/usr/share/doc/mlton/guide/Fold is in mlton-doc 20100608-5.
This file is owned by root:root, with mode 0o644.
The actual contents of the file can be viewed below.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 | <!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN" "http://www.w3.org/TR/html4/strict.dtd">
<html>
<head>
<meta name="robots" content="index,nofollow">
<title>Fold - MLton Standard ML Compiler (SML Compiler)</title>
<link rel="stylesheet" type="text/css" charset="iso-8859-1" media="all" href="common.css">
<link rel="stylesheet" type="text/css" charset="iso-8859-1" media="screen" href="screen.css">
<link rel="stylesheet" type="text/css" charset="iso-8859-1" media="print" href="print.css">
<link rel="Start" href="Home">
</head>
<body lang="en" dir="ltr">
<script src="http://www.google-analytics.com/urchin.js" type="text/javascript">
</script>
<script type="text/javascript">
_uacct = "UA-833377-1";
urchinTracker();
</script>
<table bgcolor = lightblue cellspacing = 0 style = "border: 0px;" width = 100%>
<tr>
<td style = "
border: 0px;
color: darkblue;
font-size: 150%;
text-align: left;">
<a class = mltona href="Home">MLton MLTONWIKIVERSION</a>
<td style = "
border: 0px;
font-size: 150%;
text-align: center;
width: 50%;">
Fold
<td style = "
border: 0px;
text-align: right;">
<table cellspacing = 0 style = "border: 0px">
<tr style = "vertical-align: middle;">
</table>
<tr style = "background-color: white;">
<td colspan = 3
style = "
border: 0px;
font-size:70%;
text-align: right;">
<a href = "Home">Home</a>
<a href = "TitleIndex">Index</a>
</table>
<div id="content" lang="en" dir="ltr">
This page describes a technique that enables convenient syntax for a number of language features that are not explicitly supported by <a href="StandardML">Standard ML</a>, including: variable number of arguments, <a href="OptionalArguments">optional arguments and labeled arguments</a>, <a href="ArrayLiteral">array and vector literals</a>, <a href="FunctionalRecordUpdate">functional record update</a>, and (seemingly) dependently typed functions like <a href="Printf">printf</a> and scanf. <p>
The key idea to <em>fold</em> is to define functions <tt>fold</tt>, <tt>step0</tt>, and <tt>$</tt> such that the following equation holds.
</p>
<pre class=code>
fold (a, f) (step0 h1) (step0 h2) ... (step0 hn) $
= f (hn (... (h2 (h1 a))))
</PRE>
<p>
</p>
<p>
The name <tt>fold</tt> comes because this is like a traditional list fold, where <tt>a</tt> is the <em>base element</em>, and each <em>step function</em>, <tt>step0 hi</tt>, corresponds to one element of the list and does one step of the fold. The name <tt>$</tt> is chosen to mean <em>end of arguments</em> from its common use in regular-expression syntax.
</p>
<p>
Unlike the usual list fold in which the same function is used to step over each element in the list, this fold allows the step functions to be different from each other, and even to be of different types. Also unlike the usual list fold, this fold includes a <em>finishing function</em>, <tt>f</tt>, that is applied to the result of the fold. The presence of the finishing function may seem odd because there is no analogy in list fold. However, the finishing function is essential; without it, there would be no way for the folder to perform an arbitrary computation after processing all the arguments. The examples below will make this clear.
</p>
<p>
The functions <tt>fold</tt>, <tt>step0</tt>, and <tt>$</tt> are easy to define.
</p>
<pre class=code>
<B><FONT COLOR="#A020F0">fun</FONT></B> $ (a, f) = f a
<B><FONT COLOR="#A020F0">fun</FONT></B> id x = x
<B><FONT COLOR="#0000FF">structure</FONT></B> Fold =
<B><FONT COLOR="#0000FF">struct</FONT></B>
<B><FONT COLOR="#A020F0">fun</FONT></B> fold (a, f) g = g (a, f)
<B><FONT COLOR="#A020F0">fun</FONT></B> step0 h (a, f) = fold (h a, f)
<B><FONT COLOR="#0000FF">end</FONT></B>
</PRE>
<p>
</p>
<p>
We've placed <tt>fold</tt> and <tt>step0</tt> in the <tt>Fold</tt> structure but left <tt>$</tt> at the toplevel because it is convenient in code to always have <tt>$</tt> in scope. We've also defined the identity function, <tt>id</tt>, at the toplevel since we use it so frequently.
</p>
<p>
Plugging in the definitions, it is easy to verify the equation from above.
<pre class=code>
fold (a, f) (step0 h1) (step0 h2) ... (step0 hn) $
= step0 h1 (a, f) (step0 h2) ... (step0 hn) $
= fold (h1 a, f) (step0 h2) ... (step0 hn) $
= step0 h2 (h1 a, f) ... (step0 hn) $
= fold (h2 (h1 a), f) ... (step0 hn) $
...
= fold (hn (... (h2 (h1 a))), f) $
= $ (hn (... (h2 (h1 a))), f)
= f (hn (... (h2 (h1 a))))
</PRE>
</p>
<h2 id="head-eb8e90f0113d0b7b2a22602f865ce3c364dd31e2">Example: variable number of arguments</h2>
<p>
The simplest example of fold is accepting a variable number of (curried) arguments. We'll define a function <tt>f</tt> and argument <tt>a</tt> such that all of the following expressions are valid.
</p>
<pre class=code>
f $
f a $
f a a $
f a a a $
f a a a ... a a a $ <I><FONT COLOR="#B22222">(* as many a's as we want *)</FONT></I>
</PRE>
<p>
</p>
<p>
Off-hand it may appear impossible that all of the above expressions are type correct SML -- how can a function <tt>f</tt> accept a variable number of curried arguments? What could the type of <tt>f</tt> be? We'll have more to say later on how type checking works. For now, once we have supplied the definitions below, you can check that the expressions are type correct by feeding them to your favorite SML implementation.
</p>
<p>
It is simple to define <tt>f</tt> and <tt>a</tt>. We define <tt>f</tt> as a folder whose base element is <tt>()</tt> and whose finish function does nothing. We define <tt>a</tt> as the step function that does nothing. The only trickiness is that we must <a href="EtaExpansion">eta expand</a> the definition of <tt>f</tt> and <tt>a</tt> to work around the <a href="ValueRestriction">ValueRestriction</a>; we frequently use eta expansion for this purpose without mention.
</p>
<pre class=code>
<B><FONT COLOR="#A020F0">val</FONT></B> base = ()
<B><FONT COLOR="#A020F0">fun</FONT></B> finish () = ()
<B><FONT COLOR="#A020F0">fun</FONT></B> step () = ()
<B><FONT COLOR="#A020F0">val</FONT></B> f = <B><FONT COLOR="#A020F0">fn</FONT></B> z => Fold.fold (base, finish) z
<B><FONT COLOR="#A020F0">val</FONT></B> a = <B><FONT COLOR="#A020F0">fn</FONT></B> z => Fold.step0 step z
</PRE>
<p>
</p>
<p>
One can easily apply the fold equation to verify by hand that <tt>f</tt> applied to any number of <tt>a</tt>'s evaluates to <tt>()</tt>.
</p>
<pre class=code>
f a ... a $
= finish (step (... (step base)))
= finish (step (... ()))
...
= finish ()
= ()
</PRE>
<p>
</p>
<h2 id="head-677e9d12eacb1c50946b3f79103afd8ab8eea31e">Example: variable-argument sum</h2>
<p>
Let's look at an example that computes something: a variable-argument function <tt>sum</tt> and a stepper <tt>a</tt> such that
</p>
<pre class=code>
sum (a i1) (a i2) ... (a im) $ = i1 + i2 + ... + im
</PRE>
<p>
</p>
<p>
The idea is simple -- the folder starts with a base accumulator of <tt>0</tt> and the stepper adds each element to the accumulator, <tt>s</tt>, which the folder simply returns at the end.
</p>
<pre class=code>
<B><FONT COLOR="#A020F0">val</FONT></B> sum = <B><FONT COLOR="#A020F0">fn</FONT></B> z => Fold.fold (<B><FONT COLOR="#5F9EA0">0</FONT></B>, <B><FONT COLOR="#A020F0">fn</FONT></B> s => s) z
<B><FONT COLOR="#A020F0">fun</FONT></B> a i = Fold.step0 (<B><FONT COLOR="#A020F0">fn</FONT></B> s => i + s)
</PRE>
<p>
</p>
<p>
Using the fold equation, one can verify the following.
</p>
<pre class=code>
sum (a <B><FONT COLOR="#5F9EA0">1</FONT></B>) (a <B><FONT COLOR="#5F9EA0">2</FONT></B>) (a <B><FONT COLOR="#5F9EA0">3</FONT></B>) $ = <B><FONT COLOR="#5F9EA0">6</FONT></B>
</PRE>
<p>
</p>
<h2 id="head-1480d125a98eea997802e10f81dd1a8a9384619b">Step1</h2>
<p>
It is sometimes syntactically convenient to omit the parentheses around the steps in a fold. This is easily done by defining a new function, <tt>step1</tt>, as follows.
</p>
<pre class=code>
<B><FONT COLOR="#0000FF">structure</FONT></B> Fold =
<B><FONT COLOR="#0000FF">struct</FONT></B>
<B><FONT COLOR="#0000FF">open</FONT></B> Fold
<B><FONT COLOR="#A020F0">fun</FONT></B> step1 h (a, f) b = fold (h (b, a), f)
<B><FONT COLOR="#0000FF">end</FONT></B>
</PRE>
<p>
</p>
<p>
From the definition of <tt>step1</tt>, we have the following equivalence.
</p>
<pre class=code>
fold (a, f) (step1 h) b
= step1 h (a, f) b
= fold (h (b, a), f)
</PRE>
<p>
</p>
<p>
Using the above equivalence, we can compute the following equation for <tt>step1</tt>.
</p>
<pre class=code>
fold (a, f) (step1 h1) b1 (step1 h2) b2 ... (step1 hn) bn $
= fold (h1 (b1, a), f) (step1 h2) b2 ... (step1 hn) bn $
= fold (h2 (b2, h1 (b1, a)), f) ... (step1 hn) bn $
= fold (hn (bn, ... (h2 (b2, h1 (b1, a)))), f) $
= f (hn (bn, ... (h2 (b2, h1 (b1, a)))))
</PRE>
<p>
</p>
<p>
Here is an example using <tt>step1</tt> to define a variable-argument product function, <tt>prod</tt>, with a convenient syntax.
</p>
<pre class=code>
<B><FONT COLOR="#A020F0">val</FONT></B> prod = <B><FONT COLOR="#A020F0">fn</FONT></B> z => Fold.fold (<B><FONT COLOR="#5F9EA0">1</FONT></B>, <B><FONT COLOR="#A020F0">fn</FONT></B> p => p) z
<B><FONT COLOR="#A020F0">val</FONT></B> ` = <B><FONT COLOR="#A020F0">fn</FONT></B> z => Fold.step1 (<B><FONT COLOR="#A020F0">fn</FONT></B> (i, p) => i * p) z
</PRE>
<p>
</p>
<p>
The functions <tt>prod</tt> and <tt>`</tt> satisfy the following equation.
<pre class=code>
prod `i1 `i2 ... `im $ = i1 * i2 * ... * im
</PRE>
</p>
<p>
Note that in SML, <tt>`i1</tt> is two different tokens, <tt>`</tt> and <tt>i1</tt>. We often use <tt>`</tt> for an instance of a <tt>step1</tt> function because of its syntactic unobtrusiveness and because no space is required to separate it from an alphanumeric token.
</p>
<p>
Also note that there are no parenthesis around the steps. That is, the following expression is not the same as the above one (in fact, it is not type correct).
</p>
<pre class=code>
prod (`i1) (`i2) ... (`im) $
</PRE>
<p>
</p>
<h2 id="head-970703bc37877d12c60099a1b012e155e9057ca7">Example: list literals</h2>
<p>
SML already has a syntax for list literals, e.g. <tt>[w, x, y, z]</tt>. However, using fold, we can define our own syntax.
</p>
<pre class=code>
<B><FONT COLOR="#A020F0">val</FONT></B> list = <B><FONT COLOR="#A020F0">fn</FONT></B> z => Fold.fold ([], rev) z
<B><FONT COLOR="#A020F0">val</FONT></B> ` = <B><FONT COLOR="#A020F0">fn</FONT></B> z => Fold.step1 (<B><FONT COLOR="#A020F0">op</FONT></B> ::) z
</PRE>
<p>
</p>
<p>
The idea is that the folder starts out with the empty list, the steps accumulate the elements into a list, and then the finishing function reverses the list at the end.
</p>
<p>
With these definitions one can write a list like:
</p>
<pre class=code>
list `w `x `y `z $
</PRE>
<p>
</p>
<p>
While the example is not practically useful, it does demonstrate the need for the finishing function to be incorporated in <tt>fold</tt>. Without a finishing function, every use of <tt>list</tt> would need to be wrapped in <tt>rev</tt>, as follows.
</p>
<pre class=code>
rev (list `w `x `y `z $)
</PRE>
<p>
</p>
<p>
The finishing function allows us to incorporate the reversal into the definition of <tt>list</tt>, and to treat <tt>list</tt> as a truly variable argument function, performing an arbitrary computation after receiving all of its arguments.
</p>
<p>
See <a href="ArrayLiteral">ArrayLiteral</a> for a similar use of <tt>fold</tt> that provides a syntax for array and vector literals, which are not built in to SML.
</p>
<h2 id="head-a23a2fc1bc8d5112da93328461608221b9d8cb14">Fold right</h2>
<p>
Just as <tt>fold</tt> is analogous to a fold left, in which the functions are applied to the accumulator left-to-right, we can define a variant of <tt>fold</tt> that is analogous to a fold right, in which the functions are applied to the accumulator right-to-left. That is, we can define functions <tt>foldr</tt> and <tt>step0</tt> such that the following equation holds.
</p>
<pre class=code>
foldr (a, f) (step0 h1) (step0 h2) ... (step0 hn) $
= f (h1 (h2 (... (hn a))))
</PRE>
<p>
</p>
<p>
The implementation of fold right is easy, using fold. The idea is for the fold to start with <tt>f</tt> and for each step to precompose the next <tt>hi</tt>. Then, the finisher applies the composed function to the base value, <tt>a</tt>. Here is the code.
</p>
<pre class=code>
<B><FONT COLOR="#0000FF">structure</FONT></B> Foldr =
<B><FONT COLOR="#0000FF">struct</FONT></B>
<B><FONT COLOR="#A020F0">fun</FONT></B> foldr (a, f) = Fold.fold (f, <B><FONT COLOR="#A020F0">fn</FONT></B> g => g a)
<B><FONT COLOR="#A020F0">fun</FONT></B> step0 h = Fold.step0 (<B><FONT COLOR="#A020F0">fn</FONT></B> g => g o h)
<B><FONT COLOR="#0000FF">end</FONT></B>
</PRE>
<p>
</p>
<p>
Verifying the fold-right equation is straightforward, using the fold-left equation.
</p>
<pre class=code>
foldr (a, f) (Foldr.step0 h1) (Foldr.step0 h2) ... (Foldr.step0 hn) $
= fold (f, <B><FONT COLOR="#A020F0">fn</FONT></B> g => g a)
(Fold.step0 (<B><FONT COLOR="#A020F0">fn</FONT></B> g => g o h1))
(Fold.step0 (<B><FONT COLOR="#A020F0">fn</FONT></B> g => g o h2))
...
(Fold.step0 (<B><FONT COLOR="#A020F0">fn</FONT></B> g => g o hn)) $
= (<B><FONT COLOR="#A020F0">fn</FONT></B> g => g a)
((<B><FONT COLOR="#A020F0">fn</FONT></B> g => g o hn) (... ((<B><FONT COLOR="#A020F0">fn</FONT></B> g => g o h2) ((<B><FONT COLOR="#A020F0">fn</FONT></B> g => g o h1) f))))
= (<B><FONT COLOR="#A020F0">fn</FONT></B> g => g a)
((<B><FONT COLOR="#A020F0">fn</FONT></B> g => g o hn) (... ((<B><FONT COLOR="#A020F0">fn</FONT></B> g => g o h2) (f o h1))))
= (<B><FONT COLOR="#A020F0">fn</FONT></B> g => g a) ((<B><FONT COLOR="#A020F0">fn</FONT></B> g => g o hn) (... (f o h1 o h2)))
= (<B><FONT COLOR="#A020F0">fn</FONT></B> g => g a) (f o h1 o h2 o ... o hn)
= (f o h1 o h2 o ... o hn) a
= f (h1 (h2 (... (hn a))))
</PRE>
<p>
</p>
<p>
One can also define the fold-right analogue of <tt>step1</tt>.
</p>
<pre class=code>
<B><FONT COLOR="#0000FF">structure</FONT></B> Foldr =
<B><FONT COLOR="#0000FF">struct</FONT></B>
<B><FONT COLOR="#0000FF">open</FONT></B> Foldr
<B><FONT COLOR="#A020F0">fun</FONT></B> step1 h = Fold.step1 (<B><FONT COLOR="#A020F0">fn</FONT></B> (b, g) => g o (<B><FONT COLOR="#A020F0">fn</FONT></B> a => h (b, a)))
<B><FONT COLOR="#0000FF">end</FONT></B>
</PRE>
<p>
</p>
<h2 id="head-ab8bc412c39f88e53289115eeb16ca0026e5b292">Example: list literals via fold right</h2>
<p>
Revisiting the list literal example from earlier, we can use fold right to define a syntax for list literals that doesn't do a reversal.
</p>
<pre class=code>
<B><FONT COLOR="#A020F0">val</FONT></B> list = <B><FONT COLOR="#A020F0">fn</FONT></B> z => Foldr.foldr ([], <B><FONT COLOR="#A020F0">fn</FONT></B> l => l) z
<B><FONT COLOR="#A020F0">val</FONT></B> ` = <B><FONT COLOR="#A020F0">fn</FONT></B> z => Foldr.step1 (<B><FONT COLOR="#A020F0">op</FONT></B> ::) z
</PRE>
<p>
</p>
<p>
As before, with these definitions, one can write a list like:
</p>
<pre class=code>
list `w `x `y `z $
</PRE>
<p>
</p>
<p>
The difference between the fold-left and fold-right approaches is that the fold-right approach does not have to reverse the list at the end, since it accumulates the elements in the correct order. In practice, MLton will simplify away all of the intermediate function composition, so the the fold-right approach will be more efficient.
</p>
<h2 id="head-45726afdeece4355f73ffcb615f1d992001c692e">Mixing steppers</h2>
<p>
All of the examples so far have used the same step function throughout a fold. This need not be the case. For example, consider the following.
</p>
<pre class=code>
<B><FONT COLOR="#A020F0">val</FONT></B> n = <B><FONT COLOR="#A020F0">fn</FONT></B> z => Fold.fold (<B><FONT COLOR="#5F9EA0">0</FONT></B>, <B><FONT COLOR="#A020F0">fn</FONT></B> i => i) z
<B><FONT COLOR="#A020F0">val</FONT></B> I = <B><FONT COLOR="#A020F0">fn</FONT></B> z => Fold.step0 (<B><FONT COLOR="#A020F0">fn</FONT></B> i => i * <B><FONT COLOR="#5F9EA0">2</FONT></B>) z
<B><FONT COLOR="#A020F0">val</FONT></B> O = <B><FONT COLOR="#A020F0">fn</FONT></B> z => Fold.step0 (<B><FONT COLOR="#A020F0">fn</FONT></B> i => i * <B><FONT COLOR="#5F9EA0">2</FONT></B> + <B><FONT COLOR="#5F9EA0">1</FONT></B>) z
</PRE>
<p>
</p>
<p>
Here we have one folder, <tt>n</tt>, that can be used with two different steppers, <tt>I</tt> and <tt>O</tt>. By using the fold equation, one can verify the following equations.
</p>
<pre class=code>
n O $ = <B><FONT COLOR="#5F9EA0">0</FONT></B>
n I $ = <B><FONT COLOR="#5F9EA0">1</FONT></B>
n I O $ = <B><FONT COLOR="#5F9EA0">2</FONT></B>
n I O I $ = <B><FONT COLOR="#5F9EA0">5</FONT></B>
n I I I O $ = <B><FONT COLOR="#5F9EA0">14</FONT></B>
</PRE>
<p>
</p>
<p>
That is, we've defined a syntax for writing binary integer constants.
</p>
<p>
Not only can one use different instances of <tt>step0</tt> in the same fold, one can also intermix uses of <tt>step0</tt> and <tt>step1</tt>. For example, consider the following.
</p>
<pre class=code>
<B><FONT COLOR="#A020F0">val</FONT></B> n = <B><FONT COLOR="#A020F0">fn</FONT></B> z => Fold.fold (<B><FONT COLOR="#5F9EA0">0</FONT></B>, <B><FONT COLOR="#A020F0">fn</FONT></B> i => i) z
<B><FONT COLOR="#A020F0">val</FONT></B> O = <B><FONT COLOR="#A020F0">fn</FONT></B> z => Fold.step0 (<B><FONT COLOR="#A020F0">fn</FONT></B> i => n * <B><FONT COLOR="#5F9EA0">8</FONT></B>) z
<B><FONT COLOR="#A020F0">val</FONT></B> ` = <B><FONT COLOR="#A020F0">fn</FONT></B> z => Fold.step1 (<B><FONT COLOR="#A020F0">fn</FONT></B> (i, n) => n * <B><FONT COLOR="#5F9EA0">8</FONT></B> + i) z
</PRE>
<p>
</p>
<p>
Using the straightforward generalization of the fold equation to mixed steppers, one can verify the following equations.
</p>
<pre class=code>
n <B><FONT COLOR="#5F9EA0">0</FONT></B> $ = <B><FONT COLOR="#5F9EA0">0</FONT></B>
n `<B><FONT COLOR="#5F9EA0">3</FONT></B> O $ = <B><FONT COLOR="#5F9EA0">24</FONT></B>
n `<B><FONT COLOR="#5F9EA0">1</FONT></B> O `<B><FONT COLOR="#5F9EA0">7</FONT></B> $ = <B><FONT COLOR="#5F9EA0">71</FONT></B>
</PRE>
<p>
</p>
<p>
That is, we've defined a syntax for writing octal integer constants, with a special syntax, <tt>O</tt>, for the zero digit (admittedly contrived, since one could just write <tt>`0</tt> instead of <tt>O</tt>).
</p>
<p>
See <a href="NumericLiteral">NumericLiteral</a> for a practical extension of this approach that supports numeric constants in any base and of any type.
</p>
<h2 id="head-c9f72c00a8519db78dbb19512887495769c7358d">(Seemingly) dependent types</h2>
<p>
A normal list fold always returns the same type no matter what elements are in the list or how long the list is. Variable-argument fold is more powerful, because the result type can vary based both on the arguments that are passed and on their number. This can provide the illusion of dependent types.
</p>
<p>
For example, consider the following.
</p>
<pre class=code>
<B><FONT COLOR="#A020F0">val</FONT></B> f = <B><FONT COLOR="#A020F0">fn</FONT></B> z => Fold.fold ((), id) z
<B><FONT COLOR="#A020F0">val</FONT></B> a = <B><FONT COLOR="#A020F0">fn</FONT></B> z => Fold.step0 (<B><FONT COLOR="#A020F0">fn</FONT></B> () => <B><FONT COLOR="#BC8F8F">"hello"</FONT></B>) z
<B><FONT COLOR="#A020F0">val</FONT></B> b = <B><FONT COLOR="#A020F0">fn</FONT></B> z => Fold.step0 (<B><FONT COLOR="#A020F0">fn</FONT></B> () => <B><FONT COLOR="#5F9EA0">13</FONT></B>) z
<B><FONT COLOR="#A020F0">val</FONT></B> c = <B><FONT COLOR="#A020F0">fn</FONT></B> z => Fold.step0 (<B><FONT COLOR="#A020F0">fn</FONT></B> () => (<B><FONT COLOR="#5F9EA0">1</FONT></B>, <B><FONT COLOR="#5F9EA0">2</FONT></B>)) z
</PRE>
<p>
</p>
<p>
Using the fold equation, one can verify the following equations.
</p>
<pre class=code>
f a $ = <B><FONT COLOR="#BC8F8F">"hello"</FONT></B>: string
f b $ = <B><FONT COLOR="#5F9EA0">13</FONT></B>: int
f c $ = (<B><FONT COLOR="#5F9EA0">1</FONT></B>, <B><FONT COLOR="#5F9EA0">2</FONT></B>): int * int
</PRE>
<p>
</p>
<p>
That is, <tt>f</tt> returns a value of a different type depending on whether it is applied to argument <tt>a</tt>, argument <tt>b</tt>, or argument <tt>c</tt>.
</p>
<p>
The following example shows how the type of a fold can depend on the number of arguments.
</p>
<pre class=code>
<B><FONT COLOR="#A020F0">val</FONT></B> grow = <B><FONT COLOR="#A020F0">fn</FONT></B> z => Fold.fold ([], <B><FONT COLOR="#A020F0">fn</FONT></B> l => l) z
<B><FONT COLOR="#A020F0">val</FONT></B> a = <B><FONT COLOR="#A020F0">fn</FONT></B> z => Fold.step0 (<B><FONT COLOR="#A020F0">fn</FONT></B> x => [x]) z
</PRE>
<p>
</p>
<p>
Using the fold equation, one can verify the following equations.
</p>
<pre class=code>
grow $ = []: 'a list
grow a $ = [[]]: 'a list list
grow a a $ = [[[]]]: 'a list list list
</PRE>
<p>
</p>
<p>
Clearly, the result type of a call to the variable argument <tt>grow</tt> function depends on the number of arguments that are passed.
</p>
<p>
As a reminder, this is well-typed SML. You can check it out in any implementation.
</p>
<h2 id="head-dee978912295db9e76200c094e45424c3e4d4b7c">(Seemingly) dependently-typed functional results</h2>
<p>
Fold is especially useful when it returns a curried function whose arity depends on the number of arguments. For example, consider the following.
</p>
<pre class=code>
<B><FONT COLOR="#A020F0">val</FONT></B> makeSum = <B><FONT COLOR="#A020F0">fn</FONT></B> z => Fold.fold (id, <B><FONT COLOR="#A020F0">fn</FONT></B> f => f <B><FONT COLOR="#5F9EA0">0</FONT></B>) z
<B><FONT COLOR="#A020F0">val</FONT></B> I = <B><FONT COLOR="#A020F0">fn</FONT></B> z => Fold.step0 (<B><FONT COLOR="#A020F0">fn</FONT></B> f => <B><FONT COLOR="#A020F0">fn</FONT></B> i => <B><FONT COLOR="#A020F0">fn</FONT></B> x => f (x + i)) z
</PRE>
<p>
</p>
<p>
The <tt>makeSum</tt> folder constructs a function whose arity depends on the number of <tt>I</tt> arguments and that adds together all of its arguments. For example, <tt>makeSum I $</tt> is of type <tt>int -> int</tt> and <tt>makeSum I I $</tt> is of type <tt>int -> int -> int</tt>.
</p>
<p>
One can use the fold equation to verify that the <tt>makeSum</tt> works correctly. For example, one can easily check by hand the following equations.
<pre class=code>
makeSum I $ <B><FONT COLOR="#5F9EA0">1</FONT></B> = <B><FONT COLOR="#5F9EA0">1</FONT></B>
makeSum I I $ <B><FONT COLOR="#5F9EA0">1</FONT></B> <B><FONT COLOR="#5F9EA0">2</FONT></B> = <B><FONT COLOR="#5F9EA0">3</FONT></B>
makeSum I I I $ <B><FONT COLOR="#5F9EA0">1</FONT></B> <B><FONT COLOR="#5F9EA0">2</FONT></B> <B><FONT COLOR="#5F9EA0">3</FONT></B> = <B><FONT COLOR="#5F9EA0">6</FONT></B>
</PRE>
</p>
<p>
Returning a function becomes especially interesting when there are steppers of different types. For example, the following <tt>makeSum</tt> folder constructs functions that sum integers and reals.
</p>
<pre class=code>
<B><FONT COLOR="#A020F0">val</FONT></B> makeSum = <B><FONT COLOR="#A020F0">fn</FONT></B> z => Foldr.foldr (id, <B><FONT COLOR="#A020F0">fn</FONT></B> f => f <B><FONT COLOR="#5F9EA0">0.0</FONT></B>) z
<B><FONT COLOR="#A020F0">val</FONT></B> I = <B><FONT COLOR="#A020F0">fn</FONT></B> z => Foldr.step0 (<B><FONT COLOR="#A020F0">fn</FONT></B> f => <B><FONT COLOR="#A020F0">fn</FONT></B> x => <B><FONT COLOR="#A020F0">fn</FONT></B> i => f (x + real i)) z
<B><FONT COLOR="#A020F0">val</FONT></B> R = <B><FONT COLOR="#A020F0">fn</FONT></B> z => Foldr.step0 (<B><FONT COLOR="#A020F0">fn</FONT></B> f => <B><FONT COLOR="#A020F0">fn</FONT></B> x: real => <B><FONT COLOR="#A020F0">fn</FONT></B> r => f (x + r)) z
</PRE>
<p>
</p>
<p>
With these definitions, <tt>makeSum I R $</tt> is of type <tt>int -> real -> real</tt> and <tt>makeSum R I I $</tt> is of type <tt>real -> int -> int -> real</tt>. One can use the foldr equation to check the following equations.
</p>
<pre class=code>
makeSum I $ <B><FONT COLOR="#5F9EA0">1</FONT></B> = <B><FONT COLOR="#5F9EA0">1.0</FONT></B>
makeSum I R $ <B><FONT COLOR="#5F9EA0">1</FONT></B> <B><FONT COLOR="#5F9EA0">2.5</FONT></B> = <B><FONT COLOR="#5F9EA0">3.5</FONT></B>
makeSum R I I $ <B><FONT COLOR="#5F9EA0">1.5</FONT></B> <B><FONT COLOR="#5F9EA0">2</FONT></B> <B><FONT COLOR="#5F9EA0">3</FONT></B> = <B><FONT COLOR="#5F9EA0">6.5</FONT></B>
</PRE>
<p>
</p>
<p>
We used <tt>foldr</tt> instead of <tt>fold</tt> for this so that the order in which the specifiers <tt>I</tt> and <tt>R</tt> appear is the same as the order in which the arguments appear. Had we used <tt>fold</tt>, things would have been reversed.
</p>
<p>
An extension of this idea is sufficient to define <a href="Printf">Printf</a>-like functions in SML.
</p>
<h2 id="head-82cd031b8d5489f5a7ab5198fd65d34f16cada72">An idiom for combining steps</h2>
<p>
It is sometimes useful to combine a number of steps together and name them as a single step. As a simple example, suppose that one often sees an integer follower by a real in the <tt>makeSum</tt> example above. One can define a new <em>compound step</em> <tt>IR</tt> as follows.
</p>
<pre class=code>
<B><FONT COLOR="#A020F0">val</FONT></B> IR = <B><FONT COLOR="#A020F0">fn</FONT></B> u => Fold.fold u I R
</PRE>
<p>
</p>
<p>
With this definition in place, one can verify the following.
</p>
<pre class=code>
makeSum IR IR $ <B><FONT COLOR="#5F9EA0">1</FONT></B> <B><FONT COLOR="#5F9EA0">2.2</FONT></B> <B><FONT COLOR="#5F9EA0">3</FONT></B> <B><FONT COLOR="#5F9EA0">4.4</FONT></B> = <B><FONT COLOR="#5F9EA0">10.6</FONT></B>
</PRE>
<p>
</p>
<p>
In general, one can combine steps <tt>s1</tt>, <tt>s2</tt>, ... <tt>sn</tt> as
</p>
<pre class=code>
<B><FONT COLOR="#A020F0">fn</FONT></B> u => Fold.fold u s1 s2 ... sn
</PRE>
<p>
</p>
<p>
The following calculation shows why a compound step behaves as the composition of its constituent steps.
</p>
<pre class=code>
fold u (<B><FONT COLOR="#A020F0">fn</FONT></B> u => fold u s1 s2 ... sn)
= (<B><FONT COLOR="#A020F0">fn</FONT></B> u => fold u s1 s2 ... sn) u
= fold u s1 s2 ... sn
</PRE>
<p>
</p>
<h2 id="head-ffb763362aa4b1ab8da43465778c25a2676ba059">Post composition</h2>
<p>
Suppose we already have a function defined via fold, <tt>w = fold (a, f)</tt>, and we would like to construct a new fold function that is like <tt>w</tt>, but applies <tt>g</tt> to the result produced by <tt>w</tt>. This is similar to function composition, but we can't just do <tt>g o w</tt>, because we don't want to use <tt>g</tt> until <tt>w</tt> has been applied to all of its arguments and received the end-of-arguments terminator <tt>$</tt>.
</p>
<p>
More precisely, we want to define a post-composition function <tt>post</tt> that satisfies the following equation.
</p>
<pre class=code>
post (w, g) s1 ... sn $ = g (w s1 ... sn $)
</PRE>
<p>
</p>
<p>
Here is the definition of <tt>post</tt>.
</p>
<pre class=code>
<B><FONT COLOR="#0000FF">structure</FONT></B> Fold =
<B><FONT COLOR="#0000FF">struct</FONT></B>
<B><FONT COLOR="#0000FF">open</FONT></B> Fold
<B><FONT COLOR="#A020F0">fun</FONT></B> post (w, g) s = w (<B><FONT COLOR="#A020F0">fn</FONT></B> (a, h) => s (a, g o h))
<B><FONT COLOR="#0000FF">end</FONT></B>
</PRE>
<p>
</p>
<p>
The following calculations show that <tt>post</tt> satisfies the desired equation, where <tt>w = fold (a, f)</tt>.
</p>
<pre class=code>
post (w, g) s
= w (<B><FONT COLOR="#A020F0">fn</FONT></B> (a, h) => s (a, g o h))
= fold (a, f) (<B><FONT COLOR="#A020F0">fn</FONT></B> (a, h) => s (a, g o h))
= (<B><FONT COLOR="#A020F0">fn</FONT></B> (a, h) => s (a, g o h)) (a, f)
= s (a, g o f)
= fold (a, g o f) s
</PRE>
<p>
</p>
<p>
Now, suppose <tt>si = step0 hi</tt> for <tt>i</tt> from <tt>1</tt> to <tt>n</tt>.
</p>
<pre class=code>
post (w, g) s1 s2 ... sn $
= fold (a, g o f) s1 s2 ... sn $
= (g o f) (hn (... (h1 a)))
= g (f (hn (... (h1 a))))
= g (fold (a, f) s1 ... sn $)
= g (w s1 ... sn $)
</PRE>
<p>
</p>
<p>
For a practical example of post composition, see <a href="ArrayLiteral">ArrayLiteral</a>.
</p>
<h2 id="head-efd0976a0ad531b6453a3782b50f0f3ff5363b1e">Lift</h2>
<p>
We now define a peculiar-looking function, <tt>lift0</tt>, that is, equationally speaking, equivalent to the identity function on a step function.
</p>
<pre class=code>
<B><FONT COLOR="#A020F0">fun</FONT></B> lift0 s (a, f) = fold (fold (a, id) s $, f)
</PRE>
<p>
</p>
<p>
Using the definitions, we can prove the following equation.
</p>
<pre class=code>
fold (a, f) (lift0 (step0 h)) = fold (a, f) (step0 h)
</PRE>
<p>
</p>
<p>
Here is the proof.
</p>
<pre class=code>
fold (a, f) (lift0 (step0 h))
= lift0 (step0 h) (a, f)
= fold (fold (a, id) (step0 h) $, f)
= fold (step0 h (a, id) $, f)
= fold (fold (h a, id) $, f)
= fold ($ (h a, id), f)
= fold (id (h a), f)
= fold (h a, f)
= step0 h (a, f)
= fold (a, f) (step0 h)
</PRE>
<p>
</p>
<p>
If <tt>lift0</tt> is the identity, then why even define it? The answer lies in the typing of fold expressions, which we have, until now, left unexplained.
</p>
<h2 id="head-5614fd83d7c12176ea9d23d0a9a55af15cb1297f">Typing</h2>
<p>
Perhaps the most surprising aspect of fold is that it can be checked by the SML type system. The types involved in fold expressions are complex; fortunately type inference is able to deduce them. Nevertheless, it is instructive to study the types of fold functions and steppers. More importantly, it is essential to understand the typing aspects of fold in order to write down signatures of functions defined using fold and step.
</p>
<p>
Here is the <tt>FOLD</tt> signature, and a recapitulation of the entire <tt>Fold</tt> structure, with additional type annotations.
</p>
<pre class=code>
<B><FONT COLOR="#0000FF">signature</FONT></B> FOLD =
<B><FONT COLOR="#0000FF">sig</FONT></B>
<B><FONT COLOR="#A020F0">type</FONT></B><B><FONT COLOR="#228B22"> ('a, 'b, 'c, 'd) step </FONT></B>=<B><FONT COLOR="#228B22"> 'a * ('b -> 'c) -> 'd
</FONT></B><B><FONT COLOR="#A020F0">type</FONT></B><B><FONT COLOR="#228B22"> ('a, 'b, 'c, 'd) t </FONT></B>=<B><FONT COLOR="#228B22"> ('a, 'b, 'c, 'd) step -> 'd
</FONT></B><B><FONT COLOR="#A020F0">type</FONT></B><B><FONT COLOR="#228B22"> ('a1, 'a2, 'b, 'c, 'd) step0 </FONT></B>=<B><FONT COLOR="#228B22">
('a1, 'b, 'c, ('a2, 'b, 'c, 'd) t) step
</FONT></B><B><FONT COLOR="#A020F0">type</FONT></B><B><FONT COLOR="#228B22"> ('a11, 'a12, 'a2, 'b, 'c, 'd) step1 </FONT></B>=<B><FONT COLOR="#228B22">
('a12, 'b, 'c, 'a11 -> ('a2, 'b, 'c, 'd) t) step
</FONT></B><B><FONT COLOR="#A020F0">val</FONT></B> fold: 'a * ('b -> 'c) -> ('a, 'b, 'c, 'd) t
<B><FONT COLOR="#A020F0">val</FONT></B> lift0: ('a1, 'a2, 'a2, 'a2, 'a2) step0
-> ('a1, 'a2, 'b, 'c, 'd) step0
<B><FONT COLOR="#A020F0">val</FONT></B> post: ('a, 'b, 'c1, 'd) t * ('c1 -> 'c2)
-> ('a, 'b, 'c2, 'd) t
<B><FONT COLOR="#A020F0">val</FONT></B> step0: ('a1 -> 'a2) -> ('a1, 'a2, 'b, 'c, 'd) step0
<B><FONT COLOR="#A020F0">val</FONT></B> step1: ('a11 * 'a12 -> 'a2)
-> ('a11, 'a12, 'a2, 'b, 'c, 'd) step1
<B><FONT COLOR="#0000FF">end</FONT></B>
<B><FONT COLOR="#0000FF">structure</FONT></B> Fold:> FOLD =
<B><FONT COLOR="#0000FF">struct</FONT></B>
<B><FONT COLOR="#A020F0">type</FONT></B><B><FONT COLOR="#228B22"> ('a, 'b, 'c, 'd) step </FONT></B>=<B><FONT COLOR="#228B22"> 'a * ('b -> 'c) -> 'd
</FONT></B><B><FONT COLOR="#A020F0">type</FONT></B><B><FONT COLOR="#228B22"> ('a, 'b, 'c, 'd) t </FONT></B>=<B><FONT COLOR="#228B22"> ('a, 'b, 'c, 'd) step -> 'd
</FONT></B><B><FONT COLOR="#A020F0">type</FONT></B><B><FONT COLOR="#228B22"> ('a1, 'a2, 'b, 'c, 'd) step0 </FONT></B>=<B><FONT COLOR="#228B22">
('a1, 'b, 'c, ('a2, 'b, 'c, 'd) t) step
</FONT></B><B><FONT COLOR="#A020F0">type</FONT></B><B><FONT COLOR="#228B22"> ('a11, 'a12, 'a2, 'b, 'c, 'd) step1 </FONT></B>=<B><FONT COLOR="#228B22">
('a12, 'b, 'c, 'a11 -> ('a2, 'b, 'c, 'd) t) step
</FONT></B><B><FONT COLOR="#A020F0">fun</FONT></B> fold (a: 'a, f: 'b -> 'c)
(g: ('a, 'b, 'c, 'd) step): 'd =
g (a, f)
<B><FONT COLOR="#A020F0">fun</FONT></B> step0 (h: 'a1 -> 'a2)
(a1: 'a1, f: 'b -> 'c): ('a2, 'b, 'c, 'd) t =
fold (h a1, f)
<B><FONT COLOR="#A020F0">fun</FONT></B> step1 (h: 'a11 * 'a12 -> 'a2)
(a12: 'a12, f: 'b -> 'c)
(a11: 'a11): ('a2, 'b, 'c, 'd) t =
fold (h (a11, a12), f)
<B><FONT COLOR="#A020F0">fun</FONT></B> lift0 (s: ('a1, 'a2, 'a2, 'a2, 'a2) step0)
(a: 'a1, f: 'b -> 'c): ('a2, 'b, 'c, 'd) t =
fold (fold (a, id) s $, f)
<B><FONT COLOR="#A020F0">fun</FONT></B> post (w: ('a, 'b, 'c1, 'd) t,
g: 'c1 -> 'c2)
(s: ('a, 'b, 'c2, 'd) step): 'd =
w (<B><FONT COLOR="#A020F0">fn</FONT></B> (a, h) => s (a, g o h))
<B><FONT COLOR="#0000FF">end</FONT></B>
</PRE>
<p>
</p>
<p>
That's a lot to swallow, so let's walk through it one step at a time. First, we have the definition of type <tt>Fold.step</tt>.
<pre class=code>
<B><FONT COLOR="#A020F0">type</FONT></B><B><FONT COLOR="#228B22"> ('a, 'b, 'c, 'd) step </FONT></B>=<B><FONT COLOR="#228B22"> 'a * ('b -> 'c) -> 'd
</FONT></B></PRE>
</p>
<p>
As a fold proceeds over its arguments, it maintains two things: the accumulator, of type <tt>'a</tt>, and the finishing function, of type <tt>'b -> 'c</tt>. Each step in the fold is a function that takes those two pieces (i.e. <tt>'a * ('b -> 'c)</tt> and does something to them (i.e. produces <tt>'d</tt>). The result type of the step is completely left open to be filled in by type inference, as it is an arrow type that is capable of consuming the rest of the arguments to the fold.
</p>
<p>
A folder, of type <tt>Fold.t</tt>, is a function that consumes a single step.
<pre class=code>
<B><FONT COLOR="#A020F0">type</FONT></B><B><FONT COLOR="#228B22"> ('a, 'b, 'c, 'd) t </FONT></B>=<B><FONT COLOR="#228B22"> ('a, 'b, 'c, 'd) step -> 'd
</FONT></B></PRE>
</p>
<p>
Expanding out the type, we have:
<pre class=code>
<B><FONT COLOR="#A020F0">type</FONT></B><B><FONT COLOR="#228B22"> ('a, 'b, 'c, 'd) t </FONT></B>=<B><FONT COLOR="#228B22"> ('a * ('b -> 'c) -> 'd) -> 'd
</FONT></B></PRE>
</p>
<p>
This shows that the only thing a folder does is to hand its accumulator (<tt>'a</tt>) and finisher (<tt>'b -> 'c</tt>) to the next step (<tt>'a * ('b -> 'c) -> 'd</tt>). If SML had <a href="FirstClassPolymorphism">first-class polymorphism</a>, we would write the fold type as follows.
</p>
<pre class=code>
<B><FONT COLOR="#A020F0">type</FONT></B><B><FONT COLOR="#228B22"> ('a, 'b, 'c) t </FONT></B>=<B><FONT COLOR="#228B22"> Forall 'd. ('a, 'b, 'c, 'd) step -> 'd
</FONT></B></PRE>
<p>
</p>
<p>
This type definition shows that a folder had nothing to do with the rest of the fold, it only deals with the next step.
</p>
<p>
We now can understand the type of <tt>fold</tt>, which takes the initial value of the accumulator and the finishing function, and constructs a folder, i.e. a function awaiting the next step.
<pre class=code>
<B><FONT COLOR="#A020F0">val</FONT></B> fold: 'a * ('b -> 'c) -> ('a, 'b, 'c, 'd) t
<B><FONT COLOR="#A020F0">fun</FONT></B> fold (a: 'a, f: 'b -> 'c)
(g: ('a, 'b, 'c, 'd) step): 'd =
g (a, f)
</PRE>
</p>
<p>
Continuing on, we have the type of step functions.
</p>
<pre class=code>
<B><FONT COLOR="#A020F0">type</FONT></B><B><FONT COLOR="#228B22"> ('a1, 'a2, 'b, 'c, 'd) step0 </FONT></B>=<B><FONT COLOR="#228B22">
('a1, 'b, 'c, ('a2, 'b, 'c, 'd) t) step
</FONT></B></PRE>
<p>
</p>
<p>
Expanding out the type a bit gives:
</p>
<pre class=code>
<B><FONT COLOR="#A020F0">type</FONT></B><B><FONT COLOR="#228B22"> ('a1, 'a2, 'b, 'c, 'd) step0 </FONT></B>=<B><FONT COLOR="#228B22">
'a1 * ('b -> 'c) -> ('a2, 'b, 'c, 'd) t
</FONT></B></PRE>
<p>
</p>
<p>
So, a step function takes the accumulator (<tt>'a1</tt>) and finishing function (<tt>'b -> 'c</tt>), which will be passed to it by the previous folder, and transforms them to a new folder. This new folder has a new accumulator (<tt>'a2</tt>) and the same finishing function.
</p>
<p>
Again, imagining that SML had <a href="FirstClassPolymorphism">first-class polymorphism</a> makes the type clearer.
<pre class=code>
<B><FONT COLOR="#A020F0">type</FONT></B><B><FONT COLOR="#228B22"> ('a1, 'a2) step0 </FONT></B>=<B><FONT COLOR="#228B22">
Forall ('b, 'c). ('a1, 'b, 'c, ('a2, 'b, 'c) t) step
</FONT></B></PRE>
</p>
<p>
Thus, in essence, a <tt>step0</tt> function is a wrapper around a function of type <tt>'a1 -> 'a2</tt>, which is exactly what the definition of <tt>step0</tt> does.
</p>
<pre class=code>
<B><FONT COLOR="#A020F0">val</FONT></B> step0: ('a1 -> 'a2) -> ('a1, 'a2, 'b, 'c, 'd) step0
<B><FONT COLOR="#A020F0">fun</FONT></B> step0 (h: 'a1 -> 'a2)
(a1: 'a1, f: 'b -> 'c): ('a2, 'b, 'c, 'd) t =
fold (h a1, f)
</PRE>
<p>
</p>
<p>
It is not much beyond <tt>step0</tt> to understand <tt>step1</tt>.
<pre class=code>
<B><FONT COLOR="#A020F0">type</FONT></B><B><FONT COLOR="#228B22"> ('a11, 'a12, 'a2, 'b, 'c, 'd) step1 </FONT></B>=<B><FONT COLOR="#228B22">
('a12, 'b, 'c, 'a11 -> ('a2, 'b, 'c, 'd) t) step
</FONT></B></PRE>
</p>
<p>
A <tt>step1</tt> function takes the accumulator (<tt>'a12</tt>) and finisher (<tt>'b -> 'c</tt>) passed to it by the previous folder and transforms them into a function that consumes the next argument (<tt>'a11</tt>) and produces a folder that will continue the fold with a new accumulator (<tt>'a2</tt>) and the same finisher.
</p>
<pre class=code>
<B><FONT COLOR="#A020F0">fun</FONT></B> step1 (h: 'a11 * 'a12 -> 'a2)
(a12: 'a12, f: 'b -> 'c)
(a11: 'a11): ('a2, 'b, 'c, 'd) t =
fold (h (a11, a12), f)
</PRE>
<p>
</p>
<p>
With <a href="FirstClassPolymorphism">first-class polymorphism</a>, a <tt>step1</tt> function is more clearly seen as a wrapper around a binary function of type <tt>'a11 * 'a12 -> 'a2</tt>.
</p>
<pre class=code>
<B><FONT COLOR="#A020F0">type</FONT></B><B><FONT COLOR="#228B22"> ('a11, 'a12, 'a2) step1 </FONT></B>=<B><FONT COLOR="#228B22">
Forall ('b, 'c). ('a12, 'b, 'c, 'a11 -> ('a2, 'b, 'c) t) step
</FONT></B></PRE>
<p>
</p>
<p>
The type of <tt>post</tt> is clear: it takes a folder with a finishing function that produces type <tt>'c1</tt>, and a function of type <tt>'c1 -> 'c2</tt> to postcompose onto the folder. It returns a new folder with a finishing function that produces type <tt>'c2</tt>.
</p>
<pre class=code>
<B><FONT COLOR="#A020F0">val</FONT></B> post: ('a, 'b, 'c1, 'd) t * ('c1 -> 'c2)
-> ('a, 'b, 'c2, 'd) t
<B><FONT COLOR="#A020F0">fun</FONT></B> post (w: ('a, 'b, 'c1, 'd) t,
g: 'c1 -> 'c2)
(s: ('a, 'b, 'c2, 'd) step): 'd =
w (<B><FONT COLOR="#A020F0">fn</FONT></B> (a, h) => s (a, g o h))
</PRE>
<p>
</p>
<p>
We will return to <tt>lift0</tt> after an example.
</p>
<h2 id="head-e13dce34fe568689c1beb05e209023d50ff036df">An example typing</h2>
<p>
Let's type check our simplest example, a variable-argument fold. Recall that we have a folder <tt>f</tt> and a stepper <tt>a</tt> defined as follows.
</p>
<pre class=code>
<B><FONT COLOR="#A020F0">val</FONT></B> f = <B><FONT COLOR="#A020F0">fn</FONT></B> z => Fold.fold ((), <B><FONT COLOR="#A020F0">fn</FONT></B> () => ()) z
<B><FONT COLOR="#A020F0">val</FONT></B> a = <B><FONT COLOR="#A020F0">fn</FONT></B> z => Fold.step0 (<B><FONT COLOR="#A020F0">fn</FONT></B> () => ()) z
</PRE>
<p>
</p>
<p>
Since the accumulator and finisher are uninteresting, we'll use some abbreviations to simplify things.
</p>
<pre class=code>
<B><FONT COLOR="#A020F0">type</FONT></B><B><FONT COLOR="#228B22"> 'd step </FONT></B>=<B><FONT COLOR="#228B22"> (unit, unit, unit, 'd) Fold.step
</FONT></B><B><FONT COLOR="#A020F0">type</FONT></B><B><FONT COLOR="#228B22"> 'd fold </FONT></B>=<B><FONT COLOR="#228B22"> 'd step -> 'd
</FONT></B></PRE>
<p>
</p>
<p>
With these abbreviations, <tt>f</tt> and <tt>a</tt> have the following polymorphic types.
<pre class=code>
f: 'd fold
a: 'd step
</PRE>
</p>
<p>
Suppose we want to type check
<pre class=code>
f a a a $: unit
</PRE>
As a reminder, the fully parenthesized expression is
<pre class=code>
((((f a) a) a) a) $
</PRE>
The observation that we will use repeatedly is that for any type <tt>z</tt>, if <tt>f: z fold</tt> and <tt>s: z step</tt>, then <tt>f s: z</tt>. So, if we want
<pre class=code>
(f a a a) $: unit
</PRE>
then we must have
<pre class=code>
f a a a: unit fold
$: unit step
</PRE>
Applying the observation again, we must have
<pre class=code>
f a a: unit fold fold
a: unit fold step
</PRE>
</p>
<p>
Applying the observation two more times leads to the following type derivation.
</p>
<pre class=code>
f: unit fold fold fold fold a: unit fold fold fold step
f a: unit fold fold fold a: unit fold fold step
f a a: unit fold fold a: unit fold step
f a a a: unit fold $: unit step
f a a a $: unit
</PRE>
<p>
</p>
<p>
So, each application is a fold that consumes the next step, producing a fold of one smaller type.
</p>
<p>
One can expand some of the type definitions in <tt>f</tt> to see that it is indeed a function that takes four curried arguments, each one a step function.
</p>
<pre class=code>
f: unit fold fold fold step
-> unit fold fold step
-> unit fold step
-> unit step
-> unit
</PRE>
<p>
</p>
<p>
This example shows why we must eta expand uses of <tt>fold</tt> and <tt>step0</tt> to work around the value restriction and make folders and steppers polymorphic. The type of a fold function like <tt>f</tt> depends on the number of arguments, and so will vary from use to use. Similarly, each occurrence of an argument like <tt>a</tt> has a different type, depending on the number of remaining arguments.
</p>
<p>
This example also shows that the type of a folder, when fully expanded, is exponential in the number of arguments: there are as many nested occurrences of the <tt>fold</tt> type constructor as there are arguments, and each occurrence duplicates its type argument. One can observe this exponential behavior in a type checker that doesn't share enough of the representation of types (e.g. one that represents types as trees rather than directed acyclic graphs).
</p>
<p>
Generalizing this type derivation to uses of fold where the accumulator and finisher are more interesting is straightforward. One simply includes the type of the accumulator, which may change, for each step, and the type of the finisher, which doesn't change from step to step.
</p>
<h2 id="head-f1a127be542fc4a883aa37976ca48c19faf93b98">Typing lift</h2>
<p>
The lack of <a href="FirstClassPolymorphism">first-class polymorphism</a> in SML causes problems if one wants to use a step in a first-class way. Consider the following <tt>double</tt> function, which takes a step, <tt>s</tt>, and produces a composite step that does <tt>s</tt> twice.
</p>
<pre class=code>
<B><FONT COLOR="#A020F0">fun</FONT></B> double s = <B><FONT COLOR="#A020F0">fn</FONT></B> u => Fold.fold u s s
</PRE>
<p>
</p>
<p>
The definition of <tt>double</tt> is not type correct. The problem is that the type of a step depends on the number of remaining arguments but that the parameter <tt>s</tt> is not polymorphic, and so can not be used in two different positions.
</p>
<p>
Fortunately, we can define a function, <tt>lift0</tt>, that takes a monotyped step function and <em>lifts</em> it into a polymorphic step function. This is apparent in the type of <tt>lift0</tt>.
</p>
<pre class=code>
<B><FONT COLOR="#A020F0">val</FONT></B> lift0: ('a1, 'a2, 'a2, 'a2, 'a2) step0
-> ('a1, 'a2, 'b, 'c, 'd) step0
<B><FONT COLOR="#A020F0">fun</FONT></B> lift0 (s: ('a1, 'a2, 'a2, 'a2, 'a2) step0)
(a: 'a1, f: 'b -> 'c): ('a2, 'b, 'c, 'd) t =
fold (fold (a, id) s $, f)
</PRE>
<p>
</p>
<p>
The following definition of <tt>double</tt> uses <tt>lift0</tt>, appropriately eta wrapped, to fix the problem.
<pre class=code>
<B><FONT COLOR="#A020F0">fun</FONT></B> double s =
<B><FONT COLOR="#A020F0">let</FONT></B>
<B><FONT COLOR="#A020F0">val</FONT></B> s = <B><FONT COLOR="#A020F0">fn</FONT></B> z => Fold.lift0 s z
<B><FONT COLOR="#A020F0">in</FONT></B>
<B><FONT COLOR="#A020F0">fn</FONT></B> u => Fold.fold u s s
<B><FONT COLOR="#A020F0">end</FONT></B>
</PRE>
</p>
<p>
With that definition of <tt>double</tt> in place, we can use it as in the following example.
</p>
<pre class=code>
<B><FONT COLOR="#A020F0">val</FONT></B> f = <B><FONT COLOR="#A020F0">fn</FONT></B> z => Fold.fold ((), <B><FONT COLOR="#A020F0">fn</FONT></B> () => ()) z
<B><FONT COLOR="#A020F0">val</FONT></B> a = <B><FONT COLOR="#A020F0">fn</FONT></B> z => Fold.step0 (<B><FONT COLOR="#A020F0">fn</FONT></B> () => ()) z
<B><FONT COLOR="#A020F0">val</FONT></B> a2 = <B><FONT COLOR="#A020F0">fn</FONT></B> z => double a z
<B><FONT COLOR="#A020F0">val</FONT></B> () = f a a2 a a2 $
</PRE>
<p>
</p>
<p>
Of course, we must eta wrap the call <tt>double</tt> in order to use its result, which is a step function, polymorphically.
</p>
<h2 id="head-031723e7e62f3a194de0e2e7e9aa28cd46815d0f">Hiding the type of the accumulator</h2>
<p>
For clarity and to avoid mistakes, it can be useful to hide the type of the accumulator in a fold. Reworking the simple variable-argument example to do this leads to the following.
</p>
<pre class=code>
<B><FONT COLOR="#0000FF">structure</FONT></B> S:>
<B><FONT COLOR="#0000FF">sig</FONT></B>
<B><FONT COLOR="#A020F0">type</FONT></B><B><FONT COLOR="#228B22"> ac
</FONT></B><B><FONT COLOR="#A020F0">val</FONT></B> f: (ac, ac, unit, 'd) Fold.t
<B><FONT COLOR="#A020F0">val</FONT></B> s: (ac, ac, 'b, 'c, 'd) Fold.step0
<B><FONT COLOR="#0000FF">end</FONT></B> =
<B><FONT COLOR="#0000FF">struct</FONT></B>
<B><FONT COLOR="#A020F0">type</FONT></B><B><FONT COLOR="#228B22"> ac </FONT></B>=<B><FONT COLOR="#228B22"> unit
</FONT></B><B><FONT COLOR="#A020F0">val</FONT></B> f = <B><FONT COLOR="#A020F0">fn</FONT></B> z => Fold.fold ((), <B><FONT COLOR="#A020F0">fn</FONT></B> () => ()) z
<B><FONT COLOR="#A020F0">val</FONT></B> s = <B><FONT COLOR="#A020F0">fn</FONT></B> z => Fold.step0 (<B><FONT COLOR="#A020F0">fn</FONT></B> () => ()) z
<B><FONT COLOR="#0000FF">end</FONT></B>
</PRE>
<p>
</p>
<p>
The idea is to name the accumulator type and use opaque signature matching to make it abstract. This can prevent improper manipulation of the accumulator by client code and ensure invariants that the folder and stepper would like to maintain.
</p>
<p>
For a practical example of this technique, see <a href="ArrayLiteral">ArrayLiteral</a>.
</p>
<h2 id="head-a4bc8bf5caf54b18cea9f58e83dd4acb488deb17">Also see</h2>
<p>
Fold has a number of practical applications. Here are some of them.
</p>
<ul>
<li>
<p>
<a href="ArrayLiteral">ArrayLiteral</a>
</p>
</li>
<li>
<p>
<a href="Fold01N">Fold01N</a>
</p>
</li>
<li>
<p>
<a href="FunctionalRecordUpdate">FunctionalRecordUpdate</a>
</p>
</li>
<li>
<p>
<a href="NumericLiteral">NumericLiteral</a>
</p>
</li>
<li>
<p>
<a href="OptionalArguments">OptionalArguments</a>
</p>
</li>
<li>
<p>
<a href="Printf">Printf</a>
</p>
</li>
<li>
<p>
<a href="VariableArityPolymorphism">VariableArityPolymorphism</a>
</p>
</li>
</ul>
<p>
There are a number of related techniques. Here are some of them.
</p>
<ul>
<li>
<p>
<a href="StaticSum">StaticSum</a>
</p>
</li>
<li>
<p>
<a href="TypeIndexedValues">TypeIndexedValues</a>
</p>
</li>
</ul>
</div>
<p>
<hr>
Last edited on 2009-06-10 19:19:58 by <span title="fenrir.uchicago.edu"><a href="MatthewFluet">MatthewFluet</a></span>.
</body></html>
|