character data of a <code>CHARSXP</code> returns <code>(const
char *)</code> (this also allows compilers to issue warnings about improper
use).  Since <code>CHARSXP</code>s are immutable, the same <code>CHARSXP</code> can
be shared by any <code>STRSXP</code> needing an element representing the same
string. R maintains a global cache of <code>CHARSXP</code>s so that there
is only ever one <code>CHARSXP</code> representing a given string in memory.
</p>
<a name="index-mkChar"></a>
<a name="index-mkCharLen"></a>
<p>You can obtain a <code>CHARSXP</code> by calling <code>mkChar</code> and providing a
nul-terminated C-style string.  This function will return a pre-existing
<code>CHARSXP</code> if one with a matching string already exists, otherwise
it will create a new one and add it to the cache before returning it to
you.   The variant <code>mkCharLen</code> can be used to create a
<code>CHARSXP</code> from part of a buffer and will ensure null-termination.
</p>
<p>Note that R character strings are restricted to <code>2^31 - 1</code>
bytes, and hence so should the input to <code>mkChar</code> be (C allows
longer strings on 64-bit platforms).
</p>
<hr>
<a name="Finding-and-setting-variables"></a>
<div class="header">
<p>
Next: <a href="#Some-convenience-functions" accesskey="n" rel="next">Some convenience functions</a>, Previous: <a href="#Handling-character-data" accesskey="p" rel="prev">Handling character data</a>, Up: <a href="#Handling-R-objects-in-C" accesskey="u" rel="up">Handling R objects in C</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Finding-and-setting-variables-1"></a>
<h4 class="subsection">5.9.8 Finding and setting variables</h4>
<a name="index-Finding-variables"></a>
<a name="index-Setting-variables"></a>

<p>It will be usual that all the R objects needed in our C computations
are passed as arguments to <code>.Call</code> or <code>.External</code>, but it is
possible to find the values of R objects from within the C given
their names.  The following code is the equivalent of <code>get(name,
envir = rho)</code>.
</p>
<div class="example">
<pre class="example">SEXP getvar(SEXP name, SEXP rho)
{
    SEXP ans;

    if(!isString(name) || length(name) != 1)
	error(&quot;name is not a single string&quot;);
    if(!isEnvironment(rho))
	error(&quot;rho should be an environment&quot;);
    ans = findVar(installChar(STRING_ELT(name, 0)), rho);
    Rprintf(&quot;first value is %f\n&quot;, REAL(ans)[0]);
    return R_NilValue;
}
</pre></div>

<p>The main work is done by
<a name="index-findVar"></a>
<code>findVar</code>, but to use it we need to install <code>name</code> as a name
in the symbol table.  As we wanted the value for internal use, we return
<code>NULL</code>.
</p>
<p>Similar functions with syntax
</p>
<div class="example">
<pre class="example">void defineVar(SEXP symbol, SEXP value, SEXP rho)
void setVar(SEXP symbol, SEXP value, SEXP rho)
</pre></div>
<a name="index-defineVar"></a>
<a name="index-setVar"></a>

<p>can be used to assign values to R variables.  <code>defineVar</code>
creates a new binding or changes the value of an existing binding in the
specified environment frame; it is the analogue of <code>assign(symbol,
value, envir = rho, inherits = FALSE)</code>, but unlike <code>assign</code>,
<code>defineVar</code> does not make a copy of the object
<code>value</code>.<a name="DOCF136" href="#FOOT136"><sup>136</sup></a>  <code>setVar</code> searches for an existing
binding for <code>symbol</code> in <code>rho</code> or its enclosing environments.
If a binding is found, its value is changed to <code>value</code>.  Otherwise,
a new binding with the specified value is created in the global
environment.  This corresponds to <code>assign(symbol, value, envir =
rho, inherits = TRUE)</code>.
</p>
<hr>
<a name="Some-convenience-functions"></a>
<div class="header">
<p>
Next: <a href="#Named-objects-and-copying" accesskey="n" rel="next">Named objects and copying</a>, Previous: <a href="#Finding-and-setting-variables" accesskey="p" rel="prev">Finding and setting variables</a>, Up: <a href="#Handling-R-objects-in-C" accesskey="u" rel="up">Handling R objects in C</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Some-convenience-functions-1"></a>
<h4 class="subsection">5.9.9 Some convenience functions</h4>

<p>Some operations are done so frequently that there are convenience
functions to handle them. (All these are provided via the header file
<samp>Rinternals.h</samp>.)
</p>
<p>Suppose we wanted to pass a single logical argument
<code>ignore_quotes</code>: we could use
</p>
<div class="example">
<pre class="example">    int ign = asLogical(ignore_quotes);
    if(ign == NA_LOGICAL) error(&quot;'ignore_quotes' must be TRUE or FALSE&quot;);
</pre></div>

<p>which will do any coercion needed (at least from a vector argument), and
return <code>NA_LOGICAL</code> if the value passed was <code>NA</code> or coercion
failed.  There are also <code>asInteger</code>, <code>asReal</code> and
<code>asComplex</code>.  The function <code>asChar</code> returns a <code>CHARSXP</code>.
All of these functions ignore any elements of an input vector after the
first.
</p>
<p>To return a length-one real vector we can use
</p>
<div class="example">
<pre class="example">    double x;

    ...
    return ScalarReal(x);
</pre></div>

<p>and there are versions of this for all the atomic vector types (those for
a length-one character vector being <code>ScalarString</code> with argument a
<code>CHARSXP</code> and <code>mkString</code> with argument <code>const char *</code>).
</p>
<p>Some of the <code>is<var>XXXX</var></code> functions differ from their apparent
R-level counterparts: for example <code>isVector</code> is true for any
atomic vector type (<code>isVectorAtomic</code>) and for lists and expressions
(<code>isVectorList</code>) (with no check on attributes).  <code>isMatrix</code> is
a test of a length-2 <code>&quot;dim&quot;</code> attribute.
</p>
<p>There are a series of small macros/functions to help construct pairlists
and language objects (whose internal structures just differ by
<code>SEXPTYPE</code>).  Function <code>CONS(u, v)</code> is the basic building
block: it constructs a pairlist from <code>u</code> followed by <code>v</code>
(which is a pairlist or <code>R_NilValue</code>).  <code>LCONS</code> is a variant
that constructs a language object.  Functions <code>list1</code> to
<code>list5</code> construct a pairlist from one to five items, and
<code>lang1</code> to <code>lang6</code> do the same for a language object (a
function to call plus zero to five arguments).  Functions <code>elt</code> and
<code>lastElt</code> find the <var>i</var>th element and the last element of a
pairlist, and <code>nthcdr</code> returns a pointer to the <var>n</var>th position
in the pairlist (whose <code>CAR</code> is the <var>n</var>th item).
</p>
<p>Functions <code>str2type</code> and <code>type2str</code> map R
length-one character strings to and from <code>SEXPTYPE</code> numbers, and
<code>type2char</code> maps numbers to C character strings.
</p>
<table summary="" class="menu" border="0" cellspacing="0">
<tr><td align="left" valign="top">&bull; <a href="#Semi_002dinternal-convenience-functions" accesskey="1">Semi-internal convenience functions</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
</table>

<hr>
<a name="Semi_002dinternal-convenience-functions"></a>
<div class="header">
<p>
Previous: <a href="#Some-convenience-functions" accesskey="p" rel="prev">Some convenience functions</a>, Up: <a href="#Some-convenience-functions" accesskey="u" rel="up">Some convenience functions</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Semi_002dinternal-convenience-functions-1"></a>
<h4 class="subsubsection">5.9.9.1 Semi-internal convenience functions</h4>

<p>There is quite a collection of functions that may be used in your C code
<em>if</em> you are willing to adapt to rare &ldquo;API&rdquo; changes.
These typically contain &ldquo;workhorses&rdquo; of their R counterparts.
</p>
<p>Functions <code>any_duplicated</code> and <code>any_duplicated3</code> are fast
versions of R&rsquo;s <code>any(duplicated(.))</code>.
</p>
<p>Function <code>R_compute_identical</code> corresponds to R&rsquo;s <code>identical</code> function.
</p>

<hr>
<a name="Named-objects-and-copying"></a>
<div class="header">
<p>
Previous: <a href="#Some-convenience-functions" accesskey="p" rel="prev">Some convenience functions</a>, Up: <a href="#Handling-R-objects-in-C" accesskey="u" rel="up">Handling R objects in C</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Named-objects-and-copying-1"></a>
<h4 class="subsection">5.9.10 Named objects and copying</h4>
<a name="index-duplicate"></a>
<a name="index-Copying-objects"></a>

<p>When assignments are done in R such as
</p>
<div class="example">
<pre class="example">x &lt;- 1:10
y &lt;- x
</pre></div>

<p>the named object is not necessarily copied, so after those two
assignments <code>y</code> and <code>x</code> are bound to the same <code>SEXPREC</code>
(the structure a <code>SEXP</code> points to).  This means that any code which
alters one of them has to make a copy before modifying the copy if the
usual R semantics are to apply.  Note that whereas <code>.C</code> and
<code>.Fortran</code> do copy their arguments (unless the dangerous <code>dup
= FALSE</code> is used), <code>.Call</code> and <code>.External</code> do not.  So
<code>duplicate</code> is commonly called on arguments to <code>.Call</code> before
modifying them.
</p>
<p>However, at least some of this copying is unneeded.  In the first
assignment shown, <code>x &lt;- 1:10</code>, R first creates an object with
value <code>1:10</code> and then assigns it to <code>x</code> but if <code>x</code> is
modified no copy is necessary as the temporary object with value
<code>1:10</code> cannot be referred to again.  R distinguishes between
named and unnamed objects <em>via</em> a field in a <code>SEXPREC</code> that
can be accessed <em>via</em> the macros <code>NAMED</code> and <code>SET_NAMED</code>.  This
can take values
</p>
<dl compact="compact">
<dt><code>0</code></dt>
<dd><p>The object is not bound to any symbol
</p></dd>
<dt><code>1</code></dt>
<dd><p>The object has been bound to exactly one symbol
</p></dd>
<dt><code>2</code></dt>
<dd><p>The object has potentially been bound to two or more symbols, and one
should act as if another variable is currently bound to this value.
</p></dd>
</dl>

<p>Note the past tenses: R does not do full reference counting and there
may currently be fewer bindings.
</p>
<p>It is safe to modify the value of any <code>SEXP</code> for which
<code>NAMED(foo)</code> is zero, and if <code>NAMED(foo)</code> is two, the value
should be duplicated (<em>via</em> a call to <code>duplicate</code>) before any
modification.  Note that it is the responsibility of the author of the
code making the modification to do the duplication, even if it is
<code>x</code> whose value is being modified after <code>y &lt;- x</code>.
</p>
<p>The case <code>NAMED(foo) == 1</code> allows some optimization, but it can be
ignored (and duplication done whenever <code>NAMED(foo) &gt; 0</code>).  (This
optimization is not currently usable in user code.)  It is intended
for use within replacement functions.  Suppose we used
</p>
<div class="example">
<pre class="example">x &lt;- 1:10
foo(x) &lt;- 3
</pre></div>

<p>which is computed as
</p>
<div class="example">
<pre class="example">x &lt;- 1:10
x &lt;- &quot;foo&lt;-&quot;(x, 3)
</pre></div>

<p>Then inside <code>&quot;foo&lt;-&quot;</code> the object pointing to the current value of
<code>x</code> will have <code>NAMED(foo)</code> as one, and it would be safe to
modify it as the only symbol bound to it is <code>x</code> and that will be
rebound immediately.  (Provided the remaining code in <code>&quot;foo&lt;-&quot;</code>
make no reference to <code>x</code>, and no one is going to attempt a direct
call such as <code>y &lt;- &quot;foo&lt;-&quot;(x)</code>.)
</p>
<p>This mechanism is likely to be replaced in future versions of R.
</p>

<hr>
<a name="Interface-functions-_002eCall-and-_002eExternal"></a>
<div class="header">
<p>
Next: <a href="#Evaluating-R-expressions-from-C" accesskey="n" rel="next">Evaluating R expressions from C</a>, Previous: <a href="#Handling-R-objects-in-C" accesskey="p" rel="prev">Handling R objects in C</a>, Up: <a href="#System-and-foreign-language-interfaces" accesskey="u" rel="up">System and foreign language interfaces</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Interface-functions-_002eCall-and-_002eExternal-1"></a>
<h3 class="section">5.10 Interface functions <code>.Call</code> and <code>.External</code></h3>
<a name="index-Interfaces-to-compiled-code-1"></a>

<p>In this section we consider the details of the R/C interfaces.
</p>
<p>These two interfaces have almost the same functionality. <code>.Call</code> is
based on the interface of the same name in S version 4, and
<code>.External</code> is based on R&rsquo;s <code>.Internal</code>.  <code>.External</code>
is more complex but allows a variable number of arguments.
</p>
<table summary="" class="menu" border="0" cellspacing="0">
<tr><td align="left" valign="top">&bull; <a href="#Calling-_002eCall" accesskey="1">Calling .Call</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Calling-_002eExternal" accesskey="2">Calling .External</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Missing-and-special-values" accesskey="3">Missing and special values</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
</table>

<hr>
<a name="Calling-_002eCall"></a>
<div class="header">
<p>
Next: <a href="#Calling-_002eExternal" accesskey="n" rel="next">Calling .External</a>, Previous: <a href="#Interface-functions-_002eCall-and-_002eExternal" accesskey="p" rel="prev">Interface functions .Call and .External</a>, Up: <a href="#Interface-functions-_002eCall-and-_002eExternal" accesskey="u" rel="up">Interface functions .Call and .External</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Calling-_002eCall-1"></a>
<h4 class="subsection">5.10.1 Calling <code>.Call</code></h4>

<a name="index-_002eCall-1"></a>

<p>Let us convert our finite convolution example to use <code>.Call</code>.  The
calling function in R is
</p>
<div class="example">
<pre class="example">conv &lt;- function(a, b) .Call(&quot;convolve2&quot;, a, b)
</pre></div>

<p>which could hardly be simpler, but as we shall see all the type
coercion is transferred to the C code, which is
</p>
<div class="example">
<pre class="example">#include &lt;R.h&gt;
#include &lt;Rinternals.h&gt;

SEXP convolve2(SEXP a, SEXP b)
{
    int na, nb, nab;
    double *xa, *xb, *xab;
    SEXP ab;

    a = PROTECT(coerceVector(a, REALSXP));
    b = PROTECT(coerceVector(b, REALSXP));
    na = length(a); nb = length(b); nab = na + nb - 1;
    ab = PROTECT(allocVector(REALSXP, nab));
    xa = REAL(a); xb = REAL(b); xab = REAL(ab);
    for(int i = 0; i &lt; nab; i++) xab[i] = 0.0;
    for(int i = 0; i &lt; na; i++)
	for(int j = 0; j &lt; nb; j++) xab[i + j] += xa[i] * xb[j];
    UNPROTECT(3);
    return ab;
}
</pre></div>

<hr>
<a name="Calling-_002eExternal"></a>
<div class="header">
<p>
Next: <a href="#Missing-and-special-values" accesskey="n" rel="next">Missing and special values</a>, Previous: <a href="#Calling-_002eCall" accesskey="p" rel="prev">Calling .Call</a>, Up: <a href="#Interface-functions-_002eCall-and-_002eExternal" accesskey="u" rel="up">Interface functions .Call and .External</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Calling-_002eExternal-1"></a>
<h4 class="subsection">5.10.2 Calling <code>.External</code></h4>

<a name="index-_002eExternal-1"></a>

<p>We can use the same example to illustrate <code>.External</code>.  The R
code changes only by replacing <code>.Call</code> by <code>.External</code>
</p>
<div class="example">
<pre class="example">conv &lt;- function(a, b) .External(&quot;convolveE&quot;, a, b)
</pre></div>

<p>but the main change is how the arguments are passed to the C code, this
time as a single SEXP.  The only change to the C code is how we handle
the arguments.
</p>
<div class="example">
<pre class="example">#include &lt;R.h&gt;
#include &lt;Rinternals.h&gt;

SEXP convolveE(SEXP args)
{
    int i, j, na, nb, nab;
    double *xa, *xb, *xab;
    SEXP a, b, ab;

    a = PROTECT(coerceVector(CADR(args), REALSXP));
    b = PROTECT(coerceVector(CADDR(args), REALSXP));
    ...
}
</pre></div>

<p>Once again we do not need to protect the arguments, as in the R side
of the interface they are objects that are already in use.  The macros
</p>
<div class="example">
<pre class="example">  first = CADR(args);
  second = CADDR(args);
  third = CADDDR(args);
  fourth = CAD4R(args);
</pre></div>

<p>provide convenient ways to access the first four arguments.  More
generally we can use the
<a name="index-CAR"></a>
<a name="index-CDR"></a>
<code>CDR</code> and <code>CAR</code> macros as in
</p>
<div class="example">
<pre class="example">  args = CDR(args); a = CAR(args);
  args = CDR(args); b = CAR(args);
</pre></div>

<p>which clearly allows us to extract an unlimited number of arguments
(whereas <code>.Call</code> has a limit, albeit at 65 not a small one).
</p>
<p>More usefully, the <code>.External</code> interface provides an easy way to
handle calls with a variable number of arguments, as <code>length(args)</code>
will give the number of arguments supplied (of which the first is
ignored).  We may need to know the names (&lsquo;tags&rsquo;) given to the actual
arguments, which we can by using the <code>TAG</code> macro and using
something like the following example, that prints the names and the first
value of its arguments if they are vector types.
</p>
<div class="example">
<pre class="example">SEXP showArgs(SEXP args)
{
    args = CDR(args); /* skip ‘name’ */
    for(int i = 0; args != R_NilValue; i++, args = CDR(args)) {
	const char *name =
	    isNull(TAG(args)) ? &quot;&quot; : CHAR(PRINTNAME(TAG(args)));
	SEXP el = CAR(args);
	if (length(el) == 0) {
	    Rprintf(&quot;[%d] ‘%s’ R type, length 0\n&quot;, i+1, name);
	   continue;
	}
</pre><pre class="example">	switch(TYPEOF(el)) {
	case REALSXP:
	    Rprintf(&quot;[%d] ‘%s’ %f\n&quot;, i+1, name, REAL(el)[0]);
	    break;
</pre><pre class="example">	case LGLSXP:
	case INTSXP:
	    Rprintf(&quot;[%d] ‘%s’ %d\n&quot;, i+1, name, INTEGER(el)[0]);
	    break;
</pre><pre class="example">	case CPLXSXP:
	{
	    Rcomplex cpl = COMPLEX(el)[0];
	    Rprintf(&quot;[%d] ‘%s’ %f + %fi\n&quot;, i+1, name, cpl.r, cpl.i);
	}
	    break;
</pre><pre class="example">	case STRSXP:
	    Rprintf(&quot;[%d] ‘%s’ %s\n&quot;, i+1, name,
		   CHAR(STRING_ELT(el, 0)));
	   break;
</pre><pre class="example">	default:
	    Rprintf(&quot;[%d] ‘%s’ R type\n&quot;, i+1, name);
       }
    }
    return R_NilValue;
}
</pre></div>

<p>This can be called by the wrapper function
</p>
<div class="example">
<pre class="example">showArgs &lt;- function(...) invisible(.External(&quot;showArgs&quot;, ...))
</pre></div>

<p>Note that this style of programming is convenient but not necessary, as
an alternative style is
</p>
<div class="example">
<pre class="example">showArgs1 &lt;- function(...) invisible(.Call(&quot;showArgs1&quot;, list(...)))
</pre></div>

<p>The (very similar) C code is in the scripts.
</p>
<hr>
<a name="Missing-and-special-values"></a>
<div class="header">
<p>
Previous: <a href="#Calling-_002eExternal" accesskey="p" rel="prev">Calling .External</a>, Up: <a href="#Interface-functions-_002eCall-and-_002eExternal" accesskey="u" rel="up">Interface functions .Call and .External</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Missing-and-special-values-1"></a>
<h4 class="subsection">5.10.3 Missing and special values</h4>
<a name="index-Missing-values"></a>
<a name="index-IEEE-special-values"></a>

<p>One piece of error-checking the <code>.C</code> call does (unless <code>NAOK</code>
is true) is to check for missing (<code>NA</code>) and <acronym>IEEE</acronym> special
values (<code>Inf</code>, <code>-Inf</code> and <code>NaN</code>) and give an error if any
are found.  With the <code>.Call</code> interface these will be passed to our
code.  In this example the special values are no problem, as
<acronym>IEC60559</acronym> arithmetic will handle them correctly.  In the current
implementation this is also true of <code>NA</code> as it is a type of
<code>NaN</code>, but it is unwise to rely on such details.  Thus we will
re-write the code to handle <code>NA</code>s using macros defined in
<samp>R_ext/Arith.h</samp> included by <samp>R.h</samp>.
</p>
<p>The code changes are the same in any of the versions of <code>convolve2</code>
or <code>convolveE</code>:
</p>
<div class="example">
<pre class="example">    ...
  for(int i = 0; i &lt; na; i++)
    for(int j = 0; j &lt; nb; j++)
	if(ISNA(xa[i]) || ISNA(xb[j]) || ISNA(xab[i + j]))
	    xab[i + j] = NA_REAL;
	else
	    xab[i + j] += xa[i] * xb[j];
    ...
</pre></div>

<a name="index-ISNA"></a>
<a name="index-ISNAN"></a>

<p>Note that the <code>ISNA</code> macro, and the similar macros <code>ISNAN</code>
(which checks for <code>NaN</code> or <code>NA</code>) and <code>R_FINITE</code> (which is
false for <code>NA</code> and all the special values), only apply to numeric
values of type <code>double</code>.  Missingness of integers, logicals and
character strings can be tested by equality to the constants
<code>NA_INTEGER</code>, <code>NA_LOGICAL</code> and <code>NA_STRING</code>.  These and
<code>NA_REAL</code> can be used to set elements of R vectors to <code>NA</code>.
</p>
<p>The constants <code>R_NaN</code>, <code>R_PosInf</code> and <code>R_NegInf</code> can be
used to set <code>double</code>s to the special values.
</p>
<hr>
<a name="Evaluating-R-expressions-from-C"></a>
<div class="header">
<p>
Next: <a href="#Parsing-R-code-from-C" accesskey="n" rel="next">Parsing R code from C</a>, Previous: <a href="#Interface-functions-_002eCall-and-_002eExternal" accesskey="p" rel="prev">Interface functions .Call and .External</a>, Up: <a href="#System-and-foreign-language-interfaces" accesskey="u" rel="up">System and foreign language interfaces</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Evaluating-R-expressions-from-C-1"></a>
<h3 class="section">5.11 Evaluating R expressions from C</h3>
<a name="index-Evaluating-R-expressions-from-C"></a>

<p>The main function we will use is
</p>
<div class="example">
<pre class="example">SEXP eval(SEXP expr, SEXP rho);
</pre></div>

<p>the equivalent of the interpreted R code <code>eval(expr, envir =
rho)</code> (so <code>rho</code> must be an environment), although we can also make
use of <code>findVar</code>, <code>defineVar</code> and <code>findFun</code> (which
restricts the search to functions).
</p>
<p>To see how this might be applied, here is a simplified internal version
of <code>lapply</code> for expressions, used as
</p>
<div class="example">
<pre class="example">a &lt;- list(a = 1:5, b = rnorm(10), test = runif(100))
.Call(&quot;lapply&quot;, a, quote(sum(x)), new.env())
</pre></div>

<p>with C code
</p>
<div class="example">
<pre class="example">SEXP lapply(SEXP list, SEXP expr, SEXP rho)
{
    int n = length(list);
    SEXP ans;

    if(!isNewList(list)) error(&quot;'list' must be a list&quot;);
    if(!isEnvironment(rho)) error(&quot;'rho' should be an environment&quot;);
    ans = PROTECT(allocVector(VECSXP, n));
    for(int i = 0; i &lt; n; i++) {
	defineVar(install(&quot;x&quot;), VECTOR_ELT(list, i), rho);
	SET_VECTOR_ELT(ans, i, eval(expr, rho));
    }
    setAttrib(ans, R_NamesSymbol, getAttrib(list, R_NamesSymbol));
    UNPROTECT(1);
    return ans;
}
</pre></div>

<p>It would be closer to <code>lapply</code> if we could pass in a function
rather than an expression.  One way to do this is <em>via</em> interpreted
R code as in the next example, but it is possible (if somewhat
obscure) to do this in C code.  The following is based on the code in
<samp>src/main/optimize.c</samp>.
</p>
<div class="example">
<pre class="example">SEXP lapply2(SEXP list, SEXP fn, SEXP rho)
{
    int n = length(list);
    SEXP R_fcall, ans;

    if(!isNewList(list)) error(&quot;'list' must be a list&quot;);
    if(!isFunction(fn)) error(&quot;'fn' must be a function&quot;);
    if(!isEnvironment(rho)) error(&quot;'rho' should be an environment&quot;);
    R_fcall = PROTECT(lang2(fn, R_NilValue));
    ans = PROTECT(allocVector(VECSXP, n));
    for(int i = 0; i &lt; n; i++) {
	SETCADR(R_fcall, VECTOR_ELT(list, i));
	SET_VECTOR_ELT(ans, i, eval(R_fcall, rho));
    }
    setAttrib(ans, R_NamesSymbol, getAttrib(list, R_NamesSymbol));
    UNPROTECT(2);
    return ans;
}
</pre></div>

<p>used by
</p>
<div class="example">
<pre class="example">.Call(&quot;lapply2&quot;, a, sum, new.env())
</pre></div>

<p>Function <code>lang2</code> creates an executable pairlist of two elements, but
this will only be clear to those with a knowledge of a LISP-like
language.
</p>
<p>As a more comprehensive example of constructing an R call in C code
and evaluating, consider the following fragment of
<code>printAttributes</code> in <samp>src/main/print.c</samp>.
</p>
<div class="example">
<pre class="example">    /* Need to construct a call to
       print(CAR(a), digits=digits)
       based on the R_print structure, then eval(call, env).
       See do_docall for the template for this sort of thing.
    */
    SEXP s, t;
    t = s = PROTECT(allocList(3));
    SET_TYPEOF(s, LANGSXP);
    SETCAR(t, install(&quot;print&quot;)); t = CDR(t);
    SETCAR(t,  CAR(a)); t = CDR(t);
    SETCAR(t, ScalarInteger(digits));
    SET_TAG(t, install(&quot;digits&quot;));
    eval(s, env);
    UNPROTECT(1);
</pre></div>

<p>At this point <code>CAR(a)</code> is the R object to be printed, the
current attribute.  There are three steps: the call is constructed as
a pairlist of length 3, the list is filled in, and the expression
represented by the pairlist is evaluated.
</p>
<p>A pairlist is quite distinct from a generic vector list, the only
user-visible form of list in R.  A pairlist is a linked list (with
<code>CDR(t)</code> computing the next entry), with items (accessed by
<code>CAR(t)</code>) and names or tags (set by <code>SET_TAG</code>).  In this call
there are to be three items, a symbol (pointing to the function to be
called) and two argument values, the first unnamed and the second named.
Setting the type to <code>LANGSXP</code> makes this a call which can be evaluated.
</p>
<table summary="" class="menu" border="0" cellspacing="0">
<tr><td align="left" valign="top">&bull; <a href="#Zero_002dfinding" accesskey="1">Zero-finding</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Calculating-numerical-derivatives" accesskey="2">Calculating numerical derivatives</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
</table>

<hr>
<a name="Zero_002dfinding"></a>
<div class="header">
<p>
Next: <a href="#Calculating-numerical-derivatives" accesskey="n" rel="next">Calculating numerical derivatives</a>, Previous: <a href="#Evaluating-R-expressions-from-C" accesskey="p" rel="prev">Evaluating R expressions from C</a>, Up: <a href="#Evaluating-R-expressions-from-C" accesskey="u" rel="up">Evaluating R expressions from C</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Zero_002dfinding-1"></a>
<h4 class="subsection">5.11.1 Zero-finding</h4>
<a name="index-Zero_002dfinding"></a>

<p>In this section we re-work the example of Becker, Chambers &amp; Wilks (1988,
pp.~205&ndash;10) on finding a zero of a univariate function.  The R code
and an example are
</p>
<div class="example">
<pre class="example">zero &lt;- function(f, guesses, tol = 1e-7) {
    f.check &lt;- function(x) {
	x &lt;- f(x)
	if(!is.numeric(x)) stop(&quot;Need a numeric result&quot;)
	as.double(x)
    }
    .Call(&quot;zero&quot;, body(f.check), as.double(guesses), as.double(tol),
	  new.env())
}

cube1 &lt;- function(x) (x^2 + 1) * (x - 1.5)
zero(cube1, c(0, 5))
</pre></div>

<p>where this time we do the coercion and error-checking in the R code.
The C code is
</p>
<div class="example">
<pre class="example">SEXP mkans(double x)
{
    // no need for PROTECT() here, as REAL(.) does not allocate:
    SEXP ans = allocVector(REALSXP, 1);
    REAL(ans)[0] = x;
    return ans;
}
</pre><pre class="example">

</pre><pre class="example">double feval(double x, SEXP f, SEXP rho)
{
    // a version with (too) much PROTECT()ion .. &quot;better safe than sorry&quot;
    SEXP symbol, value;
    PROTECT(symbol = install(&quot;x&quot;));
    PROTECT(value = mkans(x));
    defineVar(symbol, value, rho);
    UNPROTECT(2);
    return(REAL(eval(f, rho))[0]);
}
</pre><pre class="example">

</pre><pre class="example">SEXP zero(SEXP f, SEXP guesses, SEXP stol, SEXP rho)
{
    double x0 = REAL(guesses)[0], x1 = REAL(guesses)[1],
	   tol = REAL(stol)[0];
    double f0, f1, fc, xc;
</pre><pre class="example">

</pre><pre class="example">    if(tol &lt;= 0.0) error(&quot;non-positive tol value&quot;);
    f0 = feval(x0, f, rho); f1 = feval(x1, f, rho);
    if(f0 == 0.0) return mkans(x0);
    if(f1 == 0.0) return mkans(x1);
    if(f0*f1 &gt; 0.0) error(&quot;x[0] and x[1] have the same sign&quot;);
</pre><pre class="example">

</pre><pre class="example">    for(;;) {
	xc = 0.5*(x0+x1);
	if(fabs(x0-x1) &lt; tol) return  mkans(xc);
	fc = feval(xc, f, rho);
	if(fc == 0) return  mkans(xc);
	if(f0*fc &gt; 0.0) {
	    x0 = xc; f0 = fc;
	} else {
	    x1 = xc; f1 = fc;
	}
    }
}
</pre></div>

<hr>
<a name="Calculating-numerical-derivatives"></a>
<div class="header">
<p>
Previous: <a href="#Zero_002dfinding" accesskey="p" rel="prev">Zero-finding</a>, Up: <a href="#Evaluating-R-expressions-from-C" accesskey="u" rel="up">Evaluating R expressions from C</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Calculating-numerical-derivatives-1"></a>
<h4 class="subsection">5.11.2 Calculating numerical derivatives</h4>
<a name="index-Numerical-derivatives"></a>

<p>We will use a longer example (by Saikat DebRoy) to illustrate the use of
evaluation and <code>.External</code>.  This calculates numerical derivatives,
something that could be done as effectively in interpreted R code but
may be needed as part of a larger C calculation.
</p>
<p>An interpreted R version and an example are
</p>
<div class="example">
<pre class="example">numeric.deriv &lt;- function(expr, theta, rho=sys.frame(sys.parent()))
{
    eps &lt;- sqrt(.Machine$double.eps)
    ans &lt;- eval(substitute(expr), rho)
    grad &lt;- matrix(, length(ans), length(theta),
		   dimnames=list(NULL, theta))
    for (i in seq_along(theta)) {
	old &lt;- get(theta[i], envir=rho)
	delta &lt;- eps * max(1, abs(old))
	assign(theta[i], old+delta, envir=rho)
	ans1 &lt;- eval(substitute(expr), rho)
	assign(theta[i], old, envir=rho)
	grad[, i] &lt;- (ans1 - ans)/delta
    }
    attr(ans, &quot;gradient&quot;) &lt;- grad
    ans
}
omega &lt;- 1:5; x &lt;- 1; y &lt;- 2
numeric.deriv(sin(omega*x*y), c(&quot;x&quot;, &quot;y&quot;))
</pre></div>

<p>where <code>expr</code> is an expression, <code>theta</code> a character vector of
variable names and <code>rho</code> the environment to be used.
</p>
<p>For the compiled version the call from R will be
</p>
<div class="example">
<pre class="example">.External(&quot;numeric_deriv&quot;, <var>expr</var>, <var>theta</var>, <var>rho</var>)
</pre></div>

<p>with example usage
</p>
<div class="example">
<pre class="example">.External(&quot;numeric_deriv&quot;, quote(sin(omega*x*y)),
	  c(&quot;x&quot;, &quot;y&quot;), .GlobalEnv)
</pre></div>

<p>Note the need to quote the expression to stop it being evaluated in the
caller.
</p>
<p>Here is the complete C code which we will explain section by section.
</p>
<div class="example">
<pre class="example">#include &lt;R.h&gt; /* for DOUBLE_EPS */
#include &lt;Rinternals.h&gt;

SEXP numeric_deriv(SEXP args)
{
    SEXP theta, expr, rho, ans, ans1, gradient, par, dimnames;
    double tt, xx, delta, eps = sqrt(DOUBLE_EPS), *rgr, *rans;
    int i, start;
</pre><pre class="example">

</pre><pre class="example">    expr = CADR(args);
    if(!isString(theta = CADDR(args)))
	error(&quot;theta should be of type character&quot;);
    if(!isEnvironment(rho = CADDDR(args)))
	error(&quot;rho should be an environment&quot;);
</pre><pre class="example">

</pre><pre class="example">    ans = PROTECT(coerceVector(eval(expr, rho), REALSXP));
    gradient = PROTECT(allocMatrix(REALSXP, LENGTH(ans), LENGTH(theta)));
    rgr = REAL(gradient); rans = REAL(ans);
</pre><pre class="example">

</pre><pre class="example">    for(i = 0, start = 0; i &lt; LENGTH(theta); i++, start += LENGTH(ans)) {
	par = PROTECT(findVar(installChar(STRING_ELT(theta, i)), rho));
	tt = REAL(par)[0];
	xx = fabs(tt);
	delta = (xx &lt; 1) ? eps : xx*eps;
	REAL(par)[0] += delta;
	ans1 = PROTECT(coerceVector(eval(expr, rho), REALSXP));
	for(int j = 0; j &lt; LENGTH(ans); j++)
	    rgr[j + start] = (REAL(ans1)[j] - rans[j])/delta;
	REAL(par)[0] = tt;
	UNPROTECT(2); /* par, ans1 */
    }
</pre><pre class="example">

</pre><pre class="example">    dimnames = PROTECT(allocVector(VECSXP, 2));
    SET_VECTOR_ELT(dimnames, 1,  theta);
    dimnamesgets(gradient, dimnames);
    setAttrib(ans, install(&quot;gradient&quot;), gradient);
    UNPROTECT(3); /* ans  gradient  dimnames */
    return ans;
}
</pre></div>

<p>The code to handle the arguments is
</p>
<div class="example">
<pre class="example">    expr = CADR(args);
    if(!isString(theta = CADDR(args)))
	error(&quot;theta should be of type character&quot;);
    if(!isEnvironment(rho = CADDDR(args)))
	error(&quot;rho should be an environment&quot;);
</pre></div>

<p>Note that we check for correct types of <code>theta</code> and <code>rho</code> but
do not check the type of <code>expr</code>.  That is because <code>eval</code> can
handle many types of R objects other than <code>EXPRSXP</code>.  There is
no useful coercion we can do, so we stop with an error message if the
arguments are not of the correct mode.
</p>
<p>The first step in the code is to evaluate the expression in the
environment <code>rho</code>, by
</p>
<div class="example">
<pre class="example">    ans = PROTECT(coerceVector(eval(expr, rho), REALSXP));
</pre></div>

<p>We then allocate space for the calculated derivative by
</p>
<div class="example">
<pre class="example">    gradient = PROTECT(allocMatrix(REALSXP, LENGTH(ans), LENGTH(theta)));
</pre></div>

<p>The first argument to <code>allocMatrix</code> gives the <code>SEXPTYPE</code> of
the matrix: here we want it to be <code>REALSXP</code>.  The other two
arguments are the numbers of rows and columns.  (Note that <code>LENGTH</code>
is intended to be used for vectors: <code>length</code> is more generally
applicable.)
</p>
<div class="example">
<pre class="example">    for(i = 0, start = 0; i &lt; LENGTH(theta); i++, start += LENGTH(ans)) {
	par = PROTECT(findVar(installChar(STRING_ELT(theta, i)), rho));
</pre></div>

<p>Here, we are entering a for loop.  We loop through each of the
variables.  In the <code>for</code> loop, we first create a symbol
corresponding to the <code>i</code>&rsquo;th element of the <code>STRSXP</code>
<code>theta</code>.  Here, <code>STRING_ELT(theta, i)</code> accesses the
<code>i</code>&rsquo;th element of the <code>STRSXP</code> <code>theta</code>.  Macro
<code>CHAR()</code> extracts the actual character
representation<a name="DOCF137" href="#FOOT137"><sup>137</sup></a> of it: it returns a pointer.  We then
install the name and use <code>findVar</code> to find its value.
</p>
<div class="example">
<pre class="example">	tt = REAL(par)[0];
	xx = fabs(tt);
	delta = (xx &lt; 1) ? eps : xx*eps;
	REAL(par)[0] += delta;
	ans1 = PROTECT(coerceVector(eval(expr, rho), REALSXP));
</pre></div>

<p>We first extract the real value of the parameter, then calculate
<code>delta</code>, the increment to be used for approximating the numerical
derivative.  Then we change the value stored in <code>par</code> (in
environment <code>rho</code>) by <code>delta</code> and evaluate <code>expr</code> in
environment <code>rho</code> again.  Because we are directly dealing with
original R memory locations here, R does the evaluation for the
changed parameter value.
</p>
<div class="example">
<pre class="example">	for(int j = 0; j &lt; LENGTH(ans); j++)
	    rgr[j + start] = (REAL(ans1)[j] - rans[j])/delta;
	REAL(par)[0] = tt;
	UNPROTECT(2);
    }
</pre></div>

<p>Now, we compute the <code>i</code>&rsquo;th column of the gradient matrix.  Note how
it is accessed: R stores matrices by column (like FORTRAN).
</p>
<div class="example">
<pre class="example">    dimnames = PROTECT(allocVector(VECSXP, 2));
    SET_VECTOR_ELT(dimnames, 1, theta);
    dimnamesgets(gradient, dimnames);
    setAttrib(ans, install(&quot;gradient&quot;), gradient);
    UNPROTECT(3);
    return ans;
}
</pre></div>

<p>First we add column names to the gradient matrix.  This is done by
allocating a list (a <code>VECSXP</code>) whose first element, the row names,
is <code>NULL</code> (the default) and the second element, the column names,
is set as <code>theta</code>.  This list is then assigned as the attribute
having the symbol <code>R_DimNamesSymbol</code>.  Finally we set the gradient
matrix as the gradient attribute of <code>ans</code>, unprotect the remaining
protected locations and return the answer <code>ans</code>.
</p>
<hr>
<a name="Parsing-R-code-from-C"></a>
<div class="header">
<p>
Next: <a href="#External-pointers-and-weak-references" accesskey="n" rel="next">External pointers and weak references</a>, Previous: <a href="#Evaluating-R-expressions-from-C" accesskey="p" rel="prev">Evaluating R expressions from C</a>, Up: <a href="#System-and-foreign-language-interfaces" accesskey="u" rel="up">System and foreign language interfaces</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Parsing-R-code-from-C-1"></a>
<h3 class="section">5.12 Parsing R code from C</h3>
<a name="index-Parsing-R-code-from-C"></a>

<p>Suppose an R extension want to accept an R expression from the
user and evaluate it.  The previous section covered evaluation, but the
expression will be entered as text and needs to be parsed first.  A
small part of R&rsquo;s parse interface is declared in header file
<samp>R_ext/Parse.h</samp><a name="DOCF138" href="#FOOT138"><sup>138</sup></a>.
</p>
<p>An example of the usage can be found in the (example) Windows package
<strong>windlgs</strong> included in the R source tree.  The essential part is
</p>
<div class="example">
<pre class="example">#include &lt;R.h&gt;
#include &lt;Rinternals.h&gt;
#include &lt;R_ext/Parse.h&gt;

SEXP menu_ttest3()
{
    char cmd[256];
    SEXP cmdSexp, cmdexpr, ans = R_NilValue;
    ParseStatus status;
   ...
    if(done == 1) {
	cmdSexp = PROTECT(allocVector(STRSXP, 1));
	SET_STRING_ELT(cmdSexp, 0, mkChar(cmd));
	cmdexpr = PROTECT(R_ParseVector(cmdSexp, -1, &amp;status, R_NilValue));
	if (status != PARSE_OK) {
	    UNPROTECT(2);
	    error(&quot;invalid call %s&quot;, cmd);
	}
	/* Loop is needed here as EXPSEXP will be of length &gt; 1 */
	for(int i = 0; i &lt; length(cmdexpr); i++)
	    ans = eval(VECTOR_ELT(cmdexpr, i), R_GlobalEnv);
	UNPROTECT(2);
    }
    return ans;
}
</pre></div>
<p>Note that a single line of text may give rise to more than one R
expression.
</p>
<a name="index-R_005fParseVector"></a>
<p><code>R_ParseVector</code> is essentially the code used to implement
<code>parse(text=)</code> at R level.  The first argument is a character
vector (corresponding to <code>text</code>) and the second the maximal
number of expressions to parse (corresponding to <code>n</code>).  The third
argument is a pointer to a variable of an enumeration type, and it is
normal (as <code>parse</code> does) to regard all values other than
<code>PARSE_OK</code> as an error.  Other values which might be returned are
<code>PARSE_INCOMPLETE</code> (an incomplete expression was found) and
<code>PARSE_ERROR</code> (a syntax error), in both cases the value returned
being <code>R_NilValue</code>. The fourth argument is a length one character
vector to be used as a filename in error messages, a <code>srcfile</code>
object or the R <code>NULL</code> object (as in the example above). If a
<code>srcfile</code> object was used, a <code>srcref</code> attribute would be
attached to the result, containing a list of <code>srcref</code> objects of
the same length as the expression, to allow it to be echoed with its
original formatting.
</p>
<table summary="" class="menu" border="0" cellspacing="0">
<tr><td align="left" valign="top">&bull; <a href="#Accessing-source-references" accesskey="1">Accessing source references</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
</table>

<hr>
<a name="Accessing-source-references"></a>
<div class="header">
<p>
Previous: <a href="#Parsing-R-code-from-C" accesskey="p" rel="prev">Parsing R code from C</a>, Up: <a href="#Parsing-R-code-from-C" accesskey="u" rel="up">Parsing R code from C</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Accessing-source-references-1"></a>
<h4 class="subsection">5.12.1 Accessing source references</h4>

<p>The source references added by the parser are recorded by R&rsquo;s evaluator
as it evaluates code. Two functions
make these available to debuggers running C code:
<a name="index-R_005fSrcref"></a>
<a name="index-R_005fGetCurrentSrcref"></a>
<a name="index-R_005fGetSrcFilename"></a>
</p>
<div class="example">
<pre class="example">SEXP R_GetCurrentSrcref(int skip);
</pre></div>

<p>This function checks <code>R_Srcref</code> and the current evaluation stack
for entries that contain source reference information.  The
<code>skip</code> argument tells how many source references to skip before
returning the <code>SEXP</code> of the <code>srcref</code> object, counting from
the top of the stack.  If <code>skip &lt; 0</code>, <code>abs(skip)</code> locations
are counted up from the bottom of the stack. If too few or no source
references are found, <code>NULL</code> is returned.
</p>
<div class="example">
<pre class="example">SEXP R_GetSrcFilename(SEXP srcref);
</pre></div>

<p>This function extracts the filename from the source reference for
display, returning a length 1 character vector containing the
filename.  If no name is found, <code>&quot;&quot;</code> is returned.
</p>
<hr>
<a name="External-pointers-and-weak-references"></a>
<div class="header">
<p>
Next: <a href="#Vector-accessor-functions" accesskey="n" rel="next">Vector accessor functions</a>, Previous: <a href="#Parsing-R-code-from-C" accesskey="p" rel="prev">Parsing R code from C</a>, Up: <a href="#System-and-foreign-language-interfaces" accesskey="u" rel="up">System and foreign language interfaces</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="External-pointers-and-weak-references-1"></a>
<h3 class="section">5.13 External pointers and weak references</h3>

<p>The <code>SEXPTYPE</code>s <code>EXTPTRSXP</code> and <code>WEAKREFSXP</code> can be
encountered at R level, but are created in C code.
</p>
<a name="index-external-pointer"></a>
<p>External pointer <code>SEXP</code>s are intended to handle references to C
structures such as &lsquo;handles&rsquo;, and are used for this purpose in package
<a href="https://CRAN.R-project.org/package=RODBC"><strong>RODBC</strong></a> for example.  They are unusual in their copying semantics in
that when an R object is copied, the external pointer object is not
duplicated.  (For this reason external pointers should only be used as
part of an object with normal semantics, for example an attribute or an
element of a list.)
</p>
<p>An external pointer is created by
</p>
<div class="example">
<pre class="example">SEXP R_MakeExternalPtr(void *p, SEXP tag, SEXP prot);
</pre></div>

<p>where <code>p</code> is the pointer (and hence this cannot portably be a
function pointer), and <code>tag</code> and <code>prot</code> are references to
ordinary R objects which will remain in existence (be protected from
garbage collection) for the lifetime of the external pointer object.  A
useful convention is to use the <code>tag</code> field for some form of type
identification and the <code>prot</code> field for protecting the memory that
the external pointer represents, if that memory is allocated from the
R heap.  Both <code>tag</code> and <code>prot</code> can be <code>R_NilValue</code>,
and often are.
</p>
<p>The elements of an external pointer can be accessed and set <em>via</em>
</p>
<div class="example">
<pre class="example">void *R_ExternalPtrAddr(SEXP s);
SEXP R_ExternalPtrTag(SEXP s);
SEXP R_ExternalPtrProtected(SEXP s);
void R_ClearExternalPtr(SEXP s);
void R_SetExternalPtrAddr(SEXP s, void *p);
void R_SetExternalPtrTag(SEXP s, SEXP tag);
void R_SetExternalPtrProtected(SEXP s, SEXP p);
</pre></div>

<p>Clearing a pointer sets its value to the C <code>NULL</code> pointer.
</p>
<a name="index-finalizer"></a>
<p>An external pointer object can have a <em>finalizer</em>, a piece of code
to be run when the object is garbage collected.  This can be R code
or C code, and the various interfaces are, respectively.
</p>
<div class="example">
<pre class="example">void R_RegisterFinalizerEx(SEXP s, SEXP fun, Rboolean onexit);

typedef void (*R_CFinalizer_t)(SEXP);
void R_RegisterCFinalizerEx(SEXP s, R_CFinalizer_t fun, Rboolean onexit);
</pre></div>

<p>The R function indicated by <code>fun</code> should be a function of a
single argument, the object to be finalized.  R does not perform a
garbage collection when shutting down, and the <code>onexit</code> argument of
the extended forms can be used to ask that the finalizer be run during a
normal shutdown of the R session.  It is suggested that it is good
practice to clear the pointer on finalization.
</p>
<p>The only R level function for interacting with external pointers is
<code>reg.finalizer</code> which can be used to set a finalizer.
</p>
<p>It is probably not a good idea to allow an external pointer to be
<code>save</code>d and then reloaded, but if this happens the pointer will be
set to the C <code>NULL</code> pointer.
</p>
<p>Finalizers can be run at many places in the code base and much of it,
including the R interpreter, is not re-entrant.  So great care is
needed in choosing the code to be run in a finalizer. As from R 3.0.3
finalizers are marked to be run at garbage collection but only run at a
somewhat safe point thereafter.
</p>
<a name="index-weak-reference"></a>
<p>Weak references are used to allow the programmer to maintain information
on entities without preventing the garbage collection of the entities
once they become unreachable.
</p>
<p>A weak reference contains a key and a value.  The value is reachable is
if it either reachable directly or <em>via</em> weak references with reachable
keys.  Once a value is determined to be unreachable during garbage
collection, the key and value are set to <code>R_NilValue</code> and the
finalizer will be run later in the garbage collection.
</p>
<p>Weak reference objects are created by one of
</p>
<div class="example">
<pre class="example">SEXP R_MakeWeakRef(SEXP key, SEXP val, SEXP fin, Rboolean onexit);
SEXP R_MakeWeakRefC(SEXP key, SEXP val, R_CFinalizer_t fin,
		    Rboolean onexit);
</pre></div>

<p>where the R or C finalizer are specified in exactly the same way as
for an external pointer object (whose finalization interface is
implemented <em>via</em> weak references).
</p>
<p>The parts can be accessed <em>via</em>
</p>
<div class="example">
<pre class="example">SEXP R_WeakRefKey(SEXP w);
SEXP R_WeakRefValue(SEXP w);
void R_RunWeakRefFinalizer(SEXP w);
</pre></div>

<p>A toy example of the use of weak references can be found at
<a href="http://homepage.stat.uiowa.edu/~luke/R/references/weakfinex.html"><code>homepage.stat.uiowa.edu/~luke/R/references/weakfinex.html</code></a>,
but that is used to add finalizers to external pointers which can now be
done more directly.  At the time of writing no <acronym>CRAN</acronym> or
Bioconductor package uses weak references.
</p>

<table summary="" class="menu" border="0" cellspacing="0">
<tr><td align="left" valign="top">&bull; <a href="#An-external-pointer-example" accesskey="1">An external pointer example</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
</table>

<hr>
<a name="An-external-pointer-example"></a>
<div class="header">
<p>
Previous: <a href="#External-pointers-and-weak-references" accesskey="p" rel="prev">External pointers and weak references</a>, Up: <a href="#External-pointers-and-weak-references" accesskey="u" rel="up">External pointers and weak references</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="An-example-2"></a>
<h4 class="subsection">5.13.1 An example</h4>

<p>Package <a href="https://CRAN.R-project.org/package=RODBC"><strong>RODBC</strong></a> uses external pointers to maintain its
<em>channels</em>, connections to databases.  There can be several
connections open at once, and the status information for each is stored
in a C structure (pointed to by <code>thisHandle</code> in the code extract
below) that is returned <em>via</em> an external pointer as part of the RODBC
&lsquo;channel&rsquo; (as the <code>&quot;handle_ptr&quot;</code> attribute).  The external pointer
is created by
</p>
<div class="example">
<pre class="example">    SEXP ans, ptr;
    ans = PROTECT(allocVector(INTSXP, 1));
    ptr = R_MakeExternalPtr(thisHandle, install(&quot;RODBC_channel&quot;), R_NilValue);
    PROTECT(ptr);
    R_RegisterCFinalizerEx(ptr, chanFinalizer, TRUE);
	    ...
    /* return the channel no */
    INTEGER(ans)[0] = nChannels;
    /* and the connection string as an attribute */
    setAttrib(ans, install(&quot;connection.string&quot;), constr);
    setAttrib(ans, install(&quot;handle_ptr&quot;), ptr);
    UNPROTECT(3);
    return ans;
</pre></div>

<p>Note the symbol given to identify the usage of the external pointer, and
the use of the finalizer.  Since the final argument when registering the
finalizer is <code>TRUE</code>, the finalizer will be run at the of the
R session (unless it crashes).  This is used to close and clean up
the connection to the database.  The finalizer code is simply
</p>
<div class="example">
<pre class="example">static void chanFinalizer(SEXP ptr)
{
    if(!R_ExternalPtrAddr(ptr)) return;
    inRODBCClose(R_ExternalPtrAddr(ptr));
    R_ClearExternalPtr(ptr); /* not really needed */
}
</pre></div>

<p>Clearing the pointer and checking for a <code>NULL</code> pointer avoids any
possibility of attempting to close an already-closed channel.
</p>
<p>R&rsquo;s connections provide another example of using external pointers,
in that case purely to be able to use a finalizer to close and destroy the
connection if it is no longer is use.
</p>
<hr>
<a name="Vector-accessor-functions"></a>
<div class="header">
<p>
Next: <a href="#Character-encoding-issues" accesskey="n" rel="next">Character encoding issues</a>, Previous: <a href="#External-pointers-and-weak-references" accesskey="p" rel="prev">External pointers and weak references</a>, Up: <a href="#System-and-foreign-language-interfaces" accesskey="u" rel="up">System and foreign language interfaces</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Vector-accessor-functions-1"></a>
<h3 class="section">5.14 Vector accessor functions</h3>

<p>The vector accessors like <code>REAL</code> and <code>INTEGER</code> and
<code>VECTOR_ELT</code> are <em>functions</em> when used in R extensions.
(For efficiency they are macros when used in the R source code, apart
from <code>SET_STRING_ELT</code> and <code>SET_VECTOR_ELT</code> which are always
functions.)
</p>
<p>The accessor functions check that they are being used on an appropriate
type of <code>SEXP</code>.
</p>
<p>If efficiency is essential, the macro versions of the accessors can be
obtained by defining &lsquo;<samp>USE_RINTERNALS</samp>&rsquo; before including
<samp>Rinternals.h</samp>.  If you find it necessary to do so, please do test
that your code compiles without &lsquo;<samp>USE_RINTERNALS</samp>&rsquo; defined, as this
provides a stricter test that the accessors have been used correctly.
Note too that the use of &lsquo;<samp>USE_RINTERNALS</samp>&rsquo; when the header is
included in C++ code is not supported: doing so may use C99 features
which are not necessarily supported by the C++ compiler.  Nor is use
with <samp>Rdefines.h</samp> supported.
</p>
<hr>
<a name="Character-encoding-issues"></a>
<div class="header">
<p>
Previous: <a href="#Vector-accessor-functions" accesskey="p" rel="prev">Vector accessor functions</a>, Up: <a href="#System-and-foreign-language-interfaces" accesskey="u" rel="up">System and foreign language interfaces</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Character-encoding-issues-1"></a>
<h3 class="section">5.15 Character encoding issues</h3>

<a name="index-translateChar"></a>
<a name="index-translateCharUTF8"></a>
<p><code>CHARSXP</code>s can be marked as coming from a known encoding (Latin-1
or UTF-8).  This is mainly intended for human-readable output, and most
packages can just treat such <code>CHARSXP</code>s as a whole.  However, if
they need to be interpreted as characters or output at C level then it
would normally be correct to ensure that they are converted to the
encoding of the current locale: this can be done by accessing the data
in the <code>CHARSXP</code> by <code>translateChar</code> rather than by
<code>CHAR</code>.  If re-encoding is needed this allocates memory with
<code>R_alloc</code> which thus persists to the end of the
<code>.Call</code>/<code>.External</code> call unless <code>vmaxset</code> is used
(see <a href="#Transient-storage-allocation">Transient storage allocation</a>).
</p>
<p>There is a similar function <code>translateCharUTF8</code> which converts to
UTF-8: this has the advantage that a faithful translation is almost
always possible (whereas only a few languages can be represented in the
encoding of the current locale unless that is UTF-8).
</p>
<a name="index-getCharCE"></a>
<a name="index-mkCharCE"></a>
<p>There is a public interface to the encoding marked on <code>CHARXSXPs</code>
<em>via</em>
</p>
<div class="example">
<pre class="example">typedef enum {CE_NATIVE, CE_UTF8, CE_LATIN1, CE_SYMBOL, CE_ANY} cetype_t;
cetype_t getCharCE(SEXP);
SEXP mkCharCE(const char *, cetype_t);
</pre></div>

<p>Only <code>CE_UTF8</code> and <code>CE_LATIN1</code> are marked on <code>CHARSXPs</code>
(and so <code>Rf_getCharCE</code> will only return one of the first three),
and these should only be used on non-<acronym>ASCII</acronym> strings.  Value
<code>CE_SYMBOL</code> is used internally to indicate Adobe Symbol encoding.
Value <code>CE_ANY</code> is used to indicate a character string that will not
need re-encoding &ndash; this is used for character strings known to be in
<acronym>ASCII</acronym>, and can also be used as an input parameter where the
intention is that the string is treated as a series of bytes.  (See the
comments under <code>mkChar</code> about the length of input allowed.)
</p>
<p>Function
</p>
<a name="index-reEnc"></a>
<div class="example">
<pre class="example">const char *reEnc(const char *x, cetype_t ce_in, cetype_t ce_out,
		  int subst);
</pre></div>

<p>can be used to re-encode character strings: like <code>translateChar</code> it
returns a string allocated by <code>R_alloc</code>.  This can translate from
<code>CE_SYMBOL</code> to <code>CE_UTF8</code>, but not conversely.  Argument
<code>subst</code> controls what to do with untranslatable characters or
invalid input: this is done byte-by-byte with <code>1</code> indicates to
output hex of the form <code>&lt;a0&gt;</code>, and <code>2</code> to replace by <code>.</code>,
with any other value causing the byte to produce no output.
</p>
<a name="index-mkCharLenCE"></a>
<p>There is also
</p>
<div class="example">
<pre class="example">SEXP mkCharLenCE(const char *, size_t, cetype_t);
</pre></div>

<p>to create marked character strings of a given length.
</p>

<hr>
<a name="The-R-API"></a>
<div class="header">
<p>
Next: <a href="#Generic-functions-and-methods" accesskey="n" rel="next">Generic functions and methods</a>, Previous: <a href="#System-and-foreign-language-interfaces" accesskey="p" rel="prev">System and foreign language interfaces</a>, Up: <a href="#Top" accesskey="u" rel="up">Top</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="The-R-API_003a-entry-points-for-C-code"></a>
<h2 class="chapter">6 The R <acronym>API</acronym>: entry points for C code</h2>

<table summary="" class="menu" border="0" cellspacing="0">
<tr><td align="left" valign="top">&bull; <a href="#Memory-allocation" accesskey="1">Memory allocation</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Error-handling" accesskey="2">Error handling</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Random-numbers" accesskey="3">Random numbers</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Missing-and-IEEE-values" accesskey="4">Missing and IEEE values</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Printing" accesskey="5">Printing</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Calling-C-from-FORTRAN-and-vice-versa" accesskey="6">Calling C from FORTRAN and vice versa</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Numerical-analysis-subroutines" accesskey="7">Numerical analysis subroutines</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Optimization" accesskey="8">Optimization</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Integration" accesskey="9">Integration</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Utility-functions">Utility functions</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Re_002dencoding">Re-encoding</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Allowing-interrupts">Allowing interrupts</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Platform-and-version-information">Platform and version information</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Inlining-C-functions">Inlining C functions</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Controlling-visibility">Controlling visibility</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Standalone-Mathlib">Standalone Mathlib</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Organization-of-header-files">Organization of header files</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
</table>

<p>There are a large number of entry points in the R executable/DLL that
can be called from C code (and some that can be called from FORTRAN
code).  Only those documented here are stable enough that they will only
be changed with considerable notice.
</p>
<p>The recommended procedure to use these is to include the header file
<samp>R.h</samp> in your C code by
</p>
<div class="example">
<pre class="example">#include &lt;R.h&gt;
</pre></div>

<p>This will include several other header files from the directory
<samp><var>R_INCLUDE_DIR</var>/R_ext</samp>, and there are other header files
there that can be included too, but many of the features they contain
should be regarded as undocumented and unstable.
</p>
<p>Most of these header files, including all those included by <samp>R.h</samp>,
can be used from C++ code.
</p>
<blockquote>
<p><b>Note:</b> Because R re-maps many of its external names to avoid clashes with
user code, it is <em>essential</em> to include the appropriate header
files when using these entry points.
</p></blockquote>

<p>This remapping can cause problems<a name="DOCF139" href="#FOOT139"><sup>139</sup></a>, and can be eliminated by defining <code>R_NO_REMAP</code> and
prepending &lsquo;<samp>Rf_</samp>&rsquo; to <em>all</em> the function names used from
<samp>Rinternals.h</samp> and <samp>R_ext/Error.h</samp>.  These problems can
usually be avoided by including other headers (such as system headers
and those for external software used by the package) before <samp>R.h</samp>.
</p>
<p>We can classify the entry points as
</p>
<dl compact="compact">
<dt><em>API</em></dt>
<dd><p>Entry points which are documented in this manual and declared in an
installed header file.  These can be used in distributed packages and
will only be changed after deprecation.
</p>
</dd>
<dt><em>public</em></dt>
<dd><p>Entry points declared in an installed header file that are exported
on all R platforms but are not documented and subject to change
without notice.
</p>
</dd>
<dt><em>private</em></dt>
<dd><p>Entry points that are used when building R and exported on all R
platforms but are not declared in the installed header files.
Do not use these in distributed code.
</p>
</dd>
<dt><em>hidden</em></dt>
<dd><p>Entry points that are where possible (Windows and some modern Unix-alike
compilers/loaders when using R as a shared library) not exported.
</p></dd>
</dl>

<hr>
<a name="Memory-allocation"></a>
<div class="header">
<p>
Next: <a href="#Error-handling" accesskey="n" rel="next">Error handling</a>, Previous: <a href="#The-R-API" accesskey="p" rel="prev">The R API</a>, Up: <a href="#The-R-API" accesskey="u" rel="up">The R API</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Memory-allocation-1"></a>
<h3 class="section">6.1 Memory allocation</h3>
<a name="index-Memory-allocation-from-C"></a>

<table summary="" class="menu" border="0" cellspacing="0">
<tr><td align="left" valign="top">&bull; <a href="#Transient-storage-allocation" accesskey="1">Transient storage allocation</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#User_002dcontrolled-memory" accesskey="2">User-controlled memory</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
</table>

<p>There are two types of memory allocation available to the C programmer,
one in which R manages the clean-up and the other in which user
has full control (and responsibility).
</p>
<hr>
<a name="Transient-storage-allocation"></a>
<div class="header">
<p>
Next: <a href="#User_002dcontrolled-memory" accesskey="n" rel="next">User-controlled memory</a>, Previous: <a href="#Memory-allocation" accesskey="p" rel="prev">Memory allocation</a>, Up: <a href="#Memory-allocation" accesskey="u" rel="up">Memory allocation</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Transient-storage-allocation-1"></a>
<h4 class="subsection">6.1.1 Transient storage allocation</h4>
<a name="index-R_005falloc"></a>
<a name="index-R_005fallocLD"></a>
<a name="index-S_005falloc"></a>
<a name="index-S_005frealloc"></a>
<a name="index-vmaxget"></a>
<a name="index-vmaxset"></a>

<p>Here R will reclaim the memory at the end of the call to <code>.C</code>,
<code>.Call</code> or <code>.External</code>.  Use
</p>
<div class="example">
<pre class="example">char *R_alloc(size_t <var>n</var>, int <var>size</var>)
</pre></div>

<p>which allocates <var>n</var> units of <var>size</var> bytes each.  A typical usage
(from package <strong>stats</strong>) is
</p>
<div class="example">
<pre class="example">x = (int *) R_alloc(nrows(merge)+2, sizeof(int));
</pre></div>

<p>(<code>size_t</code> is defined in <samp>stddef.h</samp> which the header defining
<code>R_alloc</code> includes.)
</p>
<p>There is a similar call, <code>S_alloc</code> (for compatibility with older
versions of S) which zeroes the memory allocated,
</p>
<div class="example">
<pre class="example">char *S_alloc(long <var>n</var>, int <var>size</var>)
</pre></div>

<p>and
</p>
<div class="example">
<pre class="example">char *S_realloc(char *<var>p</var>, long <var>new</var>, long <var>old</var>, int <var>size</var>)
</pre></div>

<p>which changes the allocation size from <var>old</var> to <var>new</var> units, and
zeroes the additional units.
</p>
<p>For compatibility with current versions of S, header <samp>S.h</samp>
(only) defines wrapper macros equivalent to
</p>
<div class="example">
<pre class="example">type* Salloc(long <var>n</var>, int <var>type</var>)
type* Srealloc(char *<var>p</var>, long <var>new</var>, long <var>old</var>, int <var>type</var>)
</pre></div>

<p>This memory is taken from the heap, and released at the end of the
<code>.C</code>, <code>.Call</code> or <code>.External</code> call.  Users can also manage
it, by noting the current position with a call to <code>vmaxget</code> and
subsequently clearing memory allocated by a call to <code>vmaxset</code>. An
example might be
</p>
<div class="example">
<pre class="example">void *vmax = vmaxget()
// a loop involving the use of R_alloc at each iteration
vmaxset(vmax)
</pre></div>

<p>This is only recommended for experts.
</p>
<p>Note that this memory will be freed on error or user interrupt
(if allowed: see <a href="#Allowing-interrupts">Allowing interrupts</a>).
</p>

<p>The memory returned is only guaranteed to be aligned as required for
<code>double</code> pointers: take precautions if casting to a pointer which
needs more.  As from R 3.2.0 there is also
</p>
<div class="example">
<pre class="example">long double *R_allocLD(size_t <var>n</var>)
</pre></div>

<p>which is guaranteed to have the 16-byte alignment needed for <code>long
double</code> pointers on some platforms.
</p>

<p>These functions should only be used in code called by <code>.C</code> etc,
never from front-ends.  They are not thread-safe.
</p>
<hr>
<a name="User_002dcontrolled-memory"></a>
<div class="header">
<p>
Previous: <a href="#Transient-storage-allocation" accesskey="p" rel="prev">Transient storage allocation</a>, Up: <a href="#Memory-allocation" accesskey="u" rel="up">Memory allocation</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="User_002dcontrolled-memory-1"></a>
<h4 class="subsection">6.1.2 User-controlled memory</h4>
<a name="index-Calloc"></a>
<a name="index-Realloc"></a>
<a name="index-Free"></a>

<p>The other form of memory allocation is an interface to <code>malloc</code>,
the interface providing R error handling.  This memory lasts until
freed by the user and is additional to the memory allocated for the R
workspace.
</p>
<p>The interface functions are
</p>
<div class="example">
<pre class="example"><var>type</var>* Calloc(size_t <var>n</var>, <var>type</var>)
<var>type</var>* Realloc(<var>any</var> *<var>p</var>, size_t <var>n</var>, <var>type</var>)
void Free(<var>any</var> *<var>p</var>)
</pre></div>

<p>providing analogues of <code>calloc</code>, <code>realloc</code> and <code>free</code>.
If there is an error during allocation it is handled by R, so if
these routines return the memory has been successfully allocated or
freed.  <code>Free</code> will set the pointer <var>p</var> to <code>NULL</code>.  (Some
but not all versions of S do so.)
</p>
<p>Users should arrange to <code>Free</code> this memory when no longer needed,
including on error or user interrupt.  This can often be done most
conveniently from an <code>on.exit</code> action in the calling R function
&ndash; see <code>pwilcox</code> for an example.
</p>
<p>Do not assume that memory allocated by <code>Calloc</code>/<code>Realloc</code>
comes from the same pool as used by <code>malloc</code>: in particular do not
use <code>free</code> or <code>strdup</code> with it.
</p>
<p>Memory obtained by these functions should be aligned in the same way as
<code>malloc</code>, that is &lsquo;suitably aligned for any kind of variable&rsquo;.
</p>
<p>These entry points need to be prefixed by <code>R_</code> if
<code>STRICT_R_HEADERS</code> has been defined.
</p>

<hr>
<a name="Error-handling"></a>
<div class="header">
<p>
Next: <a href="#Random-numbers" accesskey="n" rel="next">Random numbers</a>, Previous: <a href="#Memory-allocation" accesskey="p" rel="prev">Memory allocation</a>, Up: <a href="#The-R-API" accesskey="u" rel="up">The R API</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Error-handling-1"></a>
<h3 class="section">6.2 Error handling</h3>
<a name="index-Error-handling-from-C"></a>

<p>The basic error handling routines are the equivalents of <code>stop</code> and
<code>warning</code> in R code, and use the same interface.
</p>
<div class="example">
<pre class="example">void error(const char * <var>format</var>, ...);
void warning(const char * <var>format</var>, ...);
</pre></div>

<p>These have the same call sequences as calls to <code>printf</code>, but in the
simplest case can be called with a single character string argument
giving the error message. (Don&rsquo;t do this if the string contains &lsquo;<samp>%</samp>&rsquo;
or might otherwise be interpreted as a format.)
</p>
<p>If <code>STRICT_R_HEADERS</code> is not defined there is also an
S-compatibility interface which uses calls of the form
</p>
<div class="example">
<pre class="example">PROBLEM ...... ERROR
MESSAGE ...... WARN
PROBLEM ...... RECOVER(NULL_ENTRY)
MESSAGE ...... WARNING(NULL_ENTRY)
</pre></div>

<p>the last two being the forms available in all S versions.  Here
&lsquo;<samp>......</samp>&rsquo; is a set of arguments to <code>printf</code>, so can be a string
or a format string followed by arguments separated by commas.
</p>
<table summary="" class="menu" border="0" cellspacing="0">
<tr><td align="left" valign="top">&bull; <a href="#Error-handling-from-FORTRAN" accesskey="1">Error handling from FORTRAN</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
</table>

<hr>
<a name="Error-handling-from-FORTRAN"></a>
<div class="header">
<p>
Previous: <a href="#Error-handling" accesskey="p" rel="prev">Error handling</a>, Up: <a href="#Error-handling" accesskey="u" rel="up">Error handling</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Error-handling-from-FORTRAN-1"></a>
<h4 class="subsection">6.2.1 Error handling from FORTRAN</h4>
<a name="index-Error-handling-from-FORTRAN"></a>

<p>There are two interface function provided to call <code>error</code> and
<code>warning</code> from FORTRAN code, in each case with a simple character
string argument.  They are defined as
</p>
<div class="example">
<pre class="example">subroutine rexit(<var>message</var>)
subroutine rwarn(<var>message</var>)
</pre></div>

<p>Messages of more than 255 characters are truncated, with a warning.
</p>

<hr>
<a name="Random-numbers"></a>
<div class="header">
<p>
Next: <a href="#Missing-and-IEEE-values" accesskey="n" rel="next">Missing and IEEE values</a>, Previous: <a href="#Error-handling" accesskey="p" rel="prev">Error handling</a>, Up: <a href="#The-R-API" accesskey="u" rel="up">The R API</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Random-number-generation"></a>
<h3 class="section">6.3 Random number generation</h3>
<a name="index-Random-numbers-in-C"></a>
<a name="index-unif_005frand"></a>
<a name="index-norm_005frand"></a>
<a name="index-exp_005frand"></a>
<a name="index-GetRNGstate"></a>
<a name="index-PutRNGstate"></a>
<a name="index-_002eRandom_002eseed"></a>
<a name="index-seed_005fin"></a>
<a name="index-seed_005fout"></a>

<p>The interface to R&rsquo;s internal random number generation routines is
</p>
<div class="example">
<pre class="example">double unif_rand();
double norm_rand();
double exp_rand();
</pre></div>

<p>giving one uniform, normal or exponential pseudo-random variate.
However, before these are used, the user must call
</p>
<div class="example">
<pre class="example">GetRNGstate();
</pre></div>

<p>and after all the required variates have been generated, call
</p>
<div class="example">
<pre class="example">PutRNGstate();
</pre></div>

<p>These essentially read in (or create) <code>.Random.seed</code> and write it
out after use.
</p>
<p>File <samp>S.h</samp> defines <code>seed_in</code> and <code>seed_out</code> for
S-compatibility rather than <code>GetRNGstate</code> and
<code>PutRNGstate</code>.  These take a <code>long *</code> argument which is
ignored.
</p>
<p>The random number generator is private to R; there is no way to
select the kind of RNG or set the seed except by evaluating calls to the
R functions.
</p>
<p>The C code behind R&rsquo;s <code>r<var>xxx</var></code> functions can be accessed by
including the header file <samp>Rmath.h</samp>; See <a href="#Distribution-functions">Distribution functions</a>.  Those calls generate a single variate and should also be
enclosed in calls to <code>GetRNGstate</code> and <code>PutRNGstate</code>.
</p>

<hr>
<a name="Missing-and-IEEE-values"></a>
<div class="header">
<p>
Next: <a href="#Printing" accesskey="n" rel="next">Printing</a>, Previous: <a href="#Random-numbers" accesskey="p" rel="prev">Random numbers</a>, Up: <a href="#The-R-API" accesskey="u" rel="up">The R API</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Missing-and-IEEE-special-values"></a>
<h3 class="section">6.4 Missing and <acronym>IEEE</acronym> special values</h3>
<a name="index-Missing-values-1"></a>
<a name="index-IEEE-special-values-1"></a>
<a name="index-ISNA-1"></a>
<a name="index-ISNAN-1"></a>
<a name="index-R_005fFINITE"></a>
<a name="index-R_005fIsNaN"></a>
<a name="index-R_005fPosInf"></a>
<a name="index-R_005fNegInf"></a>
<a name="index-NA_005fREAL"></a>

<p>A set of functions is provided to test for <code>NA</code>, <code>Inf</code>,
<code>-Inf</code> and <code>NaN</code>.  These functions are accessed <em>via</em> macros:
</p>
<div class="example">
<pre class="example">ISNA(<var>x</var>)        <span class="roman">True for R&rsquo;s <code>NA</code> only</span>
ISNAN(<var>x</var>)       <span class="roman">True for R&rsquo;s <code>NA</code> and <acronym>IEEE</acronym> <code>NaN</code></span>
R_FINITE(<var>x</var>)    <span class="roman">False for <code>Inf</code>, <code>-Inf</code>, <code>NA</code>, <code>NaN</code></span>
</pre></div>

<p>and <em>via</em> function <code>R_IsNaN</code> which is true for <code>NaN</code> but not
<code>NA</code>.
</p>
<p>Do use <code>R_FINITE</code> rather than <code>isfinite</code> or <code>finite</code>; the
latter is often mendacious and <code>isfinite</code> is only available on a
some platforms, on which <code>R_FINITE</code> is a macro expanding to
<code>isfinite</code>.
</p>
<p>Currently in C code <code>ISNAN</code> is a macro calling <code>isnan</code>.
(Since this gives problems on some C++ systems, if the R headers is
called from C++ code a function call is used.)
</p>
<p>You can check for <code>Inf</code> or <code>-Inf</code> by testing equality to
<code>R_PosInf</code> or <code>R_NegInf</code>, and set (but not test) an <code>NA</code>
as <code>NA_REAL</code>.
</p>
<p>All of the above apply to <em>double</em> variables only.  For integer
variables there is a variable accessed by the macro <code>NA_INTEGER</code>
which can used to set or test for missingness.
</p>

<hr>
<a name="Printing"></a>
<div class="header">
<p>
Next: <a href="#Calling-C-from-FORTRAN-and-vice-versa" accesskey="n" rel="next">Calling C from FORTRAN and vice versa</a>, Previous: <a href="#Missing-and-IEEE-values" accesskey="p" rel="prev">Missing and IEEE values</a>, Up: <a href="#The-R-API" accesskey="u" rel="up">The R API</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Printing-1"></a>
<h3 class="section">6.5 Printing</h3>
<a name="index-Printing-from-C"></a>
<a name="index-Rprintf"></a>
<a name="index-REprintf"></a>
<a name="index-Rvprintf"></a>
<a name="index-REvprintf"></a>

<p>The most useful function for printing from a C routine compiled into
R is <code>Rprintf</code>.  This is used in exactly the same way as
<code>printf</code>, but is guaranteed to write to R&rsquo;s output (which might
be a <acronym>GUI</acronym> console rather than a file, and can be re-directed by
<code>sink</code>).  It is wise to write complete lines (including the
<code>&quot;\n&quot;</code>) before returning to R.  It is defined in
<samp>R_ext/Print.h</samp>.
</p>
<p>The function <code>REprintf</code> is similar but writes on the error stream
(<code>stderr</code>) which may or may not be different from the standard
output stream.
</p>
<p>Functions <code>Rvprintf</code> and <code>REvprintf</code> are analogues using the
<code>vprintf</code> interface.  Because that is a C99<a name="DOCF140" href="#FOOT140"><sup>140</sup></a> interface, they are only defined by <samp>R_ext/Print.h</samp> in C++
code if the macro <code>R_USE_C99_IN_CXX</code> is defined when it is
included.
</p>
<p>Another circumstance when it may be important to use these functions is
when using parallel computation on a cluster of computational nodes, as
their output will be re-directed/logged appropriately.
</p>
<table summary="" class="menu" border="0" cellspacing="0">
<tr><td align="left" valign="top">&bull; <a href="#Printing-from-FORTRAN" accesskey="1">Printing from FORTRAN</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
</table>

<hr>
<a name="Printing-from-FORTRAN"></a>
<div class="header">
<p>
Previous: <a href="#Printing" accesskey="p" rel="prev">Printing</a>, Up: <a href="#Printing" accesskey="u" rel="up">Printing</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Printing-from-FORTRAN-1"></a>
<h4 class="subsection">6.5.1 Printing from FORTRAN</h4>
<a name="index-Printing-from-FORTRAN"></a>

<p>On many systems FORTRAN <code>write</code> and <code>print</code> statements can be
used, but the output may not interleave well with that of C, and will be
invisible on <acronym>GUI</acronym> interfaces.  They are not portable and best
avoided.
</p>
<p>Three subroutines are provided to ease the output of information from
FORTRAN code.
</p>
<div class="example">
<pre class="example">subroutine dblepr(<var>label</var>, <var>nchar</var>, <var>data</var>, <var>ndata</var>)
subroutine realpr(<var>label</var>, <var>nchar</var>, <var>data</var>, <var>ndata</var>)
subroutine intpr (<var>label</var>, <var>nchar</var>, <var>data</var>, <var>ndata</var>)
</pre></div>

<p>Here <var>label</var> is a character label of up to 255 characters,
<var>nchar</var> is its length (which can be <code>-1</code> if the whole label is
to be used), and <var>data</var> is an array of length at least <var>ndata</var>
of the appropriate type (<code>double precision</code>, <code>real</code> and
<code>integer</code> respectively).  These routines print the label on one
line and then print <var>data</var> as if it were an R vector on
subsequent line(s).  They work with zero <var>ndata</var>, and so can be used
to print a label alone.
</p>
<hr>
<a name="Calling-C-from-FORTRAN-and-vice-versa"></a>
<div class="header">
<p>
Next: <a href="#Numerical-analysis-subroutines" accesskey="n" rel="next">Numerical analysis subroutines</a>, Previous: <a href="#Printing" accesskey="p" rel="prev">Printing</a>, Up: <a href="#The-R-API" accesskey="u" rel="up">The R API</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Calling-C-from-FORTRAN-and-vice-versa-1"></a>
<h3 class="section">6.6 Calling C from FORTRAN and vice versa</h3>
<a name="index-Calling-C-from-FORTRAN-and-vice-versa"></a>

<p>Naming conventions for symbols generated by FORTRAN differ by platform:
it is not safe to assume that FORTRAN names appear to C with a trailing
underscore.  To help cover up the platform-specific differences there is
a set of macros that should be used.
</p>
<dl compact="compact">
<dt><code>F77_SUB(<var>name</var>)</code></dt>
<dd><p>to define a function in C to be called from FORTRAN
</p></dd>
<dt><code>F77_NAME(<var>name</var>)</code></dt>
<dd><p>to declare a FORTRAN routine in C before use
</p></dd>
<dt><code>F77_CALL(<var>name</var>)</code></dt>
<dd><p>to call a FORTRAN routine from C
</p></dd>
<dt><code>F77_COMDECL(<var>name</var>)</code></dt>
<dd><p>to declare a FORTRAN common block in C
</p></dd>
<dt><code>F77_COM(<var>name</var>)</code></dt>
<dd><p>to access a FORTRAN common block from C
</p></dd>
</dl>

<p>On most current platforms these are all the same, but it is unwise to
rely on this.  Note that names with underscores are not legal in FORTRAN
77, and are not portably handled by the above macros.  (Also, all
FORTRAN names for use by R are lower case, but this is not enforced
by the macros.)
</p>
<p>For example, suppose we want to call R&rsquo;s normal random numbers from
FORTRAN.  We need a C wrapper along the lines of
</p>
<a name="index-Random-numbers-in-FORTRAN"></a>
<div class="example">
<pre class="example">#include &lt;R.h&gt;

void F77_SUB(rndstart)(void) { GetRNGstate(); }
void F77_SUB(rndend)(void) { PutRNGstate(); }
double F77_SUB(normrnd)(void) { return norm_rand(); }
</pre></div>

<p>to be called from FORTRAN as in
</p>
<div class="example">
<pre class="example">      subroutine testit()
      double precision normrnd, x
      call rndstart()
      x = normrnd()
      call dblepr(&quot;X was&quot;, 5, x, 1)
      call rndend()
      end
</pre></div>

<p>Note that this is not guaranteed to be portable, for the return
conventions might not be compatible between the C and FORTRAN compilers
used.  (Passing values <em>via</em> arguments is safer.)
</p>
<p>The standard packages, for example <strong>stats</strong>, are a rich source of
further examples.
</p>
<p>Passing character strings from C to FORTRAN 77 or <em>vice versa</em> is
not portable (and to Fortran 90 or later is even less so).  We have
found that it helps to ensure that a C string to be passed is followed
by several <code>nul</code>s (and not just the one needed as a C terminator).
But for maximal portability character strings in FORTRAN should be
avoided.
</p>

<hr>
<a name="Numerical-analysis-subroutines"></a>
<div class="header">
<p>
Next: <a href="#Optimization" accesskey="n" rel="next">Optimization</a>, Previous: <a href="#Calling-C-from-FORTRAN-and-vice-versa" accesskey="p" rel="prev">Calling C from FORTRAN and vice versa</a>, Up: <a href="#The-R-API" accesskey="u" rel="up">The R API</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Numerical-analysis-subroutines-1"></a>
<h3 class="section">6.7 Numerical analysis subroutines</h3>
<a name="index-Numerical-analysis-subroutines-from-C"></a>

<p>R contains a large number of mathematical functions for its own use,
for example numerical linear algebra computations and special functions.
</p>
<p>The header files <samp>R_ext/BLAS.h</samp>, <samp>R_ext/Lapack.h</samp> and
<samp>R_ext/Linpack.h</samp> contains declarations of the BLAS, LAPACK and
LINPACK linear algebra functions included in R.  These are expressed
as calls to FORTRAN subroutines, and they will also be usable from
users&rsquo; FORTRAN code.  Although not part of the official <acronym>API</acronym>,
this set of subroutines is unlikely to change (but might be
supplemented).
</p>
<p>The header file <samp>Rmath.h</samp> lists many other functions that are
available and documented in the following subsections. Many of these are
C interfaces to the code behind R functions, so the R function
documentation may give further details.
</p>
<table summary="" class="menu" border="0" cellspacing="0">
<tr><td align="left" valign="top">&bull; <a href="#Distribution-functions" accesskey="1">Distribution functions</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Mathematical-functions" accesskey="2">Mathematical functions</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Numerical-Utilities" accesskey="3">Numerical Utilities</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Mathematical-constants" accesskey="4">Mathematical constants</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
</table>

<hr>
<a name="Distribution-functions"></a>
<div class="header">
<p>
Next: <a href="#Mathematical-functions" accesskey="n" rel="next">Mathematical functions</a>, Previous: <a href="#Numerical-analysis-subroutines" accesskey="p" rel="prev">Numerical analysis subroutines</a>, Up: <a href="#Numerical-analysis-subroutines" accesskey="u" rel="up">Numerical analysis subroutines</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Distribution-functions-1"></a>
<h4 class="subsection">6.7.1 Distribution functions</h4>
<a name="index-Distribution-functions-from-C"></a>

<p>The routines used to calculate densities, cumulative distribution
functions and quantile functions for the standard statistical
distributions are available as entry points.
</p>
<p>The arguments for the entry points follow the pattern of those for the
normal distribution:
</p>
<div class="example">
<pre class="example">double dnorm(double <var>x</var>, double <var>mu</var>, double <var>sigma</var>, int <var>give_log</var>);
double pnorm(double <var>x</var>, double <var>mu</var>, double <var>sigma</var>, int <var>lower_tail</var>,
	     int <var>give_log</var>);
double qnorm(double <var>p</var>, double <var>mu</var>, double <var>sigma</var>, int <var>lower_tail</var>,
	     int <var>log_p</var>);
double rnorm(double <var>mu</var>, double <var>sigma</var>);
</pre></div>

<p>That is, the first argument gives the position for the density and CDF
and probability for the quantile function, followed by the
distribution&rsquo;s parameters.  Argument <var>lower_tail</var> should be
<code>TRUE</code> (or <code>1</code>) for normal use, but can be <code>FALSE</code> (or
<code>0</code>) if the probability of the upper tail is desired or specified.
</p>
<p>Finally, <var>give_log</var> should be non-zero if the result is required on
log scale, and <var>log_p</var> should be non-zero if <var>p</var> has been
specified on log scale.
</p>
<p>Note that you directly get the cumulative (or &ldquo;integrated&rdquo;)
<em>hazard</em> function, H(t) = - log(1 -
F(t)), by using
</p>
<div class="example">
<pre class="example">- p<var>dist</var>(t, ..., /*lower_tail = */ FALSE, /* give_log = */ TRUE)
</pre></div>

<p>or shorter (and more cryptic) <code>- p<var>dist</var>(t, ..., 0, 1)</code>.
<a name="index-cumulative-hazard"></a>
</p>
<p>The random-variate generation routine <code>rnorm</code> returns one normal
variate. See <a href="#Random-numbers">Random numbers</a>, for the protocol in using the
random-variate routines.
<a name="index-Random-numbers-in-C-1"></a>
</p>
<p>Note that these argument sequences are (apart from the names and that
<code>rnorm</code> has no <var>n</var>) mainly the same as the corresponding R
functions of the same name, so the documentation of the R functions
can be used.  Note that the exponential and gamma distributions are
parametrized by <code>scale</code> rather than <code>rate</code>.
</p>

<p>For reference, the following table gives the basic name (to be prefixed
by &lsquo;<samp>d</samp>&rsquo;, &lsquo;<samp>p</samp>&rsquo;, &lsquo;<samp>q</samp>&rsquo; or &lsquo;<samp>r</samp>&rsquo; apart from the exceptions
noted) and distribution-specific arguments for the complete set of
distributions.
</p>
<blockquote>
<table summary="">
<tr><td width="28%">beta</td><td width="22%"><code>beta</code></td><td width="30%"><code>a</code>, <code>b</code></td></tr>
<tr><td width="28%">non-central beta</td><td width="22%"><code>nbeta</code></td><td width="30%"><code>a</code>, <code>b</code>, <code>ncp</code></td></tr>
<tr><td width="28%">binomial</td><td width="22%"><code>binom</code></td><td width="30%"><code>n</code>, <code>p</code></td></tr>
<tr><td width="28%">Cauchy</td><td width="22%"><code>cauchy</code></td><td width="30%"><code>location</code>, <code>scale</code></td></tr>
<tr><td width="28%">chi-squared</td><td width="22%"><code>chisq</code></td><td width="30%"><code>df</code></td></tr>
<tr><td width="28%">non-central chi-squared</td><td width="22%"><code>nchisq</code></td><td width="30%"><code>df</code>, <code>ncp</code></td></tr>
<tr><td width="28%">exponential</td><td width="22%"><code>exp</code></td><td width="30%"><code>scale</code> (and <strong>not</strong> <code>rate</code>)</td></tr>
<tr><td width="28%">F</td><td width="22%"><code>f</code></td><td width="30%"><code>n1</code>, <code>n2</code></td></tr>
<tr><td width="28%">non-central F</td><td width="22%"><code>nf</code></td><td width="30%"><code>n1</code>, <code>n2</code>, <code>ncp</code></td></tr>
<tr><td width="28%">gamma</td><td width="22%"><code>gamma</code></td><td width="30%"><code>shape</code>, <code>scale</code></td></tr>
<tr><td width="28%">geometric</td><td width="22%"><code>geom</code></td><td width="30%"><code>p</code></td></tr>
<tr><td width="28%">hypergeometric</td><td width="22%"><code>hyper</code></td><td width="30%"><code>NR</code>, <code>NB</code>, <code>n</code></td></tr>
<tr><td width="28%">logistic</td><td width="22%"><code>logis</code></td><td width="30%"><code>location</code>, <code>scale</code></td></tr>
<tr><td width="28%">lognormal</td><td width="22%"><code>lnorm</code></td><td width="30%"><code>logmean</code>, <code>logsd</code></td></tr>
<tr><td width="28%">negative binomial</td><td width="22%"><code>nbinom</code></td><td width="30%"><code>size</code>, <code>prob</code></td></tr>
<tr><td width="28%">normal</td><td width="22%"><code>norm</code></td><td width="30%"><code>mu</code>, <code>sigma</code></td></tr>
<tr><td width="28%">Poisson</td><td width="22%"><code>pois</code></td><td width="30%"><code>lambda</code></td></tr>
<tr><td width="28%">Student&rsquo;s t</td><td width="22%"><code>t</code></td><td width="30%"><code>n</code></td></tr>
<tr><td width="28%">non-central t</td><td width="22%"><code>nt</code></td><td width="30%"><code>df</code>, <code>delta</code></td></tr>
<tr><td width="28%">Studentized range</td><td width="22%"><code>tukey</code> (*)</td><td width="30%"><code>rr</code>, <code>cc</code>, <code>df</code></td></tr>
<tr><td width="28%">uniform</td><td width="22%"><code>unif</code></td><td width="30%"><code>a</code>, <code>b</code></td></tr>
<tr><td width="28%">Weibull</td><td width="22%"><code>weibull</code></td><td width="30%"><code>shape</code>, <code>scale</code></td></tr>
<tr><td width="28%">Wilcoxon rank sum</td><td width="22%"><code>wilcox</code></td><td width="30%"><code>m</code>, <code>n</code></td></tr>
<tr><td width="28%">Wilcoxon signed rank</td><td width="22%"><code>signrank</code></td><td width="30%"><code>n</code></td></tr>
</table>
</blockquote>

<p>Entries marked with an asterisk only have &lsquo;<samp>p</samp>&rsquo; and &lsquo;<samp>q</samp>&rsquo;
functions available, and none of the non-central distributions have
&lsquo;<samp>r</samp>&rsquo; functions.  After a call to <code>dwilcox</code>, <code>pwilcox</code> or
<code>qwilcox</code> the function <code>wilcox_free()</code> should be called, and
similarly for the signed rank functions.
</p>
<p>(If remapping is suppressed, the Normal distribution names are
<code>Rf_dnorm4</code>, <code>Rf_pnorm5</code> and <code>Rf_qnorm5</code>.)
</p>
<p>For the negative binomial distribution (&lsquo;<samp>nbinom</samp>&rsquo;), in addition to the
<code>(size, prob)</code> parametrization, the alternative <code>(size, mu)</code>
parametrization is provided as well by functions &lsquo;<samp>[dpqr]nbinom_mu()</samp>&rsquo;,
see <kbd>?NegBinomial</kbd> in R.
</p>
<p>Functions <code>dpois_raw(x, *)</code> and <code>dbinom_raw(x, *)</code> are versions of the
Poisson and binomial probability mass functions which work continuously in
<code>x</code>, whereas <code>dbinom(x,*)</code> and <code>dpois(x,*)</code> only return non
zero values for integer <code>x</code>.
</p><div class="example">
<pre class="example">double dbinom_raw(double x, double n, double p, double q, int give_log)
double dpois_raw (double x, double lambda, int give_log)
</pre></div>
<p>Note that <code>dbinom_raw()</code> gets both p and q = 1-p which
may be advantageous when one of them is close to 1.
</p>

<hr>
<a name="Mathematical-functions"></a>
<div class="header">
<p>
Next: <a href="#Numerical-Utilities" accesskey="n" rel="next">Numerical Utilities</a>, Previous: <a href="#Distribution-functions" accesskey="p" rel="prev">Distribution functions</a>, Up: <a href="#Numerical-analysis-subroutines" accesskey="u" rel="up">Numerical analysis subroutines</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Mathematical-functions-1"></a>
<h4 class="subsection">6.7.2 Mathematical functions</h4>

<a name="index-gammafn"></a>
<a name="index-lgammafn"></a>
<a name="index-digamma"></a>
<a name="index-trigamma"></a>
<a name="index-tetragamma"></a>
<a name="index-pentagamma"></a>
<a name="index-psigamma"></a>
<a name="index-Gamma-function"></a>
<dl>
<dt><a name="index-gammafn-1"></a>Function: <em>double</em> <strong>gammafn</strong> <em>(double <var>x</var>)</em></dt>
<dt><a name="index-lgammafn-1"></a>Function: <em>double</em> <strong>lgammafn</strong> <em>(double <var>x</var>)</em></dt>
<dt><a name="index-digamma-1"></a>Function: <em>double</em> <strong>digamma</strong> <em>(double <var>x</var>)</em></dt>
<dt><a name="index-trigamma-1"></a>Function: <em>double</em> <strong>trigamma</strong> <em>(double <var>x</var>)</em></dt>
<dt><a name="index-tetragamma-1"></a>Function: <em>double</em> <strong>tetragamma</strong> <em>(double <var>x</var>)</em></dt>
<dt><a name="index-pentagamma-1"></a>Function: <em>double</em> <strong>pentagamma</strong> <em>(double <var>x</var>)</em></dt>
<dt><a name="index-psigamma-1"></a>Function: <em>double</em> <strong>psigamma</strong> <em>(double <var>x</var>, double <var>deriv</var>)</em></dt>
<dd><p>The Gamma function, the natural logarithm of its absolute value and
first four derivatives and the n-th derivative of Psi, the digamma
function, which is the derivative of <code>lgammafn</code>. In other words,
<code>digamma(x)</code> is the same as <code>psigamma(x,0)</code>,
<code>trigamma(x) == psigamma(x,1)</code>, etc.
</p></dd></dl>

<a name="index-beta"></a>
<a name="index-lbeta"></a>
<a name="index-Beta-function"></a>
<dl>
<dt><a name="index-beta-1"></a>Function: <em>double</em> <strong>beta</strong> <em>(double <var>a</var>, double <var>b</var>)</em></dt>
<dt><a name="index-lbeta-1"></a>Function: <em>double</em> <strong>lbeta</strong> <em>(double <var>a</var>, double <var>b</var>)</em></dt>
<dd><p>The (complete) Beta function and its natural logarithm.
</p></dd></dl>

<a name="index-choose"></a>
<a name="index-lchoose"></a>
<dl>
<dt><a name="index-choose-1"></a>Function: <em>double</em> <strong>choose</strong> <em>(double <var>n</var>, double <var>k</var>)</em></dt>
<dt><a name="index-lchoose-1"></a>Function: <em>double</em> <strong>lchoose</strong> <em>(double <var>n</var>, double <var>k</var>)</em></dt>
<dd><p>The number of combinations of <var>k</var> items chosen from from <var>n</var> and
the natural logarithm of its absolute value, generalized to arbitrary real
<var>n</var>.  <var>k</var> is rounded to the nearest integer (with a warning if
needed).
</p></dd></dl>

<a name="index-bessel_005fi"></a>
<a name="index-bessel_005fj"></a>
<a name="index-bessel_005fk"></a>
<a name="index-bessel_005fy"></a>
<a name="index-Bessel-functions"></a>
<dl>
<dt><a name="index-bessel_005fi-1"></a>Function: <em>double</em> <strong>bessel_i</strong> <em>(double <var>x</var>, double <var>nu</var>, double <var>expo</var>)</em></dt>
<dt><a name="index-bessel_005fj-1"></a>Function: <em>double</em> <strong>bessel_j</strong> <em>(double <var>x</var>, double <var>nu</var>)</em></dt>
<dt><a name="index-bessel_005fk-1"></a>Function: <em>double</em> <strong>bessel_k</strong> <em>(double <var>x</var>, double <var>nu</var>, double <var>expo</var>)</em></dt>
<dt><a name="index-bessel_005fy-1"></a>Function: <em>double</em> <strong>bessel_y</strong> <em>(double <var>x</var>, double <var>nu</var>)</em></dt>
<dd><p>Bessel functions of types I, J, K and Y with index <var>nu</var>.  For
<code>bessel_i</code> and <code>bessel_k</code> there is the option to return
<span class="nolinebreak">exp(-</span><var>x</var>)&nbsp;I(<var>x</var>;&nbsp;<var>nu</var>)<!-- /@w --> or exp(<var>x</var>)&nbsp;K(<var>x</var>;&nbsp;<var>nu</var>)<!-- /@w --> if <var>expo</var> is 2. (Use <code><var>expo</var> == 1</code> for unscaled
values.)
</p></dd></dl>


<hr>
<a name="Numerical-Utilities"></a>
<div class="header">
<p>
Next: <a href="#Mathematical-constants" accesskey="n" rel="next">Mathematical constants</a>, Previous: <a href="#Mathematical-functions" accesskey="p" rel="prev">Mathematical functions</a>, Up: <a href="#Numerical-analysis-subroutines" accesskey="u" rel="up">Numerical analysis subroutines</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Numerical-Utilities-1"></a>
<h4 class="subsection">6.7.3 Numerical Utilities</h4>
<p>There are a few other numerical utility functions available as entry points.
</p>

<dl>
<dt><a name="index-R_005fpow"></a>Function: <em>double</em> <strong>R_pow</strong> <em>(double <var>x</var>, double <var>y</var>)</em></dt>
<dt><a name="index-R_005fpow_005fdi"></a>Function: <em>double</em> <strong>R_pow_di</strong> <em>(double <var>x</var>, int <var>i</var>)</em></dt>
<dd><p><code>R_pow(<var>x</var>, <var>y</var>)</code> and <code>R_pow_di(<var>x</var>, <var>i</var>)</code>
compute <code><var>x</var>^<var>y</var></code> and <code><var>x</var>^<var>i</var></code>, respectively
using <code>R_FINITE</code> checks and returning the proper result (the same
as R) for the cases where <var>x</var>, <var>y</var> or <var>i</var> are 0 or
missing or infinite or <code>NaN</code>.
</p></dd></dl>

<dl>
<dt><a name="index-log1p"></a>Function: <em>double</em> <strong>log1p</strong> <em>(double <var>x</var>)</em></dt>
<dd><p>Computes <code>log(1 + <var>x</var>)</code> (<em>log 1 <b>p</b>lus x</em>), accurately
even for small <var>x</var>, i.e., |x| &lt;&lt; 1.
</p>
<p>This should be provided by your platform, in which case it is not
included in <samp>Rmath.h</samp>, but is (probably) in <samp>math.h</samp> which
<samp>Rmath.h</samp> includes (except under C++, so it may not be declared for
C++98).
</p></dd></dl>

<dl>
<dt><a name="index-log1pmx"></a>Function: <em>double</em> <strong>log1pmx</strong> <em>(double <var>x</var>)</em></dt>
<dd><p>Computes <code>log(1 + <var>x</var>) - <var>x</var></code> (<em>log 1 <b>p</b>lus x <b>m</b>inus <b>x</b></em>),
accurately even for small <var>x</var>, i.e., |x| &lt;&lt; 1.
</p></dd></dl>

<dl>
<dt><a name="index-log1pexp"></a>Function: <em>double</em> <strong>log1pexp</strong> <em>(double <var>x</var>)</em></dt>
<dd><p>Computes <code>log(1 + exp(<var>x</var>))</code> (<em>log 1 <b>p</b>lus <b>exp</b></em>),
accurately, notably for large <var>x</var>, e.g., x &gt; 720.
</p></dd></dl>


<dl>
<dt><a name="index-expm1"></a>Function: <em>double</em> <strong>expm1</strong> <em>(double <var>x</var>)</em></dt>
<dd><p>Computes <code>exp(<var>x</var>) - 1</code> (<em>exp x <b>m</b>inus 1</em>), accurately
even for small <var>x</var>, i.e., |x| &lt;&lt; 1.
</p>
<p>This should be provided by your platform, in which case it is not
included in <samp>Rmath.h</samp>, but is (probably) in <samp>math.h</samp> which
<samp>Rmath.h</samp> includes (except under C++, so it may not be declared for
C++98).
</p></dd></dl>

<dl>
<dt><a name="index-lgamma1p"></a>Function: <em>double</em> <strong>lgamma1p</strong> <em>(double <var>x</var>)</em></dt>
<dd><p>Computes <code>log(gamma(<var>x</var> + 1))</code> (<em>log(gamma(1 <b>p</b>lus x))</em>),
accurately even for small <var>x</var>, i.e., 0 &lt; x &lt; 0.5.
</p></dd></dl>

<dl>
<dt><a name="index-cospi"></a>Function: <em>double</em> <strong>cospi</strong> <em>(double <var>x</var>)</em></dt>
<dd><p>Computes <code>cos(pi * x)</code> (where <code>pi</code> is 3.14159...),
accurately, notably for half integer <var>x</var>.
</p>
<p>This might be provided by your platform<a name="DOCF141" href="#FOOT141"><sup>141</sup></a>, in which case it is not included in <samp>Rmath.h</samp>, but is
in <samp>math.h</samp> which <samp>Rmath.h</samp> includes.  (Ensure that
neither <samp>math.h</samp> nor <samp>cmath</samp> is included before
<samp>Rmath.h</samp> or define
</p><div class="example">
<pre class="example">#define __STDC_WANT_IEC_60559_FUNCS_EXT__ 1
</pre></div>
<p>before the first inclusion.)
</p></dd></dl>

<dl>
<dt><a name="index-sinpi"></a>Function: <em>double</em> <strong>sinpi</strong> <em>(double <var>x</var>)</em></dt>
<dd><p>Computes <code>sin(pi * x)</code> accurately, notably for (half) integer <var>x</var>.
</p>
<p>This might be provided by your platform, in which case it is not
included in <samp>Rmath.h</samp>, but is in <samp>math.h</samp> which <samp>Rmath.h</samp>
includes (but see the comments for <code>cospi</code>).
</p></dd></dl>

<dl>
<dt><a name="index-tanpi"></a>Function: <em>double</em> <strong>tanpi</strong> <em>(double <var>x</var>)</em></dt>
<dd><p>Computes <code>tan(pi * x)</code> accurately, notably for (half) integer <var>x</var>.
</p>
<p>This might be provided by your platform, in which case it is not included
in <samp>Rmath.h</samp>, but is in <samp>math.h</samp> which <samp>Rmath.h</samp> includes
(but see the comments for <code>cospi</code>).
</p></dd></dl>

<dl>
<dt><a name="index-logspace_005fadd"></a>Function: <em>double</em> <strong>logspace_add</strong> <em>(double <var>logx</var>, double <var>logy</var>)</em></dt>
<dt><a name="index-logspace_005fsub"></a>Function: <em>double</em> <strong>logspace_sub</strong> <em>(double <var>logx</var>, double <var>logy</var>)</em></dt>
<dt><a name="index-logspace_005fsum"></a>Function: <em>double</em> <strong>logspace_sum</strong> <em>(const double* <var>logx</var>, int <var>n</var>)</em></dt>
<dd><p>Compute the log of a sum or difference from logs of terms, i.e., &ldquo;x +
y&rdquo; as <code>log (exp(<var>logx</var>) + exp(<var>logy</var>))</code> and &ldquo;x - y&rdquo; as
<code>log (exp(<var>logx</var>) - exp(<var>logy</var>))</code>,
and &ldquo;sum_i x[i]&rdquo; as <code>log (sum[i = 1:<var>n</var> exp(<var>logx</var>[i])] )</code>
without causing unnecessary overflows or throwing away too much accuracy.
</p></dd></dl>

<dl>
<dt><a name="index-imax2"></a>Function: <em>int</em> <strong>imax2</strong> <em>(int <var>x</var>, int <var>y</var>)</em></dt>
<dt><a name="index-imin2"></a>Function: <em>int</em> <strong>imin2</strong> <em>(int <var>x</var>, int <var>y</var>)</em></dt>
<dt><a name="index-fmax2"></a>Function: <em>double</em> <strong>fmax2</strong> <em>(double <var>x</var>, double <var>y</var>)</em></dt>
<dt><a name="index-fmin2"></a>Function: <em>double</em> <strong>fmin2</strong> <em>(double <var>x</var>, double <var>y</var>)</em></dt>
<dd><p>Return the larger (<code>max</code>) or smaller (<code>min</code>) of two integer or
double numbers, respectively.  Note that <code>fmax2</code> and <code>fmin2</code>
differ from C99&rsquo;s <code>fmax</code> and <code>fmin</code> when one of the arguments
is a <code>NaN</code>: these versions return <code>NaN</code>.
</p></dd></dl>

<dl>
<dt><a name="index-sign"></a>Function: <em>double</em> <strong>sign</strong> <em>(double <var>x</var>)</em></dt>
<dd><p>Compute the <em>signum</em> function, where sign(<var>x</var>) is 1, 0, or
<em>-1</em>, when <var>x</var> is positive, 0, or negative, respectively, and
<code>NaN</code> if <code>x</code> is a <code>NaN</code>.
</p></dd></dl>

<dl>
<dt><a name="index-fsign"></a>Function: <em>double</em> <strong>fsign</strong> <em>(double <var>x</var>, double <var>y</var>)</em></dt>
<dd><p>Performs &ldquo;transfer of sign&rdquo; and is defined as |x| * sign(y).
</p></dd></dl>

<dl>
<dt><a name="index-fprec"></a>Function: <em>double</em> <strong>fprec</strong> <em>(double <var>x</var>, double <var>digits</var>)</em></dt>
<dd><p>Returns the value of <var>x</var> rounded to <var>digits</var> decimal digits
(after the decimal point).
</p>
<p>This is the function used by R&rsquo;s <code>signif()</code>.
</p></dd></dl>

<dl>
<dt><a name="index-fround"></a>Function: <em>double</em> <strong>fround</strong> <em>(double <var>x</var>, double <var>digits</var>)</em></dt>
<dd><p>Returns the value of <var>x</var> rounded to <var>digits</var> <em>significant</em>
decimal digits.
</p>
<p>This is the function used by R&rsquo;s <code>round()</code>.  (Note that C99/C++11
provide a <code>round</code> function but C++98 need not.)
</p></dd></dl>

<dl>
<dt><a name="index-ftrunc"></a>Function: <em>double</em> <strong>ftrunc</strong> <em>(double <var>x</var>)</em></dt>
<dd><p>Returns the value of <var>x</var> truncated (to an integer value) towards
zero.
</p>
<p>(Note that C99/C++11 provide a <code>round</code> function but C++98 need
not.)
</p></dd></dl>

<hr>
<a name="Mathematical-constants"></a>
<div class="header">
<p>
Previous: <a href="#Numerical-Utilities" accesskey="p" rel="prev">Numerical Utilities</a>, Up: <a href="#Numerical-analysis-subroutines" accesskey="u" rel="up">Numerical analysis subroutines</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Mathematical-constants-1"></a>
<h4 class="subsection">6.7.4 Mathematical constants</h4>
<a name="index-M_005fE"></a>
<a name="index-M_005fPI"></a>

<p>R has a set of commonly used mathematical constants encompassing
constants usually found <samp>math.h</samp> and contains further ones that are
used in statistical computations.  All these are defined to (at least)
30 digits accuracy in <samp>Rmath.h</samp>.  The following definitions
use <code>ln(x)</code> for the natural logarithm (<code>log(x)</code> in R).
</p>
<blockquote>
<table summary="">
<thead><tr><th>Name</th><th>Definition (<code>ln = log</code>)</th><th>round(<em>value</em>, 7)</th></tr></thead>
<tr><td><code>M_E</code></td><td><em>e</em></td><td>2.7182818</td></tr>
<tr><td><code>M_LOG2E</code></td><td>log2(<em>e</em>)</td><td>1.4426950</td></tr>
<tr><td><code>M_LOG10E</code></td><td>log10(<em>e</em>)</td><td>0.4342945</td></tr>
<tr><td><code>M_LN2</code></td><td>ln(2)</td><td>0.6931472</td></tr>
<tr><td><code>M_LN10</code></td><td>ln(10)</td><td>2.3025851</td></tr>
<tr><td><code>M_PI</code></td><td>pi</td><td>3.1415927</td></tr>
<tr><td><code>M_PI_2</code></td><td>pi/2</td><td>1.5707963</td></tr>
<tr><td><code>M_PI_4</code></td><td>pi/4</td><td>0.7853982</td></tr>
<tr><td><code>M_1_PI</code></td><td>1/pi</td><td>0.3183099</td></tr>
<tr><td><code>M_2_PI</code></td><td>2/pi</td><td>0.6366198</td></tr>
<tr><td><code>M_2_SQRTPI</code></td><td>2/sqrt(pi)</td><td>1.1283792</td></tr>
<tr><td><code>M_SQRT2</code></td><td>sqrt(2)</td><td>1.4142136</td></tr>
<tr><td><code>M_SQRT1_2</code></td><td>1/sqrt(2)</td><td>0.7071068</td></tr>
<tr><td><code>M_SQRT_3</code></td><td>sqrt(3)</td><td>1.7320508</td></tr>
<tr><td><code>M_SQRT_32</code></td><td>sqrt(32)</td><td>5.6568542</td></tr>
<tr><td><code>M_LOG10_2</code></td><td>log10(2)</td><td>0.3010300</td></tr>
<tr><td><code>M_2PI</code></td><td>2*pi</td><td>6.2831853</td></tr>
<tr><td><code>M_SQRT_PI</code></td><td>sqrt(pi)</td><td>1.7724539</td></tr>
<tr><td><code>M_1_SQRT_2PI</code></td><td>1/sqrt(2*pi)</td><td>0.3989423</td></tr>
<tr><td><code>M_SQRT_2dPI</code></td><td>sqrt(2/pi)</td><td>0.7978846</td></tr>
<tr><td><code>M_LN_SQRT_PI</code></td><td>ln(sqrt(pi))</td><td>0.5723649</td></tr>
<tr><td><code>M_LN_SQRT_2PI</code></td><td>ln(sqrt(2*pi))</td><td>0.9189385</td></tr>
<tr><td><code>M_LN_SQRT_PId2</code></td><td>ln(sqrt(pi/2))</td><td>0.2257914</td></tr>
</table>
</blockquote>

<p>There are a set of constants (<code>PI</code>, <code>DOUBLE_EPS</code>) (and so on)
defined (unless <code>STRICT_R_HEADERS</code> is defined) in the included
header <samp>R_ext/Constants.h</samp>, mainly for compatibility with S.
</p>
<a name="index-TRUE"></a>
<a name="index-FALSE"></a>
<p>Further, the included header <samp>R_ext/Boolean.h</samp> has enumeration
constants <code>TRUE</code> and <code>FALSE</code> of type <code>Rboolean</code> in
order to provide a way of using &ldquo;logical&rdquo; variables in C consistently.
This can conflict with other software: for example it conflicts with the
headers in IJG&rsquo;s <code>jpeg-9</code> (but not earlier versions).
</p>

<hr>
<a name="Optimization"></a>
<div class="header">
<p>
Next: <a href="#Integration" accesskey="n" rel="next">Integration</a>, Previous: <a href="#Numerical-analysis-subroutines" accesskey="p" rel="prev">Numerical analysis subroutines</a>, Up: <a href="#The-R-API" accesskey="u" rel="up">The R API</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Optimization-1"></a>
<h3 class="section">6.8 Optimization</h3>
<a name="index-optimization"></a>

<p>The C code underlying <code>optim</code> can be accessed directly.  The user
needs to supply a function to compute the function to be minimized, of
the type
</p>
<div class="example">
<pre class="example">typedef double optimfn(int n, double *par, void *ex);
</pre></div>

<p>where the first argument is the number of parameters in the second
argument.  The third argument is a pointer passed down from the calling
routine, normally used to carry auxiliary information.
</p>
<p>Some of the methods also require a gradient function
</p>
<div class="example">
<pre class="example">typedef void optimgr(int n, double *par, double *gr, void *ex);
</pre></div>

<p>which passes back the gradient in the <code>gr</code> argument.  No function
is provided for finite-differencing, nor for approximating the Hessian
at the result.
</p>
<p>The interfaces (defined in header <samp>R_ext/Applic.h</samp>) are
</p>
<ul>
<li> Nelder Mead:
<a name="index-nmmin"></a>
<div class="example">
<pre class="example">void nmmin(int n, double *xin, double *x, double *Fmin, optimfn fn,
	   int *fail, double abstol, double intol, void *ex,
	   double alpha, double beta, double gamma, int trace,
	   int *fncount, int maxit);
</pre></div>

</li><li> BFGS:
<a name="index-vmmin"></a>
<div class="example">
<pre class="example">void vmmin(int n, double *x, double *Fmin,
	   optimfn fn, optimgr gr, int maxit, int trace,
	   int *mask, double abstol, double reltol, int nREPORT,
	   void *ex, int *fncount, int *grcount, int *fail);
</pre></div>

</li><li> Conjugate gradients:
<a name="index-cgmin"></a>
<div class="example">
<pre class="example">void cgmin(int n, double *xin, double *x, double *Fmin,
	   optimfn fn, optimgr gr, int *fail, double abstol,
	   double intol, void *ex, int type, int trace,
	   int *fncount, int *grcount, int maxit);
</pre></div>

</li><li> Limited-memory BFGS with bounds:
<a name="index-lbfgsb"></a>
<div class="example">
<pre class="example">void lbfgsb(int n, int lmm, double *x, double *lower,
	    double *upper, int *nbd, double *Fmin, optimfn fn,
	    optimgr gr, int *fail, void *ex, double factr,
	    double pgtol, int *fncount, int *grcount,
	    int maxit, char *msg, int trace, int nREPORT);
</pre></div>

</li><li> Simulated annealing:
<a name="index-samin"></a>
<div class="example">
<pre class="example">void samin(int n, double *x, double *Fmin, optimfn fn, int maxit,
	   int tmax, double temp, int trace, void *ex);
</pre></div>

</li></ul>

<p>Many of the arguments are common to the various methods.  <code>n</code> is
the number of parameters, <code>x</code> or <code>xin</code> is the starting
parameters on entry and <code>x</code> the final parameters on exit, with
final value returned in <code>Fmin</code>.  Most of the other parameters can
be found from the help page for <code>optim</code>: see the source code
<samp>src/appl/lbfgsb.c</samp> for the values of <code>nbd</code>, which
specifies which bounds are to be used.
</p>

<hr>
<a name="Integration"></a>
<div class="header">
<p>
Next: <a href="#Utility-functions" accesskey="n" rel="next">Utility functions</a>, Previous: <a href="#Optimization" accesskey="p" rel="prev">Optimization</a>, Up: <a href="#The-R-API" accesskey="u" rel="up">The R API</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Integration-1"></a>
<h3 class="section">6.9 Integration</h3>
<a name="index-integration"></a>

<p>The C code underlying <code>integrate</code> can be accessed directly.  The
user needs to supply a <em>vectorizing</em> C function to compute the
function to be integrated, of the type
</p>
<div class="example">
<pre class="example">typedef void integr_fn(double *x, int n, void *ex);
</pre></div>

<p>where <code>x[]</code> is both input and output and has length <code>n</code>, i.e.,
a C function, say <code>fn</code>, of type <code>integr_fn</code> must basically do
<code>for(i in 1:n) x[i] := f(x[i], ex)</code>.  The vectorization requirement
can be used to speed up the integrand instead of calling it <code>n</code>
times.  Note that in the current implementation built on QUADPACK,
<code>n</code> will be either 15 or 21.  The <code>ex</code> argument is a pointer
passed down from the calling routine, normally used to carry auxiliary
information.
</p>
<p>There are interfaces (defined in header <samp>R_ext/Applic.h</samp>) for
integrals over finite and infinite intervals (or &ldquo;ranges&rdquo; or
&ldquo;integration boundaries&rdquo;).
</p>
<ul>
<li> Finite:
<a name="index-Rdqags"></a>
<div class="example">
<pre class="example">void Rdqags(integr_fn f, void *ex, double *a, double *b,
	    double *epsabs, double *epsrel,
	    double *result, double *abserr, int *neval, int *ier,
	    int *limit, int *lenw, int *last,
	    int *iwork, double *work);
</pre></div>

</li><li> Infinite:
<a name="index-Rdqagi"></a>
<div class="example">
<pre class="example">void Rdqagi(integr_fn f, void *ex, double *bound, int *inf,
	    double *epsabs, double *epsrel,
	    double *result, double *abserr, int *neval, int *ier,
	    int *limit, int *lenw, int *last,
	    int *iwork, double *work);
</pre></div>

</li></ul>

<p>Only the 3rd and 4th argument differ for the two integrators; for the
finite range integral using <code>Rdqags</code>, <code>a</code> and <code>b</code> are the
integration interval bounds, whereas for an infinite range integral using
<code>Rdqagi</code>, <code>bound</code> is the finite bound of the integration (if
the integral is not doubly-infinite) and <code>inf</code> is a code indicating
the kind of integration range,
</p>
<dl compact="compact">
<dt><code>inf = 1</code></dt>
<dd><p>corresponds to (bound, +Inf),
</p></dd>
<dt><code>inf = -1</code></dt>
<dd><p>corresponds to (-Inf, bound),
</p></dd>
<dt><code>inf = 2</code></dt>
<dd><p>corresponds to (-Inf, +Inf),
</p></dd>
</dl>

<p><code>f</code> and <code>ex</code> define the integrand function, see above;
<code>epsabs</code> and <code>epsrel</code> specify the absolute and relative
accuracy requested, <code>result</code>, <code>abserr</code> and <code>last</code> are the
output components <code>value</code>, <code>abs.err</code> and <code>subdivisions</code>
of the R function integrate, where <code>neval</code> gives the number of
integrand function evaluations, and the error code <code>ier</code> is
translated to R&rsquo;s <code>integrate() $ message</code>, look at that function
definition.  <code>limit</code> corresponds to <code>integrate(...,
subdivisions = *)</code>.  It seems you should always define the two work
arrays and the length of the second one as
</p>
<div class="example">
<pre class="example">    lenw = 4 * limit;
    iwork =   (int *) R_alloc(limit, sizeof(int));
    work = (double *) R_alloc(lenw,  sizeof(double));
</pre></div>

<p>The comments in the source code in <samp>src/appl/integrate.c</samp> give
more details, particularly about reasons for failure (<code>ier &gt;= 1</code>).
</p>

<hr>
<a name="Utility-functions"></a>
<div class="header">
<p>
Next: <a href="#Re_002dencoding" accesskey="n" rel="next">Re-encoding</a>, Previous: <a href="#Integration" accesskey="p" rel="prev">Integration</a>, Up: <a href="#The-R-API" accesskey="u" rel="up">The R API</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Utility-functions-1"></a>
<h3 class="section">6.10 Utility functions</h3>
<a name="index-Sort-functions-from-C"></a>

<p>R has a fairly comprehensive set of sort routines which are made
available to users&rsquo; C code.
The following is declared in header file <samp>Rinternals.h</samp>.
</p>
<dl>
<dt><a name="index-R_005forderVector"></a>Function: <em>void</em> <strong>R_orderVector</strong> <em>(int* <var>indx</var>, int <var>n</var>, SEXP <var>arglist</var>, Rboolean <var>nalast</var>, Rboolean <var>decreasing</var>)</em></dt>
<dt><a name="index-R_005forderVector1"></a>Function: <em>void</em> <strong>R_orderVector1</strong> <em>(int* <var>indx</var>, int <var>n</var>, SEXP <var>x</var>, Rboolean <var>nalast</var>, Rboolean <var>decreasing</var>)</em></dt>
<dd>
<p><code>R_orderVector()</code> corresponds to R&rsquo;s <code>order(..., na.last, decreasing)</code>.
More specifically, <code>indx &lt;- order(x, y, na.last, decreasing)</code> corresponds to
<code>R_orderVector(indx, n, Rf_lang2(x, y), nalast, decreasing)</code> and for
three vectors, <code>Rf_lang3(x,y,z)</code> is used as <var>arglist</var>.
</p>
<p>Both <code>R_orderVector</code> and <code>R_orderVector1</code> assume the vector
<code>indx</code> to be allocated to length &gt;= n.  On return,
<code>indx[]</code> contains a permutation of <code>0:(n-1)</code>, i.e., 0-based C
indices (and not 1-based R indices, as R&rsquo;s <code>order()</code>).
</p>
<p>When ordering only one vector, <code>R_orderVector1</code> is faster and
corresponds (but is 0-based) to R&rsquo;s <code>indx &lt;- order(x, na.last,
decreasing)</code>.  It was added in R 3.3.0.
</p></dd></dl>

<p>All other sort routines are declared in header file
<samp>R_ext/Utils.h</samp> (included by <samp>R.h</samp>) and include the following.
</p>
<dl>
<dt><a name="index-R_005fisort"></a>Function: <em>void</em> <strong>R_isort</strong> <em>(int* <var>x</var>, int <var>n</var>)</em></dt>
<dt><a name="index-R_005frsort"></a>Function: <em>void</em> <strong>R_rsort</strong> <em>(double* <var>x</var>, int <var>n</var>)</em></dt>
<dt><a name="index-R_005fcsort"></a>Function: <em>void</em> <strong>R_csort</strong> <em>(Rcomplex* <var>x</var>, int <var>n</var>)</em></dt>
<dt><a name="index-rsort_005fwith_005findex"></a>Function: <em>void</em> <strong>rsort_with_index</strong> <em>(double* <var>x</var>, int* <var>index</var>, int <var>n</var>)</em></dt>
<dd><p>The first three sort integer, real (double) and complex data
respectively.  (Complex numbers are sorted by the real part first then
the imaginary part.)  <code>NA</code>s are sorted last.
</p>
<p><code>rsort_with_index</code> sorts on <var>x</var>, and applies the same
permutation to <var>index</var>.  <code>NA</code>s are sorted last.
</p></dd></dl>

<dl>
<dt><a name="index-revsort"></a>Function: <em>void</em> <strong>revsort</strong> <em>(double* <var>x</var>, int* <var>index</var>, int <var>n</var>)</em></dt>
<dd><p>Is similar to <code>rsort_with_index</code> but sorts into decreasing order,
and <code>NA</code>s are not handled.
</p></dd></dl>

<dl>
<dt><a name="index-iPsort"></a>Function: <em>void</em> <strong>iPsort</strong> <em>(int* <var>x</var>, int <var>n</var>, int <var>k</var>)</em></dt>
<dt><a name="index-rPsort"></a>Function: <em>void</em> <strong>rPsort</strong> <em>(double* <var>x</var>, int <var>n</var>, int <var>k</var>)</em></dt>
<dt><a name="index-cPsort"></a>Function: <em>void</em> <strong>cPsort</strong> <em>(Rcomplex* <var>x</var>, int <var>n</var>, int <var>k</var>)</em></dt>
<dd><p>These all provide (very) partial sorting: they permute <var>x</var> so that
<code><var>x</var>[<var>k</var>]</code> is in the correct place with smaller values to
the left, larger ones to the right.
</p></dd></dl>


<dl>
<dt><a name="index-R_005fqsort"></a>Function: <em>void</em> <strong>R_qsort</strong> <em>(double *<var>v</var>, size_t <var>i</var>, size_t <var>j</var>)</em></dt>
<dt><a name="index-R_005fqsort_005fI"></a>Function: <em>void</em> <strong>R_qsort_I</strong> <em>(double *<var>v</var>, int *<var>I</var>, int <var>i</var>, int <var>j</var>)</em></dt>
<dt><a name="index-R_005fqsort_005fint"></a>Function: <em>void</em> <strong>R_qsort_int</strong> <em>(int *<var>iv</var>, size_t <var>i</var>, size_t <var>j</var>)</em></dt>
<dt><a name="index-R_005fqsort_005fint_005fI"></a>Function: <em>void</em> <strong>R_qsort_int_I</strong> <em>(int *<var>iv</var>, int *<var>I</var>, int <var>i</var>, int <var>j</var>)</em></dt>
<dd>

<p>These routines sort <code><var>v</var>[<var>i</var>:<var>j</var>]</code> or
<code><var>iv</var>[<var>i</var>:<var>j</var>]</code> (using 1-indexing, i.e.,
<code><var>v</var>[1]</code> is the first element) calling the quicksort algorithm
as used by R&rsquo;s <code>sort(v, method = &quot;quick&quot;)</code> and documented on the
help page for the R function <code>sort</code>.  The <code>..._I()</code>
versions also return the <code>sort.index()</code> vector in <code>I</code>.  Note
that the ordering is <em>not</em> stable, so tied values may be permuted.
</p>
<p>Note that <code>NA</code>s are not handled (explicitly) and you should
use different sorting functions if <code>NA</code>s can be present.
</p></dd></dl>

<dl>
<dt><a name="index-qsort4"></a>Function: <em>subroutine</em> <strong>qsort4</strong> <em>(double precision <var>v</var>, integer <var>indx</var>, integer <var>ii</var>, integer <var>jj</var>)</em></dt>
<dt><a name="index-qsort3"></a>Function: <em>subroutine</em> <strong>qsort3</strong> <em>(double precision <var>v</var>, integer <var>ii</var>, integer <var>jj</var>)</em></dt>
<dd>
<p>The FORTRAN interface routines for sorting double precision vectors are
<code>qsort3</code> and <code>qsort4</code>, equivalent to <code>R_qsort</code> and
<code>R_qsort_I</code>, respectively.
</p></dd></dl>

<dl>
<dt><a name="index-R_005fmax_005fcol"></a>Function: <em>void</em> <strong>R_max_col</strong> <em>(double* <var>matrix</var>, int* <var>nr</var>, int* <var>nc</var>, int* <var>maxes</var>, int* <var>ties_meth</var>)</em></dt>
<dd><p>Given the <var>nr</var> by <var>nc</var> matrix <code>matrix</code> in column-major
(&ldquo;FORTRAN&rdquo;)
order, <code>R_max_col()</code> returns in <code><var>maxes</var>[<var>i</var>-1]</code> the
column number of the maximal element in the <var>i</var>-th row (the same as
R&rsquo;s <code>max.col()</code> function).  In the case of ties (multiple maxima),
<code>*ties_meth</code> is an integer code in <code>1:3</code> determining the method:
1 = &ldquo;random&rdquo;, 2 = &ldquo;first&rdquo; and 3 = &ldquo;last&rdquo;.
See R&rsquo;s help page <code>?max.col</code>.
</p></dd></dl>

<dl>
<dt><a name="index-findInterval"></a>Function: <em>int</em> <strong>findInterval</strong> <em>(double* <var>xt</var>, int <var>n</var>, double <var>x</var>, Rboolean <var>rightmost_closed</var>, Rboolean <var>all_inside</var>, int <var>ilo</var>, int* <var>mflag</var>)</em></dt>
<dt><a name="index-findInterval2_0028double_002a"></a>Function: <em>int</em> <strong>findInterval2(double*</strong> <em><var>xt</var>, int <var>n</var>, double <var>x</var>, Rboolean <var>rightmost_closed</var>, Rboolean <var>all_inside</var>, Rboolean <var>left_open</var>, int <var>ilo</var>, int* <var>mflag</var>)</em></dt>
<dd><p>Given the ordered vector <var>xt</var> of length <var>n</var>, return the interval
or index of <var>x</var> in <code><var>xt</var>[]</code>, typically max(<em>i</em>; 1 &lt;= i &lt;= <var>n</var> &amp; <em><var>xt</var>[i]</em> &lt;=
<var>x</var>) where we use 1-indexing as in R and FORTRAN (but not C).  If
<var>rightmost_closed</var> is true, also returns <em><var>n</var>-1</em> if <var>x</var>
equals <em><var>xt</var>[<var>n</var>]</em>.  If <var>all_inside</var> is not 0, the
result is coerced to lie in <code>1:(<var>n</var>-1)</code> even when <var>x</var> is
outside the <var>xt</var>[] range.  On return, <code>*<var>mflag</var></code> equals
<em>-1</em> if <var>x</var> &lt; <var>xt</var>[1], <em>+1</em> if <var>x</var> &gt;=
<var>xt</var>[<var>n</var>], and 0 otherwise.
</p>
<p>The algorithm is particularly fast when <var>ilo</var> is set to the last
result of <code>findInterval()</code> and <var>x</var> is a value of a sequence which
is increasing or decreasing for subsequent calls.
</p>
<p><code>findInterval2()</code> is a generalization of <code>findInterval()</code>,
with an extra <code>Rboolean</code> argument <var>left_open</var>.  Setting
<code>left_open = TRUE</code> basically replaces all left-closed right-open
intervals t) by left-open ones t], see the help page
of R function <code>findInterval</code> for details.
</p>
<p>There is also an <code>F77_CALL(interv)()</code> version of
<code>findInterval()</code> with the same arguments, but all pointers.
</p></dd></dl>

<p>A system-independent interface to produce the name of a temporary
file is provided as
</p>
<dl>
<dt><a name="index-R_005ftmpnam"></a>Function: <em>char *</em> <strong>R_tmpnam</strong> <em>(const char *<var>prefix</var>, const char *<var>tmpdir</var>)</em></dt>
<dt><a name="index-R_005ftmpnam2"></a>Function: <em>char *</em> <strong>R_tmpnam2</strong> <em>(const char *<var>prefix</var>, const char *<var>tmpdir</var>, const char *<var>fileext</var>)</em></dt>
<dd><p>Return a pathname for a temporary file with name beginning with
<var>prefix</var> and ending with <var>fileext</var> in directory <var>tmpdir</var>.
A <code>NULL</code> prefix or extension is replaced by <code>&quot;&quot;</code>.  Note that
the return value is <code>malloc</code>ed and should be <code>free</code>d when no
longer needed (unlike the system call <code>tmpnam</code>).
</p></dd></dl>


<p>There is also the internal function used to expand file names in several
R functions, and called directly by <code>path.expand</code>.
</p>
<dl>
<dt><a name="index-R_005fExpandFileName"></a>Function: <em>const char *</em> <strong>R_ExpandFileName</strong> <em>(const char *<var>fn</var>)</em></dt>
<dd><p>Expand a path name <var>fn</var> by replacing a leading tilde by the user&rsquo;s
home directory (if defined).  The precise meaning is platform-specific;
it will usually be taken from the environment variable <code>HOME</code> if
this is defined.
</p></dd></dl>

<p>For historical reasons there are FORTRAN interfaces to functions
<code>D1MACH</code> and <code>I1MACH</code>.  These can be called from C code as
e.g. <code>F77_CALL(d1mach)(4)</code>.  Note that these are emulations of
the original functions by Fox, Hall and Schryer on NetLib at
<a href="http://www.netlib.org/slatec/src/">http://www.netlib.org/slatec/src/</a> for IEC 60559 arithmetic
(required by R).
</p>
<hr>
<a name="Re_002dencoding"></a>
<div class="header">
<p>
Next: <a href="#Allowing-interrupts" accesskey="n" rel="next">Allowing interrupts</a>, Previous: <a href="#Utility-functions" accesskey="p" rel="prev">Utility functions</a>, Up: <a href="#The-R-API" accesskey="u" rel="up">The R API</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Re_002dencoding-1"></a>
<h3 class="section">6.11 Re-encoding</h3>

<p>R has its own C-level interface to the encoding conversion
capabilities provided by <code>iconv</code> because there are
incompatibilities between the declarations in different implementations
of <code>iconv</code>.
</p>
<p>These are declared in header file <samp>R_ext/Riconv.h</samp>.
</p>
<dl>
<dt><a name="index-Riconv_005fopen"></a>Function: <em>void *</em> <strong>Riconv_open</strong> <em>(const char *<var>to</var>, const char *<var>from</var>)</em></dt>
</dl>
<p>Set up a pointer to an encoding object to be used to convert between two
encodings: <code>&quot;&quot;</code> indicates the current locale.
</p>
<dl>
<dt><a name="index-Riconv"></a>Function: <em>size_t</em> <strong>Riconv</strong> <em>(void *<var>cd</var>, const char **<var>inbuf</var>, size_t *<var>inbytesleft</var>, char  **<var>outbuf</var>, size_t *<var>outbytesleft</var>)</em></dt>
</dl>
<p>Convert as much as possible of <code>inbuf</code> to <code>outbuf</code>.  Initially
the <code>int</code> variables indicate the number of bytes available in the
buffers, and they are updated (and the <code>char</code> pointers are updated
to point to the next free byte in the buffer).  The return value is the
number of characters converted, or <code>(size_t)-1</code> (beware:
<code>size_t</code> is usually an unsigned type).  It should be safe to assume
that an error condition sets <code>errno</code> to one of <code>E2BIG</code> (the
output buffer is full), <code>EILSEQ</code> (the input cannot be converted,
and might be invalid in the encoding specified) or <code>EINVAL</code> (the
input does not end with a complete multi-byte character).
</p>
<dl>
<dt><a name="index-Riconv_005fclose"></a>Function: <em>int</em> <strong>Riconv_close</strong> <em>(void * <var>cd</var>)</em></dt>
</dl>
<p>Free the resources of an encoding object.
</p>

<hr>
<a name="Allowing-interrupts"></a>
<div class="header">
<p>
Next: <a href="#Platform-and-version-information" accesskey="n" rel="next">Platform and version information</a>, Previous: <a href="#Re_002dencoding" accesskey="p" rel="prev">Re-encoding</a>, Up: <a href="#The-R-API" accesskey="u" rel="up">The R API</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Allowing-interrupts-1"></a>
<h3 class="section">6.12 Allowing interrupts</h3>
<a name="index-Interrupts"></a>

<p>No port of R can be interrupted whilst running long computations in
compiled code, so programmers should make provision for the code to be
interrupted at suitable points by calling from C
</p>
<div class="example">
<pre class="example">#include &lt;R_ext/Utils.h&gt;

void R_CheckUserInterrupt(void);
</pre></div>

<p>and from FORTRAN
</p>
<div class="example">
<pre class="example">subroutine rchkusr()
</pre></div>

<p>These check if the user has requested an interrupt, and if so branch to
R&rsquo;s error handling functions.
</p>
<p>Note that it is possible that the code behind one of the entry points
defined here if called from your C or FORTRAN code could be interruptible
or generate an error and so not return to your code.
</p>

<hr>
<a name="Platform-and-version-information"></a>
<div class="header">
<p>
Next: <a href="#Inlining-C-functions" accesskey="n" rel="next">Inlining C functions</a>, Previous: <a href="#Allowing-interrupts" accesskey="p" rel="prev">Allowing interrupts</a>, Up: <a href="#The-R-API" accesskey="u" rel="up">The R API</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Platform-and-version-information-1"></a>
<h3 class="section">6.13 Platform and version information</h3>
<a name="index-Version-information-from-C"></a>
<a name="index-OpenMP-1"></a>
<a name="index-R_005fVersion"></a>

<p>The header files define <code>USING_R</code>, which can be used to test if
the code is indeed being used with R.
</p>
<p>Header file <samp>Rconfig.h</samp> (included by <samp>R.h</samp>) is used to define
platform-specific macros that are mainly for use in other header files.
The macro <code>WORDS_BIGENDIAN</code> is defined on
big-endian<a name="DOCF142" href="#FOOT142"><sup>142</sup></a>
systems (e.g. most OSes on Sparc and PowerPC hardware) and not on
little-endian systems (nowadays all the commoner R platforms).  It
can be useful when manipulating binary files.  NB: these macros apply
only to the C compiler used to build R, not necessarily to another C
or C++ compiler.
</p>
<p>Header file <samp>Rversion.h</samp> (<strong>not</strong> included by <samp>R.h</samp>)
defines a macro <code>R_VERSION</code> giving the version number encoded as an
integer, plus a macro <code>R_Version</code> to do the encoding.  This can be
used to test if the version of R is late enough, or to include
back-compatibility features.  For protection against very old versions
of R which did not have this macro, use a construction such as
</p>
<div class="example">
<pre class="example">#if defined(R_VERSION) &amp;&amp; R_VERSION &gt;= R_Version(3, 1, 0)
  ...
#endif
</pre></div>

<p>More detailed information is available in the macros <code>R_MAJOR</code>,
<code>R_MINOR</code>, <code>R_YEAR</code>, <code>R_MONTH</code> and <code>R_DAY</code>: see the
header file <samp>Rversion.h</samp> for their format.  Note that the minor
version includes the patchlevel (as in &lsquo;<samp>2.2</samp>&rsquo;).
</p>
<p>Packages which use <code>alloca</code> need to ensure it is defined: as it is
part of neither C nor POSIX there is no standard way to do so.  As
from R 3.2.2 one can use
</p>
<div class="example">
<pre class="example">#include &lt;Rconfig.h&gt; // for HAVE_ALLOCA_H
#ifdef __GNUC__
// this covers gcc, clang, icc
# undef alloca
# define alloca(x) __builtin_alloca((x))
#elif defined(HAVE_ALLOCA_H)
// needed for native compilers on Solaris and AIX
# include &lt;alloca.h&gt;
#endif
</pre></div>

<p>(and this should be included before standard C headers such as
<samp>stdlib.h</samp>, since on some platforms these include <samp>malloc.h</samp>
which may have a conflicting definition), which suffices for known R
platforms.
</p>
<hr>
<a name="Inlining-C-functions"></a>
<div class="header">
<p>
Next: <a href="#Controlling-visibility" accesskey="n" rel="next">Controlling visibility</a>, Previous: <a href="#Platform-and-version-information" accesskey="p" rel="prev">Platform and version information</a>, Up: <a href="#The-R-API" accesskey="u" rel="up">The R API</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Inlining-C-functions-1"></a>
<h3 class="section">6.14 Inlining C functions</h3>
<a name="index-R_005fINLINE"></a>

<p>The C99 keyword <code>inline</code> should be recognized by all compilers
nowadays used to build R.  Portable code which might be used with
earlier versions of R can be written using the macro <code>R_INLINE</code>
(defined in file <samp>Rconfig.h</samp> included by <samp>R.h</samp>), as for
example from package <a href="https://CRAN.R-project.org/package=cluster"><strong>cluster</strong></a>
</p>
<div class="example">
<pre class="example">#include &lt;R.h&gt;

static R_INLINE int ind_2(int l, int j)
{
...
}
</pre></div>

<p>Be aware that using inlining with functions in more than one compilation
unit is almost impossible to do portably, see
<a href="http://www.greenend.org.uk/rjk/2003/03/inline.html">http://www.greenend.org.uk/rjk/2003/03/inline.html</a>, so this usage
is for <code>static</code> functions as in the example.  All the R
configure code has checked is that <code>R_INLINE</code> can be used in a
single C file with the compiler used to build R.  We recommend that
packages making extensive use of inlining include their own configure
code.
</p>
<hr>
<a name="Controlling-visibility"></a>
<div class="header">
<p>
Next: <a href="#Standalone-Mathlib" accesskey="n" rel="next">Standalone Mathlib</a>, Previous: <a href="#Inlining-C-functions" accesskey="p" rel="prev">Inlining C functions</a>, Up: <a href="#The-R-API" accesskey="u" rel="up">The R API</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Controlling-visibility-1"></a>
<h3 class="section">6.15 Controlling visibility</h3>
<a name="index-Visibility"></a>

<p>Header <samp>R_ext/Visibility.h</samp> has some definitions for controlling the
visibility of entry points.  These are only effective when
&lsquo;<samp>HAVE_VISIBILITY_ATTRIBUTE</samp>&rsquo; is defined &ndash; this is checked when R
is configured and recorded in header <samp>Rconfig.h</samp> (included by
<samp>R_ext/Visibility.h</samp>).  It is often defined on modern Unix-alikes
with a recent compiler<a name="DOCF143" href="#FOOT143"><sup>143</sup></a>, but not
supported on macOS nor Windows.  Minimizing the visibility of symbols in
a shared library will both speed up its loading (unlikely to be
significant) and reduce the possibility of linking to other entry points
of the same name.
</p>
<p>C/C++ entry points prefixed by <code>attribute_hidden</code> will not be
visible in the shared object.  There is no comparable mechanism for
FORTRAN entry points, but there is a more comprehensive scheme used by,
for example package <strong>stats</strong>.  Most compilers which allow control of
visibility will allow control of visibility for all symbols <em>via</em> a flag,
and where known the flag is encapsulated in the macros
&lsquo;<samp>C_VISIBILITY</samp>&rsquo; and <code>F77_VISIBILITY</code> for C and FORTRAN
compilers.  These are defined in <samp>etc/Makeconf</samp> and so available
for normal compilation of package code.  For example,
<samp>src/Makevars</samp> could include
</p>
<div class="example">
<pre class="example">PKG_CFLAGS=$(C_VISIBILITY)
PKG_FFLAGS=$(F77_VISIBILITY)
</pre></div>

<p>This would end up with <strong>no</strong> visible entry points, which would be
pointless.  However, the effect of the flags can be overridden by using
the <code>attribute_visible</code> prefix.  A shared object which registers
its entry points needs only for have one visible entry point, its
initializer, so for example package <strong>stats</strong> has
</p>
<div class="example">
<pre class="example">void attribute_visible R_init_stats(DllInfo *dll)
{
    R_registerRoutines(dll, CEntries, CallEntries, FortEntries, NULL);
    R_useDynamicSymbols(dll, FALSE);
...
}
</pre></div>

<p>The visibility mechanism is not available on Windows, but there is an
equally effective way to control which entry points are visible, by
supplying a definitions file
<samp><var>pkgnme</var>/src/<var>pkgname</var>-win.def</samp>: only entry points
listed in that file will be visible.  Again using <strong>stats</strong> as an
example, it has
</p>
<div class="example">
<pre class="example">LIBRARY stats.dll
EXPORTS
 R_init_stats
</pre></div>

<hr>
<a name="Standalone-Mathlib"></a>
<div class="header">
<p>
Next: <a href="#Organization-of-header-files" accesskey="n" rel="next">Organization of header files</a>, Previous: <a href="#Controlling-visibility" accesskey="p" rel="prev">Controlling visibility</a>, Up: <a href="#The-R-API" accesskey="u" rel="up">The R API</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Using-these-functions-in-your-own-C-code"></a>
<h3 class="section">6.16 Using these functions in your own C code</h3>

<p>It is possible to build <code>Mathlib</code>, the R set of mathematical
functions documented in <samp>Rmath.h</samp>, as a standalone library
<samp>libRmath</samp> under both Unix-alikes and Windows.  (This includes the
functions documented in <a href="#Numerical-analysis-subroutines">Numerical analysis subroutines</a> as from
that header file.)
</p>
<p>The library is not built automatically when R is installed, but can
be built in the directory <samp>src/nmath/standalone</samp> in the R
sources: see the file <samp>README</samp> there.  To use the code in your own
C program include
</p>
<div class="example">
<pre class="example">#define MATHLIB_STANDALONE
#include &lt;Rmath.h&gt;
</pre></div>

<p>and link against &lsquo;<samp>-lRmath</samp>&rsquo; (and perhaps &lsquo;<samp>-lm</samp>&rsquo;).  There is an
example file <samp>test.c</samp>.
</p>
<p>A little care is needed to use the random-number routines. You will
need to supply the uniform random number generator
</p>
<div class="example">
<pre class="example">double unif_rand(void)
</pre></div>

<p>or use the one supplied (and with a dynamic library or DLL you will have
to use the one supplied, which is the Marsaglia-multicarry with an entry
points
</p>
<div class="example">
<pre class="example">set_seed(unsigned int, unsigned int)
</pre></div>

<p>to set its seeds and
</p>
<div class="example">
<pre class="example">get_seed(unsigned int *, unsigned int *)
</pre></div>

<p>to read the seeds).
</p>
<hr>
<a name="Organization-of-header-files"></a>
<div class="header">
<p>
Previous: <a href="#Standalone-Mathlib" accesskey="p" rel="prev">Standalone Mathlib</a>, Up: <a href="#The-R-API" accesskey="u" rel="up">The R API</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Organization-of-header-files-1"></a>
<h3 class="section">6.17 Organization of header files</h3>

<p>The header files which R installs are in directory
<samp><var>R_INCLUDE_DIR</var></samp> (default <samp><var>R_HOME</var>/include</samp>).  This
currently includes
</p>
<blockquote>
<table summary="">
<tr><td width="30%"><samp>R.h</samp></td><td width="55%">includes many other files</td></tr>
<tr><td width="30%"><samp>S.h</samp></td><td width="55%">different version for code ported from S</td></tr>
<tr><td width="30%"><samp>Rinternals.h</samp></td><td width="55%">definitions for using R&rsquo;s internal
structures</td></tr>
<tr><td width="30%"><samp>Rdefines.h</samp></td><td width="55%">macros for an S-like interface to the
above (no longer maintained)</td></tr>
<tr><td width="30%"><samp>Rmath.h</samp></td><td width="55%">standalone math library</td></tr>
<tr><td width="30%"><samp>Rversion.h</samp></td><td width="55%">R version information</td></tr>
<tr><td width="30%"><samp>Rinterface.h</samp></td><td width="55%">for add-on front-ends (Unix-alikes only)</td></tr>
<tr><td width="30%"><samp>Rembedded.h</samp></td><td width="55%">for add-on front-ends</td></tr>
<tr><td width="30%"><samp>R_ext/Applic.h</samp></td><td width="55%">optimization and integration</td></tr>
<tr><td width="30%"><samp>R_ext/BLAS.h</samp></td><td width="55%">C definitions for BLAS routines</td></tr>
<tr><td width="30%"><samp>R_ext/Callbacks.h</samp></td><td width="55%">C (and R function) top-level task
handlers</td></tr>
<tr><td width="30%"><samp>R_ext/GetX11Image.h</samp></td><td width="55%">X11Image interface used by package
<strong>trkplot</strong></td></tr>
<tr><td width="30%"><samp>R_ext/Lapack.h</samp></td><td width="55%">C definitions for some LAPACK routines</td></tr>
<tr><td width="30%"><samp>R_ext/Linpack.h</samp></td><td width="55%">C definitions for some LINPACK
routines, not all of which are included in R</td></tr>
<tr><td width="30%"><samp>R_ext/Parse.h</samp></td><td width="55%">a small part of R&rsquo;s parse interface:
not part of the stable API.</td></tr>
<tr><td width="30%"><samp>R_ext/RStartup.h</samp></td><td width="55%">for add-on front-ends</td></tr>
<tr><td width="30%"><samp>R_ext/Rdynload.h</samp></td><td width="55%">needed to register compiled code in
packages</td></tr>
<tr><td width="30%"><samp>R_ext/R-ftp-http.h</samp></td><td width="55%">interface to internal method of
<code>download.file</code></td></tr>
<tr><td width="30%"><samp>R_ext/Riconv.h</samp></td><td width="55%">interface to <code>iconv</code></td></tr>
<tr><td width="30%"><samp>R_ext/Visibility.h</samp></td><td width="55%">definitions controlling visibility</td></tr>
<tr><td width="30%"><samp>R_ext/eventloop.h</samp></td><td width="55%">for add-on front-ends and for
packages that need to share in the R event loops (not Windows)</td></tr>
</table>
</blockquote>

<p>The following headers are included by <samp>R.h</samp>:
</p>
<blockquote>
<table summary="">
<tr><td width="30%"><samp>Rconfig.h</samp></td><td width="55%">configuration info that is made available</td></tr>
<tr><td width="30%"><samp>R_ext/Arith.h</samp></td><td width="55%">handling for <code>NA</code>s, <code>NaN</code>s,
<code>Inf</code>/<code>-Inf</code></td></tr>
<tr><td width="30%"><samp>R_ext/Boolean.h</samp></td><td width="55%"><code>TRUE</code>/<code>FALSE</code> type</td></tr>
<tr><td width="30%"><samp>R_ext/Complex.h</samp></td><td width="55%">C typedefs for R&rsquo;s <code>complex</code></td></tr>
<tr><td width="30%"><samp>R_ext/Constants.h</samp></td><td width="55%">constants</td></tr>
<tr><td width="30%"><samp>R_ext/Error.h</samp></td><td width="55%">error handling</td></tr>
<tr><td width="30%"><samp>R_ext/Memory.h</samp></td><td width="55%">memory allocation</td></tr>
<tr><td width="30%"><samp>R_ext/Print.h</samp></td><td width="55%"><code>Rprintf</code> and variations.</td></tr>
<tr><td width="30%"><samp>R_ext/RS.h</samp></td><td width="55%">definitions common to <samp>R.h</samp> and
<samp>S.h</samp>, including <code>F77_CALL</code> etc.</td></tr>
<tr><td width="30%"><samp>R_ext/Random.h</samp></td><td width="55%">random number generation</td></tr>
<tr><td width="30%"><samp>R_ext/Utils.h</samp></td><td width="55%">sorting and other utilities</td></tr>
<tr><td width="30%"><samp>R_ext/libextern.h</samp></td><td width="55%">definitions for exports from
<samp>R.dll</samp> on Windows.</td></tr>
</table>
</blockquote>

<p>The graphics systems are exposed in headers
<samp>R_ext/GraphicsEngine.h</samp>, <samp>R_ext/GraphicsDevice.h</samp> (which it
includes) and <samp>R_ext/QuartzDevice.h</samp>. Facilities for defining
custom connection implementations are provided in
<samp>R_ext/Connections.h</samp>, but make sure you consult the file before
use.
</p>
<p>Let us re-iterate the advice to include system headers before the R
header files, especially <samp>Rinternals.h</samp> (included by
<samp>Rdefines.h</samp>) and <samp>Rmath.h</samp>, which redefine names which may be
used in system headers (fewer if &lsquo;<samp>R_NO_REMAP</samp>&rsquo; is defined, or
&lsquo;<samp>R_NO_REMAP_RMATH</samp>&rsquo; for <samp>Rmath.h</samp>, as from R 3.1.0).
</p>
<hr>
<a name="Generic-functions-and-methods"></a>
<div class="header">
<p>
Next: <a href="#Linking-GUIs-and-other-front_002dends-to-R" accesskey="n" rel="next">Linking GUIs and other front-ends to R</a>, Previous: <a href="#The-R-API" accesskey="p" rel="prev">The R API</a>, Up: <a href="#Top" accesskey="u" rel="up">Top</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Generic-functions-and-methods-1"></a>
<h2 class="chapter">7 Generic functions and methods</h2>
<a name="index-Generic-functions"></a>
<a name="index-Method-functions"></a>

<p>R programmers will often want to add methods for existing generic
functions, and may want to add new generic functions or make existing
functions generic.  In this chapter we give guidelines for doing so,
with examples of the problems caused by not adhering to them.
</p>
<p>This chapter only covers the &lsquo;informal&rsquo; class system copied from S3,
and not with the S4 (formal) methods of package <strong>methods</strong>.
</p>
<p>First, a <em>caveat</em>: a function named <code><var>gen</var>.<var>cl</var></code> will
be invoked by the generic <code><var>gen</var></code> for class <code><var>cl</var></code>, so
do not name functions in this style unless they are intended to be
methods.
</p>
<p>The key function for methods is <code>NextMethod</code>, which dispatches the
next method.  It is quite typical for a method function to make a few
changes to its arguments, dispatch to the next method, receive the
results and modify them a little.  An example is
</p>
<div class="example">
<pre class="example">t.data.frame &lt;- function(x)
{
    x &lt;- as.matrix(x)
    NextMethod(&quot;t&quot;)
}
</pre></div>

<p>Note that the example above works because there is a <em>next</em> method,
the default method, not that a new method is selected when the class is
changed.
</p>
<p><em>Any</em> method a programmer writes may be invoked from another method
by <code>NextMethod</code>, <em>with the arguments appropriate to the
previous method</em>.  Further, the programmer cannot predict which method
<code>NextMethod</code> will pick (it might be one not yet dreamt of), and the
end user calling the generic needs to be able to pass arguments to the
next method.  For this to work
</p>
<blockquote>
<p><em>A method must have all the arguments of the generic, including
<code>&hellip;</code> if the generic does.</em>
</p></blockquote>

<p>It is a grave misunderstanding to think that a method needs only to
accept the arguments it needs.  The original S version of
<code>predict.lm</code> did not have a <code>&hellip;</code> argument, although
<code>predict</code> did.  It soon became clear that <code>predict.glm</code> needed
an argument <code>dispersion</code> to handle over-dispersion.  As
<code>predict.lm</code> had neither a <code>dispersion</code> nor a <code>&hellip;</code>
argument, <code>NextMethod</code> could no longer be used.  (The legacy, two
direct calls to <code>predict.lm</code>, lives on in <code>predict.glm</code> in
R, which is based on the workaround for S3 written by Venables &amp;
Ripley.)
</p>
<p>Further, the user is entitled to use positional matching when calling
the generic, and the arguments to a method called by <code>UseMethod</code>
are those of the call to the generic.  Thus
</p>
<blockquote>
<p><em>A method must have arguments in exactly the same order as the
generic.</em>
</p></blockquote>

<p>To see the scale of this problem, consider the generic function
<code>scale</code>, defined as
</p>
<div class="example">
<pre class="example">scale &lt;- function (x, center = TRUE, scale = TRUE)
    UseMethod(&quot;scale&quot;)
</pre></div>

<p>Suppose an unthinking package writer created methods such as
</p>
<div class="example">
<pre class="example">scale.foo &lt;- function(x, scale = FALSE, ...) { }
</pre></div>

<p>Then for <code>x</code> of class <code>&quot;foo&quot;</code> the calls
</p>
<div class="example">
<pre class="example">scale(x, , TRUE)
scale(x, scale = TRUE)
</pre></div>

<p>would do most likely do different things, to the justifiable
consternation of the end user.
</p>
<p>To add a further twist, which default is used when a user calls
<code>scale(x)</code> in our example?  What if
</p>
<div class="example">
<pre class="example">scale.bar &lt;- function(x, center, scale = TRUE) NextMethod(&quot;scale&quot;)
</pre></div>

<p>and <code>x</code> has class <code>c(&quot;bar&quot;, &quot;foo&quot;)</code>?  It is the default
specified in the method that is used, but the default
specified in the generic may be the one the user sees.
This leads to the recommendation:
</p>
<blockquote>
<p><em>If the generic specifies defaults, all methods should use the same defaults.</em>
</p></blockquote>

<p>An easy way to follow these recommendations is to always keep generics
simple, e.g.
</p>
<div class="example">
<pre class="example">scale &lt;- function(x, ...) UseMethod(&quot;scale&quot;)
</pre></div>

<p>Only add parameters and defaults to the generic if they make sense in
all possible methods implementing it.
</p>
<table summary="" class="menu" border="0" cellspacing="0">
<tr><td align="left" valign="top">&bull; <a href="#Adding-new-generics" accesskey="1">Adding new generics</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
</table>

<hr>
<a name="Adding-new-generics"></a>
<div class="header">
<p>
Previous: <a href="#Generic-functions-and-methods" accesskey="p" rel="prev">Generic functions and methods</a>, Up: <a href="#Generic-functions-and-methods" accesskey="u" rel="up">Generic functions and methods</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Adding-new-generics-1"></a>
<h3 class="section">7.1 Adding new generics</h3>

<p>When creating a new generic function, bear in mind that its argument
list will be the maximal set of arguments for methods, including those
written elsewhere years later.  So choosing a good set of arguments may
well be an important design issue, and there need to be good arguments
<em>not</em> to include a <code>&hellip;</code> argument.
</p>
<p>If a <code>&hellip;</code> argument is supplied, some thought should be given
to its position in the argument sequence.  Arguments which follow
<code>&hellip;</code> must be named in calls to the function, and they must be
named in full (partial matching is suppressed after <code>&hellip;</code>).
Formal arguments before <code>&hellip;</code> can be partially matched, and so
may &lsquo;swallow&rsquo; actual arguments intended for <code>&hellip;</code>.  Although it
is commonplace to make the <code>&hellip;</code> argument the last one, that is
not always the right choice.
</p>
<p>Sometimes package writers want to make generic a function in the base
package, and request a change in R.  This may be justifiable, but
making a function generic with the old definition as the default method
does have a small performance cost.  It is never necessary, as a package
can take over a function in the base package and make it generic by
something like
</p>
<div class="example">
<pre class="example">foo &lt;- function(object, ...) UseMethod(&quot;foo&quot;)
foo.default &lt;- function(object, ...) base::foo(object)
</pre></div>

<p>Earlier versions of this manual suggested assigning <code>foo.default &lt;-
base::foo</code>.  This is <strong>not</strong> a good idea, as it captures the base
function at the time of installation and it might be changed as R is
patched or updated.
</p>
<p>The same idea can be applied for functions in other packages with namespaces.
</p>
<hr>
<a name="Linking-GUIs-and-other-front_002dends-to-R"></a>
<div class="header">
<p>
Next: <a href="#Function-and-variable-index" accesskey="n" rel="next">Function and variable index</a>, Previous: <a href="#Generic-functions-and-methods" accesskey="p" rel="prev">Generic functions and methods</a>, Up: <a href="#Top" accesskey="u" rel="up">Top</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Linking-GUIs-and-other-front_002dends-to-R-1"></a>
<h2 class="chapter">8 Linking GUIs and other front-ends to R</h2>

<p>There are a number of ways to build front-ends to R: we take this to
mean a GUI or other application that has the ability to submit commands
to R and perhaps to receive results back (not necessarily in a text
format).  There are other routes besides those described here, for
example the package <a href="https://CRAN.R-project.org/package=Rserve"><strong>Rserve</strong></a> (from <acronym>CRAN</acronym>, see also
<a href="https://www.rforge.net/Rserve/">https://www.rforge.net/Rserve/</a>) and connections to Java in
&lsquo;<samp>JRI</samp>&rsquo; (part of the <a href="https://CRAN.R-project.org/package=rJava"><strong>rJava</strong></a> package on <acronym>CRAN</acronym>) and
the Omegahat/Bioconductor package &lsquo;<samp>SJava</samp>&rsquo;.
</p>
<p>Note that the APIs described in this chapter are only intended to be
used in an alternative front-end: they are not part of the API made
available for R packages and can be dangerous to use in a
conventional package (although packages may contain alternative
front-ends).  Conversely some of the functions from the API (such as
<code>R_alloc</code>) should not be used in front-ends.
</p>
<table summary="" class="menu" border="0" cellspacing="0">
<tr><td align="left" valign="top">&bull; <a href="#Embedding-R-under-Unix_002dalikes" accesskey="1">Embedding R under Unix-alikes</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Embedding-R-under-Windows" accesskey="2">Embedding R under Windows</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
</table>

<hr>
<a name="Embedding-R-under-Unix_002dalikes"></a>
<div class="header">
<p>
Next: <a href="#Embedding-R-under-Windows" accesskey="n" rel="next">Embedding R under Windows</a>, Previous: <a href="#Linking-GUIs-and-other-front_002dends-to-R" accesskey="p" rel="prev">Linking GUIs and other front-ends to R</a>, Up: <a href="#Linking-GUIs-and-other-front_002dends-to-R" accesskey="u" rel="up">Linking GUIs and other front-ends to R</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Embedding-R-under-Unix_002dalikes-1"></a>
<h3 class="section">8.1 Embedding R under Unix-alikes</h3>

<p>R can be built as a shared library<a name="DOCF144" href="#FOOT144"><sup>144</sup></a> if configured with <samp>--enable-R-shlib</samp>.  This
shared library can be used to run R from alternative front-end
programs.  We will assume this has been done for the rest of this
section.  Also, it can be built as a static library if configured with
<samp>--enable-R-static-lib</samp>, and that can be used in a very similar
way (at least on Linux: on other platforms one needs to ensure that all
the symbols exported by <samp>libR.a</samp> are linked into the front-end).
</p>
<p>The command-line R front-end, <samp><var>R_HOME</var>/bin/exec/R</samp>, is one
such example, and the former <acronym>GNOME</acronym> (see package <strong>gnomeGUI</strong>
on <acronym>CRAN</acronym>&rsquo;s &lsquo;<samp>Archive</samp>&rsquo; area) and macOS consoles are others.
The source for <samp><var>R_HOME</var>/bin/exec/R</samp> is in file
<samp>src/main/Rmain.c</samp> and is very simple
</p>
<div class="example">
<pre class="example">int Rf_initialize_R(int ac, char **av); /* in ../unix/system.c */
void Rf_mainloop();                     /* in main.c */

extern int R_running_as_main_program;   /* in ../unix/system.c */

int main(int ac, char **av)
{
    R_running_as_main_program = 1;
    Rf_initialize_R(ac, av);
    Rf_mainloop(); /* does not return */
    return 0;
}
</pre></div>

<p>indeed, misleadingly simple.  Remember that
<samp><var>R_HOME</var>/bin/exec/R</samp> is run from a shell script
<samp><var>R_HOME</var>/bin/R</samp> which sets up the environment for the
executable, and this is used for
</p>
<ul>
<li> Setting <code>R_HOME</code> and checking it is valid, as well as the path
<code>R_SHARE_DIR</code> and <code>R_DOC_DIR</code> to the installed <samp>share</samp> and
<samp>doc</samp> directory trees.  Also setting <code>R_ARCH</code> if needed.

</li><li> Setting <code>LD_LIBRARY_PATH</code> to include the directories used in linking
R.  This is recorded as the default setting of
<code>R_LD_LIBRARY_PATH</code> in the shell script
<samp><var>R_HOME</var>/etc<var>R_ARCH</var>/ldpaths</samp>.

</li><li> Processing some of the arguments, for example to run R under a
debugger and to launch alternative front-ends to provide GUIs.
</li></ul>

<p>The first two of these can be achieved for your front-end by running it
<em>via</em> <code>R CMD</code>. So, for example
</p>
<div class="example">
<pre class="example">R CMD /usr/local/lib/R/bin/exec/R
R CMD exec/R
</pre></div>

<p>will both work in a standard R installation. (<code>R CMD</code> looks
first for executables in <samp><var>R_HOME</var>/bin</samp>.  These command-lines
need modification if a sub-architecture is in use.) If you do not want
to run your front-end in this way, you need to ensure that <code>R_HOME</code>
is set and <code>LD_LIBRARY_PATH</code> is suitable.  (The latter might well
be, but modern Unix/Linux systems do not normally include
<samp>/usr/local/lib</samp> (<samp>/usr/local/lib64</samp> on some architectures),
and R does look there for system components.)
</p>
<p>The other senses in which this example is too simple are that all the
internal defaults are used and that control is handed over to the
R main loop.  There are a number of small examples<a name="DOCF145" href="#FOOT145"><sup>145</sup></a> in the
<samp>tests/Embedding</samp> directory.  These make use of
<code>Rf_initEmbeddedR</code> in <samp>src/main/Rembedded.c</samp>, and essentially
use
</p><div class="example">
<pre class="example">#include &lt;Rembedded.h&gt;

int main(int ac, char **av)
{
    /* do some setup */
    Rf_initEmbeddedR(argc, argv);
    /* do some more setup */

    /* submit some code to R, which is done interactively via
	run_Rmainloop();

	A possible substitute for a pseudo-console is

	R_ReplDLLinit();
	while(R_ReplDLLdo1() &gt; 0) {
	/* add user actions here if desired */
       }

     */
    Rf_endEmbeddedR(0);
    /* final tidying up after R is shutdown */
    return 0;
}
</pre></div>

<p>If you do not want to pass R arguments, you can fake an <code>argv</code>
array, for example by
</p>
<div class="example">
<pre class="example">    char *argv[]= {&quot;REmbeddedPostgres&quot;, &quot;--silent&quot;};
    Rf_initEmbeddedR(sizeof(argv)/sizeof(argv[0]), argv);
</pre></div>

<p>However, to make a GUI we usually do want to run <code>run_Rmainloop</code>
after setting up various parts of R to talk to our GUI, and arranging
for our GUI callbacks to be called during the R mainloop.
</p>
<p>One issue to watch is that on some platforms <code>Rf_initEmbeddedR</code> and
<code>Rf_endEmbeddedR</code> change the settings of the FPU (e.g. to allow
errors to be trapped and to make use of extended precision registers).
</p>
<p>The standard code sets up a session temporary directory in the usual
way, <em>unless</em> <code>R_TempDir</code> is set to a non-NULL value before
<code>Rf_initEmbeddedR</code> is called.  In that case the value is assumed to
contain an existing writable directory (no check is done), and it is not
cleaned up when R is shut down.
</p>
<p><code>Rf_initEmbeddedR</code> sets R to be in interactive mode: you can set
<code>R_Interactive</code> (defined in <samp>Rinterface.h</samp>) subsequently to
change this.
</p>
<p>Note that R expects to be run with the locale category
&lsquo;<samp>LC_NUMERIC</samp>&rsquo; set to its default value of <code>C</code>, and so should
not be embedded into an application which changes that.
</p>
<p>It is the user&rsquo;s responsibility to attempt to initialize only once.  To
protect the R interpreter, <code>Rf_initialize_R</code> will exit the
process if re-initialization is attempted.
</p>
<table summary="" class="menu" border="0" cellspacing="0">
<tr><td align="left" valign="top">&bull; <a href="#Compiling-against-the-R-library" accesskey="1">Compiling against the R library</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Setting-R-callbacks" accesskey="2">Setting R callbacks</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Registering-symbols" accesskey="3">Registering symbols</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Meshing-event-loops" accesskey="4">Meshing event loops</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Threading-issues" accesskey="5">Threading issues</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
</table>

<hr>
<a name="Compiling-against-the-R-library"></a>
<div class="header">
<p>
Next: <a href="#Setting-R-callbacks" accesskey="n" rel="next">Setting R callbacks</a>, Previous: <a href="#Embedding-R-under-Unix_002dalikes" accesskey="p" rel="prev">Embedding R under Unix-alikes</a>, Up: <a href="#Embedding-R-under-Unix_002dalikes" accesskey="u" rel="up">Embedding R under Unix-alikes</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Compiling-against-the-R-library-1"></a>
<h4 class="subsection">8.1.1 Compiling against the R library</h4>

<p>Suitable flags to compile and link against the R (shared or static)
library can be found by
</p>
<div class="example">
<pre class="example">R CMD config --cppflags
R CMD config --ldflags
</pre></div>

<p>(These apply only to an uninstalled copy or a standard install.)
</p>
<p>If R is installed, <code>pkg-config</code> is available and neither
sub-architectures nor a macOS framework have been used, alternatives for
a shared R library are
</p>
<div class="example">
<pre class="example">pkg-config --cflags libR
pkg-config --libs libR
</pre></div>

<p>and for a static R library
</p>
<div class="example">
<pre class="example">pkg-config --cflags libR
pkg-config --libs --static libR
</pre></div>

<p>(This may work for an installed OS framework if <code>pkg-config</code> is
taught where to look for <samp>libR.pc</samp>: it is installed inside the
framework.)
</p>
<p>However, a more comprehensive way is to set up a <samp>Makefile</samp> to
compile the front-end.  Suppose file <samp>myfe.c</samp> is to be compiled to
<samp>myfe</samp>. A suitable <samp>Makefile</samp> might be
</p>
<div class="example">
<pre class="example">include ${R_HOME}/etc${R_ARCH}/Makeconf
all: myfe

## The following is not needed, but avoids PIC flags.
myfe.o: myfe.c
        $(CC) $(ALL_CPPFLAGS) $(CFLAGS) -c myfe.c -o $@

## replace $(LIBR) $(LIBS) by $(STATIC_LIBR) if R was build with a static libR
myfe: myfe.o
        $(MAIN_LINK) -o $@ myfe.o $(LIBR) $(LIBS)
</pre></div>

<p>invoked as
</p>
<div class="example">
<pre class="example">R CMD make
R CMD myfe
</pre></div>

<p>Additional flags which <code>$(MAIN_LINK)</code> includes are, amongst others,
those to select OpenMP and <samp>--export-dynamic</samp> for the GNU linker
on some platforms.  In principle <code>$(LIBS)</code> is not needed
when using a shared R library as <samp>libR</samp> is linked against
those libraries, but some platforms need the executable also linked
against them.
</p>
<hr>
<a name="Setting-R-callbacks"></a>
<div class="header">
<p>
Next: <a href="#Registering-symbols" accesskey="n" rel="next">Registering symbols</a>, Previous: <a href="#Compiling-against-the-R-library" accesskey="p" rel="prev">Compiling against the R library</a>, Up: <a href="#Embedding-R-under-Unix_002dalikes" accesskey="u" rel="up">Embedding R under Unix-alikes</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Setting-R-callbacks-1"></a>
<h4 class="subsection">8.1.2 Setting R callbacks</h4>

<p>For Unix-alikes there is a public header file <samp>Rinterface.h</samp> that
makes it possible to change the standard callbacks used by R in a
documented way.  This defines pointers (if <code>R_INTERFACE_PTRS</code> is
defined)
</p>
<div class="example">
<pre class="example">extern void (*ptr_R_Suicide)(const char *);
extern void (*ptr_R_ShowMessage)(const char *);
extern int  (*ptr_R_ReadConsole)(const char *, unsigned char *, int, int);
extern void (*ptr_R_WriteConsole)(const char *, int);
extern void (*ptr_R_WriteConsoleEx)(const char *, int, int);
extern void (*ptr_R_ResetConsole)();
extern void (*ptr_R_FlushConsole)();
extern void (*ptr_R_ClearerrConsole)();
extern void (*ptr_R_Busy)(int);
extern void (*ptr_R_CleanUp)(SA_TYPE, int, int);
extern int  (*ptr_R_ShowFiles)(int, const char **, const char **,
                               const char *, Rboolean, const char *);
extern int  (*ptr_R_ChooseFile)(int, char *, int);
extern int  (*ptr_R_EditFile)(const char *);
extern void (*ptr_R_loadhistory)(SEXP, SEXP, SEXP, SEXP);
extern void (*ptr_R_savehistory)(SEXP, SEXP, SEXP, SEXP);
extern void (*ptr_R_addhistory)(SEXP, SEXP, SEXP, SEXP);
// added in R 3.0.0
extern int  (*ptr_R_EditFiles)(int, const char **, const char **, const char *);
extern SEXP (*ptr_do_selectlist)(SEXP, SEXP, SEXP, SEXP);
extern SEXP (*ptr_do_dataentry)(SEXP, SEXP, SEXP, SEXP);
extern SEXP (*ptr_do_dataviewer)(SEXP, SEXP, SEXP, SEXP);
extern void (*ptr_R_ProcessEvents)();
</pre></div>

<p>which allow standard R callbacks to be redirected to your GUI.  What
these do is generally documented in the file <samp>src/unix/system.txt</samp>.
</p>
<dl>
<dt><a name="index-R_005fShowMessage"></a>Function: <em>void</em> <strong>R_ShowMessage</strong> <em>(char *<var>message</var>)</em></dt>
<dd><p>This should display the message, which may have multiple lines:  it
should be brought to the user&rsquo;s attention immediately.
</p></dd></dl>

<dl>
<dt><a name="index-R_005fBusy"></a>Function: <em>void</em> <strong>R_Busy</strong> <em>(int <var>which</var>)</em></dt>
<dd><p>This function invokes actions (such as change of cursor) when R
embarks on an extended computation (<code><var>which</var>=1</code>) and when such
a state terminates (<code><var>which</var>=0</code>).
</p></dd></dl>

<dl>
<dt><a name="index-R_005fReadConsole"></a>Function: <em>int</em> <strong>R_ReadConsole</strong> <em>(const char *<var>prompt</var>, unsigned char *<var>buf</var>,   int <var>buflen</var>, int <var>hist</var>)</em></dt>
<dt><a name="index-R_005fWriteConsole"></a>Function: <em>void</em> <strong>R_WriteConsole</strong> <em>(const char *<var>buf</var>, int <var>buflen</var>)</em></dt>
<dt><a name="index-R_005fWriteConsoleEx"></a>Function: <em>void</em> <strong>R_WriteConsoleEx</strong> <em>(const char *<var>buf</var>, int <var>buflen</var>, int <var>otype</var>)</em></dt>
<dt><a name="index-R_005fResetConsole"></a>Function: <em>void</em> <strong>R_ResetConsole</strong> <em>()</em></dt>
<dt><a name="index-R_005fFlushConsole"></a>Function: <em>void</em> <strong>R_FlushConsole</strong> <em>()</em></dt>
<dt><a name="index-R_005fClearErrConsole"></a>Function: <em>void</em> <strong>R_ClearErrConsole</strong> <em>()</em></dt>
<dd>
<p>These functions interact with a console.
</p>
<p><code>R_ReadConsole</code> prints the given prompt at the console and then
does a <code>fgets(3)</code>&ndash;like operation, transferring up to <var>buflen</var>
characters into the buffer <var>buf</var>. The last two bytes should be
set to &lsquo;<samp>&quot;\n\0&quot;</samp>&rsquo; to preserve sanity.  If <var>hist</var> is non-zero,
then the line should be added to any command history which is being
maintained.  The return value is 0 is no input is available and &gt;0
otherwise.
</p>
<p><code>R_WriteConsoleEx</code> writes the given buffer to the console,
<var>otype</var> specifies the output type (regular output or
warning/error). Call to <code>R_WriteConsole(buf, buflen)</code> is equivalent
to <code>R_WriteConsoleEx(buf, buflen, 0)</code>. To ensure backward
compatibility of the callbacks, <code>ptr_R_WriteConsoleEx</code> is used only
if <code>ptr_R_WriteConsole</code> is set to <code>NULL</code>.  To ensure that
<code>stdout()</code> and <code>stderr()</code> connections point to the console,
set the corresponding files to <code>NULL</code> <em>via</em>
</p><div class="example">
<pre class="example">      R_Outputfile = NULL;
      R_Consolefile = NULL;
</pre></div>

<p><code>R_ResetConsole</code> is called when the system is reset after an error.
<code>R_FlushConsole</code> is called to flush any pending output to the
system console.  <code>R_ClearerrConsole</code> clears any errors associated
with reading from the console.
</p></dd></dl>

<dl>
<dt><a name="index-R_005fShowFiles"></a>Function: <em>int</em> <strong>R_ShowFiles</strong> <em>(int <var>nfile</var>, const char **<var>file</var>,   const char **<var>headers</var>, const char *<var>wtitle</var>, Rboolean <var>del</var>,   const char *<var>pager</var>)</em></dt>
<dd>
<p>This function is used to display the contents of files.
</p></dd></dl>

<dl>
<dt><a name="index-R_005fChooseFile"></a>Function: <em>int</em> <strong>R_ChooseFile</strong> <em>(int <var>new</var>, char *<var>buf</var>,   int <var>len</var>)</em></dt>
<dd>
<p>Choose a file and return its name in <var>buf</var> of length <var>len</var>.
Return value is 0 for success, &gt; 0 otherwise.
</p></dd></dl>

<dl>
<dt><a name="index-R_005fEditFile"></a>Function: <em>int</em> <strong>R_EditFile</strong> <em>(const char *<var>buf</var>)</em></dt>
<dd><p>Send a file to an editor window.
</p></dd></dl>

<dl>
<dt><a name="index-R_005fEditFiles"></a>Function: <em>int</em> <strong>R_EditFiles</strong> <em>(int <var>nfile</var>, const char **<var>file</var>, const char **<var>title</var>, const char *<var>editor</var>)</em></dt>
<dd><p>Send <var>nfile</var> files to an editor, with titles possibly to be used for
the editor window(s).
</p></dd></dl>

<dl>
<dt><a name="index-R_005floadhistory"></a>Function: <em>SEXP</em> <strong>R_loadhistory</strong> <em>(SEXP, SEXP, SEXP, SEXP);</em></dt>
<dt><a name="index-R_005fsavehistory"></a>Function: <em>SEXP</em> <strong>R_savehistory</strong> <em>(SEXP, SEXP, SEXP, SEXP);</em></dt>
<dt><a name="index-R_005faddhistory"></a>Function: <em>SEXP</em> <strong>R_addhistory</strong> <em>(SEXP, SEXP, SEXP, SEXP);</em></dt>
<dd>
<p><code>.Internal</code> functions for <code>loadhistory</code>, <code>savehistory</code>
and <code>timestamp</code>.
</p>
<p>If the console has no history mechanism these can be as
simple as
</p>
<div class="example">
<pre class="example">SEXP R_loadhistory (SEXP call, SEXP op, SEXP args, SEXP env)
{
    errorcall(call, &quot;loadhistory is not implemented&quot;);
    return R_NilValue;
}
SEXP R_savehistory (SEXP call, SEXP op , SEXP args, SEXP env)
{
    errorcall(call, &quot;savehistory is not implemented&quot;);
    return R_NilValue;
}
SEXP R_addhistory (SEXP call, SEXP op , SEXP args, SEXP env)
{
    return R_NilValue;
}
</pre></div>

<p>The <code>R_addhistory</code> function should return silently if no history
mechanism is present, as a user may be calling <code>timestamp</code> purely
to write the time stamp to the console.
</p></dd></dl>

<dl>
<dt><a name="index-R_005fSuicide"></a>Function: <em>void</em> <strong>R_Suicide</strong> <em>(const char *<var>message</var>)</em></dt>
<dd><p>This should abort R as rapidly as possible, displaying the message.
A possible implementation is
</p>
<div class="example">
<pre class="example">void R_Suicide (const char *message)
{
    char  pp[1024];
    snprintf(pp, 1024, &quot;Fatal error: %s\n&quot;, s);
    R_ShowMessage(pp);
    R_CleanUp(SA_SUICIDE, 2, 0);
}
</pre></div>
</dd></dl>

<dl>
<dt><a name="index-R_005fCleanUp"></a>Function: <em>void</em> <strong>R_CleanUp</strong> <em>(SA_TYPE <var>saveact</var>, int <var>status</var>,   int <var>RunLast</var>)</em></dt>
<dd>
<p>This function invokes any actions which occur at system termination.
It needs to be quite complex:
</p>
<div class="example">
<pre class="example">#include &lt;Rinterface.h&gt;
#include &lt;Rembedded.h&gt;    /* for Rf_KillAllDevices */

void R_CleanUp (SA_TYPE saveact, int status, int RunLast)
{
    if(saveact == SA_DEFAULT) saveact = SaveAction;
    if(saveact == SA_SAVEASK) {
       /* ask what to do and set saveact */
    }
    switch (saveact) {
    case SA_SAVE:
        if(runLast) R_dot_Last();
        if(R_DirtyImage) R_SaveGlobalEnv();
        /* save the console history in R_HistoryFile */
        break;
    case SA_NOSAVE:
        if(runLast) R_dot_Last();
        break;
    case SA_SUICIDE:
    default:
        break;
    }

    R_RunExitFinalizers();
    /* clean up after the editor e.g. CleanEd() */

    R_CleanTempDir();

    /* close all the graphics devices */
    if(saveact != SA_SUICIDE) Rf_KillAllDevices();
    fpu_setup(FALSE);

    exit(status);
}
</pre></div>
</dd></dl>

<p>These callbacks should never be changed in a running R session (and
hence cannot be called from an extension package).
</p>
<dl>
<dt><a name="index-R_005fdataentry"></a>Function: <em>SEXP</em> <strong>R_dataentry</strong> <em>(SEXP, SEXP, SEXP, SEXP);</em></dt>
<dt><a name="index-R_005fdataviewer"></a>Function: <em>SEXP</em> <strong>R_dataviewer</strong> <em>(SEXP, SEXP, SEXP, SEXP);</em></dt>
<dt><a name="index-R_005fselectlist"></a>Function: <em>SEXP</em> <strong>R_selectlist</strong> <em>(SEXP, SEXP, SEXP, SEXP);</em></dt>
<dd>
<p><code>.External</code> functions for <code>dataentry</code> (and <code>edit</code> on
matrices and data frames), <code>View</code> and <code>select.list</code>.  These
can be changed if they are not currently in use.
</p></dd></dl>


<hr>
<a name="Registering-symbols"></a>
<div class="header">
<p>
Next: <a href="#Meshing-event-loops" accesskey="n" rel="next">Meshing event loops</a>, Previous: <a href="#Setting-R-callbacks" accesskey="p" rel="prev">Setting R callbacks</a>, Up: <a href="#Embedding-R-under-Unix_002dalikes" accesskey="u" rel="up">Embedding R under Unix-alikes</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Registering-symbols-1"></a>
<h4 class="subsection">8.1.3 Registering symbols</h4>

<p>An application embedding R needs a different way of registering
symbols because it is not a dynamic library loaded by R as would be
the case with a package.  Therefore R reserves a special
<code>DllInfo</code> entry for the embedding application such that it can
register symbols to be used with <code>.C</code>, <code>.Call</code> etc.  This
entry can be obtained by calling <code>getEmbeddingDllInfo</code>, so a
typical use is
</p>
<div class="example">
<pre class="example">DllInfo *info = R_getEmbeddingDllInfo();
R_registerRoutines(info, cMethods, callMethods, NULL, NULL);
</pre></div>

<p>The native routines defined by <code>cMethods</code> and <code>callMethods</code>
should be present in the embedding application.  See <a href="#Registering-native-routines">Registering native routines</a> for details on registering symbols in general.
</p>

<hr>
<a name="Meshing-event-loops"></a>
<div class="header">
<p>
Next: <a href="#Threading-issues" accesskey="n" rel="next">Threading issues</a>, Previous: <a href="#Registering-symbols" accesskey="p" rel="prev">Registering symbols</a>, Up: <a href="#Embedding-R-under-Unix_002dalikes" accesskey="u" rel="up">Embedding R under Unix-alikes</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Meshing-event-loops-1"></a>
<h4 class="subsection">8.1.4 Meshing event loops</h4>

<p>One of the most difficult issues in interfacing R to a front-end is
the handling of event loops, at least if a single thread is used.  R
uses events and timers for
</p>
<ul>
<li> Running X11 windows such as the graphics device and data editor, and
interacting with them (e.g., using <code>locator()</code>).

</li><li> Supporting Tcl/Tk events for the <strong>tcltk</strong> package (for at least the
X11 version of Tk).

</li><li> Preparing input.

</li><li> Timing operations, for example for profiling R code and
<code>Sys.sleep()</code>.

</li><li> Interrupts, where permitted.
</li></ul>

<p>Specifically, the Unix-alike command-line version of R runs separate
event loops for
</p>
<ul>
<li> Preparing input at the console command-line, in file
<samp>src/unix/sys-unix.c</samp>.

</li><li> Waiting for a response from a socket in the internal functions
underlying FTP and HTTP transfers in <code>download.file()</code> and for
direct socket access, in files
<samp>src/modules/internet/nanoftp.c</samp>,
<samp>src/modules/internet/nanohttp.c</samp> and
<samp>src/modules/internet/Rsock.c</samp>

</li><li> Mouse and window events when displaying the X11-based dataentry window,
in file <samp>src/modules/X11/dataentry.c</samp>.  This is regarded as
<em>modal</em>, and no other events are serviced whilst it is active.
</li></ul>

<p>There is a protocol for adding event handlers to the first two types of
event loops, using types and functions declared in the header
<samp>R_ext/eventloop.h</samp> and described in comments in file
<samp>src/unix/sys-std.c</samp>.  It is possible to add (or remove) an input
handler for events on a particular file descriptor, or to set a polling
interval (<em>via</em> <code>R_wait_usec</code>) and a function to be called
periodically <em>via</em> <code>R_PolledEvents</code>: the polling mechanism is used by
the <strong>tcltk</strong> package.
</p>
<p>It is not intended that these facilities are used by packages, but if
they are needed exceptionally, the package should ensure that it cleans
up and removes its handlers when its namespace is unloaded.  Note that
the header <samp>sys/select.h</samp> is required: users should check this is
available and define <code>HAVE_SYS_SELECT_H</code> before including
<samp>R_ext/eventloop.h</samp>.  (It is often the case that another header
will include <samp>sys/select.h</samp> before <samp>eventloop.h</samp> is processed,
but this should not be relied on.)
</p>
<p>An alternative front-end needs both to make provision for other R
events whilst waiting for input, and to ensure that it is not frozen out
during events of the second type.  The ability to add a polled handler
as <code>R_timeout_handler</code> is used by the <strong>tcltk</strong> package.
</p>

<hr>
<a name="Threading-issues"></a>
<div class="header">
<p>
Previous: <a href="#Meshing-event-loops" accesskey="p" rel="prev">Meshing event loops</a>, Up: <a href="#Embedding-R-under-Unix_002dalikes" accesskey="u" rel="up">Embedding R under Unix-alikes</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Threading-issues-1"></a>
<h4 class="subsection">8.1.5 Threading issues</h4>

<p>Embedded R is designed to be run in the main thread, and all the
testing is done in that context.  There is a potential issue with the
stack-checking mechanism where threads are involved.  This uses two
variables declared in <samp>Rinterface.h</samp> (if <code>CSTACK_DEFNS</code> is
defined) as
</p>
<div class="example">
<pre class="example">extern uintptr_t R_CStackLimit; /* C stack limit */
extern uintptr_t R_CStackStart; /* Initial stack address */
</pre></div>

<p>Note that <code>uintptr_t</code> is an optional C99 type for which a
substitute is defined in R, so your code needs to define
<code>HAVE_UINTPTR_T</code> appropriately.  To do so, test if the type is
defined in C header <samp>stdint.h</samp> or C++ header <samp>cstdint</samp> and if
so include the header and define <code>HAVE_UINTPTR_T</code> before including
<samp>Rinterface.h</samp>.
</p>
<p>These will be set<a name="DOCF146" href="#FOOT146"><sup>146</sup></a> when <code>Rf_initialize_R</code> is called, to values appropriate to the
main thread.  Stack-checking can be disabled by setting
<code>R_CStackLimit = (uintptr_t)-1</code> immediately after
<code>Rf_initialize_R</code> is called, but it is better to if possible set
appropriate values.  (What these are and how to determine them are
OS-specific, and the stack size limit may differ for secondary threads.
If you have a choice of stack size, at least 10Mb is recommended.)
</p>
<p>You may also want to consider how signals are handled: R sets signal
handlers for several signals, including <code>SIGINT</code>, <code>SIGSEGV</code>,
<code>SIGPIPE</code>, <code>SIGUSR1</code> and <code>SIGUSR2</code>, but these can all be
suppressed by setting the variable <code>R_SignalHandlers</code> (declared in
<samp>Rinterface.h</samp>) to <code>0</code>.
</p>
<p>Note that these variables must not be changed by an R
<strong>package</strong>: a package should not call R internals which
makes use of the stack-checking mechanism on a secondary thread.
</p>
<hr>
<a name="Embedding-R-under-Windows"></a>
<div class="header">
<p>
Previous: <a href="#Embedding-R-under-Unix_002dalikes" accesskey="p" rel="prev">Embedding R under Unix-alikes</a>, Up: <a href="#Linking-GUIs-and-other-front_002dends-to-R" accesskey="u" rel="up">Linking GUIs and other front-ends to R</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Embedding-R-under-Windows-1"></a>
<h3 class="section">8.2 Embedding R under Windows</h3>

<p>All Windows interfaces to R call entry points in the DLL
<samp>R.dll</samp>, directly or indirectly.  Simpler applications may find it
easier to use the indirect route <em>via</em> <acronym>(D)COM</acronym>.
</p>
<table summary="" class="menu" border="0" cellspacing="0">
<tr><td align="left" valign="top">&bull; <a href="#Using-_0028D_0029COM" accesskey="1">Using (D)COM</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Calling-R_002edll-directly" accesskey="2">Calling R.dll directly</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
<tr><td align="left" valign="top">&bull; <a href="#Finding-R_005fHOME" accesskey="3">Finding R_HOME</a>:</td><td>&nbsp;&nbsp;</td><td align="left" valign="top">
</td></tr>
</table>

<hr>
<a name="Using-_0028D_0029COM"></a>
<div class="header">
<p>
Next: <a href="#Calling-R_002edll-directly" accesskey="n" rel="next">Calling R.dll directly</a>, Previous: <a href="#Embedding-R-under-Windows" accesskey="p" rel="prev">Embedding R under Windows</a>, Up: <a href="#Embedding-R-under-Windows" accesskey="u" rel="up">Embedding R under Windows</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Using-_0028D_0029COM-1"></a>
<h4 class="subsection">8.2.1 Using (D)COM</h4>

<p><acronym>(D)COM</acronym> is a standard Windows mechanism used for communication
between Windows applications.  One application (here R) is run as COM
server which offers services to clients, here the front-end calling
application.  The services are described in a &lsquo;Type Library&rsquo; and are
(more or less) language-independent, so the calling application can be
written in C or C++ or Visual Basic or Perl or Python and so on.
The &lsquo;D&rsquo; in (D)COM refers to &lsquo;distributed&rsquo;, as the client and server can
be running on different machines.
</p>
<p>The basic R distribution is not a (D)COM server, but two addons are
currently available that interface directly with R and provide a
(D)COM server:
</p><ul>
<li> There is a (D)COM server called <code>StatConnector</code> written by Thomas
Baier available <em>via</em> <a href="http://sunsite.univie.ac.at/rcom/">http://sunsite.univie.ac.at/rcom/</a>,
which works with R packages to support transfer of data to and from
R and remote execution of R commands, as well as embedding of an
R graphics window.

<p>Recent versions have usage restrictions.
</p>
</li><li> Another (D)COM server, <code>RDCOMServer</code>, may be available from Omegahat,
<a href="http://www.omegahat.net/">http://www.omegahat.net/</a>.  Its philosophy is discussed in
<a href="http://www.omegahat.net/RDCOMServer/Docs/Paradigm.html">http://www.omegahat.net/RDCOMServer/Docs/Paradigm.html</a> and is
very different from the purpose of this section.
</li></ul>
<hr>
<a name="Calling-R_002edll-directly"></a>
<div class="header">
<p>
Next: <a href="#Finding-R_005fHOME" accesskey="n" rel="next">Finding R_HOME</a>, Previous: <a href="#Using-_0028D_0029COM" accesskey="p" rel="prev">Using (D)COM</a>, Up: <a href="#Embedding-R-under-Windows" accesskey="u" rel="up">Embedding R under Windows</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Calling-R_002edll-directly-1"></a>
<h4 class="subsection">8.2.2 Calling R.dll directly</h4>

<p>The <code>R</code> DLL is mainly written in C and has <code>_cdecl</code> entry
points.  Calling it directly will be tricky except from C code (or C++
with a little care).
</p>
<p>There is a version of the Unix-alike interface calling
</p>
<div class="example">
<pre class="example">int Rf_initEmbeddedR(int ac, char **av);
void Rf_endEmbeddedR(int fatal);
</pre></div>

<p>which is an entry point in <samp>R.dll</samp>.  Examples of its use (and a
suitable <samp>Makefile.win</samp>) can be found in the <samp>tests/Embedding</samp>
directory of the sources.  You may need to ensure that
<samp><var>R_HOME</var>/bin</samp> is in your <code>PATH</code> so the R DLLs are found.
</p>
<p>Examples of calling <samp>R.dll</samp> directly are provided in the directory
<samp>src/gnuwin32/front-ends</samp>, including a simple command-line
front end <samp>rtest.c</samp> whose code is
</p>
<div class="smallexample">
<pre class="smallexample">#define Win32
#include &lt;windows.h&gt;
#include &lt;stdio.h&gt;
#include &lt;Rversion.h&gt;
#define LibExtern __declspec(dllimport) extern
#include &lt;Rembedded.h&gt;
#include &lt;R_ext/RStartup.h&gt;
/* for askok and askyesnocancel */
#include &lt;graphapp.h&gt;

/* for signal-handling code */
#include &lt;psignal.h&gt;

/* simple input, simple output */

/* This version blocks all events: a real one needs to call ProcessEvents
   frequently. See rterm.c and ../system.c for one approach using
   a separate thread for input.
*/
int myReadConsole(const char *prompt, char *buf, int len, int addtohistory)
{
    fputs(prompt, stdout);
    fflush(stdout);
    if(fgets(buf, len, stdin)) return 1; else return 0;
}

void myWriteConsole(const char *buf, int len)
{
    printf(&quot;%s&quot;, buf);
}

void myCallBack(void)
{
    /* called during i/o, eval, graphics in ProcessEvents */
}

void myBusy(int which)
{
    /* set a busy cursor ... if which = 1, unset if which = 0 */
}

static void my_onintr(int sig) { UserBreak = 1; }

int main (int argc, char **argv)
{
    structRstart rp;
    Rstart Rp = &amp;rp;
    char Rversion[25], *RHome;

    sprintf(Rversion, &quot;%s.%s&quot;, R_MAJOR, R_MINOR);
    if(strcmp(getDLLVersion(), Rversion) != 0) {
        fprintf(stderr, &quot;Error: R.DLL version does not match\n&quot;);
        exit(1);
    }

    R_setStartTime();
    R_DefParams(Rp);
    if((RHome = get_R_HOME()) == NULL) {
        fprintf(stderr, &quot;R_HOME must be set in the environment or Registry\n&quot;);
        exit(1);
    }
    Rp-&gt;rhome = RHome;
    Rp-&gt;home = getRUser();
    Rp-&gt;CharacterMode = LinkDLL;
    Rp-&gt;ReadConsole = myReadConsole;
    Rp-&gt;WriteConsole = myWriteConsole;
    Rp-&gt;CallBack = myCallBack;
    Rp-&gt;ShowMessage = askok;
    Rp-&gt;YesNoCancel = askyesnocancel;
    Rp-&gt;Busy = myBusy;

    Rp-&gt;R_Quiet = TRUE;        /* Default is FALSE */
    Rp-&gt;R_Interactive = FALSE; /* Default is TRUE */
    Rp-&gt;RestoreAction = SA_RESTORE;
    Rp-&gt;SaveAction = SA_NOSAVE;
    R_SetParams(Rp);
    R_set_command_line_arguments(argc, argv);

    FlushConsoleInputBuffer(GetStdHandle(STD_INPUT_HANDLE));

    signal(SIGBREAK, my_onintr);
    GA_initapp(0, 0);
    readconsolecfg();
    setup_Rmainloop();
#ifdef SIMPLE_CASE
    run_Rmainloop();
#else
    R_ReplDLLinit();
    while(R_ReplDLLdo1() &gt; 0) {
/* add user actions here if desired */
    }
/* only get here on EOF (not q()) */
#endif
    Rf_endEmbeddedR(0);
    return 0;
}
</pre></div>

<p>The ideas are
</p>
<ul>
<li> Check that the front-end and the linked <samp>R.dll</samp> match &ndash; other
front-ends may allow a looser match.

</li><li> Find and set the R home directory and the user&rsquo;s home directory.  The
former may be available from the Windows Registry: it will be in
<code>HKEY_LOCAL_MACHINE\Software\R-core\R\InstallPath</code> from an
administrative install and
<code>HKEY_CURRENT_USER\Software\R-core\R\InstallPath</code> otherwise, if
selected during installation (as it is by default).

</li><li> Define startup conditions and callbacks <em>via</em> the <code>Rstart</code> structure.
<code>R_DefParams</code> sets the defaults, and <code>R_SetParams</code> sets
updated values.

</li><li> Record the command-line arguments used by
<code>R_set_command_line_arguments</code> for use by the R function
<code>commandArgs()</code>.

</li><li> Set up the signal handler and the basic user interface.

</li><li> Run the main R loop, possibly with our actions intermeshed.

</li><li> Arrange to clean up.
</li></ul>

<p>An underlying theme is the need to keep the GUI &lsquo;alive&rsquo;, and this has
not been done in this example.  The R callback <code>R_ProcessEvents</code>
needs to be called frequently to ensure that Windows events in R
windows are handled expeditiously.  Conversely, R needs to allow the
GUI code (which is running in the same process) to update itself as
needed &ndash; two ways are provided to allow this:
</p>
<ul>
<li> <code>R_ProcessEvents</code> calls the callback registered by
<code>Rp-&gt;callback</code>.  A version of this is used to run package Tcl/Tk
for <strong>tcltk</strong> under Windows, for the code is

<div class="example">
<pre class="example">void R_ProcessEvents(void)
{
    while (peekevent()) doevent(); /* Windows events for GraphApp */
    if (UserBreak) { UserBreak = FALSE; onintr(); }
    R_CallBackHook();
    if(R_tcldo) R_tcldo();
}
</pre></div>

</li><li> The mainloop can be split up to allow the calling application to take
some action after each line of input has been dealt with: see the
alternative code below <code>#ifdef SIMPLE_CASE</code>.
</li></ul>

<p>It may be that no R GraphApp windows need to be considered, although
these include pagers, the <code>windows()</code> graphics device, the R
data and script editors and various popups such as <code>choose.file()</code>
and <code>select.list()</code>.  It would be possible to replace all of these,
but it seems easier to allow GraphApp to handle most of them.
</p>
<p>It is possible to run R in a GUI in a single thread (as
<samp>RGui.exe</samp> shows) but it will normally be easier<a name="DOCF147" href="#FOOT147"><sup>147</sup></a> to
use multiple threads.
</p>
<p>Note that R&rsquo;s own front ends use a stack size of 10Mb, whereas MinGW
executables default to 2Mb, and Visual C++ ones to 1Mb.  The latter
stack sizes are too small for a number of R applications, so
general-purpose front-ends should use a larger stack size.
</p>

<hr>
<a name="Finding-R_005fHOME"></a>
<div class="header">
<p>
Previous: <a href="#Calling-R_002edll-directly" accesskey="p" rel="prev">Calling R.dll directly</a>, Up: <a href="#Embedding-R-under-Windows" accesskey="u" rel="up">Embedding R under Windows</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Finding-R_005fHOME-1"></a>
<h4 class="subsection">8.2.3 Finding R_HOME</h4>

<p>Both applications which embed R and those which use a <code>system</code>
call to invoke R (as <code>Rscript.exe</code>, <code>Rterm.exe</code> or
<code>R.exe</code>) need to be able to find the R <samp>bin</samp> directory.
The simplest way to do so is the ask the user to set an environment
variable <code>R_HOME</code> and use that, but naive users may be flummoxed as
to how to do so or what value to use.
</p>
<p>The R for Windows installers have for a long time allowed the value
of <code>R_HOME</code> to be recorded in the Windows Registry: this is
optional but selected by default.  <em>Where</em> it is recorded has
changed over the years to allow for multiple versions of R to be
installed at once, and to allow 32- and 64-bit versions of R to be
installed on the same machine.
</p>
<p>The basic Registry location is <code>Software\R-core\R</code>.  For an
administrative install this is under <code>HKEY_LOCAL_MACHINE</code> and on a
64-bit OS <code>HKEY_LOCAL_MACHINE\Software\R-core\R</code> is by default
redirected for a 32-bit application, so a 32-bit application will see
the information for the last 32-bit install, and a 64-bit application
that for the last 64-bit install.  For a personal install, the
information is under <code>HKEY_CURRENT_USER\Software\R-core\R</code> which is
seen by both 32-bit and 64-bit applications and so records the last
install of either architecture.  To circumvent this, there are locations
<code>Software\R-core\R32</code> and <code>Software\R-core\R64</code> which always
refer to one architecture.
</p>
<p>When R is installed and recording is not disabled then two string
values are written at that location for keys <code>InstallPath</code> and
<code>Current Version</code>, and these keys are removed when R is
uninstalled.  To allow information about other installed versions to be
retained, there is also a key named something like <code>3.0.0</code> or
<code>3.0.0 patched</code> or <code>3.1.0 Pre-release</code> with a value for
<code>InstallPath</code>.
</p>
<p>So a comprehensive algorithm to search for <code>R_HOME</code> is something
like
</p>
<ul>
<li> Decide which of personal or administrative installs should have
precedence.  There are arguments both ways: we find that with roaming
profiles that <code>HKEY_CURRENT_USER\Software</code> often gets reverted to
an earlier version.  Do the following for one or both of
<code>HKEY_CURRENT_USER</code> and <code>HKEY_LOCAL_MACHINE</code>.

</li><li> If the desired architecture is known, look in <code>Software\R-core\R32</code>
or <code>Software\R-core\R64</code>, and if that does not exist or the
architecture is immaterial, in <code>Software\R-core\R</code>.

</li><li> If key <code>InstallPath</code> exists then this is <code>R_HOME</code> (recorded
using backslashes).  If it does not, look for version-specific keys like
<code>2.11.0 alpha</code>, pick the latest (which is of itself a complicated
algorithm as <code>2.11.0 patched &gt; 2.11.0 &gt; 2.11.0 alpha &gt; 2.8.1</code>) and
use its value for <code>InstallPath</code>.
</li></ul>

<hr>
<a name="Function-and-variable-index"></a>
<div class="header">
<p>
Next: <a href="#Concept-index" accesskey="n" rel="next">Concept index</a>, Previous: <a href="#Linking-GUIs-and-other-front_002dends-to-R" accesskey="p" rel="prev">Linking GUIs and other front-ends to R</a>, Up: <a href="#Top" accesskey="u" rel="up">Top</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Function-and-variable-index-1"></a>
<h2 class="unnumbered">Function and variable index</h2>

<table summary=""><tr><th valign="top">Jump to: &nbsp; </th><td><a class="summary-letter" href="#Function-and-variable-index_vr_symbol-1"><b>.</b></a>
 &nbsp; 
<a class="summary-letter" href="#Function-and-variable-index_vr_symbol-2"><b>\</b></a>
 &nbsp; 
<br>
<a class="summary-letter" href="#Function-and-variable-index_vr_letter-A"><b>A</b></a>
 &nbsp; 
<a class="summary-letter" href="#Function-and-variable-index_vr_letter-B"><b>B</b></a>
 &nbsp; 
<a class="summary-letter" href="#Function-and-variable-index_vr_letter-C"><b>C</b></a>
 &nbsp; 
<a class="summary-letter" href="#Function-and-variable-index_vr_letter-D"><b>D</b></a>
 &nbsp; 
<a class="summary-letter" href="#Function-and-variable-index_vr_letter-E"><b>E</b></a>
 &nbsp; 
<a class="summary-letter" href="#Function-and-variable-index_vr_letter-F"><b>F</b></a>
 &nbsp; 
<a class="summary-letter" href="#Function-and-variable-index_vr_letter-G"><b>G</b></a>
 &nbsp; 
<a class="summary-letter" href="#Function-and-variable-index_vr_letter-I"><b>I</b></a>
 &nbsp; 
<a class="summary-letter" href="#Function-and-variable-index_vr_letter-L"><b>L</b></a>
 &nbsp; 
<a class="summary-letter" href="#Function-and-variable-index_vr_letter-M"><b>M</b></a>
 &nbsp; 
<a class="summary-letter" href="#Function-and-variable-index_vr_letter-N"><b>N</b></a>
 &nbsp; 
<a class="summary-letter" href="#Function-and-variable-index_vr_letter-O"><b>O</b></a>
 &nbsp; 
<a class="summary-letter" href="#Function-and-variable-index_vr_letter-P"><b>P</b></a>
 &nbsp; 
<a class="summary-letter" href="#Function-and-variable-index_vr_letter-Q"><b>Q</b></a>
 &nbsp; 
<a class="summary-letter" href="#Function-and-variable-index_vr_letter-R"><b>R</b></a>
 &nbsp; 
<a class="summary-letter" href="#Function-and-variable-index_vr_letter-S"><b>S</b></a>
 &nbsp; 
<a class="summary-letter" href="#Function-and-variable-index_vr_letter-T"><b>T</b></a>
 &nbsp; 
<a class="summary-letter" href="#Function-and-variable-index_vr_letter-U"><b>U</b></a>
 &nbsp; 
<a class="summary-letter" href="#Function-and-variable-index_vr_letter-V"><b>V</b></a>
 &nbsp; 
</td></tr></table>
<table summary="" class="index-vr" border="0">
<tr><td></td><th align="left">Index Entry</th><td>&nbsp;</td><th align="left"> Section</th></tr>
<tr><td colspan="4"> <hr></td></tr>
<tr><th><a name="Function-and-variable-index_vr_symbol-1">.</a></th><td></td><td></td></tr>
<tr><td></td><td valign="top"><a href="#index-_002eC"><code>.C</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Interface-functions-_002eC-and-_002eFortran">Interface functions .C and .Fortran</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_002eCall"><code>.Call</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Handling-R-objects-in-C">Handling R objects in C</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_002eCall-1"><code>.Call</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Calling-_002eCall">Calling .Call</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_002eExternal"><code>.External</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Handling-R-objects-in-C">Handling R objects in C</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_002eExternal-1"><code>.External</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Calling-_002eExternal">Calling .External</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_002eFortran"><code>.Fortran</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Interface-functions-_002eC-and-_002eFortran">Interface functions .C and .Fortran</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_002eLast_002elib"><code>.Last.lib</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Load-hooks">Load hooks</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_002eonAttach"><code>.onAttach</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Load-hooks">Load hooks</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_002eonDetach"><code>.onDetach</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Load-hooks">Load hooks</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_002eonLoad"><code>.onLoad</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Load-hooks">Load hooks</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_002eonUnload"><code>.onUnload</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Load-hooks">Load hooks</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_002eRandom_002eseed"><code>.Random.seed</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Random-numbers">Random numbers</a></td></tr>
<tr><td colspan="4"> <hr></td></tr>
<tr><th><a name="Function-and-variable-index_vr_symbol-2">\</a></th><td></td><td></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005cacronym"><code>\acronym</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Marking-text">Marking text</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005calias"><code>\alias</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Documenting-functions">Documenting functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005carguments"><code>\arguments</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Documenting-functions">Documenting functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005cauthor"><code>\author</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Documenting-functions">Documenting functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005cbold"><code>\bold</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Marking-text">Marking text</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005ccite"><code>\cite</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Marking-text">Marking text</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005ccode"><code>\code</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Marking-text">Marking text</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005ccommand"><code>\command</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Marking-text">Marking text</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005cconcept"><code>\concept</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Indices">Indices</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005ccr"><code>\cr</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Sectioning">Sectioning</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005cCRANpkg_007bpkg_007d"><code>\CRANpkg{<var>pkg</var>}</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#User_002ddefined-macros">User-defined macros</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005cdeqn"><code>\deqn</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Mathematics">Mathematics</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005cdescribe"><code>\describe</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Lists-and-tables">Lists and tables</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005cdescription"><code>\description</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Documenting-functions">Documenting functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005cdetails"><code>\details</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Documenting-functions">Documenting functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005cdfn"><code>\dfn</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Marking-text">Marking text</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005cdoi_007bnumbers_007d"><code>\doi{<var>numbers</var>}</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#User_002ddefined-macros">User-defined macros</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005cdontrun"><code>\dontrun</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Documenting-functions">Documenting functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005cdontshow"><code>\dontshow</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Documenting-functions">Documenting functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005cdots"><code>\dots</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Insertions">Insertions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005cdQuote"><code>\dQuote</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Marking-text">Marking text</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005cemail"><code>\email</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Marking-text">Marking text</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005cemph"><code>\emph</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Marking-text">Marking text</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005cenc"><code>\enc</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Insertions">Insertions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005cenumerate"><code>\enumerate</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Lists-and-tables">Lists and tables</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005cenv"><code>\env</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Marking-text">Marking text</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005ceqn"><code>\eqn</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Mathematics">Mathematics</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005cexamples"><code>\examples</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Documenting-functions">Documenting functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005cfigure"><code>\figure</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Figures">Figures</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005cfile"><code>\file</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Marking-text">Marking text</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005cformat"><code>\format</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Documenting-data-sets">Documenting data sets</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005chref"><code>\href</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Marking-text">Marking text</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005cif"><code>\if</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Conditional-text">Conditional text</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005cifelse"><code>\ifelse</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Conditional-text">Conditional text</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005citemize"><code>\itemize</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Lists-and-tables">Lists and tables</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005ckbd"><code>\kbd</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Marking-text">Marking text</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005ckeyword"><code>\keyword</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Documenting-functions">Documenting functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005cldots"><code>\ldots</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Insertions">Insertions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005clink"><code>\link</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Cross_002dreferences">Cross-references</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005cmethod"><code>\method</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Documenting-functions">Documenting functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005cname"><code>\name</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Documenting-functions">Documenting functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005cnewcommand"><code>\newcommand</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#User_002ddefined-macros">User-defined macros</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005cnote"><code>\note</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Documenting-functions">Documenting functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005coption"><code>\option</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Marking-text">Marking text</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005cout"><code>\out</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Conditional-text">Conditional text</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005cpackageAuthor"><code><code>\packageAuthor</code></code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#User_002ddefined-macros">User-defined macros</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005cpackageDescription"><code><code>\packageDescription</code></code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#User_002ddefined-macros">User-defined macros</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005cpackageDESCRIPTION"><code><code>\packageDESCRIPTION</code></code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#User_002ddefined-macros">User-defined macros</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005cpackageIndices"><code><code>\packageIndices</code></code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#User_002ddefined-macros">User-defined macros</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005cpackageMaintainer"><code><code>\packageMaintainer</code></code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#User_002ddefined-macros">User-defined macros</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005cpackageTitle"><code><code>\packageTitle</code></code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#User_002ddefined-macros">User-defined macros</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005cpkg"><code>\pkg</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Marking-text">Marking text</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005cpreformatted"><code>\preformatted</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Marking-text">Marking text</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005cR"><code>\R</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Insertions">Insertions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005cRdOpts"><code>\RdOpts</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Dynamic-pages">Dynamic pages</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005creferences"><code>\references</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Documenting-functions">Documenting functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005crenewcommand"><code>\renewcommand</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#User_002ddefined-macros">User-defined macros</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005cS3method"><code>\S3method</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Documenting-functions">Documenting functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005csamp"><code>\samp</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Marking-text">Marking text</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005csection"><code>\section</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Sectioning">Sectioning</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005cseealso"><code>\seealso</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Documenting-functions">Documenting functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005cSexpr"><code>\Sexpr</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Dynamic-pages">Dynamic pages</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005csource"><code>\source</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Documenting-data-sets">Documenting data sets</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005csQuote"><code>\sQuote</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Marking-text">Marking text</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005csspace"><code>\sspace</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#User_002ddefined-macros">User-defined macros</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005cstrong"><code>\strong</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Marking-text">Marking text</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005ctabular"><code>\tabular</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Lists-and-tables">Lists and tables</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005ctitle"><code>\title</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Documenting-functions">Documenting functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005curl"><code>\url</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Marking-text">Marking text</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005cusage"><code>\usage</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Documenting-functions">Documenting functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005cvalue"><code>\value</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Documenting-functions">Documenting functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005cvar"><code>\var</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Marking-text">Marking text</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005cverb"><code>\verb</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Marking-text">Marking text</a></td></tr>
<tr><td colspan="4"> <hr></td></tr>
<tr><th><a name="Function-and-variable-index_vr_letter-A">A</a></th><td></td><td></td></tr>
<tr><td></td><td valign="top"><a href="#index-allocVector"><code>allocVector</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Allocating-storage">Allocating storage</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-AUTHORS"><code>AUTHORS</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Package-subdirectories">Package subdirectories</a></td></tr>
<tr><td colspan="4"> <hr></td></tr>
<tr><th><a name="Function-and-variable-index_vr_letter-B">B</a></th><td></td><td></td></tr>
<tr><td></td><td valign="top"><a href="#index-bessel_005fi"><code>bessel_i</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Mathematical-functions">Mathematical functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-bessel_005fi-1"><code>bessel_i</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Mathematical-functions">Mathematical functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-bessel_005fj"><code>bessel_j</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Mathematical-functions">Mathematical functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-bessel_005fj-1"><code>bessel_j</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Mathematical-functions">Mathematical functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-bessel_005fk"><code>bessel_k</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Mathematical-functions">Mathematical functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-bessel_005fk-1"><code>bessel_k</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Mathematical-functions">Mathematical functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-bessel_005fy"><code>bessel_y</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Mathematical-functions">Mathematical functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-bessel_005fy-1"><code>bessel_y</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Mathematical-functions">Mathematical functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-beta"><code>beta</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Mathematical-functions">Mathematical functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-beta-1"><code>beta</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Mathematical-functions">Mathematical functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-BLAS_005fLIBS"><code>BLAS_LIBS</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Using-Makevars">Using Makevars</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-browser"><code>browser</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Browsing">Browsing</a></td></tr>
<tr><td colspan="4"> <hr></td></tr>
<tr><th><a name="Function-and-variable-index_vr_letter-C">C</a></th><td></td><td></td></tr>
<tr><td></td><td valign="top"><a href="#index-Calloc"><code>Calloc</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#User_002dcontrolled-memory">User-controlled memory</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-CAR"><code>CAR</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Calling-_002eExternal">Calling .External</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-CDR"><code>CDR</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Calling-_002eExternal">Calling .External</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-cgmin"><code>cgmin</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Optimization">Optimization</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-choose"><code>choose</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Mathematical-functions">Mathematical functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-choose-1"><code>choose</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Mathematical-functions">Mathematical functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-CITATION"><code>CITATION</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Package-subdirectories">Package subdirectories</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-CITATION-1"><code>CITATION</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Preparing-translations">Preparing translations</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-COPYRIGHTS"><code>COPYRIGHTS</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#The-DESCRIPTION-file">The DESCRIPTION file</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-COPYRIGHTS-1"><code>COPYRIGHTS</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Package-subdirectories">Package subdirectories</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-cospi"><code>cospi</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Numerical-Utilities">Numerical Utilities</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-cPsort"><code>cPsort</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Utility-functions">Utility functions</a></td></tr>
<tr><td colspan="4"> <hr></td></tr>
<tr><th><a name="Function-and-variable-index_vr_letter-D">D</a></th><td></td><td></td></tr>
<tr><td></td><td valign="top"><a href="#index-debug"><code>debug</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Debugging-R-code">Debugging R code</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-debugger"><code>debugger</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Debugging-R-code">Debugging R code</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-defineVar"><code>defineVar</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Finding-and-setting-variables">Finding and setting variables</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-digamma"><code>digamma</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Mathematical-functions">Mathematical functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-digamma-1"><code>digamma</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Mathematical-functions">Mathematical functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-dump_002eframes"><code>dump.frames</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Debugging-R-code">Debugging R code</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-duplicate"><code>duplicate</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Named-objects-and-copying">Named objects and copying</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-dyn_002eload"><code>dyn.load</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#dyn_002eload-and-dyn_002eunload">dyn.load and dyn.unload</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-dyn_002eunload"><code>dyn.unload</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#dyn_002eload-and-dyn_002eunload">dyn.load and dyn.unload</a></td></tr>
<tr><td colspan="4"> <hr></td></tr>
<tr><th><a name="Function-and-variable-index_vr_letter-E">E</a></th><td></td><td></td></tr>
<tr><td></td><td valign="top"><a href="#index-expm1"><code>expm1</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Numerical-Utilities">Numerical Utilities</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-export"><code>export</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Specifying-imports-and-exports">Specifying imports and exports</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-exportClasses"><code>exportClasses</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Namespaces-with-S4-classes-and-methods">Namespaces with S4 classes and methods</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-exportClassPattern"><code>exportClassPattern</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Namespaces-with-S4-classes-and-methods">Namespaces with S4 classes and methods</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-exportMethods"><code>exportMethods</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Namespaces-with-S4-classes-and-methods">Namespaces with S4 classes and methods</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-exportPattern"><code>exportPattern</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Specifying-imports-and-exports">Specifying imports and exports</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-exportPattern-1"><code>exportPattern</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Namespaces-with-S4-classes-and-methods">Namespaces with S4 classes and methods</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-exp_005frand"><code>exp_rand</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Random-numbers">Random numbers</a></td></tr>
<tr><td colspan="4"> <hr></td></tr>
<tr><th><a name="Function-and-variable-index_vr_letter-F">F</a></th><td></td><td></td></tr>
<tr><td></td><td valign="top"><a href="#index-FALSE"><code>FALSE</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Mathematical-constants">Mathematical constants</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-findInterval"><code>findInterval</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Utility-functions">Utility functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-findInterval2_0028double_002a"><code>findInterval2(double*</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Utility-functions">Utility functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-findVar"><code>findVar</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Finding-and-setting-variables">Finding and setting variables</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-FLIBS"><code>FLIBS</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Using-Makevars">Using Makevars</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-fmax2"><code>fmax2</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Numerical-Utilities">Numerical Utilities</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-fmin2"><code>fmin2</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Numerical-Utilities">Numerical Utilities</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-fprec"><code>fprec</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Numerical-Utilities">Numerical Utilities</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-Free"><code>Free</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#User_002dcontrolled-memory">User-controlled memory</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-fround"><code>fround</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Numerical-Utilities">Numerical Utilities</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-fsign"><code>fsign</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Numerical-Utilities">Numerical Utilities</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-ftrunc"><code>ftrunc</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Numerical-Utilities">Numerical Utilities</a></td></tr>
<tr><td colspan="4"> <hr></td></tr>
<tr><th><a name="Function-and-variable-index_vr_letter-G">G</a></th><td></td><td></td></tr>
<tr><td></td><td valign="top"><a href="#index-gammafn"><code>gammafn</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Mathematical-functions">Mathematical functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-gammafn-1"><code>gammafn</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Mathematical-functions">Mathematical functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-gctorture"><code>gctorture</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Using-gctorture">Using gctorture</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-getAttrib"><code>getAttrib</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Attributes">Attributes</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-getCharCE"><code>getCharCE</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Character-encoding-issues">Character encoding issues</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-GetRNGstate"><code>GetRNGstate</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Random-numbers">Random numbers</a></td></tr>
<tr><td colspan="4"> <hr></td></tr>
<tr><th><a name="Function-and-variable-index_vr_letter-I">I</a></th><td></td><td></td></tr>
<tr><td></td><td valign="top"><a href="#index-imax2"><code>imax2</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Numerical-Utilities">Numerical Utilities</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-imin2"><code>imin2</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Numerical-Utilities">Numerical Utilities</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-import"><code>import</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Specifying-imports-and-exports">Specifying imports and exports</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-importClassesFrom"><code>importClassesFrom</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Namespaces-with-S4-classes-and-methods">Namespaces with S4 classes and methods</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-importFrom"><code>importFrom</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Specifying-imports-and-exports">Specifying imports and exports</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-importMethodsFrom"><code>importMethodsFrom</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Namespaces-with-S4-classes-and-methods">Namespaces with S4 classes and methods</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-install"><code>install</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Attributes">Attributes</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-iPsort"><code>iPsort</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Utility-functions">Utility functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-ISNA"><code>ISNA</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Missing-and-special-values">Missing and special values</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-ISNA-1"><code>ISNA</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Missing-and-IEEE-values">Missing and IEEE values</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-ISNAN"><code>ISNAN</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Missing-and-special-values">Missing and special values</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-ISNAN-1"><code>ISNAN</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Missing-and-IEEE-values">Missing and IEEE values</a></td></tr>
<tr><td colspan="4"> <hr></td></tr>
<tr><th><a name="Function-and-variable-index_vr_letter-L">L</a></th><td></td><td></td></tr>
<tr><td></td><td valign="top"><a href="#index-LAPACK_005fLIBS"><code>LAPACK_LIBS</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Using-Makevars">Using Makevars</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-lbeta"><code>lbeta</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Mathematical-functions">Mathematical functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-lbeta-1"><code>lbeta</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Mathematical-functions">Mathematical functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-lbfgsb"><code>lbfgsb</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Optimization">Optimization</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-lchoose"><code>lchoose</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Mathematical-functions">Mathematical functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-lchoose-1"><code>lchoose</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Mathematical-functions">Mathematical functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-lgamma1p"><code>lgamma1p</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Numerical-Utilities">Numerical Utilities</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-lgammafn"><code>lgammafn</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Mathematical-functions">Mathematical functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-lgammafn-1"><code>lgammafn</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Mathematical-functions">Mathematical functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-library_002edynam"><code>library.dynam</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Package-subdirectories">Package subdirectories</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-library_002edynam-1"><code>library.dynam</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#dyn_002eload-and-dyn_002eunload">dyn.load and dyn.unload</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-log1p"><code>log1p</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Numerical-Utilities">Numerical Utilities</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-log1pexp"><code>log1pexp</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Numerical-Utilities">Numerical Utilities</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-log1pmx"><code>log1pmx</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Numerical-Utilities">Numerical Utilities</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-logspace_005fadd"><code>logspace_add</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Numerical-Utilities">Numerical Utilities</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-logspace_005fsub"><code>logspace_sub</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Numerical-Utilities">Numerical Utilities</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-logspace_005fsum"><code>logspace_sum</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Numerical-Utilities">Numerical Utilities</a></td></tr>
<tr><td colspan="4"> <hr></td></tr>
<tr><th><a name="Function-and-variable-index_vr_letter-M">M</a></th><td></td><td></td></tr>
<tr><td></td><td valign="top"><a href="#index-mkChar"><code>mkChar</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Handling-character-data">Handling character data</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-mkCharCE"><code>mkCharCE</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Character-encoding-issues">Character encoding issues</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-mkCharLen"><code>mkCharLen</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Handling-character-data">Handling character data</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-mkCharLenCE"><code>mkCharLenCE</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Character-encoding-issues">Character encoding issues</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-M_005fE"><code>M_E</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Mathematical-constants">Mathematical constants</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-M_005fPI"><code>M_PI</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Mathematical-constants">Mathematical constants</a></td></tr>
<tr><td colspan="4"> <hr></td></tr>
<tr><th><a name="Function-and-variable-index_vr_letter-N">N</a></th><td></td><td></td></tr>
<tr><td></td><td valign="top"><a href="#index-NA_005fREAL"><code>NA_REAL</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Missing-and-IEEE-values">Missing and IEEE values</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-NEWS_002eRd"><code>NEWS.Rd</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Package-subdirectories">Package subdirectories</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-nmmin"><code>nmmin</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Optimization">Optimization</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-norm_005frand"><code>norm_rand</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Random-numbers">Random numbers</a></td></tr>
<tr><td colspan="4"> <hr></td></tr>
<tr><th><a name="Function-and-variable-index_vr_letter-O">O</a></th><td></td><td></td></tr>
<tr><td></td><td valign="top"><a href="#index-OBJECTS"><code>OBJECTS</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Using-Makevars">Using Makevars</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-OBJECTS-1"><code>OBJECTS</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Creating-shared-objects">Creating shared objects</a></td></tr>
<tr><td colspan="4"> <hr></td></tr>
<tr><th><a name="Function-and-variable-index_vr_letter-P">P</a></th><td></td><td></td></tr>
<tr><td></td><td valign="top"><a href="#index-pentagamma"><code>pentagamma</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Mathematical-functions">Mathematical functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-pentagamma-1"><code>pentagamma</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Mathematical-functions">Mathematical functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-PKG_005fCFLAGS"><code>PKG_CFLAGS</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Creating-shared-objects">Creating shared objects</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-PKG_005fCPPFLAGS"><code>PKG_CPPFLAGS</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Creating-shared-objects">Creating shared objects</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-PKG_005fCXXFLAGS"><code>PKG_CXXFLAGS</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Creating-shared-objects">Creating shared objects</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-PKG_005fFCFLAGS"><code>PKG_FCFLAGS</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Creating-shared-objects">Creating shared objects</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-PKG_005fFFLAGS"><code>PKG_FFLAGS</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Creating-shared-objects">Creating shared objects</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-PKG_005fLIBS"><code>PKG_LIBS</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Creating-shared-objects">Creating shared objects</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-PKG_005fOBJCFLAGS"><code>PKG_OBJCFLAGS</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Creating-shared-objects">Creating shared objects</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-PKG_005fOBJCXXFLAGS"><code>PKG_OBJCXXFLAGS</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Creating-shared-objects">Creating shared objects</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-prompt"><code>prompt</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Rd-format">Rd format</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-PROTECT"><code>PROTECT</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Garbage-Collection">Garbage Collection</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-PROTECT_005fWITH_005fINDEX"><code>PROTECT_WITH_INDEX</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Garbage-Collection">Garbage Collection</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-psigamma"><code>psigamma</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Mathematical-functions">Mathematical functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-psigamma-1"><code>psigamma</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Mathematical-functions">Mathematical functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-PutRNGstate"><code>PutRNGstate</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Random-numbers">Random numbers</a></td></tr>
<tr><td colspan="4"> <hr></td></tr>
<tr><th><a name="Function-and-variable-index_vr_letter-Q">Q</a></th><td></td><td></td></tr>
<tr><td></td><td valign="top"><a href="#index-qsort3"><code>qsort3</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Utility-functions">Utility functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-qsort4"><code>qsort4</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Utility-functions">Utility functions</a></td></tr>
<tr><td colspan="4"> <hr></td></tr>
<tr><th><a name="Function-and-variable-index_vr_letter-R">R</a></th><td></td><td></td></tr>
<tr><td></td><td valign="top"><a href="#index-R-CMD-build"><code>R CMD build</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Building-package-tarballs">Building package tarballs</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-R-CMD-check"><code>R CMD check</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Checking-packages">Checking packages</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-R-CMD-config"><code>R CMD config</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Configure-and-cleanup">Configure and cleanup</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-R-CMD-Rd2pdf"><code>R CMD Rd2pdf</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Processing-documentation-files">Processing documentation files</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-R-CMD-Rdconv"><code>R CMD Rdconv</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Processing-documentation-files">Processing documentation files</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-R-CMD-SHLIB"><code>R CMD SHLIB</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Creating-shared-objects">Creating shared objects</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-R-CMD-Stangle"><code>R CMD Stangle</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Processing-documentation-files">Processing documentation files</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-R-CMD-Sweave"><code>R CMD Sweave</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Processing-documentation-files">Processing documentation files</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-Rdqagi"><code>Rdqagi</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Integration">Integration</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-Rdqags"><code>Rdqags</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Integration">Integration</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-Realloc"><code>Realloc</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#User_002dcontrolled-memory">User-controlled memory</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-recover"><code>recover</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Debugging-R-code">Debugging R code</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-reEnc"><code>reEnc</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Character-encoding-issues">Character encoding issues</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-REprintf"><code>REprintf</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Printing">Printing</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-REPROTECT"><code>REPROTECT</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Garbage-Collection">Garbage Collection</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-REvprintf"><code>REvprintf</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Printing">Printing</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-revsort"><code>revsort</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Utility-functions">Utility functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-Riconv"><code>Riconv</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Re_002dencoding">Re-encoding</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-Riconv_005fclose"><code>Riconv_close</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Re_002dencoding">Re-encoding</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-Riconv_005fopen"><code>Riconv_open</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Re_002dencoding">Re-encoding</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-Rprintf"><code>Rprintf</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Printing">Printing</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-Rprof"><code>Rprof</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Profiling-R-code-for-speed">Profiling R code for speed</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-Rprof-1"><code>Rprof</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Memory-statistics-from-Rprof">Memory statistics from Rprof</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-Rprofmem"><code>Rprofmem</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Tracking-memory-allocations">Tracking memory allocations</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-rPsort"><code>rPsort</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Utility-functions">Utility functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-rsort_005fwith_005findex"><code>rsort_with_index</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Utility-functions">Utility functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-Rvprintf"><code>Rvprintf</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Printing">Printing</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-R_005faddhistory"><code>R_addhistory</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Setting-R-callbacks">Setting R callbacks</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-R_005falloc"><code>R_alloc</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Transient-storage-allocation">Transient storage allocation</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-R_005fallocLD"><code>R_allocLD</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Transient-storage-allocation">Transient storage allocation</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-R_005fBusy"><code>R_Busy</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Setting-R-callbacks">Setting R callbacks</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-R_005fChooseFile"><code>R_ChooseFile</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Setting-R-callbacks">Setting R callbacks</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-R_005fCleanUp"><code>R_CleanUp</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Setting-R-callbacks">Setting R callbacks</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-R_005fClearErrConsole"><code>R_ClearErrConsole</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Setting-R-callbacks">Setting R callbacks</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-R_005fcsort"><code>R_csort</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Utility-functions">Utility functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-R_005fdataentry"><code>R_dataentry</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Setting-R-callbacks">Setting R callbacks</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-R_005fdataviewer"><code>R_dataviewer</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Setting-R-callbacks">Setting R callbacks</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-R_005fEditFile"><code>R_EditFile</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Setting-R-callbacks">Setting R callbacks</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-R_005fEditFiles"><code>R_EditFiles</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Setting-R-callbacks">Setting R callbacks</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-R_005fExpandFileName"><code>R_ExpandFileName</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Utility-functions">Utility functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-R_005fFINITE"><code>R_FINITE</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Missing-and-IEEE-values">Missing and IEEE values</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-R_005fFlushConsole"><code>R_FlushConsole</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Setting-R-callbacks">Setting R callbacks</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-R_005fforceSymbols"><code>R_forceSymbols</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Registering-native-routines">Registering native routines</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-R_005fGetCCallable"><code>R_GetCCallable</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Linking-to-native-routines-in-other-packages">Linking to native routines in other packages</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-R_005fGetCurrentSrcref"><code>R_GetCurrentSrcref</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Accessing-source-references">Accessing source references</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-R_005fGetSrcFilename"><code>R_GetSrcFilename</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Accessing-source-references">Accessing source references</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-R_005fINLINE"><code>R_INLINE</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Inlining-C-functions">Inlining C functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-R_005fIsNaN"><code>R_IsNaN</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Missing-and-IEEE-values">Missing and IEEE values</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-R_005fisort"><code>R_isort</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Utility-functions">Utility functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-R_005fLIBRARY_005fDIR"><code>R_LIBRARY_DIR</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Configure-and-cleanup">Configure and cleanup</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-R_005floadhistory"><code>R_loadhistory</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Setting-R-callbacks">Setting R callbacks</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-R_005fmax_005fcol"><code>R_max_col</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Utility-functions">Utility functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-R_005fNegInf"><code>R_NegInf</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Missing-and-IEEE-values">Missing and IEEE values</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-R_005forderVector"><code>R_orderVector</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Utility-functions">Utility functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-R_005forderVector1"><code>R_orderVector1</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Utility-functions">Utility functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-R_005fPACKAGE_005fDIR"><code>R_PACKAGE_DIR</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Configure-and-cleanup">Configure and cleanup</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-R_005fPACKAGE_005fNAME"><code>R_PACKAGE_NAME</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Configure-and-cleanup">Configure and cleanup</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-R_005fParseVector"><code>R_ParseVector</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Parsing-R-code-from-C">Parsing R code from C</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-R_005fPosInf"><code>R_PosInf</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Missing-and-IEEE-values">Missing and IEEE values</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-R_005fpow"><code>R_pow</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Numerical-Utilities">Numerical Utilities</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-R_005fpow_005fdi"><code>R_pow_di</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Numerical-Utilities">Numerical Utilities</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-R_005fPreserveObject"><code>R_PreserveObject</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Garbage-Collection">Garbage Collection</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-R_005fqsort"><code>R_qsort</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Utility-functions">Utility functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-R_005fqsort_005fI"><code>R_qsort_I</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Utility-functions">Utility functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-R_005fqsort_005fint"><code>R_qsort_int</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Utility-functions">Utility functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-R_005fqsort_005fint_005fI"><code>R_qsort_int_I</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Utility-functions">Utility functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-R_005fReadConsole"><code>R_ReadConsole</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Setting-R-callbacks">Setting R callbacks</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-R_005fRegisterCCallable"><code>R_RegisterCCallable</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Linking-to-native-routines-in-other-packages">Linking to native routines in other packages</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-R_005fregisterRoutines"><code>R_registerRoutines</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Registering-native-routines">Registering native routines</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-R_005fReleaseObject"><code>R_ReleaseObject</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Garbage-Collection">Garbage Collection</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-R_005fResetConsole"><code>R_ResetConsole</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Setting-R-callbacks">Setting R callbacks</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-R_005frsort"><code>R_rsort</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Utility-functions">Utility functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-R_005fsavehistory"><code>R_savehistory</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Setting-R-callbacks">Setting R callbacks</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-R_005fselectlist"><code>R_selectlist</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Setting-R-callbacks">Setting R callbacks</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-R_005fShowFiles"><code>R_ShowFiles</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Setting-R-callbacks">Setting R callbacks</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-R_005fShowMessage"><code>R_ShowMessage</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Setting-R-callbacks">Setting R callbacks</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-R_005fSrcref"><code>R_Srcref</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Accessing-source-references">Accessing source references</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-R_005fSuicide"><code>R_Suicide</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Setting-R-callbacks">Setting R callbacks</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-R_005ftmpnam"><code>R_tmpnam</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Utility-functions">Utility functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-R_005ftmpnam2"><code>R_tmpnam2</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Utility-functions">Utility functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-R_005fuseDynamicSymbols"><code>R_useDynamicSymbols</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Registering-native-routines">Registering native routines</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-R_005fVersion"><code>R_Version</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Platform-and-version-information">Platform and version information</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-R_005fWriteConsole"><code>R_WriteConsole</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Setting-R-callbacks">Setting R callbacks</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-R_005fWriteConsoleEx"><code>R_WriteConsoleEx</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Setting-R-callbacks">Setting R callbacks</a></td></tr>
<tr><td colspan="4"> <hr></td></tr>
<tr><th><a name="Function-and-variable-index_vr_letter-S">S</a></th><td></td><td></td></tr>
<tr><td></td><td valign="top"><a href="#index-S3method"><code>S3method</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Registering-S3-methods">Registering S3 methods</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-SAFE_005fFFLAGS"><code>SAFE_FFLAGS</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Using-Makevars">Using Makevars</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-samin"><code>samin</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Optimization">Optimization</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-seed_005fin"><code>seed_in</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Random-numbers">Random numbers</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-seed_005fout"><code>seed_out</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Random-numbers">Random numbers</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-setAttrib"><code>setAttrib</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Attributes">Attributes</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-setVar"><code>setVar</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Finding-and-setting-variables">Finding and setting variables</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-sign"><code>sign</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Numerical-Utilities">Numerical Utilities</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-sinpi"><code>sinpi</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Numerical-Utilities">Numerical Utilities</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-summaryRprof"><code>summaryRprof</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Memory-statistics-from-Rprof">Memory statistics from Rprof</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-system"><code>system</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Operating-system-access">Operating system access</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-system_002etime"><code>system.time</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Operating-system-access">Operating system access</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-system2"><code>system2</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Operating-system-access">Operating system access</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-S_005falloc"><code>S_alloc</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Transient-storage-allocation">Transient storage allocation</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-S_005frealloc"><code>S_realloc</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Transient-storage-allocation">Transient storage allocation</a></td></tr>
<tr><td colspan="4"> <hr></td></tr>
<tr><th><a name="Function-and-variable-index_vr_letter-T">T</a></th><td></td><td></td></tr>
<tr><td></td><td valign="top"><a href="#index-tanpi"><code>tanpi</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Numerical-Utilities">Numerical Utilities</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-tetragamma"><code>tetragamma</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Mathematical-functions">Mathematical functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-tetragamma-1"><code>tetragamma</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Mathematical-functions">Mathematical functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-trace"><code>trace</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Debugging-R-code">Debugging R code</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-traceback"><code>traceback</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Debugging-R-code">Debugging R code</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-tracemem"><code>tracemem</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Tracing-copies-of-an-object">Tracing copies of an object</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-translateChar"><code>translateChar</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Character-encoding-issues">Character encoding issues</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-translateCharUTF8"><code>translateCharUTF8</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Character-encoding-issues">Character encoding issues</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-trigamma"><code>trigamma</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Mathematical-functions">Mathematical functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-trigamma-1"><code>trigamma</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Mathematical-functions">Mathematical functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-TRUE"><code>TRUE</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Mathematical-constants">Mathematical constants</a></td></tr>
<tr><td colspan="4"> <hr></td></tr>
<tr><th><a name="Function-and-variable-index_vr_letter-U">U</a></th><td></td><td></td></tr>
<tr><td></td><td valign="top"><a href="#index-undebug"><code>undebug</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Debugging-R-code">Debugging R code</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-unif_005frand"><code>unif_rand</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Random-numbers">Random numbers</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-UNPROTECT"><code>UNPROTECT</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Garbage-Collection">Garbage Collection</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-UNPROTECT_005fPTR"><code>UNPROTECT_PTR</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Garbage-Collection">Garbage Collection</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-untracemem"><code>untracemem</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Tracing-copies-of-an-object">Tracing copies of an object</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-useDynLib"><code>useDynLib</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#useDynLib">useDynLib</a></td></tr>
<tr><td colspan="4"> <hr></td></tr>
<tr><th><a name="Function-and-variable-index_vr_letter-V">V</a></th><td></td><td></td></tr>
<tr><td></td><td valign="top"><a href="#index-vmaxget"><code>vmaxget</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Transient-storage-allocation">Transient storage allocation</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-vmaxset"><code>vmaxset</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Transient-storage-allocation">Transient storage allocation</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-vmmin"><code>vmmin</code></a>:</td><td>&nbsp;</td><td valign="top"><a href="#Optimization">Optimization</a></td></tr>
<tr><td colspan="4"> <hr></td></tr>
</table>
<table summary=""><tr><th valign="top">Jump to: &nbsp; </th><td><a class="summary-letter" href="#Function-and-variable-index_vr_symbol-1"><b>.</b></a>
 &nbsp; 
<a class="summary-letter" href="#Function-and-variable-index_vr_symbol-2"><b>\</b></a>
 &nbsp; 
<br>
<a class="summary-letter" href="#Function-and-variable-index_vr_letter-A"><b>A</b></a>
 &nbsp; 
<a class="summary-letter" href="#Function-and-variable-index_vr_letter-B"><b>B</b></a>
 &nbsp; 
<a class="summary-letter" href="#Function-and-variable-index_vr_letter-C"><b>C</b></a>
 &nbsp; 
<a class="summary-letter" href="#Function-and-variable-index_vr_letter-D"><b>D</b></a>
 &nbsp; 
<a class="summary-letter" href="#Function-and-variable-index_vr_letter-E"><b>E</b></a>
 &nbsp; 
<a class="summary-letter" href="#Function-and-variable-index_vr_letter-F"><b>F</b></a>
 &nbsp; 
<a class="summary-letter" href="#Function-and-variable-index_vr_letter-G"><b>G</b></a>
 &nbsp; 
<a class="summary-letter" href="#Function-and-variable-index_vr_letter-I"><b>I</b></a>
 &nbsp; 
<a class="summary-letter" href="#Function-and-variable-index_vr_letter-L"><b>L</b></a>
 &nbsp; 
<a class="summary-letter" href="#Function-and-variable-index_vr_letter-M"><b>M</b></a>
 &nbsp; 
<a class="summary-letter" href="#Function-and-variable-index_vr_letter-N"><b>N</b></a>
 &nbsp; 
<a class="summary-letter" href="#Function-and-variable-index_vr_letter-O"><b>O</b></a>
 &nbsp; 
<a class="summary-letter" href="#Function-and-variable-index_vr_letter-P"><b>P</b></a>
 &nbsp; 
<a class="summary-letter" href="#Function-and-variable-index_vr_letter-Q"><b>Q</b></a>
 &nbsp; 
<a class="summary-letter" href="#Function-and-variable-index_vr_letter-R"><b>R</b></a>
 &nbsp; 
<a class="summary-letter" href="#Function-and-variable-index_vr_letter-S"><b>S</b></a>
 &nbsp; 
<a class="summary-letter" href="#Function-and-variable-index_vr_letter-T"><b>T</b></a>
 &nbsp; 
<a class="summary-letter" href="#Function-and-variable-index_vr_letter-U"><b>U</b></a>
 &nbsp; 
<a class="summary-letter" href="#Function-and-variable-index_vr_letter-V"><b>V</b></a>
 &nbsp; 
</td></tr></table>

<hr>
<a name="Concept-index"></a>
<div class="header">
<p>
Previous: <a href="#Function-and-variable-index" accesskey="p" rel="prev">Function and variable index</a>, Up: <a href="#Top" accesskey="u" rel="up">Top</a> &nbsp; [<a href="#SEC_Contents" title="Table of contents" rel="contents">Contents</a>][<a href="#Function-and-variable-index" title="Index" rel="index">Index</a>]</p>
</div>
<a name="Concept-index-1"></a>
<h2 class="unnumbered">Concept index</h2>

<table summary=""><tr><th valign="top">Jump to: &nbsp; </th><td><a class="summary-letter" href="#Concept-index_cp_symbol-1"><b>.</b></a>
 &nbsp; 
<a class="summary-letter" href="#Concept-index_cp_symbol-2"><b>\</b></a>
 &nbsp; 
<br>
<a class="summary-letter" href="#Concept-index_cp_letter-A"><b>A</b></a>
 &nbsp; 
<a class="summary-letter" href="#Concept-index_cp_letter-B"><b>B</b></a>
 &nbsp; 
<a class="summary-letter" href="#Concept-index_cp_letter-C"><b>C</b></a>
 &nbsp; 
<a class="summary-letter" href="#Concept-index_cp_letter-D"><b>D</b></a>
 &nbsp; 
<a class="summary-letter" href="#Concept-index_cp_letter-E"><b>E</b></a>
 &nbsp; 
<a class="summary-letter" href="#Concept-index_cp_letter-F"><b>F</b></a>
 &nbsp; 
<a class="summary-letter" href="#Concept-index_cp_letter-G"><b>G</b></a>
 &nbsp; 
<a class="summary-letter" href="#Concept-index_cp_letter-H"><b>H</b></a>
 &nbsp; 
<a class="summary-letter" href="#Concept-index_cp_letter-I"><b>I</b></a>
 &nbsp; 
<a class="summary-letter" href="#Concept-index_cp_letter-L"><b>L</b></a>
 &nbsp; 
<a class="summary-letter" href="#Concept-index_cp_letter-M"><b>M</b></a>
 &nbsp; 
<a class="summary-letter" href="#Concept-index_cp_letter-N"><b>N</b></a>
 &nbsp; 
<a class="summary-letter" href="#Concept-index_cp_letter-O"><b>O</b></a>
 &nbsp; 
<a class="summary-letter" href="#Concept-index_cp_letter-P"><b>P</b></a>
 &nbsp; 
<a class="summary-letter" href="#Concept-index_cp_letter-R"><b>R</b></a>
 &nbsp; 
<a class="summary-letter" href="#Concept-index_cp_letter-S"><b>S</b></a>
 &nbsp; 
<a class="summary-letter" href="#Concept-index_cp_letter-T"><b>T</b></a>
 &nbsp; 
<a class="summary-letter" href="#Concept-index_cp_letter-U"><b>U</b></a>
 &nbsp; 
<a class="summary-letter" href="#Concept-index_cp_letter-V"><b>V</b></a>
 &nbsp; 
<a class="summary-letter" href="#Concept-index_cp_letter-W"><b>W</b></a>
 &nbsp; 
<a class="summary-letter" href="#Concept-index_cp_letter-Z"><b>Z</b></a>
 &nbsp; 
</td></tr></table>
<table summary="" class="index-cp" border="0">
<tr><td></td><th align="left">Index Entry</th><td>&nbsp;</td><th align="left"> Section</th></tr>
<tr><td colspan="4"> <hr></td></tr>
<tr><th><a name="Concept-index_cp_symbol-1">.</a></th><td></td><td></td></tr>
<tr><td></td><td valign="top"><a href="#index-_002einstall_005fextras-file">.install_extras file</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Writing-package-vignettes">Writing package vignettes</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_002eRbuildignore-file">.Rbuildignore file</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Building-package-tarballs">Building package tarballs</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-_002eRinstignore-file">.Rinstignore file</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Package-subdirectories">Package subdirectories</a></td></tr>
<tr><td colspan="4"> <hr></td></tr>
<tr><th><a name="Concept-index_cp_symbol-2">\</a></th><td></td><td></td></tr>
<tr><td></td><td valign="top"><a href="#index-_005clinkS4class">\linkS4class</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Cross_002dreferences">Cross-references</a></td></tr>
<tr><td colspan="4"> <hr></td></tr>
<tr><th><a name="Concept-index_cp_letter-A">A</a></th><td></td><td></td></tr>
<tr><td></td><td valign="top"><a href="#index-Allocating-storage">Allocating storage</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Allocating-storage">Allocating storage</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-Attributes">Attributes</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Attributes">Attributes</a></td></tr>
<tr><td colspan="4"> <hr></td></tr>
<tr><th><a name="Concept-index_cp_letter-B">B</a></th><td></td><td></td></tr>
<tr><td></td><td valign="top"><a href="#index-Bessel-functions">Bessel functions</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Mathematical-functions">Mathematical functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-Beta-function">Beta function</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Mathematical-functions">Mathematical functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-Building-binary-packages">Building binary packages</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Building-binary-packages">Building binary packages</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-Building-source-packages">Building source packages</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Building-package-tarballs">Building package tarballs</a></td></tr>
<tr><td colspan="4"> <hr></td></tr>
<tr><th><a name="Concept-index_cp_letter-C">C</a></th><td></td><td></td></tr>
<tr><td></td><td valign="top"><a href="#index-C_002b_002b-code_002c-interfacing">C++ code, interfacing</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Interfacing-C_002b_002b-code">Interfacing C++ code</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-Calling-C-from-FORTRAN-and-vice-versa">Calling C from FORTRAN and vice versa</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Calling-C-from-FORTRAN-and-vice-versa">Calling C from FORTRAN and vice versa</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-Checking-packages">Checking packages</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Checking-packages">Checking packages</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-citation">citation</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Package-subdirectories">Package subdirectories</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-citation-1">citation</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Preparing-translations">Preparing translations</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-Classes">Classes</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Classes">Classes</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-cleanup-file">cleanup file</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Package-structure">Package structure</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-conditionals">conditionals</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Conditional-text">Conditional text</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-configure-file">configure file</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Package-structure">Package structure</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-Copying-objects">Copying objects</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Named-objects-and-copying">Named objects and copying</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-CRAN">CRAN</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Creating-R-packages">Creating R packages</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-Creating-packages">Creating packages</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Creating-R-packages">Creating R packages</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-Creating-shared-objects">Creating shared objects</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Creating-shared-objects">Creating shared objects</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-Cross_002dreferences-in-documentation">Cross-references in documentation</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Cross_002dreferences">Cross-references</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-cumulative-hazard">cumulative hazard</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Distribution-functions">Distribution functions</a></td></tr>
<tr><td colspan="4"> <hr></td></tr>
<tr><th><a name="Concept-index_cp_letter-D">D</a></th><td></td><td></td></tr>
<tr><td></td><td valign="top"><a href="#index-Debugging">Debugging</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Debugging-compiled-code">Debugging compiled code</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-DESCRIPTION-file">DESCRIPTION file</a>:</td><td>&nbsp;</td><td valign="top"><a href="#The-DESCRIPTION-file">The DESCRIPTION file</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-Details-of-R-types">Details of R types</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Details-of-R-types">Details of R types</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-Distribution-functions-from-C">Distribution functions from C</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Distribution-functions">Distribution functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-Documentation_002c-writing">Documentation, writing</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Writing-R-documentation-files">Writing R documentation files</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-Dynamic-loading">Dynamic loading</a>:</td><td>&nbsp;</td><td valign="top"><a href="#dyn_002eload-and-dyn_002eunload">dyn.load and dyn.unload</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-dynamic-pages">dynamic pages</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Dynamic-pages">Dynamic pages</a></td></tr>
<tr><td colspan="4"> <hr></td></tr>
<tr><th><a name="Concept-index_cp_letter-E">E</a></th><td></td><td></td></tr>
<tr><td></td><td valign="top"><a href="#index-Editing-Rd-files">Editing Rd files</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Editing-Rd-files">Editing Rd files</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-encoding">encoding</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Encoding">Encoding</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-Error-handling-from-C">Error handling from C</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Error-handling">Error handling</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-Error-handling-from-FORTRAN">Error handling from FORTRAN</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Error-handling-from-FORTRAN">Error handling from FORTRAN</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-Evaluating-R-expressions-from-C">Evaluating R expressions from C</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Evaluating-R-expressions-from-C">Evaluating R expressions from C</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-external-pointer">external pointer</a>:</td><td>&nbsp;</td><td valign="top"><a href="#External-pointers-and-weak-references">External pointers and weak references</a></td></tr>
<tr><td colspan="4"> <hr></td></tr>
<tr><th><a name="Concept-index_cp_letter-F">F</a></th><td></td><td></td></tr>
<tr><td></td><td valign="top"><a href="#index-Figures-in-documentation">Figures in documentation</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Figures">Figures</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-finalizer">finalizer</a>:</td><td>&nbsp;</td><td valign="top"><a href="#External-pointers-and-weak-references">External pointers and weak references</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-Finding-variables">Finding variables</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Finding-and-setting-variables">Finding and setting variables</a></td></tr>
<tr><td colspan="4"> <hr></td></tr>
<tr><th><a name="Concept-index_cp_letter-G">G</a></th><td></td><td></td></tr>
<tr><td></td><td valign="top"><a href="#index-Gamma-function">Gamma function</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Mathematical-functions">Mathematical functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-Garbage-collection">Garbage collection</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Garbage-Collection">Garbage Collection</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-Generic-functions">Generic functions</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Generic-functions-and-methods">Generic functions and methods</a></td></tr>
<tr><td colspan="4"> <hr></td></tr>
<tr><th><a name="Concept-index_cp_letter-H">H</a></th><td></td><td></td></tr>
<tr><td></td><td valign="top"><a href="#index-handling-character-data">handling character data</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Handling-character-data">Handling character data</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-Handling-lists">Handling lists</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Handling-lists">Handling lists</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-Handling-R-objects-in-C">Handling R objects in C</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Handling-R-objects-in-C">Handling R objects in C</a></td></tr>
<tr><td colspan="4"> <hr></td></tr>
<tr><th><a name="Concept-index_cp_letter-I">I</a></th><td></td><td></td></tr>
<tr><td></td><td valign="top"><a href="#index-IEEE-special-values">IEEE special values</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Missing-and-special-values">Missing and special values</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-IEEE-special-values-1">IEEE special values</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Missing-and-IEEE-values">Missing and IEEE values</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-INDEX-file">INDEX file</a>:</td><td>&nbsp;</td><td valign="top"><a href="#The-INDEX-file">The INDEX file</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-Indices">Indices</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Indices">Indices</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-Inspecting-R-objects-when-debugging">Inspecting R objects when debugging</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Inspecting-R-objects">Inspecting R objects</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-integration">integration</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Integration">Integration</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-Interfaces-to-compiled-code">Interfaces to compiled code</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Interface-functions-_002eC-and-_002eFortran">Interface functions .C and .Fortran</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-Interfaces-to-compiled-code-1">Interfaces to compiled code</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Interface-functions-_002eCall-and-_002eExternal">Interface functions .Call and .External</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-Interfacing-C_002b_002b-code">Interfacing C++ code</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Interfacing-C_002b_002b-code">Interfacing C++ code</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-Interrupts">Interrupts</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Allowing-interrupts">Allowing interrupts</a></td></tr>
<tr><td colspan="4"> <hr></td></tr>
<tr><th><a name="Concept-index_cp_letter-L">L</a></th><td></td><td></td></tr>
<tr><td></td><td valign="top"><a href="#index-LICENCE-file">LICENCE file</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Licensing">Licensing</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-LICENSE-file">LICENSE file</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Licensing">Licensing</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-Lists-and-tables-in-documentation">Lists and tables in documentation</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Lists-and-tables">Lists and tables</a></td></tr>
<tr><td colspan="4"> <hr></td></tr>
<tr><th><a name="Concept-index_cp_letter-M">M</a></th><td></td><td></td></tr>
<tr><td></td><td valign="top"><a href="#index-Marking-text-in-documentation">Marking text in documentation</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Marking-text">Marking text</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-Mathematics-in-documentation">Mathematics in documentation</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Mathematics">Mathematics</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-Memory-allocation-from-C">Memory allocation from C</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Memory-allocation">Memory allocation</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-Memory-use">Memory use</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Profiling-R-code-for-memory-use">Profiling R code for memory use</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-Method-functions">Method functions</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Generic-functions-and-methods">Generic functions and methods</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-Missing-values">Missing values</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Missing-and-special-values">Missing and special values</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-Missing-values-1">Missing values</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Missing-and-IEEE-values">Missing and IEEE values</a></td></tr>
<tr><td colspan="4"> <hr></td></tr>
<tr><th><a name="Concept-index_cp_letter-N">N</a></th><td></td><td></td></tr>
<tr><td></td><td valign="top"><a href="#index-namespaces">namespaces</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Package-namespaces">Package namespaces</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-news">news</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Package-subdirectories">Package subdirectories</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-Numerical-analysis-subroutines-from-C">Numerical analysis subroutines from C</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Numerical-analysis-subroutines">Numerical analysis subroutines</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-Numerical-derivatives">Numerical derivatives</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Calculating-numerical-derivatives">Calculating numerical derivatives</a></td></tr>
<tr><td colspan="4"> <hr></td></tr>
<tr><th><a name="Concept-index_cp_letter-O">O</a></th><td></td><td></td></tr>
<tr><td></td><td valign="top"><a href="#index-OpenMP">OpenMP</a>:</td><td>&nbsp;</td><td valign="top"><a href="#OpenMP-support">OpenMP support</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-OpenMP-1">OpenMP</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Platform-and-version-information">Platform and version information</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-Operating-system-access">Operating system access</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Operating-system-access">Operating system access</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-optimization">optimization</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Optimization">Optimization</a></td></tr>
<tr><td colspan="4"> <hr></td></tr>
<tr><th><a name="Concept-index_cp_letter-P">P</a></th><td></td><td></td></tr>
<tr><td></td><td valign="top"><a href="#index-Package-builder">Package builder</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Building-package-tarballs">Building package tarballs</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-Package-structure">Package structure</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Package-structure">Package structure</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-Package-subdirectories">Package subdirectories</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Package-subdirectories">Package subdirectories</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-Packages">Packages</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Creating-R-packages">Creating R packages</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-Parsing-R-code-from-C">Parsing R code from C</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Parsing-R-code-from-C">Parsing R code from C</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-Platform_002dspecific-documentation">Platform-specific documentation</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Platform_002dspecific-sections">Platform-specific sections</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-Printing-from-C">Printing from C</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Printing">Printing</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-Printing-from-FORTRAN">Printing from FORTRAN</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Printing-from-FORTRAN">Printing from FORTRAN</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-Processing-Rd-format">Processing Rd format</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Processing-documentation-files">Processing documentation files</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-Profiling">Profiling</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Profiling-R-code-for-speed">Profiling R code for speed</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-Profiling-1">Profiling</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Profiling-R-code-for-memory-use">Profiling R code for memory use</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-Profiling-2">Profiling</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Profiling-compiled-code">Profiling compiled code</a></td></tr>
<tr><td colspan="4"> <hr></td></tr>
<tr><th><a name="Concept-index_cp_letter-R">R</a></th><td></td><td></td></tr>
<tr><td></td><td valign="top"><a href="#index-Random-numbers-in-C">Random numbers in C</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Random-numbers">Random numbers</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-Random-numbers-in-C-1">Random numbers in C</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Distribution-functions">Distribution functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-Random-numbers-in-FORTRAN">Random numbers in FORTRAN</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Calling-C-from-FORTRAN-and-vice-versa">Calling C from FORTRAN and vice versa</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-Registering-native-routines">Registering native routines</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Registering-native-routines">Registering native routines</a></td></tr>
<tr><td colspan="4"> <hr></td></tr>
<tr><th><a name="Concept-index_cp_letter-S">S</a></th><td></td><td></td></tr>
<tr><td></td><td valign="top"><a href="#index-Setting-variables">Setting variables</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Finding-and-setting-variables">Finding and setting variables</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-Sort-functions-from-C">Sort functions from C</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Utility-functions">Utility functions</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-Sweave">Sweave</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Writing-package-vignettes">Writing package vignettes</a></td></tr>
<tr><td colspan="4"> <hr></td></tr>
<tr><th><a name="Concept-index_cp_letter-T">T</a></th><td></td><td></td></tr>
<tr><td></td><td valign="top"><a href="#index-tarballs">tarballs</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Building-package-tarballs">Building package tarballs</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-Tidying-R-code">Tidying R code</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Tidying-R-code">Tidying R code</a></td></tr>
<tr><td colspan="4"> <hr></td></tr>
<tr><th><a name="Concept-index_cp_letter-U">U</a></th><td></td><td></td></tr>
<tr><td></td><td valign="top"><a href="#index-user_002ddefined-macros">user-defined macros</a>:</td><td>&nbsp;</td><td valign="top"><a href="#User_002ddefined-macros">User-defined macros</a></td></tr>
<tr><td colspan="4"> <hr></td></tr>
<tr><th><a name="Concept-index_cp_letter-V">V</a></th><td></td><td></td></tr>
<tr><td></td><td valign="top"><a href="#index-Version-information-from-C">Version information from C</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Platform-and-version-information">Platform and version information</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-vignettes">vignettes</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Writing-package-vignettes">Writing package vignettes</a></td></tr>
<tr><td></td><td valign="top"><a href="#index-Visibility">Visibility</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Controlling-visibility">Controlling visibility</a></td></tr>
<tr><td colspan="4"> <hr></td></tr>
<tr><th><a name="Concept-index_cp_letter-W">W</a></th><td></td><td></td></tr>
<tr><td></td><td valign="top"><a href="#index-weak-reference">weak reference</a>:</td><td>&nbsp;</td><td valign="top"><a href="#External-pointers-and-weak-references">External pointers and weak references</a></td></tr>
<tr><td colspan="4"> <hr></td></tr>
<tr><th><a name="Concept-index_cp_letter-Z">Z</a></th><td></td><td></td></tr>
<tr><td></td><td valign="top"><a href="#index-Zero_002dfinding">Zero-finding</a>:</td><td>&nbsp;</td><td valign="top"><a href="#Zero_002dfinding">Zero-finding</a></td></tr>
<tr><td colspan="4"> <hr></td></tr>
</table>
<table summary=""><tr><th valign="top">Jump to: &nbsp; </th><td><a class="summary-letter" href="#Concept-index_cp_symbol-1"><b>.</b></a>
 &nbsp; 
<a class="summary-letter" href="#Concept-index_cp_symbol-2"><b>\</b></a>
 &nbsp; 
<br>
<a class="summary-letter" href="#Concept-index_cp_letter-A"><b>A</b></a>
 &nbsp; 
<a class="summary-letter" href="#Concept-index_cp_letter-B"><b>B</b></a>
 &nbsp; 
<a class="summary-letter" href="#Concept-index_cp_letter-C"><b>C</b></a>
 &nbsp; 
<a class="summary-letter" href="#Concept-index_cp_letter-D"><b>D</b></a>
 &nbsp; 
<a class="summary-letter" href="#Concept-index_cp_letter-E"><b>E</b></a>
 &nbsp; 
<a class="summary-letter" href="#Concept-index_cp_letter-F"><b>F</b></a>
 &nbsp; 
<a class="summary-letter" href="#Concept-index_cp_letter-G"><b>G</b></a>
 &nbsp; 
<a class="summary-letter" href="#Concept-index_cp_letter-H"><b>H</b></a>
 &nbsp; 
<a class="summary-letter" href="#Concept-index_cp_letter-I"><b>I</b></a>
 &nbsp; 
<a class="summary-letter" href="#Concept-index_cp_letter-L"><b>L</b></a>
 &nbsp; 
<a class="summary-letter" href="#Concept-index_cp_letter-M"><b>M</b></a>
 &nbsp; 
<a class="summary-letter" href="#Concept-index_cp_letter-N"><b>N</b></a>
 &nbsp; 
<a class="summary-letter" href="#Concept-index_cp_letter-O"><b>O</b></a>
 &nbsp; 
<a class="summary-letter" href="#Concept-index_cp_letter-P"><b>P</b></a>
 &nbsp; 
<a class="summary-letter" href="#Concept-index_cp_letter-R"><b>R</b></a>
 &nbsp; 
<a class="summary-letter" href="#Concept-index_cp_letter-S"><b>S</b></a>
 &nbsp; 
<a class="summary-letter" href="#Concept-index_cp_letter-T"><b>T</b></a>
 &nbsp; 
<a class="summary-letter" href="#Concept-index_cp_letter-U"><b>U</b></a>
 &nbsp; 
<a class="summary-letter" href="#Concept-index_cp_letter-V"><b>V</b></a>
 &nbsp; 
<a class="summary-letter" href="#Concept-index_cp_letter-W"><b>W</b></a>
 &nbsp; 
<a class="summary-letter" href="#Concept-index_cp_letter-Z"><b>Z</b></a>
 &nbsp; 
</td></tr></table>

<div class="footnote">
<hr>
<h4 class="footnotes-heading">Footnotes</h4>

<h3><a name="FOOT1" href="#DOCF1">(1)</a></h3>
<p>although this is a persistent
mis-usage.  It seems to stem from S, whose analogues of R&rsquo;s packages
were officially known as <em>library sections</em> and later as
<em>chapters</em>, but almost always referred to as <em>libraries</em>.</p>
<h3><a name="FOOT2" href="#DOCF2">(2)</a></h3>
<p>This
seems to be commonly used for a file in &lsquo;markdown&rsquo; format.  Be aware
that most users of R will not know that, nor know how to view such a
file: platforms such as macOS and Windows do not have a default viewer
set in their file associations.  The <acronym>CRAN</acronym> package web pages
render such files in <acronym>HTML</acronym>: the converter used expects the file to be
encoded in UTF-8.</p>
<h3><a name="FOOT3" href="#DOCF3">(3)</a></h3>
<p>currently, top-level files
<samp>.Rbuildignore</samp> and <samp>.Rinstignore</samp>, and
<samp>vignettes/.install_extras</samp>.</p>
<h3><a name="FOOT4" href="#DOCF4">(4)</a></h3>
<p>false positives are possible, but only a handful have been
seen so far.</p>
<h3><a name="FOOT5" href="#DOCF5">(5)</a></h3>
<p>at least if this
is done in a locale which matches the package encoding.</p>
<h3><a name="FOOT6" href="#DOCF6">(6)</a></h3>
<p>and
required by <acronym>CRAN</acronym>, so checked by <code>R CMD check
--as-cran</code>.</p>
<h3><a name="FOOT7" href="#DOCF7">(7)</a></h3>
<p>But it is checked for Open Source packages
by <code>R CMD check --as-cran</code>.</p>
<h3><a name="FOOT8" href="#DOCF8">(8)</a></h3>
<p>Duplicate definitions may
trigger a warning: see <a href="#User_002ddefined-macros">User-defined macros</a>.</p>
<h3><a name="FOOT9" href="#DOCF9">(9)</a></h3>
<p>even one wrapped in <code>\donttest</code>.</p>
<h3><a name="FOOT10" href="#DOCF10">(10)</a></h3>
<p>This includes all packages
directly called by <code>library</code> and <code>require</code> calls, as well as
data obtained <em>via</em> <code>data(theirdata, package = &quot;somepkg&quot;)</code>
calls: <code>R CMD check</code> will warn about all of these.  But there
are subtler uses which it will not detect: e.g. if package A uses
package B and makes use of functionality in package B which uses package
C which package B suggests or enhances, then package C needs to be in
the &lsquo;<samp>Suggests</samp>&rsquo; list for package A.  Nor will undeclared uses in
included files be reported, nor unconditional uses of packages listed
under &lsquo;<samp>Enhances</samp>&rsquo;.</p>
<h3><a name="FOOT11" href="#DOCF11">(11)</a></h3>
<p>Extensions
<samp>.S</samp> and <samp>.s</samp> arise from code originally written for S(-PLUS),
but are commonly used for assembler code.  Extension <samp>.q</samp> was used
for S, which at one time was tentatively called QPE.</p>
<h3><a name="FOOT12" href="#DOCF12">(12)</a></h3>
<p>but they should be in the encoding
declared in the <samp>DESCRIPTION</samp> file.</p>
<h3><a name="FOOT13" href="#DOCF13">(13)</a></h3>
<p>This is true for OSes which
implement the &lsquo;<samp>C</samp>&rsquo; locale: Windows&rsquo; idea of the &lsquo;<samp>C</samp>&rsquo; locale uses
the WinAnsi charset.</p>
<h3><a name="FOOT14" href="#DOCF14">(14)</a></h3>
<p>More precisely, they can
contain the English alphanumeric characters and the symbols
&lsquo;<samp>$ - _ . + ! ' ( ) , ; &nbsp;= &amp;</samp>&rsquo;.</p>
<h3><a name="FOOT15" href="#DOCF15">(15)</a></h3>
<p>Note that Ratfor is not supported.
If you have Ratfor source code, you need to convert it to FORTRAN.  Only
FORTRAN 77 (which we write in upper case) is supported on all platforms,
but most also support Fortran-95 (for which we use title case).  If you
want to ship Ratfor source files, please do so in a subdirectory of
<samp>src</samp> and not in the main subdirectory.</p>
<h3><a name="FOOT16" href="#DOCF16">(16)</a></h3>
<p>either or both of which may not be supported on particular
platforms</p>
<h3><a name="FOOT17" href="#DOCF17">(17)</a></h3>
<p>Using <samp>.hpp</samp> is not guaranteed to be
portable.</p>
<h3><a name="FOOT18" href="#DOCF18">(18)</a></h3>
<p>There
is also &lsquo;<samp>__APPLE_CC__</samp>&rsquo;, but that indicates a compiler with
Apple-specific features, not the OS.  It is used in
<samp>Rinlinedfuns.h</samp>.</p>
<h3><a name="FOOT19" href="#DOCF19">(19)</a></h3>
<p>the POSIX
terminology, called &lsquo;make variables&rsquo; by GNU make.</p>
<h3><a name="FOOT20" href="#DOCF20">(20)</a></h3>
<p>on all platforms from R 3.1.0</p>
<h3><a name="FOOT21" href="#DOCF21">(21)</a></h3>
<p>The best way to generate such a
file is to copy the <samp>.Rout</samp> from a successful run of <code>R CMD
check</code>.  If you want to generate it separately, do run R with options
<samp>--vanilla --slave</samp> and with environment variable
<code>LANGUAGE=en</code> set to get messages in English.  Be careful not to use
output with the option <samp>--timings</samp> (and note that
<samp>--as-cran</samp> sets it).</p>
<h3><a name="FOOT22" href="#DOCF22">(22)</a></h3>
<p>e.g.
<a href="https://tools.ietf.org/html/rfc4180">https://tools.ietf.org/html/rfc4180</a>.</p>
<h3><a name="FOOT23" href="#DOCF23">(23)</a></h3>
<p>People who have trouble with
case are advised to use <samp>.rda</samp> as a common error is to refer to
<samp>abc.RData</samp> as <samp>abc.Rdata</samp>!</p>
<h3><a name="FOOT24" href="#DOCF24">(24)</a></h3>
<p>The script
should only assume a POSIX-compliant <code>/bin/sh</code> &ndash; see
<a href="http://pubs.opengroup.org/onlinepubs/9699919799/utilities/V3_chap02.html">http://pubs.opengroup.org/onlinepubs/9699919799/utilities/V3_chap02.html</a>.
In particular <code>bash</code> extensions must not be used, and not all
R platforms have a <code>bash</code> command, let alone one at
<samp>/bin/bash</samp>.  All known shells used with R support the use of
backticks, but not all support &lsquo;<samp>$(<var>cmd</var>)</samp>&rsquo;.</p>
<h3><a name="FOOT25" href="#DOCF25">(25)</a></h3>
<p>in POSIX parlance: GNU <code>make</code>
calls these &lsquo;make variables&rsquo;.</p>
<h3><a name="FOOT26" href="#DOCF26">(26)</a></h3>
<p>at least on Unix-alikes: the Windows build currently
resolves such dependencies to a static FORTRAN library when
<samp>Rblas.dll</samp> is built.</p>
<h3><a name="FOOT27" href="#DOCF27">(27)</a></h3>
<p><a href="http://www.openmp.org/">http://www.openmp.org/</a>,
<a href="https://en.wikipedia.org/wiki/OpenMP">https://en.wikipedia.org/wiki/OpenMP</a>,
<a href="https://computing.llnl.gov/tutorials/openMP/">https://computing.llnl.gov/tutorials/openMP/</a></p>
<h3><a name="FOOT28" href="#DOCF28">(28)</a></h3>
<p>Default builds of
<code>clang</code> 3.8.0 and later have support for OpenMP, but the
<code>libomp</code> run-time library may not be installed.</p>
<h3><a name="FOOT29" href="#DOCF29">(29)</a></h3>
<p>In most implementations the <code>_OPENMP</code>
macro has value a date which can be mapped to an OpenMP version: for
example, value <code>201307</code> is the date of version 4.0 (July
2013). However this may be used to denote the latest version which is
partially supported, not that which is fully implemented.</p>
<h3><a name="FOOT30" href="#DOCF30">(30)</a></h3>
<p>GCC since 4.7, hence R builds
for Windows since R 3.3.0, which also support OpenMP 4.0.</p>
<h3><a name="FOOT31" href="#DOCF31">(31)</a></h3>
<p>People do use older
versions of OSes such as Ubuntu 12.04LTS and Debian Wheezy LTS which
have GCC 4.4.</p>
<h3><a name="FOOT32" href="#DOCF32">(32)</a></h3>
<p>as did the
GCC-based Apple implementation, but not the Intel/LLVM OpenMP runtime
on macOS.</p>
<h3><a name="FOOT33" href="#DOCF33">(33)</a></h3>
<p>Windows default, not MinGW-w64 default.</p>
<h3><a name="FOOT34" href="#DOCF34">(34)</a></h3>
<p>Which it was at the time of writing with GCC,
Oracle, Intel and Clang compilers.</p>
<h3><a name="FOOT35" href="#DOCF35">(35)</a></h3>
<p>some Windows toolchains had the
typo &lsquo;<samp>_REENTRANCE</samp>&rsquo; instead.</p>
<h3><a name="FOOT36" href="#DOCF36">(36)</a></h3>
<p>A few OSes (AIX, IRIX, Windows) do not
need special flags for such code, but most do&mdash;although compilers will
often generate PIC code when not asked to do so.</p>
<h3><a name="FOOT37" href="#DOCF37">(37)</a></h3>
<p>Cygwin used <code>g77</code> up to 2011, and some pre-built
versions of R for Unix OSes still do.</p>
<h3><a name="FOOT38" href="#DOCF38">(38)</a></h3>
<p>The changes are linked from
<a href="https://isocpp.org/std/standing-documents/sd-6-sg10-feature-test-recommendations">https://isocpp.org/std/standing-documents/sd-6-sg10-feature-test-recommendations</a>.</p>
<h3><a name="FOOT39" href="#DOCF39">(39)</a></h3>
<p>Values <code>199711</code>, <code>201103L</code> and
<code>201402L</code> are most commonly used for C++98, C++11 and C++14
respectively, but some compilers set <code>1L</code>.</p>
<h3><a name="FOOT40" href="#DOCF40">(40)</a></h3>
<p>Some distributions, notably Debian, have supplied a
build of <code>clang</code> with <code>g++</code>&rsquo;s headers and library.
Conversely, Apple&rsquo;s command named <code>g++</code> is based on
<code>clang</code> using <code>libc++</code>.</p>
<h3><a name="FOOT41" href="#DOCF41">(41)</a></h3>
<p>For
details of these and related macros, see file <samp>config.site</samp> in
the R sources.</p>
<h3><a name="FOOT42" href="#DOCF42">(42)</a></h3>
<p>On systems which use sub-architectures,
architecture-specific versions such as <samp>~/.R/check.Renviron.i386</samp>
take precedence.</p>
<h3><a name="FOOT43" href="#DOCF43">(43)</a></h3>
<p>A suitable <code>file.exe</code> is
part of the Windows toolset: it checks for <code>gfile</code> if a suitable
<code>file</code> is not found: the latter is available in the OpenCSW
collection for Solaris at <a href="http://www.opencsw.org">http://www.opencsw.org</a>.  The source
repository is <a href="ftp://ftp.astron.com/pub/file/">ftp://ftp.astron.com/pub/file/</a>.</p>
<h3><a name="FOOT44" href="#DOCF44">(44)</a></h3>
<p>An exception is made
for subdirectories with names starting &lsquo;<samp>win</samp>&rsquo; or &lsquo;<samp>Win</samp>&rsquo;.</p>
<h3><a name="FOOT45" href="#DOCF45">(45)</a></h3>
<p>on most other platforms such runtime
libraries are dynamic, but static libraries are currently used on
Windows because the toolchain is not a standard part of the OS.</p>
<h3><a name="FOOT46" href="#DOCF46">(46)</a></h3>
<p>or if option <samp>--use-valgrind</samp> is
used or environment variable <code>_R_CHECK_ALWAYS_LOG_VIGNETTE_OUTPUT_</code>
is set to a true value or if there are differences from a target output
file</p>
<h3><a name="FOOT47" href="#DOCF47">(47)</a></h3>
<p>For example, in early
2014 <a href="https://CRAN.R-project.org/package=gdata"><strong>gdata</strong></a> declared &lsquo;<samp>Imports: gtools</samp>&rsquo; and <a href="https://CRAN.R-project.org/package=gtools"><strong>gtools</strong></a>
declared &lsquo;<samp>Imports: gdata</samp>&rsquo;.</p>
<h3><a name="FOOT48" href="#DOCF48">(48)</a></h3>
<p>loading, examples,
tests, running vignette code</p>
<h3><a name="FOOT49" href="#DOCF49">(49)</a></h3>
<p>on all platforms from
R 3.1.0.</p>
<h3><a name="FOOT50" href="#DOCF50">(50)</a></h3>
<p>called <samp>CVS</samp> or <samp>.svn</samp> or
<samp>.arch-ids</samp> or <samp>.bzr</samp> or <samp>.git</samp> (but not files called
<samp>.git</samp>) or <samp>.hg</samp>.</p>
<h3><a name="FOOT51" href="#DOCF51">(51)</a></h3>
<p>called
<samp>.metadata</samp>.</p>
<h3><a name="FOOT52" href="#DOCF52">(52)</a></h3>
<p>which is
an error: GNU make uses <samp>GNUmakefile</samp>.</p>
<h3><a name="FOOT53" href="#DOCF53">(53)</a></h3>
<p>and to avoid problems with case-insensitive file
systems, lower-case versions of all these extensions.</p>
<h3><a name="FOOT54" href="#DOCF54">(54)</a></h3>
<p>unless inhibited by using
&lsquo;<samp>BuildVignettes: no</samp>&rsquo; in the <samp>DESCRIPTION</samp> file.</p>
<h3><a name="FOOT55" href="#DOCF55">(55)</a></h3>
<p>provided the conditions of the
package&rsquo;s license are met: many, including <acronym>CRAN</acronym>, see the
omission of source components as incompatible with an Open Source
license.</p>
<h3><a name="FOOT56" href="#DOCF56">(56)</a></h3>
<p><code>R_HOME/bin</code> is prepended to the
<code>PATH</code> so that references to <code>R</code> or <code>Rscript</code> in the
<samp>Makefile</samp> do make use of the currently running version of R.</p>
<h3><a name="FOOT57" href="#DOCF57">(57)</a></h3>
<p>Note that
lazy-loaded datasets are <em>not</em> in the package&rsquo;s namespace so need
to be accessed <em>via</em> <code>::</code>, e.g.
<code>survival::survexp.us</code>.</p>
<h3><a name="FOOT58" href="#DOCF58">(58)</a></h3>
<p>they will be called
with two unnamed arguments, in that order.</p>
<h3><a name="FOOT59" href="#DOCF59">(59)</a></h3>
<p>NB: this will only be read in all versions of R if
the package contains R code in a <samp>R</samp> directory.</p>
<h3><a name="FOOT60" href="#DOCF60">(60)</a></h3>
<p>Note that this is the
basename of the shared object, and the appropriate extension (<samp>.so</samp>
or <samp>.dll</samp>) will be added.</p>
<h3><a name="FOOT61" href="#DOCF61">(61)</a></h3>
<p>This was necessary at least prior to
R 3.0.2 as the <strong>methods</strong> package looked for its own R code on
the search path.</p>
<h3><a name="FOOT62" href="#DOCF62">(62)</a></h3>
<p>This defaults to the same
pattern as <code>exportPattern</code>: use something like
<code>exportClassPattern(&quot;^$&quot;)</code> to override this.</p>
<h3><a name="FOOT63" href="#DOCF63">(63)</a></h3>
<p>if it does, there will be opaque warnings about
replacing imports if the classes/methods are also imported.</p>
<h3><a name="FOOT64" href="#DOCF64">(64)</a></h3>
<p>People use <code>dev.new()</code> to open a
device at a particular size: that is not portable but using
<code>dev.new(noRStudioGD = TRUE)</code> helps.</p>
<h3><a name="FOOT65" href="#DOCF65">(65)</a></h3>
<p>Solaris <code>make</code> does not accept
CRLF-terminated Makefiles; Solaris warns about and some other
<code>make</code>s ignore incomplete final lines.</p>
<h3><a name="FOOT66" href="#DOCF66">(66)</a></h3>
<p>This was
apparently introduced in SunOS 4, and is available elsewhere
<em>provided</em> it is surrounded by spaces.</p>
<h3><a name="FOOT67" href="#DOCF67">(67)</a></h3>
<p>GNU make,
BSD make formerly in FreeBSD and macOS, AT&amp;T make as implemented on
Solaris, <code>pmake</code> in FreeBSD, &lsquo;Distributed Make&rsquo; (<code>dmake</code>),
part of Oracle Studio and available in other versions.</p>
<h3><a name="FOOT68" href="#DOCF68">(68)</a></h3>
<p>For example, <code>test</code>
options <samp>-a</samp> and <samp>-e</samp> are not portable, and not supported
in the AT&amp;T Bourne shell used on Solaris 10/11, even though they are in
the 2008 POSIX standard.  Nor does Solaris support &lsquo;<samp>$(<var>cmd</var>)</samp>&rsquo;.</p>
<h3><a name="FOOT69" href="#DOCF69">(69)</a></h3>
<p><a href="http://fortranwiki.org/fortran/show/Modernizing+Old+Fortran">http://fortranwiki.org/fortran/show/Modernizing+Old+Fortran</a>
may help explain some of the warnings from <code>gfortran -Wall
-pedantic</code>.</p>
<h3><a name="FOOT70" href="#DOCF70">(70)</a></h3>
<p>See
<a href="http://people.ds.cam.ac.uk/nmm1/fortran/paper_07.pdf">http://people.ds.cam.ac.uk/nmm1/fortran/paper_07.pdf</a>.</p>
<h3><a name="FOOT71" href="#DOCF71">(71)</a></h3>
<p>but note that <code>long long</code> is not a standard C++98 type,
and C++ compilers set up for strict checking will reject it.</p>
<h3><a name="FOOT72" href="#DOCF72">(72)</a></h3>
<p>or where supported the variants <code>_Exit</code> and
<code>_exit</code>.</p>
<h3><a name="FOOT73" href="#DOCF73">(73)</a></h3>
<p>This and
<code>srandom</code> are in any case not portable.  They are in POSIX but not
in the C99 standard, and not available on Windows.</p>
<h3><a name="FOOT74" href="#DOCF74">(74)</a></h3>
<p>in <samp>libselinux</samp>.</p>
<h3><a name="FOOT75" href="#DOCF75">(75)</a></h3>
<p>At least Linux and Windows, but not macOS.</p>
<h3><a name="FOOT76" href="#DOCF76">(76)</a></h3>
<p>except perhaps the simplest kind as used by
<code>download.file()</code> in non-interactive use.</p>
<h3><a name="FOOT77" href="#DOCF77">(77)</a></h3>
<p>Whereas the GNU linker reorders so <samp>-L</samp> options
are processed first, the Solaris one does not.</p>
<h3><a name="FOOT78" href="#DOCF78">(78)</a></h3>
<p>This will change in R 3.4.0.</p>
<h3><a name="FOOT79" href="#DOCF79">(79)</a></h3>
<p>some versions of macOS did not.</p>
<h3><a name="FOOT80" href="#DOCF80">(80)</a></h3>
<p>For example, the ability to handle &lsquo;<samp>https://</samp>&rsquo;
URLs, which even the build in some major Linux distributions in 2017 did
not possess.</p>
<h3><a name="FOOT81" href="#DOCF81">(81)</a></h3>
<p>Not doing so is the
default on Windows, overridden for the R executables.  It is also the
default on some Solaris compilers.</p>
<h3><a name="FOOT82" href="#DOCF82">(82)</a></h3>
<p>These are not needed for the default compiler settings
on &lsquo;<samp>x86_64</samp>&rsquo; but are likely to be needed on &lsquo;<samp>ix86</samp>&rsquo;.</p>
<h3><a name="FOOT83" href="#DOCF83">(83)</a></h3>
<p>Select &lsquo;Save as&rsquo;, and select
&lsquo;Reduce file size&rsquo; from the &lsquo;Quartz filter&rsquo; menu&rsquo;: this can be accessed
in other ways, for example by Automator.</p>
<h3><a name="FOOT84" href="#DOCF84">(84)</a></h3>
<p>except perhaps some
special characters such as backslash and hash which may be taken over
for currency symbols.</p>
<h3><a name="FOOT85" href="#DOCF85">(85)</a></h3>
<p>Typically on a Unix-alike this is done by telling
<code>fontconfig</code> where to find suitable fonts to select glyphs
from.</p>
<h3><a name="FOOT86" href="#DOCF86">(86)</a></h3>
<p>This is seen on
Linux, Solaris and FreeBSD, although each has other ways to turn on all
extensions, e.g. defining <code>_GNU_SOURCE</code>, <code>__EXTENSIONS__</code> or
<code>_BSD_SOURCE</code>: the GCC compilers by default define
<code>_GNU_SOURCE</code> unless a strict standard such as <samp>-std=c99</samp> is
used.  On macOS extensions are declared unless one of these macros is
given too small a value.</p>
<h3><a name="FOOT87" href="#DOCF87">(87)</a></h3>
<p>Solaris 10 does not
recognize this value of <code>_POSIX_C_SOURCE</code>, nor values of
<code>_XOPEN_SOURCE</code> beyond 600.</p>
<h3><a name="FOOT88" href="#DOCF88">(88)</a></h3>
<p>E.g. <code>gcc 5.3</code> in C++11 mode.</p>
<h3><a name="FOOT89" href="#DOCF89">(89)</a></h3>
<p>There is a portable way to do this in Fortran 2003
(<code>ieee_is_nan()</code> in module <code>ieee_arithmetic</code>), but ironically
that is not supported in the commonly-used versions 4.x of GNU Fortran.
A pretty robust alternative is to test <code>if(my_var /= my_var)</code>.</p>
<h3><a name="FOOT90" href="#DOCF90">(90)</a></h3>
<p>which often is the same as the header included by
the C compiler, but some compilers have wrappers for some of the C
headers.</p>
<h3><a name="FOOT91" href="#DOCF91">(91)</a></h3>
<p>e.g. <code>\alias</code>, <code>\keyword</code> and
<code>\note</code> sections.</p>
<h3><a name="FOOT92" href="#DOCF92">(92)</a></h3>
<p>There can be exceptions: for example
<samp>Rd</samp> files are not allowed to start with a dot, and have to be
uniquely named on a case-insensitive file system.</p>
<h3><a name="FOOT93" href="#DOCF93">(93)</a></h3>
<p>in the current locale, and with special
treatment for LaTeX special characters and with any
&lsquo;<samp><var>pkgname</var>-package</samp>&rsquo; topic moved to the top of the list.</p>
<h3><a name="FOOT94" href="#DOCF94">(94)</a></h3>
<p>Text between or after list items is discouraged.</p>
<h3><a name="FOOT95" href="#DOCF95">(95)</a></h3>
<p>as defined by the R function
<code>trimws</code>.</p>
<h3><a name="FOOT96" href="#DOCF96">(96)</a></h3>
<p>Currently it is
rendered differently only in <acronym>HTML</acronym> conversions, and LaTeX conversion
outside &lsquo;<samp>\usage</samp>&rsquo; and &lsquo;<samp>\examples</samp>&rsquo; environments.</p>
<h3><a name="FOOT97" href="#DOCF97">(97)</a></h3>
<p>a common
example in <acronym>CRAN</acronym> packages is <code>\link[mgcv]{gam}</code>.</p>
<h3><a name="FOOT98" href="#DOCF98">(98)</a></h3>
<p>There is only a fine
distinction between <code>\dots</code> and <code>\ldots</code>.  It is technically
incorrect to use <code>\ldots</code> in code blocks and <code>tools::checkRd</code>
will warn about this&mdash;on the other hand the current converters treat
them the same way in code blocks, and elsewhere apart from the small
distinction between the two in LaTeX.</p>
<h3><a name="FOOT99" href="#DOCF99">(99)</a></h3>
<p>See the
examples section in the file <samp>Paren.Rd</samp> for an example.</p>
<h3><a name="FOOT100" href="#DOCF100">(100)</a></h3>
<p>R
2.9.0 added support for UTF-8 Cyrillic characters in LaTeX, but on
some OSes this will need Cyrillic support added to LaTeX, so
environment variable <code>_R_CYRILLIC_TEX_</code> may need to be set to a
non-empty value to enable this.</p>
<h3><a name="FOOT101" href="#DOCF101">(101)</a></h3>
<p>R
has to be built to enable this, but the option
<samp>--enable-R-profiling</samp> is the default.</p>
<h3><a name="FOOT102" href="#DOCF102">(102)</a></h3>
<p>For Unix-alikes these are intervals of CPU
time, and for Windows of elapsed time.</p>
<h3><a name="FOOT103" href="#DOCF103">(103)</a></h3>
<p>With the exceptions of the commands
listed below: an object of such a name can be printed <em>via</em> an
explicit call to <code>print</code>.</p>
<h3><a name="FOOT104" href="#DOCF104">(104)</a></h3>
<p>at the time of writing mainly for 10.9 with some
support for 10.8, none for the current 10.10.</p>
<h3><a name="FOOT105" href="#DOCF105">(105)</a></h3>
<p>Those in some numeric, logical,
integer, raw, complex vectors and in memory allocated by
<code>R_alloc</code>.</p>
<h3><a name="FOOT106" href="#DOCF106">(106)</a></h3>
<p>including using the data sections of R vectors after
they are freed.</p>
<h3><a name="FOOT107" href="#DOCF107">(107)</a></h3>
<p>small
fixed-size arrays by default in <code>gfortran</code>, for example.</p>
<h3><a name="FOOT108" href="#DOCF108">(108)</a></h3>
<p>currently only on &lsquo;<samp>ix86</samp>&rsquo;/&lsquo;<samp>x86_64</samp>&rsquo; Linux
and macOS (including the builds in Xcode 7 but not earlier Apple
releases).  On some platforms the runtime library, <strong>libasan</strong>, needs
to be installed separately, and for checking C++ you may also need
<strong>libubsan</strong>.</p>
<h3><a name="FOOT109" href="#DOCF109">(109)</a></h3>
<p>see
<a href="http://llvm.org/devmtg/2014-04/PDFs/LightningTalks/EuroLLVM%202014%20--%20container%20overflow.pdf">http://llvm.org/devmtg/2014-04/PDFs/LightningTalks/EuroLLVM%202014%20--%20container%20overflow.pdf</a>.</p>
<h3><a name="FOOT110" href="#DOCF110">(110)</a></h3>
<p>part of the LLVM project and
in distributed in <code>llvm</code> RPMs and <code>.deb</code>s on Linux.  It is not
currently shipped by Apple.</p>
<h3><a name="FOOT111" href="#DOCF111">(111)</a></h3>
<p>as Ubuntu is said to do.</p>
<h3><a name="FOOT112" href="#DOCF112">(112)</a></h3>
<p>installed on some Linux systems as
<code>asan_symbolize</code>, and obtainable from
<a href="https://llvm.org/svn/llvm-project/compiler-rt/trunk/lib/asan/scripts/asan_symbolize.py">https://llvm.org/svn/llvm-project/compiler-rt/trunk/lib/asan/scripts/asan_symbolize.py</a>:
it makes use of <code>llvm-symbolizer</code> if available.</p>
<h3><a name="FOOT113" href="#DOCF113">(113)</a></h3>
<p>This is expected to be in <code>gcc</code> 7 and
<code>clang</code> 4.0.0.  Currently for <code>gcc</code> it is implied by
<samp>-fsanitize=address</samp>.</p>
<h3><a name="FOOT114" href="#DOCF114">(114)</a></h3>
<p>On some
platforms the runtime library, <strong>libubsan</strong>, needs to be installed
separately.</p>
<h3><a name="FOOT115" href="#DOCF115">(115)</a></h3>
<p>e.g. <samp>src/main/dotcode.c</samp> and parts of the
<a href="https://CRAN.R-project.org/package=Matrix"><strong>Matrix</strong></a> sources with <code>clang</code> 3.7.0 amd later.</p>
<h3><a name="FOOT116" href="#DOCF116">(116)</a></h3>
<p>or
the user manual for your version of <code>clang</code>, e.g.
<a href="http://llvm.org/releases/3.8.1/tools/docs/UsersManual.html">http://llvm.org/releases/3.8.1/tools/docs/UsersManual.html</a> or
<a href="http://llvm.org/releases/3.8.0/tools/clang/docs/UsersManual.html">http://llvm.org/releases/3.8.0/tools/clang/docs/UsersManual.html</a>.</p>
<h3><a name="FOOT117" href="#DOCF117">(117)</a></h3>
<p>This includes the C++
UBSAN handlers, despite its name.</p>
<h3><a name="FOOT118" href="#DOCF118">(118)</a></h3>
<p>but
works better if inlining and frame pointer optimizations are disabled.</p>
<h3><a name="FOOT119" href="#DOCF119">(119)</a></h3>
<p>possibly after some platform-specific
translation, e.g. adding leading or trailing underscores.</p>
<h3><a name="FOOT120" href="#DOCF120">(120)</a></h3>
<p>Note that this is then not checked for over-runs by
option <code>CBoundsCheck = TRUE</code>.</p>
<h3><a name="FOOT121" href="#DOCF121">(121)</a></h3>
<p>Strictly this is OS-specific, but no exceptions have
been seen for many years.</p>
<h3><a name="FOOT122" href="#DOCF122">(122)</a></h3>
<p>For calls from within a namespace the search is confined to
the DLL loaded for that package.</p>
<h3><a name="FOOT123" href="#DOCF123">(123)</a></h3>
<p>For unregistered entry points the OS&rsquo;s <code>dlsym</code>
routine is used to find addresses.  Its performance varies considerably
by OS and even in the best case it will need to search a much larger
symbol table than, say, the table of <code>.Call</code> entry points.</p>
<h3><a name="FOOT124" href="#DOCF124">(124)</a></h3>
<p>Because it is a standard
package, one would need to rename it before attempting to reproduce the
account here.</p>
<h3><a name="FOOT125" href="#DOCF125">(125)</a></h3>
<p>whether or not &lsquo;<samp>LinkingTo</samp>&rsquo; is used.</p>
<h3><a name="FOOT126" href="#DOCF126">(126)</a></h3>
<p>so there needs to be a corresponding <code>import</code> or
<code>importFrom</code> entry in the <samp>NAMESPACE</samp> file.</p>
<h3><a name="FOOT127" href="#DOCF127">(127)</a></h3>
<p>Even including C system headers in
such a block has caused compilation errors.</p>
<h3><a name="FOOT128" href="#DOCF128">(128)</a></h3>
<p>with an exception for the
Solaris C++ compiler to be removed in R 3.4.0.</p>
<h3><a name="FOOT129" href="#DOCF129">(129)</a></h3>
<p><a href="https://en.wikipedia.org/wiki/Application_binary_interface">https://en.wikipedia.org/wiki/Application_binary_interface</a>.</p>
<h3><a name="FOOT130" href="#DOCF130">(130)</a></h3>
<p>For
example, &lsquo;<samp>_GLIBCXX_USE_CXX11_ABI</samp>&rsquo; in <code>g++</code> 5.1 and later:
<a href="https://gcc.gnu.org/onlinedocs/libstdc++/manual/using_dual_abi.html">https://gcc.gnu.org/onlinedocs/libstdc++/manual/using_dual_abi.html</a>.</p>
<h3><a name="FOOT131" href="#DOCF131">(131)</a></h3>
<p><code>dyld</code> on macOS,
and <code>DYLD_LIBRARY_PATHS</code> below.</p>
<h3><a name="FOOT132" href="#DOCF132">(132)</a></h3>
<p>That is,
similar to those defined in S version 4 from the 1990s: these are
not kept up to date and are not recommended for new projects.  Prior to
R 3.3.0 it was not compatible with defining <code>R_NO_REMAP</code>.</p>
<h3><a name="FOOT133" href="#DOCF133">(133)</a></h3>
<p>see <a href="#The-R-API">The R API</a>: note that these are not all part of
the API.</p>
<h3><a name="FOOT134" href="#DOCF134">(134)</a></h3>
<p>SEXP is an acronym for <em>S</em>imple
<em>EXP</em>ression, common in LISP-like language syntaxes.</p>
<h3><a name="FOOT135" href="#DOCF135">(135)</a></h3>
<p>If no coercion was required, <code>coerceVector</code> would
have passed the old object through unchanged.</p>
<h3><a name="FOOT136" href="#DOCF136">(136)</a></h3>
<p>You can assign a <em>copy</em> of the object in the
environment frame <code>rho</code> using <code>defineVar(symbol,
duplicate(value), rho)</code>).</p>
<h3><a name="FOOT137" href="#DOCF137">(137)</a></h3>
<p>see <a href="#Character-encoding-issues">Character encoding issues</a> for why this
might not be what is required.</p>
<h3><a name="FOOT138" href="#DOCF138">(138)</a></h3>
<p>This is only guaranteed to show the
current interface: it is liable to change.</p>
<h3><a name="FOOT139" href="#DOCF139">(139)</a></h3>
<p>Known problems are redefining
<code>LENGTH</code>, <code>error</code>, <code>length</code>, <code>vector</code> and
<code>warning</code></p>
<h3><a name="FOOT140" href="#DOCF140">(140)</a></h3>
<p>also part of
C++11.</p>
<h3><a name="FOOT141" href="#DOCF141">(141)</a></h3>
<p>It is an optional C11
extension.</p>
<h3><a name="FOOT142" href="#DOCF142">(142)</a></h3>
<p><a href="https://en.wikipedia.org/wiki/Endianness">https://en.wikipedia.org/wiki/Endianness</a>.</p>
<h3><a name="FOOT143" href="#DOCF143">(143)</a></h3>
<p>It is defined by the Intel compilers,
but also hides unsatisfied references and so cannot be used with R.
It is not supported by the AIX nor Solaris compilers.</p>
<h3><a name="FOOT144" href="#DOCF144">(144)</a></h3>
<p>In the parlance of macOS
this is a <em>dynamic</em> library, and is the normal way to build R on
that platform.</p>
<h3><a name="FOOT145" href="#DOCF145">(145)</a></h3>
<p>but these
are not part of the automated test procedures and so little tested.</p>
<h3><a name="FOOT146" href="#DOCF146">(146)</a></h3>
<p>at least on platforms where the values are
available, that is having <code>getrlimit</code> and on Linux or having
<code>sysctl</code> supporting <code>KERN_USRSTACK</code>, including FreeBSD and OS
X.</p>
<h3><a name="FOOT147" href="#DOCF147">(147)</a></h3>
<p>An
attempt to use only threads in the late 1990s failed to work correctly
under Windows 95, the predominant version of Windows at that time.</p>
</div>
<hr>



</body>
</html>
r-doc-html 3.3.3-1 / usr / share / R / doc / manual / R-exts.html
