1: \input texinfo @c -*-texinfo-*-
2: @comment The source is gforth.ds, from which gforth.texi is generated
3:
4: @comment TODO: nac29jan99 - a list of things to add in the next edit:
5: @comment 1. x-ref all ambiguous or implementation-defined features?
6: @comment 2. Describe the use of Auser Avariable AConstant A, etc.
7: @comment 3. words in miscellaneous section need a home.
8: @comment 4. search for TODO for other minor and major works required.
9: @comment 5. [rats] change all @var to @i in Forth source so that info
10: @comment file looks decent.
11: @c Not an improvement IMO - anton
12: @c and anyway, this should be taken up
13: @c with Karl Berry (the texinfo guy) - anton
14: @comment .. would be useful to have a word that identified all deferred words
15: @comment should semantics stuff in intro be moved to another section
16:
17: @c POSTPONE, COMPILE, [COMPILE], LITERAL should have their own section
18:
19: @comment %**start of header (This is for running Texinfo on a region.)
20: @setfilename gforth.info
21: @settitle Gforth Manual
22: @dircategory GNU programming tools
23: @direntry
24: * Gforth: (gforth). A fast interpreter for the Forth language.
25: @end direntry
26: @c The Texinfo manual also recommends doing this, but for Gforth it may
27: @c not make much sense
28: @c @dircategory Individual utilities
29: @c @direntry
30: @c * Gforth: (gforth)Invoking Gforth. gforth, gforth-fast, gforthmi
31: @c @end direntry
32:
33: @comment @setchapternewpage odd
34: @comment TODO this gets left in by HTML converter
35: @macro progstyle {}
36: Programming style note:
37: @end macro
38:
39: @macro assignment {}
40: @table @i
41: @item Assignment:
42: @end macro
43: @macro endassignment {}
44: @end table
45: @end macro
46:
47: @comment %**end of header (This is for running Texinfo on a region.)
48:
49:
50: @comment ----------------------------------------------------------
51: @comment macros for beautifying glossary entries
52: @comment if these are used, need to strip them out for HTML converter
53: @comment else they get repeated verbatim in HTML output.
54: @comment .. not working yet.
55:
56: @macro GLOSS-START {}
57: @iftex
58: @ninerm
59: @end iftex
60: @end macro
61:
62: @macro GLOSS-END {}
63: @iftex
64: @rm
65: @end iftex
66: @end macro
67:
68: @comment ----------------------------------------------------------
69:
70:
71: @include version.texi
72:
73: @ifnottex
74: This file documents Gforth @value{VERSION}
75:
76: Copyright @copyright{} 1995--2000 Free Software Foundation, Inc.
77:
78: Permission is granted to make and distribute verbatim copies of
79: this manual provided the copyright notice and this permission notice
80: are preserved on all copies.
81:
82: @ignore
83: Permission is granted to process this file through TeX and print the
84: results, provided the printed document carries a copying permission
85: notice identical to this one except for the removal of this paragraph
86: (this paragraph not being relevant to the printed manual).
87:
88: @end ignore
89: Permission is granted to copy and distribute modified versions of this
90: manual under the conditions for verbatim copying, provided also that the
91: sections entitled "Distribution" and "General Public License" are
92: included exactly as in the original, and provided that the entire
93: resulting derived work is distributed under the terms of a permission
94: notice identical to this one.
95:
96: Permission is granted to copy and distribute translations of this manual
97: into another language, under the above conditions for modified versions,
98: except that the sections entitled "Distribution" and "General Public
99: License" may be included in a translation approved by the author instead
100: of in the original English.
101: @end ifnottex
102:
103: @finalout
104: @titlepage
105: @sp 10
106: @center @titlefont{Gforth Manual}
107: @sp 2
108: @center for version @value{VERSION}
109: @sp 2
110: @center Neal Crook
111: @center Anton Ertl
112: @center Bernd Paysan
113: @center Jens Wilke
114: @sp 3
115: @center This manual is permanently under construction and was last updated on 15-Mar-2000
116:
117: @comment The following two commands start the copyright page.
118: @page
119: @vskip 0pt plus 1filll
120: Copyright @copyright{} 1995--2000 Free Software Foundation, Inc.
121:
122: @comment !! Published by ... or You can get a copy of this manual ...
123:
124: Permission is granted to make and distribute verbatim copies of
125: this manual provided the copyright notice and this permission notice
126: are preserved on all copies.
127:
128: Permission is granted to copy and distribute modified versions of this
129: manual under the conditions for verbatim copying, provided also that the
130: sections entitled "Distribution" and "General Public License" are
131: included exactly as in the original, and provided that the entire
132: resulting derived work is distributed under the terms of a permission
133: notice identical to this one.
134:
135: Permission is granted to copy and distribute translations of this manual
136: into another language, under the above conditions for modified versions,
137: except that the sections entitled "Distribution" and "General Public
138: License" may be included in a translation approved by the author instead
139: of in the original English.
140: @end titlepage
141:
142: @node Top, License, (dir), (dir)
143: @ifnottex
144: Gforth is a free implementation of ANS Forth available on many
145: personal machines. This manual corresponds to version @value{VERSION}.
146: @end ifnottex
147:
148: @menu
149: * License:: The GPL
150: * Goals:: About the Gforth Project
151: * Gforth Environment:: Starting (and exiting) Gforth
152: * Tutorial:: Hands-on Forth Tutorial
153: * Introduction:: An introduction to ANS Forth
154: * Words:: Forth words available in Gforth
155: * Error messages:: How to interpret them
156: * Tools:: Programming tools
157: * ANS conformance:: Implementation-defined options etc.
158: * Standard vs Extensions:: Should I use extensions?
159: * Model:: The abstract machine of Gforth
160: * Integrating Gforth:: Forth as scripting language for applications
161: * Emacs and Gforth:: The Gforth Mode
162: * Image Files:: @code{.fi} files contain compiled code
163: * Engine:: The inner interpreter and the primitives
164: * Binding to System Library::
165: * Cross Compiler:: The Cross Compiler
166: * Bugs:: How to report them
167: * Origin:: Authors and ancestors of Gforth
168: * Forth-related information:: Books and places to look on the WWW
169: * Word Index:: An item for each Forth word
170: * Name Index:: Forth words, only names listed
171: * Concept Index:: A menu covering many topics
172:
173: @detailmenu --- The Detailed Node Listing ---
174:
175: Gforth Environment
176:
177: * Invoking Gforth:: Getting in
178: * Leaving Gforth:: Getting out
179: * Command-line editing::
180: * Environment variables:: that affect how Gforth starts up
181: * Gforth Files:: What gets installed and where
182: * Startup speed:: When 35ms is not fast enough ...
183:
184: Forth Tutorial
185:
186: * Starting Gforth Tutorial::
187: * Syntax Tutorial::
188: * Crash Course Tutorial::
189: * Stack Tutorial::
190: * Arithmetics Tutorial::
191: * Stack Manipulation Tutorial::
192: * Using files for Forth code Tutorial::
193: * Comments Tutorial::
194: * Colon Definitions Tutorial::
195: * Decompilation Tutorial::
196: * Stack-Effect Comments Tutorial::
197: * Types Tutorial::
198: * Factoring Tutorial::
199: * Designing the stack effect Tutorial::
200: * Local Variables Tutorial::
201: * Conditional execution Tutorial::
202: * Flags and Comparisons Tutorial::
203: * General Loops Tutorial::
204: * Counted loops Tutorial::
205: * Recursion Tutorial::
206: * Leaving definitions or loops Tutorial::
207: * Return Stack Tutorial::
208: * Memory Tutorial::
209: * Characters and Strings Tutorial::
210: * Alignment Tutorial::
211: * Interpretation and Compilation Semantics and Immediacy Tutorial::
212: * Execution Tokens Tutorial::
213: * Exceptions Tutorial::
214: * Defining Words Tutorial::
215: * Arrays and Records Tutorial::
216: * POSTPONE Tutorial::
217: * Literal Tutorial::
218: * Advanced macros Tutorial::
219: * Compilation Tokens Tutorial::
220: * Wordlists and Search Order Tutorial::
221:
222: An Introduction to ANS Forth
223:
224: * Introducing the Text Interpreter::
225: * Stacks and Postfix notation::
226: * Your first definition::
227: * How does that work?::
228: * Forth is written in Forth::
229: * Review - elements of a Forth system::
230: * Where to go next::
231: * Exercises::
232:
233: Forth Words
234:
235: * Notation::
236: * Case insensitivity::
237: * Comments::
238: * Boolean Flags::
239: * Arithmetic::
240: * Stack Manipulation::
241: * Memory::
242: * Control Structures::
243: * Defining Words::
244: * Interpretation and Compilation Semantics::
245: * Tokens for Words::
246: * The Text Interpreter::
247: * Word Lists::
248: * Environmental Queries::
249: * Files::
250: * Blocks::
251: * Other I/O::
252: * Locals::
253: * Structures::
254: * Object-oriented Forth::
255: * Programming Tools::
256: * Assembler and Code Words::
257: * Threading Words::
258: * Passing Commands to the OS::
259: * Keeping track of Time::
260: * Miscellaneous Words::
261:
262: Arithmetic
263:
264: * Single precision::
265: * Double precision:: Double-cell integer arithmetic
266: * Bitwise operations::
267: * Numeric comparison::
268: * Mixed precision:: Operations with single and double-cell integers
269: * Floating Point::
270:
271: Stack Manipulation
272:
273: * Data stack::
274: * Floating point stack::
275: * Return stack::
276: * Locals stack::
277: * Stack pointer manipulation::
278:
279: Memory
280:
281: * Memory model::
282: * Dictionary allocation::
283: * Heap Allocation::
284: * Memory Access::
285: * Address arithmetic::
286: * Memory Blocks::
287:
288: Control Structures
289:
290: * Selection:: IF ... ELSE ... ENDIF
291: * Simple Loops:: BEGIN ...
292: * Counted Loops:: DO
293: * Arbitrary control structures::
294: * Calls and returns::
295: * Exception Handling::
296:
297: Defining Words
298:
299: * CREATE::
300: * Variables:: Variables and user variables
301: * Constants::
302: * Values:: Initialised variables
303: * Colon Definitions::
304: * Anonymous Definitions:: Definitions without names
305: * Supplying names:: Passing definition names as strings
306: * User-defined Defining Words::
307: * Deferred words:: Allow forward references
308: * Aliases::
309:
310: User-defined Defining Words
311:
312: * CREATE..DOES> applications::
313: * CREATE..DOES> details::
314: * Advanced does> usage example::
315:
316: Interpretation and Compilation Semantics
317:
318: * Combined words::
319:
320: Tokens for Words
321:
322: * Execution token:: represents execution/interpretation semantics
323: * Compilation token:: represents compilation semantics
324: * Name token:: represents named words
325:
326: The Text Interpreter
327:
328: * Input Sources::
329: * Number Conversion::
330: * Interpret/Compile states::
331: * Literals::
332: * Interpreter Directives::
333:
334: Word Lists
335:
336: * Vocabularies::
337: * Why use word lists?::
338: * Word list example::
339:
340: Files
341:
342: * Forth source files::
343: * General files::
344: * Search Paths::
345:
346: Search Paths
347:
348: * Source Search Paths::
349: * General Search Paths::
350:
351: Other I/O
352:
353: * Simple numeric output:: Predefined formats
354: * Formatted numeric output:: Formatted (pictured) output
355: * String Formats:: How Forth stores strings in memory
356: * Displaying characters and strings:: Other stuff
357: * Input:: Input
358:
359: Locals
360:
361: * Gforth locals::
362: * ANS Forth locals::
363:
364: Gforth locals
365:
366: * Where are locals visible by name?::
367: * How long do locals live?::
368: * Locals programming style::
369: * Locals implementation::
370:
371: Structures
372:
373: * Why explicit structure support?::
374: * Structure Usage::
375: * Structure Naming Convention::
376: * Structure Implementation::
377: * Structure Glossary::
378:
379: Object-oriented Forth
380:
381: * Why object-oriented programming?::
382: * Object-Oriented Terminology::
383: * Objects::
384: * OOF::
385: * Mini-OOF::
386: * Comparison with other object models::
387:
388: The @file{objects.fs} model
389:
390: * Properties of the Objects model::
391: * Basic Objects Usage::
392: * The Objects base class::
393: * Creating objects::
394: * Object-Oriented Programming Style::
395: * Class Binding::
396: * Method conveniences::
397: * Classes and Scoping::
398: * Dividing classes::
399: * Object Interfaces::
400: * Objects Implementation::
401: * Objects Glossary::
402:
403: The @file{oof.fs} model
404:
405: * Properties of the OOF model::
406: * Basic OOF Usage::
407: * The OOF base class::
408: * Class Declaration::
409: * Class Implementation::
410:
411: The @file{mini-oof.fs} model
412:
413: * Basic Mini-OOF Usage::
414: * Mini-OOF Example::
415: * Mini-OOF Implementation::
416:
417: Programming Tools
418:
419: * Examining::
420: * Forgetting words::
421: * Debugging:: Simple and quick.
422: * Assertions:: Making your programs self-checking.
423: * Singlestep Debugger:: Executing your program word by word.
424:
425: Assembler and Code Words
426:
427: * Code and ;code::
428: * Common Assembler:: Assembler Syntax
429: * Common Disassembler::
430: * 386 Assembler:: Deviations and special cases
431: * Alpha Assembler:: Deviations and special cases
432: * MIPS assembler:: Deviations and special cases
433: * Other assemblers:: How to write them
434:
435: Tools
436:
437: * ANS Report:: Report the words used, sorted by wordset.
438:
439: ANS conformance
440:
441: * The Core Words::
442: * The optional Block word set::
443: * The optional Double Number word set::
444: * The optional Exception word set::
445: * The optional Facility word set::
446: * The optional File-Access word set::
447: * The optional Floating-Point word set::
448: * The optional Locals word set::
449: * The optional Memory-Allocation word set::
450: * The optional Programming-Tools word set::
451: * The optional Search-Order word set::
452:
453: The Core Words
454:
455: * core-idef:: Implementation Defined Options
456: * core-ambcond:: Ambiguous Conditions
457: * core-other:: Other System Documentation
458:
459: The optional Block word set
460:
461: * block-idef:: Implementation Defined Options
462: * block-ambcond:: Ambiguous Conditions
463: * block-other:: Other System Documentation
464:
465: The optional Double Number word set
466:
467: * double-ambcond:: Ambiguous Conditions
468:
469: The optional Exception word set
470:
471: * exception-idef:: Implementation Defined Options
472:
473: The optional Facility word set
474:
475: * facility-idef:: Implementation Defined Options
476: * facility-ambcond:: Ambiguous Conditions
477:
478: The optional File-Access word set
479:
480: * file-idef:: Implementation Defined Options
481: * file-ambcond:: Ambiguous Conditions
482:
483: The optional Floating-Point word set
484:
485: * floating-idef:: Implementation Defined Options
486: * floating-ambcond:: Ambiguous Conditions
487:
488: The optional Locals word set
489:
490: * locals-idef:: Implementation Defined Options
491: * locals-ambcond:: Ambiguous Conditions
492:
493: The optional Memory-Allocation word set
494:
495: * memory-idef:: Implementation Defined Options
496:
497: The optional Programming-Tools word set
498:
499: * programming-idef:: Implementation Defined Options
500: * programming-ambcond:: Ambiguous Conditions
501:
502: The optional Search-Order word set
503:
504: * search-idef:: Implementation Defined Options
505: * search-ambcond:: Ambiguous Conditions
506:
507: Image Files
508:
509: * Image Licensing Issues:: Distribution terms for images.
510: * Image File Background:: Why have image files?
511: * Non-Relocatable Image Files:: don't always work.
512: * Data-Relocatable Image Files:: are better.
513: * Fully Relocatable Image Files:: better yet.
514: * Stack and Dictionary Sizes:: Setting the default sizes for an image.
515: * Running Image Files:: @code{gforth -i @i{file}} or @i{file}.
516: * Modifying the Startup Sequence:: and turnkey applications.
517:
518: Fully Relocatable Image Files
519:
520: * gforthmi:: The normal way
521: * cross.fs:: The hard way
522:
523: Engine
524:
525: * Portability::
526: * Threading::
527: * Primitives::
528: * Performance::
529:
530: Threading
531:
532: * Scheduling::
533: * Direct or Indirect Threaded?::
534: * DOES>::
535:
536: Primitives
537:
538: * Automatic Generation::
539: * TOS Optimization::
540: * Produced code::
541:
542: Cross Compiler
543:
544: * Using the Cross Compiler::
545: * How the Cross Compiler Works::
546:
547: Other Forth-related information
548:
549: * Internet resources::
550: * Books::
551: * The Forth Interest Group::
552: * Conferences::
553:
554: @end detailmenu
555: @end menu
556:
557: @node License, Goals, Top, Top
558: @unnumbered GNU GENERAL PUBLIC LICENSE
559: @center Version 2, June 1991
560:
561: @display
562: Copyright @copyright{} 1989, 1991 Free Software Foundation, Inc.
563: 675 Mass Ave, Cambridge, MA 02139, USA
564:
565: Everyone is permitted to copy and distribute verbatim copies
566: of this license document, but changing it is not allowed.
567: @end display
568:
569: @unnumberedsec Preamble
570:
571: The licenses for most software are designed to take away your
572: freedom to share and change it. By contrast, the GNU General Public
573: License is intended to guarantee your freedom to share and change free
574: software---to make sure the software is free for all its users. This
575: General Public License applies to most of the Free Software
576: Foundation's software and to any other program whose authors commit to
577: using it. (Some other Free Software Foundation software is covered by
578: the GNU Library General Public License instead.) You can apply it to
579: your programs, too.
580:
581: When we speak of free software, we are referring to freedom, not
582: price. Our General Public Licenses are designed to make sure that you
583: have the freedom to distribute copies of free software (and charge for
584: this service if you wish), that you receive source code or can get it
585: if you want it, that you can change the software or use pieces of it
586: in new free programs; and that you know you can do these things.
587:
588: To protect your rights, we need to make restrictions that forbid
589: anyone to deny you these rights or to ask you to surrender the rights.
590: These restrictions translate to certain responsibilities for you if you
591: distribute copies of the software, or if you modify it.
592:
593: For example, if you distribute copies of such a program, whether
594: gratis or for a fee, you must give the recipients all the rights that
595: you have. You must make sure that they, too, receive or can get the
596: source code. And you must show them these terms so they know their
597: rights.
598:
599: We protect your rights with two steps: (1) copyright the software, and
600: (2) offer you this license which gives you legal permission to copy,
601: distribute and/or modify the software.
602:
603: Also, for each author's protection and ours, we want to make certain
604: that everyone understands that there is no warranty for this free
605: software. If the software is modified by someone else and passed on, we
606: want its recipients to know that what they have is not the original, so
607: that any problems introduced by others will not reflect on the original
608: authors' reputations.
609:
610: Finally, any free program is threatened constantly by software
611: patents. We wish to avoid the danger that redistributors of a free
612: program will individually obtain patent licenses, in effect making the
613: program proprietary. To prevent this, we have made it clear that any
614: patent must be licensed for everyone's free use or not licensed at all.
615:
616: The precise terms and conditions for copying, distribution and
617: modification follow.
618:
619: @iftex
620: @unnumberedsec TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION
621: @end iftex
622: @ifnottex
623: @center TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION
624: @end ifnottex
625:
626: @enumerate 0
627: @item
628: This License applies to any program or other work which contains
629: a notice placed by the copyright holder saying it may be distributed
630: under the terms of this General Public License. The ``Program'', below,
631: refers to any such program or work, and a ``work based on the Program''
632: means either the Program or any derivative work under copyright law:
633: that is to say, a work containing the Program or a portion of it,
634: either verbatim or with modifications and/or translated into another
635: language. (Hereinafter, translation is included without limitation in
636: the term ``modification''.) Each licensee is addressed as ``you''.
637:
638: Activities other than copying, distribution and modification are not
639: covered by this License; they are outside its scope. The act of
640: running the Program is not restricted, and the output from the Program
641: is covered only if its contents constitute a work based on the
642: Program (independent of having been made by running the Program).
643: Whether that is true depends on what the Program does.
644:
645: @item
646: You may copy and distribute verbatim copies of the Program's
647: source code as you receive it, in any medium, provided that you
648: conspicuously and appropriately publish on each copy an appropriate
649: copyright notice and disclaimer of warranty; keep intact all the
650: notices that refer to this License and to the absence of any warranty;
651: and give any other recipients of the Program a copy of this License
652: along with the Program.
653:
654: You may charge a fee for the physical act of transferring a copy, and
655: you may at your option offer warranty protection in exchange for a fee.
656:
657: @item
658: You may modify your copy or copies of the Program or any portion
659: of it, thus forming a work based on the Program, and copy and
660: distribute such modifications or work under the terms of Section 1
661: above, provided that you also meet all of these conditions:
662:
663: @enumerate a
664: @item
665: You must cause the modified files to carry prominent notices
666: stating that you changed the files and the date of any change.
667:
668: @item
669: You must cause any work that you distribute or publish, that in
670: whole or in part contains or is derived from the Program or any
671: part thereof, to be licensed as a whole at no charge to all third
672: parties under the terms of this License.
673:
674: @item
675: If the modified program normally reads commands interactively
676: when run, you must cause it, when started running for such
677: interactive use in the most ordinary way, to print or display an
678: announcement including an appropriate copyright notice and a
679: notice that there is no warranty (or else, saying that you provide
680: a warranty) and that users may redistribute the program under
681: these conditions, and telling the user how to view a copy of this
682: License. (Exception: if the Program itself is interactive but
683: does not normally print such an announcement, your work based on
684: the Program is not required to print an announcement.)
685: @end enumerate
686:
687: These requirements apply to the modified work as a whole. If
688: identifiable sections of that work are not derived from the Program,
689: and can be reasonably considered independent and separate works in
690: themselves, then this License, and its terms, do not apply to those
691: sections when you distribute them as separate works. But when you
692: distribute the same sections as part of a whole which is a work based
693: on the Program, the distribution of the whole must be on the terms of
694: this License, whose permissions for other licensees extend to the
695: entire whole, and thus to each and every part regardless of who wrote it.
696:
697: Thus, it is not the intent of this section to claim rights or contest
698: your rights to work written entirely by you; rather, the intent is to
699: exercise the right to control the distribution of derivative or
700: collective works based on the Program.
701:
702: In addition, mere aggregation of another work not based on the Program
703: with the Program (or with a work based on the Program) on a volume of
704: a storage or distribution medium does not bring the other work under
705: the scope of this License.
706:
707: @item
708: You may copy and distribute the Program (or a work based on it,
709: under Section 2) in object code or executable form under the terms of
710: Sections 1 and 2 above provided that you also do one of the following:
711:
712: @enumerate a
713: @item
714: Accompany it with the complete corresponding machine-readable
715: source code, which must be distributed under the terms of Sections
716: 1 and 2 above on a medium customarily used for software interchange; or,
717:
718: @item
719: Accompany it with a written offer, valid for at least three
720: years, to give any third party, for a charge no more than your
721: cost of physically performing source distribution, a complete
722: machine-readable copy of the corresponding source code, to be
723: distributed under the terms of Sections 1 and 2 above on a medium
724: customarily used for software interchange; or,
725:
726: @item
727: Accompany it with the information you received as to the offer
728: to distribute corresponding source code. (This alternative is
729: allowed only for noncommercial distribution and only if you
730: received the program in object code or executable form with such
731: an offer, in accord with Subsection b above.)
732: @end enumerate
733:
734: The source code for a work means the preferred form of the work for
735: making modifications to it. For an executable work, complete source
736: code means all the source code for all modules it contains, plus any
737: associated interface definition files, plus the scripts used to
738: control compilation and installation of the executable. However, as a
739: special exception, the source code distributed need not include
740: anything that is normally distributed (in either source or binary
741: form) with the major components (compiler, kernel, and so on) of the
742: operating system on which the executable runs, unless that component
743: itself accompanies the executable.
744:
745: If distribution of executable or object code is made by offering
746: access to copy from a designated place, then offering equivalent
747: access to copy the source code from the same place counts as
748: distribution of the source code, even though third parties are not
749: compelled to copy the source along with the object code.
750:
751: @item
752: You may not copy, modify, sublicense, or distribute the Program
753: except as expressly provided under this License. Any attempt
754: otherwise to copy, modify, sublicense or distribute the Program is
755: void, and will automatically terminate your rights under this License.
756: However, parties who have received copies, or rights, from you under
757: this License will not have their licenses terminated so long as such
758: parties remain in full compliance.
759:
760: @item
761: You are not required to accept this License, since you have not
762: signed it. However, nothing else grants you permission to modify or
763: distribute the Program or its derivative works. These actions are
764: prohibited by law if you do not accept this License. Therefore, by
765: modifying or distributing the Program (or any work based on the
766: Program), you indicate your acceptance of this License to do so, and
767: all its terms and conditions for copying, distributing or modifying
768: the Program or works based on it.
769:
770: @item
771: Each time you redistribute the Program (or any work based on the
772: Program), the recipient automatically receives a license from the
773: original licensor to copy, distribute or modify the Program subject to
774: these terms and conditions. You may not impose any further
775: restrictions on the recipients' exercise of the rights granted herein.
776: You are not responsible for enforcing compliance by third parties to
777: this License.
778:
779: @item
780: If, as a consequence of a court judgment or allegation of patent
781: infringement or for any other reason (not limited to patent issues),
782: conditions are imposed on you (whether by court order, agreement or
783: otherwise) that contradict the conditions of this License, they do not
784: excuse you from the conditions of this License. If you cannot
785: distribute so as to satisfy simultaneously your obligations under this
786: License and any other pertinent obligations, then as a consequence you
787: may not distribute the Program at all. For example, if a patent
788: license would not permit royalty-free redistribution of the Program by
789: all those who receive copies directly or indirectly through you, then
790: the only way you could satisfy both it and this License would be to
791: refrain entirely from distribution of the Program.
792:
793: If any portion of this section is held invalid or unenforceable under
794: any particular circumstance, the balance of the section is intended to
795: apply and the section as a whole is intended to apply in other
796: circumstances.
797:
798: It is not the purpose of this section to induce you to infringe any
799: patents or other property right claims or to contest validity of any
800: such claims; this section has the sole purpose of protecting the
801: integrity of the free software distribution system, which is
802: implemented by public license practices. Many people have made
803: generous contributions to the wide range of software distributed
804: through that system in reliance on consistent application of that
805: system; it is up to the author/donor to decide if he or she is willing
806: to distribute software through any other system and a licensee cannot
807: impose that choice.
808:
809: This section is intended to make thoroughly clear what is believed to
810: be a consequence of the rest of this License.
811:
812: @item
813: If the distribution and/or use of the Program is restricted in
814: certain countries either by patents or by copyrighted interfaces, the
815: original copyright holder who places the Program under this License
816: may add an explicit geographical distribution limitation excluding
817: those countries, so that distribution is permitted only in or among
818: countries not thus excluded. In such case, this License incorporates
819: the limitation as if written in the body of this License.
820:
821: @item
822: The Free Software Foundation may publish revised and/or new versions
823: of the General Public License from time to time. Such new versions will
824: be similar in spirit to the present version, but may differ in detail to
825: address new problems or concerns.
826:
827: Each version is given a distinguishing version number. If the Program
828: specifies a version number of this License which applies to it and ``any
829: later version'', you have the option of following the terms and conditions
830: either of that version or of any later version published by the Free
831: Software Foundation. If the Program does not specify a version number of
832: this License, you may choose any version ever published by the Free Software
833: Foundation.
834:
835: @item
836: If you wish to incorporate parts of the Program into other free
837: programs whose distribution conditions are different, write to the author
838: to ask for permission. For software which is copyrighted by the Free
839: Software Foundation, write to the Free Software Foundation; we sometimes
840: make exceptions for this. Our decision will be guided by the two goals
841: of preserving the free status of all derivatives of our free software and
842: of promoting the sharing and reuse of software generally.
843:
844: @iftex
845: @heading NO WARRANTY
846: @end iftex
847: @ifnottex
848: @center NO WARRANTY
849: @end ifnottex
850:
851: @item
852: BECAUSE THE PROGRAM IS LICENSED FREE OF CHARGE, THERE IS NO WARRANTY
853: FOR THE PROGRAM, TO THE EXTENT PERMITTED BY APPLICABLE LAW. EXCEPT WHEN
854: OTHERWISE STATED IN WRITING THE COPYRIGHT HOLDERS AND/OR OTHER PARTIES
855: PROVIDE THE PROGRAM ``AS IS'' WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESSED
856: OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
857: MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. THE ENTIRE RISK AS
858: TO THE QUALITY AND PERFORMANCE OF THE PROGRAM IS WITH YOU. SHOULD THE
859: PROGRAM PROVE DEFECTIVE, YOU ASSUME THE COST OF ALL NECESSARY SERVICING,
860: REPAIR OR CORRECTION.
861:
862: @item
863: IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING
864: WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MAY MODIFY AND/OR
865: REDISTRIBUTE THE PROGRAM AS PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES,
866: INCLUDING ANY GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING
867: OUT OF THE USE OR INABILITY TO USE THE PROGRAM (INCLUDING BUT NOT LIMITED
868: TO LOSS OF DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY
869: YOU OR THIRD PARTIES OR A FAILURE OF THE PROGRAM TO OPERATE WITH ANY OTHER
870: PROGRAMS), EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE
871: POSSIBILITY OF SUCH DAMAGES.
872: @end enumerate
873:
874: @iftex
875: @heading END OF TERMS AND CONDITIONS
876: @end iftex
877: @ifnottex
878: @center END OF TERMS AND CONDITIONS
879: @end ifnottex
880:
881: @page
882: @unnumberedsec How to Apply These Terms to Your New Programs
883:
884: If you develop a new program, and you want it to be of the greatest
885: possible use to the public, the best way to achieve this is to make it
886: free software which everyone can redistribute and change under these terms.
887:
888: To do so, attach the following notices to the program. It is safest
889: to attach them to the start of each source file to most effectively
890: convey the exclusion of warranty; and each file should have at least
891: the ``copyright'' line and a pointer to where the full notice is found.
892:
893: @smallexample
894: @var{one line to give the program's name and a brief idea of what it does.}
895: Copyright (C) 19@var{yy} @var{name of author}
896:
897: This program is free software; you can redistribute it and/or modify
898: it under the terms of the GNU General Public License as published by
899: the Free Software Foundation; either version 2 of the License, or
900: (at your option) any later version.
901:
902: This program is distributed in the hope that it will be useful,
903: but WITHOUT ANY WARRANTY; without even the implied warranty of
904: MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
905: GNU General Public License for more details.
906:
907: You should have received a copy of the GNU General Public License
908: along with this program; if not, write to the Free Software
909: Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
910: @end smallexample
911:
912: Also add information on how to contact you by electronic and paper mail.
913:
914: If the program is interactive, make it output a short notice like this
915: when it starts in an interactive mode:
916:
917: @smallexample
918: Gnomovision version 69, Copyright (C) 19@var{yy} @var{name of author}
919: Gnomovision comes with ABSOLUTELY NO WARRANTY; for details
920: type `show w'.
921: This is free software, and you are welcome to redistribute it
922: under certain conditions; type `show c' for details.
923: @end smallexample
924:
925: The hypothetical commands @samp{show w} and @samp{show c} should show
926: the appropriate parts of the General Public License. Of course, the
927: commands you use may be called something other than @samp{show w} and
928: @samp{show c}; they could even be mouse-clicks or menu items---whatever
929: suits your program.
930:
931: You should also get your employer (if you work as a programmer) or your
932: school, if any, to sign a ``copyright disclaimer'' for the program, if
933: necessary. Here is a sample; alter the names:
934:
935: @smallexample
936: Yoyodyne, Inc., hereby disclaims all copyright interest in the program
937: `Gnomovision' (which makes passes at compilers) written by James Hacker.
938:
939: @var{signature of Ty Coon}, 1 April 1989
940: Ty Coon, President of Vice
941: @end smallexample
942:
943: This General Public License does not permit incorporating your program into
944: proprietary programs. If your program is a subroutine library, you may
945: consider it more useful to permit linking proprietary applications with the
946: library. If this is what you want to do, use the GNU Library General
947: Public License instead of this License.
948:
949: @iftex
950: @unnumbered Preface
951: @cindex Preface
952: This manual documents Gforth. Some introductory material is provided for
953: readers who are unfamiliar with Forth or who are migrating to Gforth
954: from other Forth compilers. However, this manual is primarily a
955: reference manual.
956: @end iftex
957:
958: @comment TODO much more blurb here.
959:
960: @c ******************************************************************
961: @node Goals, Gforth Environment, License, Top
962: @comment node-name, next, previous, up
963: @chapter Goals of Gforth
964: @cindex goals of the Gforth project
965: The goal of the Gforth Project is to develop a standard model for
966: ANS Forth. This can be split into several subgoals:
967:
968: @itemize @bullet
969: @item
970: Gforth should conform to the ANS Forth Standard.
971: @item
972: It should be a model, i.e. it should define all the
973: implementation-dependent things.
974: @item
975: It should become standard, i.e. widely accepted and used. This goal
976: is the most difficult one.
977: @end itemize
978:
979: To achieve these goals Gforth should be
980: @itemize @bullet
981: @item
982: Similar to previous models (fig-Forth, F83)
983: @item
984: Powerful. It should provide for all the things that are considered
985: necessary today and even some that are not yet considered necessary.
986: @item
987: Efficient. It should not get the reputation of being exceptionally
988: slow.
989: @item
990: Free.
991: @item
992: Available on many machines/easy to port.
993: @end itemize
994:
995: Have we achieved these goals? Gforth conforms to the ANS Forth
996: standard. It may be considered a model, but we have not yet documented
997: which parts of the model are stable and which parts we are likely to
998: change. It certainly has not yet become a de facto standard, but it
999: appears to be quite popular. It has some similarities to and some
1000: differences from previous models. It has some powerful features, but not
1001: yet everything that we envisioned. We certainly have achieved our
1002: execution speed goals (@pxref{Performance})@footnote{However, in 1998
1003: the bar was raised when the major commercial Forth vendors switched to
1004: native code compilers.}. It is free and available on many machines.
1005:
1006: @c ******************************************************************
1007: @node Gforth Environment, Tutorial, Goals, Top
1008: @chapter Gforth Environment
1009: @cindex Gforth environment
1010:
1011: Note: ultimately, the Gforth man page will be auto-generated from the
1012: material in this chapter.
1013:
1014: @menu
1015: * Invoking Gforth:: Getting in
1016: * Leaving Gforth:: Getting out
1017: * Command-line editing::
1018: * Environment variables:: that affect how Gforth starts up
1019: * Gforth Files:: What gets installed and where
1020: * Startup speed:: When 35ms is not fast enough ...
1021: @end menu
1022:
1023: For related information about the creation of images see @ref{Image Files}.
1024:
1025: @comment ----------------------------------------------
1026: @node Invoking Gforth, Leaving Gforth, Gforth Environment, Gforth Environment
1027: @section Invoking Gforth
1028: @cindex invoking Gforth
1029: @cindex running Gforth
1030: @cindex command-line options
1031: @cindex options on the command line
1032: @cindex flags on the command line
1033:
1034: Gforth is made up of two parts; an executable ``engine'' (named
1035: @file{gforth} or @file{gforth-fast}) and an image file. To start it, you
1036: will usually just say @code{gforth} -- this automatically loads the
1037: default image file @file{gforth.fi}. In many other cases the default
1038: Gforth image will be invoked like this:
1039: @example
1040: gforth [file | -e forth-code] ...
1041: @end example
1042: @noindent
1043: This interprets the contents of the files and the Forth code in the order they
1044: are given.
1045:
1046: In addition to the @file{gforth} engine, there is also an engine called
1047: @file{gforth-fast}, which is faster, but gives less informative error
1048: messages (@pxref{Error messages}).
1049:
1050: In general, the command line looks like this:
1051:
1052: @example
1053: gforth[-fast] [engine options] [image options]
1054: @end example
1055:
1056: The engine options must come before the rest of the command
1057: line. They are:
1058:
1059: @table @code
1060: @cindex -i, command-line option
1061: @cindex --image-file, command-line option
1062: @item --image-file @i{file}
1063: @itemx -i @i{file}
1064: Loads the Forth image @i{file} instead of the default
1065: @file{gforth.fi} (@pxref{Image Files}).
1066:
1067: @cindex --appl-image, command-line option
1068: @item --appl-image @i{file}
1069: Loads the image @i{file} and leaves all further command-line arguments
1070: to the image (instead of processing them as engine options). This is
1071: useful for building executable application images on Unix, built with
1072: @code{gforthmi --application ...}.
1073:
1074: @cindex --path, command-line option
1075: @cindex -p, command-line option
1076: @item --path @i{path}
1077: @itemx -p @i{path}
1078: Uses @i{path} for searching the image file and Forth source code files
1079: instead of the default in the environment variable @code{GFORTHPATH} or
1080: the path specified at installation time (e.g.,
1081: @file{/usr/local/share/gforth/0.2.0:.}). A path is given as a list of
1082: directories, separated by @samp{:} (on Unix) or @samp{;} (on other OSs).
1083:
1084: @cindex --dictionary-size, command-line option
1085: @cindex -m, command-line option
1086: @cindex @i{size} parameters for command-line options
1087: @cindex size of the dictionary and the stacks
1088: @item --dictionary-size @i{size}
1089: @itemx -m @i{size}
1090: Allocate @i{size} space for the Forth dictionary space instead of
1091: using the default specified in the image (typically 256K). The
1092: @i{size} specification for this and subsequent options consists of
1093: an integer and a unit (e.g.,
1094: @code{4M}). The unit can be one of @code{b} (bytes), @code{e} (element
1095: size, in this case Cells), @code{k} (kilobytes), @code{M} (Megabytes),
1096: @code{G} (Gigabytes), and @code{T} (Terabytes). If no unit is specified,
1097: @code{e} is used.
1098:
1099: @cindex --data-stack-size, command-line option
1100: @cindex -d, command-line option
1101: @item --data-stack-size @i{size}
1102: @itemx -d @i{size}
1103: Allocate @i{size} space for the data stack instead of using the
1104: default specified in the image (typically 16K).
1105:
1106: @cindex --return-stack-size, command-line option
1107: @cindex -r, command-line option
1108: @item --return-stack-size @i{size}
1109: @itemx -r @i{size}
1110: Allocate @i{size} space for the return stack instead of using the
1111: default specified in the image (typically 15K).
1112:
1113: @cindex --fp-stack-size, command-line option
1114: @cindex -f, command-line option
1115: @item --fp-stack-size @i{size}
1116: @itemx -f @i{size}
1117: Allocate @i{size} space for the floating point stack instead of
1118: using the default specified in the image (typically 15.5K). In this case
1119: the unit specifier @code{e} refers to floating point numbers.
1120:
1121: @cindex --locals-stack-size, command-line option
1122: @cindex -l, command-line option
1123: @item --locals-stack-size @i{size}
1124: @itemx -l @i{size}
1125: Allocate @i{size} space for the locals stack instead of using the
1126: default specified in the image (typically 14.5K).
1127:
1128: @cindex -h, command-line option
1129: @cindex --help, command-line option
1130: @item --help
1131: @itemx -h
1132: Print a message about the command-line options
1133:
1134: @cindex -v, command-line option
1135: @cindex --version, command-line option
1136: @item --version
1137: @itemx -v
1138: Print version and exit
1139:
1140: @cindex --debug, command-line option
1141: @item --debug
1142: Print some information useful for debugging on startup.
1143:
1144: @cindex --offset-image, command-line option
1145: @item --offset-image
1146: Start the dictionary at a slightly different position than would be used
1147: otherwise (useful for creating data-relocatable images,
1148: @pxref{Data-Relocatable Image Files}).
1149:
1150: @cindex --no-offset-im, command-line option
1151: @item --no-offset-im
1152: Start the dictionary at the normal position.
1153:
1154: @cindex --clear-dictionary, command-line option
1155: @item --clear-dictionary
1156: Initialize all bytes in the dictionary to 0 before loading the image
1157: (@pxref{Data-Relocatable Image Files}).
1158:
1159: @cindex --die-on-signal, command-line-option
1160: @item --die-on-signal
1161: Normally Gforth handles most signals (e.g., the user interrupt SIGINT,
1162: or the segmentation violation SIGSEGV) by translating it into a Forth
1163: @code{THROW}. With this option, Gforth exits if it receives such a
1164: signal. This option is useful when the engine and/or the image might be
1165: severely broken (such that it causes another signal before recovering
1166: from the first); this option avoids endless loops in such cases.
1167: @end table
1168:
1169: @cindex loading files at startup
1170: @cindex executing code on startup
1171: @cindex batch processing with Gforth
1172: As explained above, the image-specific command-line arguments for the
1173: default image @file{gforth.fi} consist of a sequence of filenames and
1174: @code{-e @var{forth-code}} options that are interpreted in the sequence
1175: in which they are given. The @code{-e @var{forth-code}} or
1176: @code{--evaluate @var{forth-code}} option evaluates the Forth
1177: code. This option takes only one argument; if you want to evaluate more
1178: Forth words, you have to quote them or use @code{-e} several times. To exit
1179: after processing the command line (instead of entering interactive mode)
1180: append @code{-e bye} to the command line.
1181:
1182: @cindex versions, invoking other versions of Gforth
1183: If you have several versions of Gforth installed, @code{gforth} will
1184: invoke the version that was installed last. @code{gforth-@i{version}}
1185: invokes a specific version. If your environment contains the variable
1186: @code{GFORTHPATH}, you may want to override it by using the
1187: @code{--path} option.
1188:
1189: Not yet implemented:
1190: On startup the system first executes the system initialization file
1191: (unless the option @code{--no-init-file} is given; note that the system
1192: resulting from using this option may not be ANS Forth conformant). Then
1193: the user initialization file @file{.gforth.fs} is executed, unless the
1194: option @code{--no-rc} is given; this file is searched for in @file{.},
1195: then in @file{~}, then in the normal path (see above).
1196:
1197:
1198:
1199: @comment ----------------------------------------------
1200: @node Leaving Gforth, Command-line editing, Invoking Gforth, Gforth Environment
1201: @section Leaving Gforth
1202: @cindex Gforth - leaving
1203: @cindex leaving Gforth
1204:
1205: You can leave Gforth by typing @code{bye} or @kbd{Ctrl-d} (at the start
1206: of a line) or (if you invoked Gforth with the @code{--die-on-signal}
1207: option) @kbd{Ctrl-c}. When you leave Gforth, all of your definitions and
1208: data are discarded. For ways of saving the state of the system before
1209: leaving Gforth see @ref{Image Files}.
1210:
1211: doc-bye
1212:
1213:
1214: @comment ----------------------------------------------
1215: @node Command-line editing, Environment variables, Leaving Gforth, Gforth Environment
1216: @section Command-line editing
1217: @cindex command-line editing
1218:
1219: Gforth maintains a history file that records every line that you type to
1220: the text interpreter. This file is preserved between sessions, and is
1221: used to provide a command-line recall facility; if you type @kbd{Ctrl-P}
1222: repeatedly you can recall successively older commands from this (or
1223: previous) session(s). The full list of command-line editing facilities is:
1224:
1225: @itemize @bullet
1226: @item
1227: @kbd{Ctrl-p} (``previous'') (or up-arrow) to recall successively older
1228: commands from the history buffer.
1229: @item
1230: @kbd{Ctrl-n} (``next'') (or down-arrow) to recall successively newer commands
1231: from the history buffer.
1232: @item
1233: @kbd{Ctrl-f} (or right-arrow) to move the cursor right, non-destructively.
1234: @item
1235: @kbd{Ctrl-b} (or left-arrow) to move the cursor left, non-destructively.
1236: @item
1237: @kbd{Ctrl-h} (backspace) to delete the character to the left of the cursor,
1238: closing up the line.
1239: @item
1240: @kbd{Ctrl-k} to delete (``kill'') from the cursor to the end of the line.
1241: @item
1242: @kbd{Ctrl-a} to move the cursor to the start of the line.
1243: @item
1244: @kbd{Ctrl-e} to move the cursor to the end of the line.
1245: @item
1246: @key{RET} (@kbd{Ctrl-m}) or @key{LFD} (@kbd{Ctrl-j}) to submit the current
1247: line.
1248: @item
1249: @key{TAB} to step through all possible full-word completions of the word
1250: currently being typed.
1251: @item
1252: @kbd{Ctrl-d} on an empty line line to terminate Gforth (gracefully,
1253: using @code{bye}).
1254: @item
1255: @kbd{Ctrl-x} (or @code{Ctrl-d} on a non-empty line) to delete the
1256: character under the cursor.
1257: @end itemize
1258:
1259: When editing, displayable characters are inserted to the left of the
1260: cursor position; the line is always in ``insert'' (as opposed to
1261: ``overstrike'') mode.
1262:
1263: @cindex history file
1264: @cindex @file{.gforth-history}
1265: On Unix systems, the history file is @file{~/.gforth-history} by
1266: default@footnote{i.e. it is stored in the user's home directory.}. You
1267: can find out the name and location of your history file using:
1268:
1269: @example
1270: history-file type \ Unix-class systems
1271:
1272: history-file type \ Other systems
1273: history-dir type
1274: @end example
1275:
1276: If you enter long definitions by hand, you can use a text editor to
1277: paste them out of the history file into a Forth source file for reuse at
1278: a later time.
1279:
1280: Gforth never trims the size of the history file, so you should do this
1281: periodically, if necessary.
1282:
1283: @comment this is all defined in history.fs
1284: @comment NAC TODO the ctrl-D behaviour can either do a bye or a beep.. how is that option
1285: @comment chosen?
1286:
1287:
1288: @comment ----------------------------------------------
1289: @node Environment variables, Gforth Files, Command-line editing, Gforth Environment
1290: @section Environment variables
1291: @cindex environment variables
1292:
1293: Gforth uses these environment variables:
1294:
1295: @itemize @bullet
1296: @item
1297: @cindex @code{GFORTHHIST} -- environment variable
1298: @code{GFORTHHIST} -- (Unix systems only) specifies the directory in which to
1299: open/create the history file, @file{.gforth-history}. Default:
1300: @code{$HOME}.
1301:
1302: @item
1303: @cindex @code{GFORTHPATH} -- environment variable
1304: @code{GFORTHPATH} -- specifies the path used when searching for the gforth image file and
1305: for Forth source-code files.
1306:
1307: @item
1308: @cindex @code{GFORTH} -- environment variable
1309: @code{GFORTH} -- used by @file{gforthmi}, @xref{gforthmi}.
1310:
1311: @item
1312: @cindex @code{GFORTHD} -- environment variable
1313: @code{GFORTHD} -- used by @file{gforthmi}, @xref{gforthmi}.
1314:
1315: @item
1316: @cindex @code{TMP}, @code{TEMP} - environment variable
1317: @code{TMP}, @code{TEMP} - (non-Unix systems only) used as a potential
1318: location for the history file.
1319: @end itemize
1320:
1321: @comment also POSIXELY_CORRECT LINES COLUMNS HOME but no interest in
1322: @comment mentioning these.
1323:
1324: All the Gforth environment variables default to sensible values if they
1325: are not set.
1326:
1327:
1328: @comment ----------------------------------------------
1329: @node Gforth Files, Startup speed, Environment variables, Gforth Environment
1330: @section Gforth files
1331: @cindex Gforth files
1332:
1333: When you install Gforth on a Unix system, it installs files in these
1334: locations by default:
1335:
1336: @itemize @bullet
1337: @item
1338: @file{/usr/local/bin/gforth}
1339: @item
1340: @file{/usr/local/bin/gforthmi}
1341: @item
1342: @file{/usr/local/man/man1/gforth.1} - man page.
1343: @item
1344: @file{/usr/local/info} - the Info version of this manual.
1345: @item
1346: @file{/usr/local/lib/gforth/<version>/...} - Gforth @file{.fi} files.
1347: @item
1348: @file{/usr/local/share/gforth/<version>/TAGS} - Emacs TAGS file.
1349: @item
1350: @file{/usr/local/share/gforth/<version>/...} - Gforth source files.
1351: @item
1352: @file{.../emacs/site-lisp/gforth.el} - Emacs gforth mode.
1353: @end itemize
1354:
1355: You can select different places for installation by using
1356: @code{configure} options (listed with @code{configure --help}).
1357:
1358: @comment ----------------------------------------------
1359: @node Startup speed, , Gforth Files, Gforth Environment
1360: @section Startup speed
1361: @cindex Startup speed
1362: @cindex speed, startup
1363:
1364: If Gforth is used for CGI scripts or in shell scripts, its startup
1365: speed may become a problem. On a 300MHz 21064a under Linux-2.2.13 with
1366: glibc-2.0.7, @code{gforth -e bye} takes about 24.6ms user and 11.3ms
1367: system time.
1368:
1369: If startup speed is a problem, you may consider the following ways to
1370: improve it; or you may consider ways to reduce the number of startups
1371: (for example, by using Fast-CGI).
1372:
1373: The first step to improve startup speed is to statically link Gforth, by
1374: building it with @code{XLDFLAGS=-static}. This requires more memory for
1375: the code and will therefore slow down the first invocation, but
1376: subsequent invocations avoid the dynamic linking overhead. Another
1377: disadvantage is that Gforth won't profit from library upgrades. As a
1378: result, @code{gforth-static -e bye} takes about 17.1ms user and
1379: 8.2ms system time.
1380:
1381: The next step to improve startup speed is to use a non-relocatable image
1382: (@pxref{Non-Relocatable Image Files}). You can create this image with
1383: @code{gforth -e "savesystem gforthnr.fi bye"} and later use it with
1384: @code{gforth -i gforthnr.fi ...}. This avoids the relocation overhead
1385: and a part of the copy-on-write overhead. The disadvantage is that the
1386: non-relocatable image does not work if the OS gives Gforth a different
1387: address for the dictionary, for whatever reason; so you better provide a
1388: fallback on a relocatable image. @code{gforth-static -i gforthnr.fi -e
1389: bye} takes about 15.3ms user and 7.5ms system time.
1390:
1391: The final step is to disable dictionary hashing in Gforth. Gforth
1392: builds the hash table on startup, which takes much of the startup
1393: overhead. You can do this by commenting out the @code{include hash.fs}
1394: in @file{startup.fs} and everything that requires @file{hash.fs} (at the
1395: moment @file{table.fs} and @file{ekey.fs}) and then doing @code{make}.
1396: The disadvantages are that functionality like @code{table} and
1397: @code{ekey} is missing and that text interpretation (e.g., compiling)
1398: now takes much longer. So, you should only use this method if there is
1399: no significant text interpretation to perform (the script should be
1400: compiled into the image, amongst other things). @code{gforth-static -i
1401: gforthnrnh.fi -e bye} takes about 2.1ms user and 6.1ms system time.
1402:
1403: @c ******************************************************************
1404: @node Tutorial, Introduction, Gforth Environment, Top
1405: @chapter Forth Tutorial
1406: @cindex Tutorial
1407: @cindex Forth Tutorial
1408:
1409: @c Topics from nac's Introduction that could be mentioned:
1410: @c press <ret> after each line
1411: @c Prompt
1412: @c numbers vs. words in dictionary on text interpretation
1413: @c what happens on redefinition
1414: @c parsing words (in particular, defining words)
1415:
1416: This tutorial can be used with any ANS-compliant Forth; any
1417: Gforth-specific features are marked as such and you can skip them if you
1418: work with another Forth. This tutorial does not explain all features of
1419: Forth, just enough to get you started and give you some ideas about the
1420: facilities available in Forth. Read the rest of the manual and the
1421: standard when you are through this.
1422:
1423: The intended way to use this tutorial is that you work through it while
1424: sitting in front of the console, take a look at the examples and predict
1425: what they will do, then try them out; if the outcome is not as expected,
1426: find out why (e.g., by trying out variations of the example), so you
1427: understand what's going on. There are also some assignments that you
1428: should solve.
1429:
1430: This tutorial assumes that you have programmed before and know what,
1431: e.g., a loop is.
1432:
1433: @c !! explain compat library
1434:
1435: @menu
1436: * Starting Gforth Tutorial::
1437: * Syntax Tutorial::
1438: * Crash Course Tutorial::
1439: * Stack Tutorial::
1440: * Arithmetics Tutorial::
1441: * Stack Manipulation Tutorial::
1442: * Using files for Forth code Tutorial::
1443: * Comments Tutorial::
1444: * Colon Definitions Tutorial::
1445: * Decompilation Tutorial::
1446: * Stack-Effect Comments Tutorial::
1447: * Types Tutorial::
1448: * Factoring Tutorial::
1449: * Designing the stack effect Tutorial::
1450: * Local Variables Tutorial::
1451: * Conditional execution Tutorial::
1452: * Flags and Comparisons Tutorial::
1453: * General Loops Tutorial::
1454: * Counted loops Tutorial::
1455: * Recursion Tutorial::
1456: * Leaving definitions or loops Tutorial::
1457: * Return Stack Tutorial::
1458: * Memory Tutorial::
1459: * Characters and Strings Tutorial::
1460: * Alignment Tutorial::
1461: * Interpretation and Compilation Semantics and Immediacy Tutorial::
1462: * Execution Tokens Tutorial::
1463: * Exceptions Tutorial::
1464: * Defining Words Tutorial::
1465: * Arrays and Records Tutorial::
1466: * POSTPONE Tutorial::
1467: * Literal Tutorial::
1468: * Advanced macros Tutorial::
1469: * Compilation Tokens Tutorial::
1470: * Wordlists and Search Order Tutorial::
1471: @end menu
1472:
1473: @node Starting Gforth Tutorial, Syntax Tutorial, Tutorial, Tutorial
1474: @section Starting Gforth
1475: @cindex starting Gforth tutorial
1476: You can start Gforth by typing its name:
1477:
1478: @example
1479: gforth
1480: @end example
1481:
1482: That puts you into interactive mode; you can leave Gforth by typing
1483: @code{bye}. While in Gforth, you can edit the command line and access
1484: the command line history with cursor keys, similar to bash.
1485:
1486:
1487: @node Syntax Tutorial, Crash Course Tutorial, Starting Gforth Tutorial, Tutorial
1488: @section Syntax
1489: @cindex syntax tutorial
1490:
1491: A @dfn{word} is a sequence of arbitrary characters (expcept white
1492: space). Words are separated by white space. E.g., each of the
1493: following lines contains exactly one word:
1494:
1495: @example
1496: word
1497: !@@#$%^&*()
1498: 1234567890
1499: 5!a
1500: @end example
1501:
1502: A frequent beginner's error is to leave away necessary white space,
1503: resulting in an error like @samp{Undefined word}; so if you see such an
1504: error, check if you have put spaces wherever necessary.
1505:
1506: @example
1507: ." hello, world" \ correct
1508: ."hello, world" \ gives an "Undefined word" error
1509: @end example
1510:
1511: Gforth and most other Forth systems ignore differences in case (they are
1512: case-insensitive), i.e., @samp{word} is the same as @samp{Word}. If
1513: your system is case-sensitive, you may have to type all the examples
1514: given here in upper case.
1515:
1516:
1517: @node Crash Course Tutorial, Stack Tutorial, Syntax Tutorial, Tutorial
1518: @section Crash Course
1519:
1520: Type
1521:
1522: @example
1523: 0 0 !
1524: here execute
1525: ' catch >body 20 erase abort
1526: ' (quit) >body 20 erase
1527: @end example
1528:
1529: The last two examples are guaranteed to destroy parts of Gforth (and
1530: most other systems), so you better leave Gforth afterwards (if it has
1531: not finished by itself). On some systems you may have to kill gforth
1532: from outside (e.g., in Unix with @code{kill}).
1533:
1534: Now that you know how to produce crashes (and that there's not much to
1535: them), let's learn how to produce meaningful programs.
1536:
1537:
1538: @node Stack Tutorial, Arithmetics Tutorial, Crash Course Tutorial, Tutorial
1539: @section Stack
1540: @cindex stack tutorial
1541:
1542: The most obvious feature of Forth is the stack. When you type in a
1543: number, it is pushed on the stack. You can display the content of the
1544: stack with @code{.s}.
1545:
1546: @example
1547: 1 2 .s
1548: 3 .s
1549: @end example
1550:
1551: @code{.s} displays the top-of-stack to the right, i.e., the numbers
1552: appear in @code{.s} output as they appeared in the input.
1553:
1554: You can print the top of stack element with @code{.}.
1555:
1556: @example
1557: 1 2 3 . . .
1558: @end example
1559:
1560: In general, words consume their stack arguments (@code{.s} is an
1561: exception).
1562:
1563: @assignment
1564: What does the stack contain after @code{5 6 7 .}?
1565: @endassignment
1566:
1567:
1568: @node Arithmetics Tutorial, Stack Manipulation Tutorial, Stack Tutorial, Tutorial
1569: @section Arithmetics
1570: @cindex arithmetics tutorial
1571:
1572: The words @code{+}, @code{-}, @code{*}, @code{/}, and @code{mod} always
1573: operate on the top two stack items:
1574:
1575: @example
1576: 2 2 .s
1577: + .s
1578: .
1579: 2 1 - .
1580: 7 3 mod .
1581: @end example
1582:
1583: The operands of @code{-}, @code{/}, and @code{mod} are in the same order
1584: as in the corresponding infix expression (this is generally the case in
1585: Forth).
1586:
1587: Parentheses are superfluous (and not available), because the order of
1588: the words unambiguously determines the order of evaluation and the
1589: operands:
1590:
1591: @example
1592: 3 4 + 5 * .
1593: 3 4 5 * + .
1594: @end example
1595:
1596: @assignment
1597: What are the infix expressions corresponding to the Forth code above?
1598: Write @code{6-7*8+9} in Forth notation@footnote{This notation is also
1599: known as Postfix or RPN (Reverse Polish Notation).}.
1600: @endassignment
1601:
1602: To change the sign, use @code{negate}:
1603:
1604: @example
1605: 2 negate .
1606: @end example
1607:
1608: @assignment
1609: Convert -(-3)*4-5 to Forth.
1610: @endassignment
1611:
1612: @code{/mod} performs both @code{/} and @code{mod}.
1613:
1614: @example
1615: 7 3 /mod . .
1616: @end example
1617:
1618: Reference: @ref{Arithmetic}.
1619:
1620:
1621: @node Stack Manipulation Tutorial, Using files for Forth code Tutorial, Arithmetics Tutorial, Tutorial
1622: @section Stack Manipulation
1623: @cindex stack manipulation tutorial
1624:
1625: Stack manipulation words rearrange the data on the stack.
1626:
1627: @example
1628: 1 .s drop .s
1629: 1 .s dup .s drop drop .s
1630: 1 2 .s over .s drop drop drop
1631: 1 2 .s swap .s drop drop
1632: 1 2 3 .s rot .s drop drop drop
1633: @end example
1634:
1635: These are the most important stack manipulation words. There are also
1636: variants that manipulate twice as many stack items:
1637:
1638: @example
1639: 1 2 3 4 .s 2swap .s 2drop 2drop
1640: @end example
1641:
1642: Two more stack manipulation words are:
1643:
1644: @example
1645: 1 2 .s nip .s drop
1646: 1 2 .s tuck .s 2drop drop
1647: @end example
1648:
1649: @assignment
1650: Replace @code{nip} and @code{tuck} with combinations of other stack
1651: manipulation words.
1652:
1653: @example
1654: Given: How do you get:
1655: 1 2 3 3 2 1
1656: 1 2 3 1 2 3 2
1657: 1 2 3 1 2 3 3
1658: 1 2 3 1 3 3
1659: 1 2 3 2 1 3
1660: 1 2 3 4 4 3 2 1
1661: 1 2 3 1 2 3 1 2 3
1662: 1 2 3 4 1 2 3 4 1 2
1663: 1 2 3
1664: 1 2 3 1 2 3 4
1665: 1 2 3 1 3
1666: @end example
1667: @endassignment
1668:
1669: @example
1670: 5 dup * .
1671: @end example
1672:
1673: @assignment
1674: Write 17^3 and 17^4 in Forth, without writing @code{17} more than once.
1675: Write a piece of Forth code that expects two numbers on the stack
1676: (@var{a} and @var{b}, with @var{b} on top) and computes
1677: @code{(a-b)(a+1)}.
1678: @endassignment
1679:
1680: Reference: @ref{Stack Manipulation}.
1681:
1682:
1683: @node Using files for Forth code Tutorial, Comments Tutorial, Stack Manipulation Tutorial, Tutorial
1684: @section Using files for Forth code
1685: @cindex loading Forth code, tutorial
1686: @cindex files containing Forth code, tutorial
1687:
1688: While working at the Forth command line is convenient for one-line
1689: examples and short one-off code, you probably want to store your source
1690: code in files for convenient editing and persistence. You can use your
1691: favourite editor (Gforth includes Emacs support, @pxref{Emacs and
1692: Gforth}) to create @var{file} and use
1693:
1694: @example
1695: s" @var{file}" included
1696: @end example
1697:
1698: to load it into your Forth system. The file name extension I use for
1699: Forth files is @samp{.fs}.
1700:
1701: You can easily start Gforth with some files loaded like this:
1702:
1703: @example
1704: gforth @var{file1} @var{file2}
1705: @end example
1706:
1707: If an error occurs during loading these files, Gforth terminates,
1708: whereas an error during @code{INCLUDED} within Gforth usually gives you
1709: a Gforth command line. Starting the Forth system every time gives you a
1710: clean start every time, without interference from the results of earlier
1711: tries.
1712:
1713: I often put all the tests in a file, then load the code and run the
1714: tests with
1715:
1716: @example
1717: gforth @var{code} @var{tests} -e bye
1718: @end example
1719:
1720: (often by performing this command with @kbd{C-x C-e} in Emacs). The
1721: @code{-e bye} ensures that Gforth terminates afterwards so that I can
1722: restart this command without ado.
1723:
1724: The advantage of this approach is that the tests can be repeated easily
1725: every time the program ist changed, making it easy to catch bugs
1726: introduced by the change.
1727:
1728: Reference: @ref{Forth source files}.
1729:
1730:
1731: @node Comments Tutorial, Colon Definitions Tutorial, Using files for Forth code Tutorial, Tutorial
1732: @section Comments
1733: @cindex comments tutorial
1734:
1735: @example
1736: \ That's a comment; it ends at the end of the line
1737: ( Another comment; it ends here: ) .s
1738: @end example
1739:
1740: @code{\} and @code{(} are ordinary Forth words and therefore have to be
1741: separated with white space from the following text.
1742:
1743: @example
1744: \This gives an "Undefined word" error
1745: @end example
1746:
1747: The first @code{)} ends a comment started with @code{(}, so you cannot
1748: nest @code{(}-comments; and you cannot comment out text containing a
1749: @code{)} with @code{( ... )}@footnote{therefore it's a good idea to
1750: avoid @code{)} in word names.}.
1751:
1752: I use @code{\}-comments for descriptive text and for commenting out code
1753: of one or more line; I use @code{(}-comments for describing the stack
1754: effect, the stack contents, or for commenting out sub-line pieces of
1755: code.
1756:
1757: The Emacs mode @file{gforth.el} (@pxref{Emacs and Gforth}) supports
1758: these uses by commenting out a region with @kbd{C-x \}, uncommenting a
1759: region with @kbd{C-u C-x \}, and filling a @code{\}-commented region
1760: with @kbd{M-q}.
1761:
1762: Reference: @ref{Comments}.
1763:
1764:
1765: @node Colon Definitions Tutorial, Decompilation Tutorial, Comments Tutorial, Tutorial
1766: @section Colon Definitions
1767: @cindex colon definitions, tutorial
1768: @cindex definitions, tutorial
1769: @cindex procedures, tutorial
1770: @cindex functions, tutorial
1771:
1772: are similar to procedures and functions in other programming languages.
1773:
1774: @example
1775: : squared ( n -- n^2 )
1776: dup * ;
1777: 5 squared .
1778: 7 squared .
1779: @end example
1780:
1781: @code{:} starts the colon definition; its name is @code{squared}. The
1782: following comment describes its stack effect. The words @code{dup *}
1783: are not executed, but compiled into the definition. @code{;} ends the
1784: colon definition.
1785:
1786: The newly-defined word can be used like any other word, including using
1787: it in other definitions:
1788:
1789: @example
1790: : cubed ( n -- n^3 )
1791: dup squared * ;
1792: -5 cubed .
1793: : fourth-power ( n -- n^4 )
1794: squared squared ;
1795: 3 fourth-power .
1796: @end example
1797:
1798: @assignment
1799: Write colon definitions for @code{nip}, @code{tuck}, @code{negate}, and
1800: @code{/mod} in terms of other Forth words, and check if they work (hint:
1801: test your tests on the originals first). Don't let the
1802: @samp{redefined}-Messages spook you, they are just warnings.
1803: @endassignment
1804:
1805: Reference: @ref{Colon Definitions}.
1806:
1807:
1808: @node Decompilation Tutorial, Stack-Effect Comments Tutorial, Colon Definitions Tutorial, Tutorial
1809: @section Decompilation
1810: @cindex decompilation tutorial
1811: @cindex see tutorial
1812:
1813: You can decompile colon definitions with @code{see}:
1814:
1815: @example
1816: see squared
1817: see cubed
1818: @end example
1819:
1820: In Gforth @code{see} shows you a reconstruction of the source code from
1821: the executable code. Informations that were present in the source, but
1822: not in the executable code, are lost (e.g., comments).
1823:
1824: You can also decompile the predefined words:
1825:
1826: @example
1827: see .
1828: see +
1829: @end example
1830:
1831:
1832: @node Stack-Effect Comments Tutorial, Types Tutorial, Decompilation Tutorial, Tutorial
1833: @section Stack-Effect Comments
1834: @cindex stack-effect comments, tutorial
1835: @cindex --, tutorial
1836: By convention the comment after the name of a definition describes the
1837: stack effect: The part in from of the @samp{--} describes the state of
1838: the stack before the execution of the definition, i.e., the parameters
1839: that are passed into the colon definition; the part behind the @samp{--}
1840: is the state of the stack after the execution of the definition, i.e.,
1841: the results of the definition. The stack comment only shows the top
1842: stack items that the definition accesses and/or changes.
1843:
1844: You should put a correct stack effect on every definition, even if it is
1845: just @code{( -- )}. You should also add some descriptive comment to
1846: more complicated words (I usually do this in the lines following
1847: @code{:}). If you don't do this, your code becomes unreadable (because
1848: you have to work through every definition before you can undertsand
1849: any).
1850:
1851: @assignment
1852: The stack effect of @code{swap} can be written like this: @code{x1 x2 --
1853: x2 x1}. Describe the stack effect of @code{-}, @code{drop}, @code{dup},
1854: @code{over}, @code{rot}, @code{nip}, and @code{tuck}. Hint: When you
1855: are done, you can compare your stack effects to those in this manual
1856: (@pxref{Word Index}).
1857: @endassignment
1858:
1859: Sometimes programmers put comments at various places in colon
1860: definitions that describe the contents of the stack at that place (stack
1861: comments); i.e., they are like the first part of a stack-effect
1862: comment. E.g.,
1863:
1864: @example
1865: : cubed ( n -- n^3 )
1866: dup squared ( n n^2 ) * ;
1867: @end example
1868:
1869: In this case the stack comment is pretty superfluous, because the word
1870: is simple enough. If you think it would be a good idea to add such a
1871: comment to increase readability, you should also consider factoring the
1872: word into several simpler words (@pxref{Factoring Tutorial,,
1873: Factoring}), which typically eliminates the need for the stack comment;
1874: however, if you decide not to refactor it, then having such a comment is
1875: better than not having it.
1876:
1877: The names of the stack items in stack-effect and stack comments in the
1878: standard, in this manual, and in many programs specify the type through
1879: a type prefix, similar to Fortran and Hungarian notation. The most
1880: frequent prefixes are:
1881:
1882: @table @code
1883: @item n
1884: signed integer
1885: @item u
1886: unsigned integer
1887: @item c
1888: character
1889: @item f
1890: Boolean flags, i.e. @code{false} or @code{true}.
1891: @item a-addr,a-
1892: Cell-aligned address
1893: @item c-addr,c-
1894: Char-aligned address (note that a Char may have two bytes in Windows NT)
1895: @item xt
1896: Execution token, same size as Cell
1897: @item w,x
1898: Cell, can contain an integer or an address. It usually takes 32, 64 or
1899: 16 bits (depending on your platform and Forth system). A cell is more
1900: commonly known as machine word, but the term @emph{word} already means
1901: something different in Forth.
1902: @item d
1903: signed double-cell integer
1904: @item ud
1905: unsigned double-cell integer
1906: @item r
1907: Float (on the FP stack)
1908: @end table
1909:
1910: You can find a more complete list in @ref{Notation}.
1911:
1912: @assignment
1913: Write stack-effect comments for all definitions you have written up to
1914: now.
1915: @endassignment
1916:
1917:
1918: @node Types Tutorial, Factoring Tutorial, Stack-Effect Comments Tutorial, Tutorial
1919: @section Types
1920: @cindex types tutorial
1921:
1922: In Forth the names of the operations are not overloaded; so similar
1923: operations on different types need different names; e.g., @code{+} adds
1924: integers, and you have to use @code{f+} to add floating-point numbers.
1925: The following prefixes are often used for related operations on
1926: different types:
1927:
1928: @table @code
1929: @item (none)
1930: signed integer
1931: @item u
1932: unsigned integer
1933: @item c
1934: character
1935: @item d
1936: signed double-cell integer
1937: @item ud, du
1938: unsigned double-cell integer
1939: @item 2
1940: two cells (not-necessarily double-cell numbers)
1941: @item m, um
1942: mixed single-cell and double-cell operations
1943: @item f
1944: floating-point (note that in stack comments @samp{f} represents flags,
1945: and @samp{r} represents FP numbers).
1946: @end table
1947:
1948: If there are no differences between the signed and the unsigned variant
1949: (e.g., for @code{+}), there is only the prefix-less variant.
1950:
1951: Forth does not perform type checking, neither at compile time, nor at
1952: run time. If you use the wrong oeration, the data are interpreted
1953: incorrectly:
1954:
1955: @example
1956: -1 u.
1957: @end example
1958:
1959: If you have only experience with type-checked languages until now, and
1960: have heard how important type-checking is, don't panic! In my
1961: experience (and that of other Forthers), type errors in Forth code are
1962: usually easy to find (once you get used to it), the increased vigilance
1963: of the programmer tends to catch some harder errors in addition to most
1964: type errors, and you never have to work around the type system, so in
1965: most situations the lack of type-checking seems to be a win (projects to
1966: add type checking to Forth have not caught on).
1967:
1968:
1969: @node Factoring Tutorial, Designing the stack effect Tutorial, Types Tutorial, Tutorial
1970: @section Factoring
1971: @cindex factoring tutorial
1972:
1973: If you try to write longer definitions, you will soon find it hard to
1974: keep track of the stack contents. Therefore, good Forth programmers
1975: tend to write only short definitions (e.g., three lines). The art of
1976: finding meaningful short definitions is known as factoring (as in
1977: factoring polynomials).
1978:
1979: Well-factored programs offer additional advantages: smaller, more
1980: general words, are easier to test and debug and can be reused more and
1981: better than larger, specialized words.
1982:
1983: So, if you run into difficulties with stack management, when writing
1984: code, try to define meaningful factors for the word, and define the word
1985: in terms of those. Even if a factor contains only two words, it is
1986: often helpful.
1987:
1988: Good factoring is not easy, and it takes some practice to get the knack
1989: for it; but even experienced Forth programmers often don't find the
1990: right solution right away, but only when rewriting the program. So, if
1991: you don't come up with a good solution immediately, keep trying, don't
1992: despair.
1993:
1994: @c example !!
1995:
1996:
1997: @node Designing the stack effect Tutorial, Local Variables Tutorial, Factoring Tutorial, Tutorial
1998: @section Designing the stack effect
1999: @cindex Stack effect design, tutorial
2000: @cindex design of stack effects, tutorial
2001:
2002: In other languages you can use an arbitrary order of parameters for a
2003: function; and since there is only one result, you don't have to deal with
2004: the order of results, either.
2005:
2006: In Forth (and other stack-based languages, e.g., Postscript) the
2007: parameter and result order of a definition is important and should be
2008: designed well. The general guideline is to design the stack effect such
2009: that the word is simple to use in most cases, even if that complicates
2010: the implementation of the word. Some concrete rules are:
2011:
2012: @itemize @bullet
2013:
2014: @item
2015: Words consume all of their parameters (e.g., @code{.}).
2016:
2017: @item
2018: If there is a convention on the order of parameters (e.g., from
2019: mathematics or another programming language), stick with it (e.g.,
2020: @code{-}).
2021:
2022: @item
2023: If one parameter usually requires only a short computation (e.g., it is
2024: a constant), pass it on the top of the stack. Conversely, parameters
2025: that usually require a long sequence of code to compute should be passed
2026: as the bottom (i.e., first) parameter. This makes the code easier to
2027: read, because reader does not need to keep track of the bottom item
2028: through a long sequence of code (or, alternatively, through stack
2029: manipulations). E.g., @code{!} (store, @pxref{Memory}) expects the
2030: address on top of the stack because it is usually simpler to compute
2031: than the stored value (often the address is just a variable).
2032:
2033: @item
2034: Similarly, results that are usually consumed quickly should be returned
2035: on the top of stack, whereas a result that is often used in long
2036: computations should be passed as bottom result. E.g., the file words
2037: like @code{open-file} return the error code on the top of stack, because
2038: it is usually consumed quickly by @code{throw}; moreover, the error code
2039: has to be checked before doing anything with the other results.
2040:
2041: @end itemize
2042:
2043: These rules are just general guidelines, don't lose sight of the overall
2044: goal to make the words easy to use. E.g., if the convention rule
2045: conflicts with the computation-length rule, you might decide in favour
2046: of the convention if the word will be used rarely, and in favour of the
2047: computation-length rule if the word will be used frequently (because
2048: with frequent use the cost of breaking the computation-length rule would
2049: be quite high, and frequent use makes it easier to remember an
2050: unconventional order).
2051:
2052: @c example !! structure package
2053:
2054:
2055: @node Local Variables Tutorial, Conditional execution Tutorial, Designing the stack effect Tutorial, Tutorial
2056: @section Local Variables
2057: @cindex local variables, tutorial
2058:
2059: You can define local variables (@emph{locals}) in a colon definition:
2060:
2061: @example
2062: : swap @{ a b -- b a @}
2063: b a ;
2064: 1 2 swap .s 2drop
2065: @end example
2066:
2067: (If your Forth system does not support this syntax, include
2068: @file{compat/anslocals.fs} first).
2069:
2070: In this example @code{@{ a b -- b a @}} is the locals definition; it
2071: takes two cells from the stack, puts the top of stack in @code{b} and
2072: the next stack element in @code{a}. @code{--} starts a comment ending
2073: with @code{@}}. After the locals definition, using the name of the
2074: local will push its value on the stack. You can leave the comment
2075: part (@code{-- b a}) away:
2076:
2077: @example
2078: : swap ( x1 x2 -- x2 x1 )
2079: @{ a b @} b a ;
2080: @end example
2081:
2082: In Gforth you can have several locals definitions, anywhere in a colon
2083: definition; in contrast, in a standard program you can have only one
2084: locals definition per colon definition, and that locals definition must
2085: be outside any controll structure.
2086:
2087: With locals you can write slightly longer definitions without running
2088: into stack trouble. However, I recommend trying to write colon
2089: definitions without locals for exercise purposes to help you gain the
2090: essential factoring skills.
2091:
2092: @assignment
2093: Rewrite your definitions until now with locals
2094: @endassignment
2095:
2096: Reference: @ref{Locals}.
2097:
2098:
2099: @node Conditional execution Tutorial, Flags and Comparisons Tutorial, Local Variables Tutorial, Tutorial
2100: @section Conditional execution
2101: @cindex conditionals, tutorial
2102: @cindex if, tutorial
2103:
2104: In Forth you can use control structures only inside colon definitions.
2105: An @code{if}-structure looks like this:
2106:
2107: @example
2108: : abs ( n1 -- +n2 )
2109: dup 0 < if
2110: negate
2111: endif ;
2112: 5 abs .
2113: -5 abs .
2114: @end example
2115:
2116: @code{if} takes a flag from the stack. If the flag is non-zero (true),
2117: the following code is performed, otherwise execution continues after the
2118: @code{endif} (or @code{else}). @code{<} compares the top two stack
2119: elements and prioduces a flag:
2120:
2121: @example
2122: 1 2 < .
2123: 2 1 < .
2124: 1 1 < .
2125: @end example
2126:
2127: Actually the standard name for @code{endif} is @code{then}. This
2128: tutorial presents the examples using @code{endif}, because this is often
2129: less confusing for people familiar with other programming languages
2130: where @code{then} has a different meaning. If your system does not have
2131: @code{endif}, define it with
2132:
2133: @example
2134: : endif postpone then ; immediate
2135: @end example
2136:
2137: You can optionally use an @code{else}-part:
2138:
2139: @example
2140: : min ( n1 n2 -- n )
2141: 2dup < if
2142: drop
2143: else
2144: nip
2145: endif ;
2146: 2 3 min .
2147: 3 2 min .
2148: @end example
2149:
2150: @assignment
2151: Write @code{min} without @code{else}-part (hint: what's the definition
2152: of @code{nip}?).
2153: @endassignment
2154:
2155: Reference: @ref{Selection}.
2156:
2157:
2158: @node Flags and Comparisons Tutorial, General Loops Tutorial, Conditional execution Tutorial, Tutorial
2159: @section Flags and Comparisons
2160: @cindex flags tutorial
2161: @cindex comparison tutorial
2162:
2163: In a false-flag all bits are clear (0 when interpreted as integer). In
2164: a canonical true-flag all bits are set (-1 as a twos-complement signed
2165: integer); in many contexts (e.g., @code{if}) any non-zero value is
2166: treated as true flag.
2167:
2168: @example
2169: false .
2170: true .
2171: true hex u. decimal
2172: @end example
2173:
2174: Comparison words produce canonical flags:
2175:
2176: @example
2177: 1 1 = .
2178: 1 0= .
2179: 0 1 < .
2180: 0 0 < .
2181: -1 1 u< . \ type error, u< interprets -1 as large unsigned number
2182: -1 1 < .
2183: @end example
2184:
2185: Gforth supports all combinations of the prefixes @code{0 u d d0 du f f0}
2186: (or none) and the comparisons @code{= <> < > <= >=}. Only a part of
2187: these combinations are standard (for details see the standard,
2188: @ref{Numeric comparison}, @ref{Floating Point} or @ref{Word Index}).
2189:
2190: You can use @code{and or xor invert} can be used as operations on
2191: canonical flags. Actually they are bitwise operations:
2192:
2193: @example
2194: 1 2 and .
2195: 1 2 or .
2196: 1 3 xor .
2197: 1 invert .
2198: @end example
2199:
2200: You can convert a zero/non-zero flag into a canonical flag with
2201: @code{0<>} (and complement it on the way with @code{0=}).
2202:
2203: @example
2204: 1 0= .
2205: 1 0<> .
2206: @end example
2207:
2208: You can use the all-bits-set feature of canonical flags and the bitwise
2209: operation of the Boolean operations to avoid @code{if}s:
2210:
2211: @example
2212: : foo ( n1 -- n2 )
2213: 0= if
2214: 14
2215: else
2216: 0
2217: endif ;
2218: 0 foo .
2219: 1 foo .
2220:
2221: : foo ( n1 -- n2 )
2222: 0= 14 and ;
2223: 0 foo .
2224: 1 foo .
2225: @end example
2226:
2227: @assignment
2228: Write @code{min} without @code{if}.
2229: @endassignment
2230:
2231: For reference, see @ref{Boolean Flags}, @ref{Numeric comparison}, and
2232: @ref{Bitwise operations}.
2233:
2234:
2235: @node General Loops Tutorial, Counted loops Tutorial, Flags and Comparisons Tutorial, Tutorial
2236: @section General Loops
2237: @cindex loops, indefinite, tutorial
2238:
2239: The endless loop is the most simple one:
2240:
2241: @example
2242: : endless ( -- )
2243: 0 begin
2244: dup . 1+
2245: again ;
2246: endless
2247: @end example
2248:
2249: Terminate this loop by pressing @kbd{Ctrl-C} (in Gforth). @code{begin}
2250: does nothing at run-time, @code{again} jumps back to @code{begin}.
2251:
2252: A loop with one exit at any place looks like this:
2253:
2254: @example
2255: : log2 ( +n1 -- n2 )
2256: \ logarithmus dualis of n1>0, rounded down to the next integer
2257: assert( dup 0> )
2258: 2/ 0 begin
2259: over 0> while
2260: 1+ swap 2/ swap
2261: repeat
2262: nip ;
2263: 7 log2 .
2264: 8 log2 .
2265: @end example
2266:
2267: At run-time @code{while} consumes a flag; if it is 0, execution
2268: continues behind the @code{repeat}; if the flag is non-zero, execution
2269: continues behind the @code{while}. @code{Repeat} jumps back to
2270: @code{begin}, just like @code{again}.
2271:
2272: In Forth there are many combinations/abbreviations, like @code{1+}.
2273: However, @code{2/} is not one of them; it shifts it's argument right by
2274: one bit (arithmetic shift right):
2275:
2276: @example
2277: -5 2 / .
2278: -5 2/ .
2279: @end example
2280:
2281: @code{assert(} is no standard word, but you can get it on systems other
2282: then Gforth by including @file{compat/assert.fs}. You can see what it
2283: does by trying
2284:
2285: @example
2286: 0 log2 .
2287: @end example
2288:
2289: Here's a loop with an exit at the end:
2290:
2291: @example
2292: : log2 ( +n1 -- n2 )
2293: \ logarithmus dualis of n1>0, rounded down to the next integer
2294: assert( dup 0 > )
2295: -1 begin
2296: 1+ swap 2/ swap
2297: over 0 <=
2298: until
2299: nip ;
2300: @end example
2301:
2302: @code{Until} consumes a flag; if it is non-zero, execution continues at
2303: the @code{begin}, otherwise after the @code{until}.
2304:
2305: @assignment
2306: Write a definition for computing the greatest common divisor.
2307: @endassignment
2308:
2309: Reference: @ref{Simple Loops}.
2310:
2311:
2312: @node Counted loops Tutorial, Recursion Tutorial, General Loops Tutorial, Tutorial
2313: @section Counted loops
2314: @cindex loops, counted, tutorial
2315:
2316: @example
2317: : ^ ( n1 u -- n )
2318: \ n = the uth power of u1
2319: 1 swap 0 u+do
2320: over *
2321: loop
2322: nip ;
2323: 3 2 ^ .
2324: 4 3 ^ .
2325: @end example
2326:
2327: @code{U+do} (from @file{compat/loops.fs}, if your Forth system doesn't
2328: have it) takes two numbers of the stack @code{( u3 u4 -- )}, and then
2329: performs the code between @code{u+do} and @code{loop} for @code{u3-u4}
2330: times (or not at all, if @code{u3-u4<0}).
2331:
2332: You can see the stack effect design rules at work in the stack effect of
2333: the loop start words: Since the start value of the loop is more
2334: frequently constant than the end value, the start value is passed on
2335: the top-of-stack.
2336:
2337: You can access the counter of a counted loop with @code{i}:
2338:
2339: @example
2340: : fac ( u -- u! )
2341: 1 swap 1+ 1 u+do
2342: i *
2343: loop ;
2344: 5 fac .
2345: 7 fac .
2346: @end example
2347:
2348: There is also @code{+do}, which expects signed numbers (important for
2349: deciding whether to enter the loop).
2350:
2351: @assignment
2352: Write a definition for computing the nth Fibonacci number.
2353: @endassignment
2354:
2355: You can also use increments other than 1:
2356:
2357: @example
2358: : up2 ( n1 n2 -- )
2359: +do
2360: i .
2361: 2 +loop ;
2362: 10 0 up2
2363:
2364: : down2 ( n1 n2 -- )
2365: -do
2366: i .
2367: 2 -loop ;
2368: 0 10 down2
2369: @end example
2370:
2371: Reference: @ref{Counted Loops}.
2372:
2373:
2374: @node Recursion Tutorial, Leaving definitions or loops Tutorial, Counted loops Tutorial, Tutorial
2375: @section Recursion
2376: @cindex recursion tutorial
2377:
2378: Usually the name of a definition is not visible in the definition; but
2379: earlier definitions are usually visible:
2380:
2381: @example
2382: 1 0 / . \ "Floating-point unidentified fault" in Gforth on most platforms
2383: : / ( n1 n2 -- n )
2384: dup 0= if
2385: -10 throw \ report division by zero
2386: endif
2387: / \ old version
2388: ;
2389: 1 0 /
2390: @end example
2391:
2392: For recursive definitions you can use @code{recursive} (non-standard) or
2393: @code{recurse}:
2394:
2395: @example
2396: : fac1 ( n -- n! ) recursive
2397: dup 0> if
2398: dup 1- fac1 *
2399: else
2400: drop 1
2401: endif ;
2402: 7 fac1 .
2403:
2404: : fac2 ( n -- n! )
2405: dup 0> if
2406: dup 1- recurse *
2407: else
2408: drop 1
2409: endif ;
2410: 8 fac2 .
2411: @end example
2412:
2413: @assignment
2414: Write a recursive definition for computing the nth Fibonacci number.
2415: @endassignment
2416:
2417: Reference (including indirect recursion): @xref{Calls and returns}.
2418:
2419:
2420: @node Leaving definitions or loops Tutorial, Return Stack Tutorial, Recursion Tutorial, Tutorial
2421: @section Leaving definitions or loops
2422: @cindex leaving definitions, tutorial
2423: @cindex leaving loops, tutorial
2424:
2425: @code{EXIT} exits the current definition right away. For every counted
2426: loop that is left in this way, an @code{UNLOOP} has to be performed
2427: before the @code{EXIT}:
2428:
2429: @c !! real examples
2430: @example
2431: : ...
2432: ... u+do
2433: ... if
2434: ... unloop exit
2435: endif
2436: ...
2437: loop
2438: ... ;
2439: @end example
2440:
2441: @code{LEAVE} leaves the innermost counted loop right away:
2442:
2443: @example
2444: : ...
2445: ... u+do
2446: ... if
2447: ... leave
2448: endif
2449: ...
2450: loop
2451: ... ;
2452: @end example
2453:
2454: @c !! example
2455:
2456: Reference: @ref{Calls and returns}, @ref{Counted Loops}.
2457:
2458:
2459: @node Return Stack Tutorial, Memory Tutorial, Leaving definitions or loops Tutorial, Tutorial
2460: @section Return Stack
2461: @cindex return stack tutorial
2462:
2463: In addition to the data stack Forth also has a second stack, the return
2464: stack; most Forth systems store the return addresses of procedure calls
2465: there (thus its name). Programmers can also use this stack:
2466:
2467: @example
2468: : foo ( n1 n2 -- )
2469: .s
2470: >r .s
2471: r@@ .
2472: >r .s
2473: r@@ .
2474: r> .
2475: r@@ .
2476: r> . ;
2477: 1 2 foo
2478: @end example
2479:
2480: @code{>r} takes an element from the data stack and pushes it onto the
2481: return stack; conversely, @code{r>} moves an elementm from the return to
2482: the data stack; @code{r@@} pushes a copy of the top of the return stack
2483: on the return stack.
2484:
2485: Forth programmers usually use the return stack for storing data
2486: temporarily, if using the data stack alone would be too complex, and
2487: factoring and locals are not an option:
2488:
2489: @example
2490: : 2swap ( x1 x2 x3 x4 -- x3 x4 x1 x2 )
2491: rot >r rot r> ;
2492: @end example
2493:
2494: The return address of the definition and the loop control parameters of
2495: counted loops usually reside on the return stack, so you have to take
2496: all items, that you have pushed on the return stack in a colon
2497: definition or counted loop, from the return stack before the definition
2498: or loop ends. You cannot access items that you pushed on the return
2499: stack outside some definition or loop within the definition of loop.
2500:
2501: If you miscount the return stack items, this usually ends in a crash:
2502:
2503: @example
2504: : crash ( n -- )
2505: >r ;
2506: 5 crash
2507: @end example
2508:
2509: You cannot mix using locals and using the return stack (according to the
2510: standard; Gforth has no problem). However, they solve the same
2511: problems, so this shouldn't be an issue.
2512:
2513: @assignment
2514: Can you rewrite any of the definitions you wrote until now in a better
2515: way using the return stack?
2516: @endassignment
2517:
2518: Reference: @ref{Return stack}.
2519:
2520:
2521: @node Memory Tutorial, Characters and Strings Tutorial, Return Stack Tutorial, Tutorial
2522: @section Memory
2523: @cindex memory access/allocation tutorial
2524:
2525: You can create a global variable @code{v} with
2526:
2527: @example
2528: variable v ( -- addr )
2529: @end example
2530:
2531: @code{v} pushes the address of a cell in memory on the stack. This cell
2532: was reserved by @code{variable}. You can use @code{!} (store) to store
2533: values into this cell and @code{@@} (fetch) to load the value from the
2534: stack into memory:
2535:
2536: @example
2537: v .
2538: 5 v ! .s
2539: v @@ .
2540: @end example
2541:
2542: You can see a raw dump of memory with @code{dump}:
2543:
2544: @example
2545: v 1 cells .s dump
2546: @end example
2547:
2548: @code{Cells ( n1 -- n2 )} gives you the number of bytes (or, more
2549: generally, address units (aus)) that @code{n1 cells} occupy. You can
2550: also reserve more memory:
2551:
2552: @example
2553: create v2 20 cells allot
2554: v2 20 cells dump
2555: @end example
2556:
2557: creates a word @code{v2} and reserves 20 uninitialized cells; the
2558: address pushed by @code{v2} points to the start of these 20 cells. You
2559: can use address arithmetic to access these cells:
2560:
2561: @example
2562: 3 v2 5 cells + !
2563: v2 20 cells dump
2564: @end example
2565:
2566: You can reserve and initialize memory with @code{,}:
2567:
2568: @example
2569: create v3
2570: 5 , 4 , 3 , 2 , 1 ,
2571: v3 @@ .
2572: v3 cell+ @@ .
2573: v3 2 cells + @@ .
2574: v3 5 cells dump
2575: @end example
2576:
2577: @assignment
2578: Write a definition @code{vsum ( addr u -- n )} that computes the sum of
2579: @code{u} cells, with the first of these cells at @code{addr}, the next
2580: one at @code{addr cell+} etc.
2581: @endassignment
2582:
2583: You can also reserve memory without creating a new word:
2584:
2585: @example
2586: here 10 cells allot .
2587: here .
2588: @end example
2589:
2590: @code{Here} pushes the start address of the memory area. You should
2591: store it somewhere, or you will have a hard time finding the memory area
2592: again.
2593:
2594: @code{Allot} manages dictionary memory. The dictionary memory contains
2595: the system's data structures for words etc. on Gforth and most other
2596: Forth systems. It is managed like a stack: You can free the memory that
2597: you have just @code{allot}ed with
2598:
2599: @example
2600: -10 cells allot
2601: here .
2602: @end example
2603:
2604: Note that you cannot do this if you have created a new word in the
2605: meantime (because then your @code{allot}ed memory is no longer on the
2606: top of the dictionary ``stack'').
2607:
2608: Alternatively, you can use @code{allocate} and @code{free} which allow
2609: freeing memory in any order:
2610:
2611: @example
2612: 10 cells allocate throw .s
2613: 20 cells allocate throw .s
2614: swap
2615: free throw
2616: free throw
2617: @end example
2618:
2619: The @code{throw}s deal with errors (e.g., out of memory).
2620:
2621: And there is also a
2622: @uref{http://www.complang.tuwien.ac.at/forth/garbage-collection.zip,
2623: garbage collector}, which eliminates the need to @code{free} memory
2624: explicitly.
2625:
2626: Reference: @ref{Memory}.
2627:
2628:
2629: @node Characters and Strings Tutorial, Alignment Tutorial, Memory Tutorial, Tutorial
2630: @section Characters and Strings
2631: @cindex strings tutorial
2632: @cindex characters tutorial
2633:
2634: On the stack characters take up a cell, like numbers. In memory they
2635: have their own size (one 8-bit byte on most systems), and therefore
2636: require their own words for memory access:
2637:
2638: @example
2639: create v4
2640: 104 c, 97 c, 108 c, 108 c, 111 c,
2641: v4 4 chars + c@@ .
2642: v4 5 chars dump
2643: @end example
2644:
2645: The preferred representation of strings on the stack is @code{addr
2646: u-count}, where @code{addr} is the address of the first character and
2647: @code{u-count} is the number of characters in the string.
2648:
2649: @example
2650: v4 5 type
2651: @end example
2652:
2653: You get a string constant with
2654:
2655: @example
2656: s" hello, world" .s
2657: type
2658: @end example
2659:
2660: Make sure you have a space between @code{s"} and the string; @code{s"}
2661: is a normal Forth word and must be delimited with white space (try what
2662: happens when you remove the space).
2663:
2664: However, this interpretive use of @code{s"} is quite restricted: the
2665: string exists only until the next call of @code{s"} (some Forth systems
2666: keep more than one of these strings, but usually they still have a
2667: limited lifetime).
2668:
2669: @example
2670: s" hello," s" world" .s
2671: type
2672: type
2673: @end example
2674:
2675: You can also use @code{s"} in a definition, and the resulting
2676: strings then live forever (well, for as long as the definition):
2677:
2678: @example
2679: : foo s" hello," s" world" ;
2680: foo .s
2681: type
2682: type
2683: @end example
2684:
2685: @assignment
2686: @code{Emit ( c -- )} types @code{c} as character (not a number).
2687: Implement @code{type ( addr u -- )}.
2688: @endassignment
2689:
2690: Reference: @ref{Memory Blocks}.
2691:
2692:
2693: @node Alignment Tutorial, Interpretation and Compilation Semantics and Immediacy Tutorial, Characters and Strings Tutorial, Tutorial
2694: @section Alignment
2695: @cindex alignment tutorial
2696: @cindex memory alignment tutorial
2697:
2698: On many processors cells have to be aligned in memory, if you want to
2699: access them with @code{@@} and @code{!} (and even if the processor does
2700: not require alignment, access to aligned cells is faster).
2701:
2702: @code{Create} aligns @code{here} (i.e., the place where the next
2703: allocation will occur, and that the @code{create}d word points to).
2704: Likewise, the memory produced by @code{allocate} starts at an aligned
2705: address. Adding a number of @code{cells} to an aligned address produces
2706: another aligned address.
2707:
2708: However, address arithmetic involving @code{char+} and @code{chars} can
2709: create an address that is not cell-aligned. @code{Aligned ( addr --
2710: a-addr )} produces the next aligned address:
2711:
2712: @example
2713: v3 char+ aligned .s @@ .
2714: v3 char+ .s @@ .
2715: @end example
2716:
2717: Similarly, @code{align} advances @code{here} to the next aligned
2718: address:
2719:
2720: @example
2721: create v5 97 c,
2722: here .
2723: align here .
2724: 1000 ,
2725: @end example
2726:
2727: Note that you should use aligned addresses even if your processor does
2728: not require them, if you want your program to be portable.
2729:
2730: Reference: @ref{Address arithmetic}.
2731:
2732:
2733: @node Interpretation and Compilation Semantics and Immediacy Tutorial, Execution Tokens Tutorial, Alignment Tutorial, Tutorial
2734: @section Interpretation and Compilation Semantics and Immediacy
2735: @cindex semantics tutorial
2736: @cindex interpretation semantics tutorial
2737: @cindex compilation semantics tutorial
2738: @cindex immediate, tutorial
2739:
2740: When a word is compiled, it behaves differently from being interpreted.
2741: E.g., consider @code{+}:
2742:
2743: @example
2744: 1 2 + .
2745: : foo + ;
2746: @end example
2747:
2748: These two behaviours are known as compilation and interpretation
2749: semantics. For normal words (e.g., @code{+}), the compilation semantics
2750: is to append the interpretation semantics to the currently defined word
2751: (@code{foo} in the example above). I.e., when @code{foo} is executed
2752: later, the interpretation semantics of @code{+} (i.e., adding two
2753: numbers) will be performed.
2754:
2755: However, there are words with non-default compilation semantics, e.g.,
2756: the control-flow words like @code{if}. You can use @code{immediate} to
2757: change the compilation semantics of the last defined word to be equal to
2758: the interpretation semantics:
2759:
2760: @example
2761: : [FOO] ( -- )
2762: 5 . ; immediate
2763:
2764: [FOO]
2765: : bar ( -- )
2766: [FOO] ;
2767: bar
2768: see bar
2769: @end example
2770:
2771: Two conventions to mark words with non-default compilation semnatics are
2772: names with brackets (more frequently used) and to write them all in
2773: upper case (less frequently used).
2774:
2775: In Gforth (and many other systems) you can also remove the
2776: interpretation semantics with @code{compile-only} (the compilation
2777: semantics is derived from the original interpretation semantics):
2778:
2779: @example
2780: : flip ( -- )
2781: 6 . ; compile-only \ but not immediate
2782: flip
2783:
2784: : flop ( -- )
2785: flip ;
2786: flop
2787: @end example
2788:
2789: In this example the interpretation semantics of @code{flop} is equal to
2790: the original interpretation semantics of @code{flip}.
2791:
2792: The text interpreter has two states: in interpret state, it performs the
2793: interpretation semantics of words it encounters; in compile state, it
2794: performs the compilation semantics of these words.
2795:
2796: Among other things, @code{:} switches into compile state, and @code{;}
2797: switches back to interpret state. They contain the factors @code{]}
2798: (switch to compile state) and @code{[} (switch to interpret state), that
2799: do nothing but switch the state.
2800:
2801: @example
2802: : xxx ( -- )
2803: [ 5 . ]
2804: ;
2805:
2806: xxx
2807: see xxx
2808: @end example
2809:
2810: These brackets are also the source of the naming convention mentioned
2811: above.
2812:
2813: Reference: @ref{Interpretation and Compilation Semantics}.
2814:
2815:
2816: @node Execution Tokens Tutorial, Exceptions Tutorial, Interpretation and Compilation Semantics and Immediacy Tutorial, Tutorial
2817: @section Execution Tokens
2818: @cindex execution tokens tutorial
2819: @cindex XT tutorial
2820:
2821: @code{' word} gives you the execution token (XT) of a word. The XT is a
2822: cell representing the interpretation semantics of a word. You can
2823: execute this semantics with @code{execute}:
2824:
2825: @example
2826: ' + .s
2827: 1 2 rot execute .
2828: @end example
2829:
2830: The XT is similar to a function pointer in C. However, parameter
2831: passing through the stack makes it a little more flexible:
2832:
2833: @example
2834: : map-array ( ... addr u xt -- ... )
2835: \ executes xt ( ... x -- ... ) for every element of the array starting
2836: \ at addr and containing u elements
2837: @{ xt @}
2838: cells over + swap ?do
2839: i @@ xt execute
2840: 1 cells +loop ;
2841:
2842: create a 3 , 4 , 2 , -1 , 4 ,
2843: a 5 ' . map-array .s
2844: 0 a 5 ' + map-array .
2845: s" max-n" environment? drop .s
2846: a 5 ' min map-array .
2847: @end example
2848:
2849: You can use map-array with the XTs of words that consume one element
2850: more than they produce. In theory you can also use it with other XTs,
2851: but the stack effect then depends on the size of the array, which is
2852: hard to understand.
2853:
2854: Since XTs are cell-sized, you can store them in memory and manipulate
2855: them on the stack like other cells. You can also compile the XT into a
2856: word with @code{compile,}:
2857:
2858: @example
2859: : foo1 ( n1 n2 -- n )
2860: [ ' + compile, ] ;
2861: see foo
2862: @end example
2863:
2864: This is non-standard, because @code{compile,} has no compilation
2865: semantics in the standard, but it works in good Forth systems. For the
2866: broken ones, use
2867:
2868: @example
2869: : [compile,] compile, ; immediate
2870:
2871: : foo1 ( n1 n2 -- n )
2872: [ ' + ] [compile,] ;
2873: see foo
2874: @end example
2875:
2876: @code{'} is a word with default compilation semantics; it parses the
2877: next word when its interpretation semantics are executed, not during
2878: compilation:
2879:
2880: @example
2881: : foo ( -- xt )
2882: ' ;
2883: see foo
2884: : bar ( ... "word" -- ... )
2885: ' execute ;
2886: see bar
2887: 1 2 bar + .
2888: @end example
2889:
2890: You often want to parse a word during compilation and compile its XT so
2891: it will be pushed on the stack at run-time. @code{[']} does this:
2892:
2893: @example
2894: : xt-+ ( -- xt )
2895: ['] + ;
2896: see xt-+
2897: 1 2 xt-+ execute .
2898: @end example
2899:
2900: Many programmers tend to see @code{'} and the word it parses as one
2901: unit, and expect it to behave like @code{[']} when compiled, and are
2902: confused by the actual behaviour. If you are, just remember that the
2903: Forth system just takes @code{'} as one unit and has no idea that it is
2904: a parsing word (attempts to convenience programmers in this issue have
2905: usually resulted in even worse pitfalls, see
2906: @uref{http://www.complang.tuwien.ac.at/papers/ertl98.ps.gz,
2907: @code{State}-smartness---Why it is evil and How to Exorcise it}).
2908:
2909: Note that the state of the interpreter does not come into play when
2910: creating and executing XTs. I.e., even when you execute @code{'} in
2911: compile state, it still gives you the interpretation semantics. And
2912: whatever that state is, @code{execute} performs the semantics
2913: represented by the XT (i.e., for XTs produced with @code{'} the
2914: interpretation semantics).
2915:
2916: Reference: @ref{Tokens for Words}.
2917:
2918:
2919: @node Exceptions Tutorial, Defining Words Tutorial, Execution Tokens Tutorial, Tutorial
2920: @section Exceptions
2921: @cindex exceptions tutorial
2922:
2923: @code{throw ( n -- )} causes an exception unless n is zero.
2924:
2925: @example
2926: 100 throw .s
2927: 0 throw .s
2928: @end example
2929:
2930: @code{catch ( ... xt -- ... n )} behaves similar to @code{execute}, but
2931: it catches exceptions and pushes the number of the exception on the
2932: stack (or 0, if the xt executed without exception). If there was an
2933: exception, the stacks have the same depth as when entering @code{catch}:
2934:
2935: @example
2936: .s
2937: 3 0 ' / catch .s
2938: 3 2 ' / catch .s
2939: @end example
2940:
2941: @assignment
2942: Try the same with @code{execute} instead of @code{catch}.
2943: @endassignment
2944:
2945: @code{Throw} always jumps to the dynamically next enclosing
2946: @code{catch}, even if it has to leave several call levels to achieve
2947: this:
2948:
2949: @example
2950: : foo 100 throw ;
2951: : foo1 foo ." after foo" ;
2952: : bar ['] foo1 catch ;
2953: bar .
2954: @end example
2955:
2956: It is often important to restore a value upon leaving a definition, even
2957: if the definition is left through an exception. You can ensure this
2958: like this:
2959:
2960: @example
2961: : ...
2962: save-x
2963: ['] word-changing-x catch ( ... n )
2964: restore-x
2965: ( ... n ) throw ;
2966: @end example
2967:
2968: Gforth provides an alternative syntax in addition to @code{catch}:
2969: @code{try ... recover ... endtry}. If the code between @code{try} and
2970: @code{recover} has an exception, the stack depths are restored, the
2971: exception number is pushed on the stack, and the code between
2972: @code{recover} and @code{endtry} is performed. E.g., the definition for
2973: @code{catch} is
2974:
2975: @example
2976: : catch ( x1 .. xn xt -- y1 .. ym 0 / z1 .. zn error ) \ exception
2977: try
2978: execute 0
2979: recover
2980: nip
2981: endtry ;
2982: @end example
2983:
2984: The equivalent to the restoration code above is
2985:
2986: @example
2987: : ...
2988: save-x
2989: try
2990: word-changing-x
2991: end-try
2992: restore-x
2993: throw ;
2994: @end example
2995:
2996: As you can see, the @code{recover} part is optional.
2997:
2998: Reference: @ref{Exception Handling}.
2999:
3000:
3001: @node Defining Words Tutorial, Arrays and Records Tutorial, Exceptions Tutorial, Tutorial
3002: @section Defining Words
3003: @cindex defining words tutorial
3004: @cindex does> tutorial
3005: @cindex create...does> tutorial
3006:
3007: @c before semantics?
3008:
3009: @code{:}, @code{create}, and @code{variable} are definition words: They
3010: define other words. @code{Constant} is another definition word:
3011:
3012: @example
3013: 5 constant foo
3014: foo .
3015: @end example
3016:
3017: You can also use the prefixes @code{2} (double-cell) and @code{f}
3018: (floating point) with @code{variable} and @code{constant}.
3019:
3020: You can also define your own defining words. E.g.:
3021:
3022: @example
3023: : variable ( "name" -- )
3024: create 0 , ;
3025: @end example
3026:
3027: You can also define defining words that create words that do something
3028: other than just producing their address:
3029:
3030: @example
3031: : constant ( n "name" -- )
3032: create ,
3033: does> ( -- n )
3034: ( addr ) @@ ;
3035:
3036: 5 constant foo
3037: foo .
3038: @end example
3039:
3040: The definition of @code{constant} above ends at the @code{does>}; i.e.,
3041: @code{does>} replaces @code{;}, but it also does something else: It
3042: changes the last defined word such that it pushes the address of the
3043: body of the word and then performs the code after the @code{does>}
3044: whenever it is called.
3045:
3046: In the example above, @code{constant} uses @code{,} to store 5 into the
3047: body of @code{foo}. When @code{foo} executes, it pushes the address of
3048: the body onto the stack, then (in the code after the @code{does>})
3049: fetches the 5 from there.
3050:
3051: The stack comment near the @code{does>} reflects the stack effect of the
3052: defined word, not the stack effect of the code after the @code{does>}
3053: (the difference is that the code expects the address of the body that
3054: the stack comment does not show).
3055:
3056: You can use these definition words to do factoring in cases that involve
3057: (other) definition words. E.g., a field offset is always added to an
3058: address. Instead of defining
3059:
3060: @example
3061: 2 cells constant offset-field1
3062: @end example
3063:
3064: and using this like
3065:
3066: @example
3067: ( addr ) offset-field1 +
3068: @end example
3069:
3070: you can define a definition word
3071:
3072: @example
3073: : simple-field ( n "name" -- )
3074: create ,
3075: does> ( n1 -- n1+n )
3076: ( addr ) @@ + ;
3077: @end example
3078:
3079: Definition and use of field offsets now look like this:
3080:
3081: @example
3082: 2 cells simple-field field1
3083: create mystruct 4 cells allot
3084: mystruct .s field1 .s drop
3085: @end example
3086:
3087: If you want to do something with the word without performing the code
3088: after the @code{does>}, you can access the body of a @code{create}d word
3089: with @code{>body ( xt -- addr )}:
3090:
3091: @example
3092: : value ( n "name" -- )
3093: create ,
3094: does> ( -- n1 )
3095: @@ ;
3096: : to ( n "name" -- )
3097: ' >body ! ;
3098:
3099: 5 value foo
3100: foo .
3101: 7 to foo
3102: foo .
3103: @end example
3104:
3105: @assignment
3106: Define @code{defer ( "name" -- )}, which creates a word that stores an
3107: XT (at the start the XT of @code{abort}), and upon execution
3108: @code{execute}s the XT. Define @code{is ( xt "name" -- )} that stores
3109: @code{xt} into @code{name}, a word defined with @code{defer}. Indirect
3110: recursion is one application of @code{defer}.
3111: @endassignment
3112:
3113: Reference: @ref{User-defined Defining Words}.
3114:
3115:
3116: @node Arrays and Records Tutorial, POSTPONE Tutorial, Defining Words Tutorial, Tutorial
3117: @section Arrays and Records
3118: @cindex arrays tutorial
3119: @cindex records tutorial
3120: @cindex structs tutorial
3121:
3122: Forth has no standard words for defining data structures such as arrays
3123: and records (structs in C terminology), but you can build them yourself
3124: based on address arithmetic. You can also define words for defining
3125: arrays and records (@pxref{Defining Words Tutorial,, Defining Words}).
3126:
3127: One of the first projects a Forth newcomer sets out upon when learning
3128: about defining words is an array defining word (possibly for
3129: n-dimensional arrays). Go ahead and do it, I did it, too; you will
3130: learn something from it. However, don't be disappointed when you later
3131: learn that you have little use for these words (inappropriate use would
3132: be even worse). I have not yet found a set of useful array words yet;
3133: the needs are just too diverse, and named, global arrays (the result of
3134: naive use of defining words) are often not flexible enough (e.g.,
3135: consider how to pass them as parameters). Another such project is a set
3136: of words to help dealing with strings.
3137:
3138: On the other hand, there is a useful set of record words, and it has
3139: been defined in @file{compat/struct.fs}; these words are predefined in
3140: Gforth. They are explained in depth elsewhere in this manual (see
3141: @pxref{Structures}). The @code{simple-field} example above is
3142: simplified variant of fields in this package.
3143:
3144:
3145: @node POSTPONE Tutorial, Literal Tutorial, Arrays and Records Tutorial, Tutorial
3146: @section @code{POSTPONE}
3147: @cindex postpone tutorial
3148:
3149: You can compile the compilation semantics (instead of compiling the
3150: interpretation semantics) of a word with @code{POSTPONE}:
3151:
3152: @example
3153: : MY-+ ( Compilation: -- ; Run-time of compiled code: n1 n2 -- n )
3154: POSTPONE + ; immediate
3155: : foo ( n1 n2 -- n )
3156: MY-+ ;
3157: 1 2 foo .
3158: see foo
3159: @end example
3160:
3161: During the definition of @code{foo} the text interpreter performs the
3162: compilation semantics of @code{MY-+}, which performs the compilation
3163: semantics of @code{+}, i.e., it compiles @code{+} into @code{foo}.
3164:
3165: This example also displays separate stack comments for the compilation
3166: semantics and for the stack effect of the compiled code. For words with
3167: default compilation semantics these stack effects are usually not
3168: displayed; the stack effect of the compilation semantics is always
3169: @code{( -- )} for these words, the stack effect for the compiled code is
3170: the stack effect of the interpretation semantics.
3171:
3172: Note that the state of the interpreter does not come into play when
3173: performing the compilation semantics in this way. You can also perform
3174: it interpretively, e.g.:
3175:
3176: @example
3177: : foo2 ( n1 n2 -- n )
3178: [ MY-+ ] ;
3179: 1 2 foo .
3180: see foo
3181: @end example
3182:
3183: However, there are some broken Forth systems where this does not always
3184: work, and therefore this practice was been declared non-standard in
3185: 1999.
3186: @c !! repair.fs
3187:
3188: Here is another example for using @code{POSTPONE}:
3189:
3190: @example
3191: : MY-- ( Compilation: -- ; Run-time of compiled code: n1 n2 -- n )
3192: POSTPONE negate POSTPONE + ; immediate compile-only
3193: : bar ( n1 n2 -- n )
3194: MY-- ;
3195: 2 1 bar .
3196: see bar
3197: @end example
3198:
3199: You can define @code{ENDIF} in this way:
3200:
3201: @example
3202: : ENDIF ( Compilation: orig -- )
3203: POSTPONE then ; immediate
3204: @end example
3205:
3206: @assignment
3207: Write @code{MY-2DUP} that has compilation semantics equivalent to
3208: @code{2dup}, but compiles @code{over over}.
3209: @endassignment
3210:
3211: @c !! @xref{Macros} for reference
3212:
3213:
3214: @node Literal Tutorial, Advanced macros Tutorial, POSTPONE Tutorial, Tutorial
3215: @section @code{Literal}
3216: @cindex literal tutorial
3217:
3218: You cannot @code{POSTPONE} numbers:
3219:
3220: @example
3221: : [FOO] POSTPONE 500 ; immediate
3222: @end example
3223:
3224: Instead, you can use @code{LITERAL (compilation: n --; run-time: -- n )}:
3225:
3226: @example
3227: : [FOO] ( compilation: --; run-time: -- n )
3228: 500 POSTPONE literal ; immediate
3229:
3230: : flip [FOO] ;
3231: flip .
3232: see flip
3233: @end example
3234:
3235: @code{LITERAL} consumes a number at compile-time (when it's compilation
3236: semantics are executed) and pushes it at run-time (when the code it
3237: compiled is executed). A frequent use of @code{LITERAL} is to compile a
3238: number computed at compile time into the current word:
3239:
3240: @example
3241: : bar ( -- n )
3242: [ 2 2 + ] literal ;
3243: see bar
3244: @end example
3245:
3246: @assignment
3247: Write @code{]L} which allows writing the example above as @code{: bar (
3248: -- n ) [ 2 2 + ]L ;}
3249: @endassignment
3250:
3251: @c !! @xref{Macros} for reference
3252:
3253:
3254: @node Advanced macros Tutorial, Compilation Tokens Tutorial, Literal Tutorial, Tutorial
3255: @section Advanced macros
3256: @cindex macros, advanced tutorial
3257: @cindex run-time code generation, tutorial
3258:
3259: Reconsider @code{map-array} from @ref{Execution Tokens Tutorial,,
3260: Execution Tokens}. It frequently performs @code{execute}, a relatively
3261: expensive operation in some Forth implementations. You can use
3262: @code{compile,} and @code{POSTPONE} to eliminate these @code{execute}s
3263: and produce a word that contains the word to be performed directly:
3264:
3265: @c use ]] ... [[
3266: @example
3267: : compile-map-array ( compilation: xt -- ; run-time: ... addr u -- ... )
3268: \ at run-time, execute xt ( ... x -- ... ) for each element of the
3269: \ array beginning at addr and containing u elements
3270: @{ xt @}
3271: POSTPONE cells POSTPONE over POSTPONE + POSTPONE swap POSTPONE ?do
3272: POSTPONE i POSTPONE @@ xt compile,
3273: 1 cells POSTPONE literal POSTPONE +loop ;
3274:
3275: : sum-array ( addr u -- n )
3276: 0 rot rot [ ' + compile-map-array ] ;
3277: see sum-array
3278: a 5 sum-array .
3279: @end example
3280:
3281: You can use the full power of Forth for generating the code; here's an
3282: example where the code is generated in a loop:
3283:
3284: @example
3285: : compile-vmul-step ( compilation: n --; run-time: n1 addr1 -- n2 addr2 )
3286: \ n2=n1+(addr1)*n, addr2=addr1+cell
3287: POSTPONE tuck POSTPONE @@
3288: POSTPONE literal POSTPONE * POSTPONE +
3289: POSTPONE swap POSTPONE cell+ ;
3290:
3291: : compile-vmul ( compilation: addr1 u -- ; run-time: addr2 -- n )
3292: \ n=v1*v2 (inner product), where the v_i are represented as addr_i u
3293: 0 postpone literal postpone swap
3294: [ ' compile-vmul-step compile-map-array ]
3295: postpone drop ;
3296: see compile-vmul
3297:
3298: : a-vmul ( addr -- n )
3299: \ n=a*v, where v is a vector that's as long as a and starts at addr
3300: [ a 5 compile-vmul ] ;
3301: see a-vmul
3302: a a-vmul .
3303: @end example
3304:
3305: This example uses @code{compile-map-array} to show off, but you could
3306: also use @code{map-array} instead (try it now!).
3307:
3308: You can use this technique for efficient multiplication of large
3309: matrices. In matrix multiplication, you multiply every line of one
3310: matrix with every column of the other matrix. You can generate the code
3311: for one line once, and use it for every column. The only downside of
3312: this technique is that it is cumbersome to recover the memory consumed
3313: by the generated code when you are done (and in more complicated cases
3314: it is not possible portably).
3315:
3316: @c !! @xref{Macros} for reference
3317:
3318:
3319: @node Compilation Tokens Tutorial, Wordlists and Search Order Tutorial, Advanced macros Tutorial, Tutorial
3320: @section Compilation Tokens
3321: @cindex compilation tokens, tutorial
3322: @cindex CT, tutorial
3323:
3324: This section is Gforth-specific. You can skip it.
3325:
3326: @code{' word compile,} compiles the interpretation semantics. For words
3327: with default compilation semantics this is the same as performing the
3328: compilation semantics. To represent the compilation semantics of other
3329: words (e.g., words like @code{if} that have no interpretation
3330: semantics), Gforth has the concept of a compilation token (CT,
3331: consisting of two cells), and words @code{comp'} and @code{[comp']}.
3332: You can perform the compilation semantics represented by a CT with
3333: @code{execute}:
3334:
3335: @example
3336: : foo2 ( n1 n2 -- n )
3337: [ comp' + execute ] ;
3338: see foo
3339: @end example
3340:
3341: You can compile the compilation semantics represented by a CT with
3342: @code{postpone,}:
3343:
3344: @example
3345: : foo3 ( -- )
3346: [ comp' + postpone, ] ;
3347: see foo3
3348: @end example
3349:
3350: @code{[ comp' word postpone, ]} is equivalent to @code{POSTPONE word}.
3351: @code{comp'} is particularly useful for words that have no
3352: interpretation semantics:
3353:
3354: @example
3355: ' if
3356: comp' if .s 2drop
3357: @end example
3358:
3359: Reference: @ref{Tokens for Words}.
3360:
3361:
3362: @node Wordlists and Search Order Tutorial, , Compilation Tokens Tutorial, Tutorial
3363: @section Wordlists and Search Order
3364: @cindex wordlists tutorial
3365: @cindex search order, tutorial
3366:
3367: The dictionary is not just a memory area that allows you to allocate
3368: memory with @code{allot}, it also contains the Forth words, arranged in
3369: several wordlists. When searching for a word in a wordlist,
3370: conceptually you start searching at the youngest and proceed towards
3371: older words (in reality most systems nowadays use hash-tables); i.e., if
3372: you define a word with the same name as an older word, the new word
3373: shadows the older word.
3374:
3375: Which wordlists are searched in which order is determined by the search
3376: order. You can display the search order with @code{order}. It displays
3377: first the search order, starting with the wordlist searched first, then
3378: it displays the wordlist that will contain newly defined words.
3379:
3380: You can create a new, empty wordlist with @code{wordlist ( -- wid )}:
3381:
3382: @example
3383: wordlist constant mywords
3384: @end example
3385:
3386: @code{Set-current ( wid -- )} sets the wordlist that will contain newly
3387: defined words (the @emph{current} wordlist):
3388:
3389: @example
3390: mywords set-current
3391: order
3392: @end example
3393:
3394: Gforth does not display a name for the wordlist in @code{mywords}
3395: because this wordlist was created anonymously with @code{wordlist}.
3396:
3397: You can get the current wordlist with @code{get-current ( -- wid)}. If
3398: you want to put something into a specific wordlist without overall
3399: effect on the current wordlist, this typically looks like this:
3400:
3401: @example
3402: get-current mywords set-current ( wid )
3403: create someword
3404: ( wid ) set-current
3405: @end example
3406:
3407: You can write the search order with @code{set-order ( wid1 .. widn n --
3408: )} and read it with @code{get-order ( -- wid1 .. widn n )}. The first
3409: searched wordlist is topmost.
3410:
3411: @example
3412: get-order mywords swap 1+ set-order
3413: order
3414: @end example
3415:
3416: Yes, the order of wordlists in the output of @code{order} is reversed
3417: from stack comments and the output of @code{.s} and thus unintuitive.
3418:
3419: @assignment
3420: Define @code{>order ( wid -- )} with adds @code{wid} as first searched
3421: wordlist to the search order. Define @code{previous ( -- )}, which
3422: removes the first searched wordlist from the search order. Experiment
3423: with boundary conditions (you will see some crashes or situations that
3424: are hard or impossible to leave).
3425: @endassignment
3426:
3427: The search order is a powerful foundation for providing features similar
3428: to Modula-2 modules and C++ namespaces. However, trying to modularize
3429: programs in this way has disadvantages for debugging and reuse/factoring
3430: that overcome the advantages in my experience (I don't do huge projects,
3431: though). These disadvantages are not so clear in other
3432: languages/programming environments, because these langauges are not so
3433: strong in debugging and reuse.
3434:
3435: @c !! example
3436:
3437: Reference: @ref{Word Lists}.
3438:
3439: @c ******************************************************************
3440: @node Introduction, Words, Tutorial, Top
3441: @comment node-name, next, previous, up
3442: @chapter An Introduction to ANS Forth
3443: @cindex Forth - an introduction
3444:
3445: The primary purpose of this manual is to document Gforth. However, since
3446: Forth is not a widely-known language and there is a lack of up-to-date
3447: teaching material, it seems worthwhile to provide some introductory
3448: material. For other sources of Forth-related
3449: information, see @ref{Forth-related information}.
3450:
3451: The examples in this section should work on any ANS Forth; the
3452: output shown was produced using Gforth. Each example attempts to
3453: reproduce the exact output that Gforth produces. If you try out the
3454: examples (and you should), what you should type is shown @kbd{like this}
3455: and Gforth's response is shown @code{like this}. The single exception is
3456: that, where the example shows @key{RET} it means that you should
3457: press the ``carriage return'' key. Unfortunately, some output formats for
3458: this manual cannot show the difference between @kbd{this} and
3459: @code{this} which will make trying out the examples harder (but not
3460: impossible).
3461:
3462: Forth is an unusual language. It provides an interactive development
3463: environment which includes both an interpreter and compiler. Forth
3464: programming style encourages you to break a problem down into many
3465: @cindex factoring
3466: small fragments (@dfn{factoring}), and then to develop and test each
3467: fragment interactively. Forth advocates assert that breaking the
3468: edit-compile-test cycle used by conventional programming languages can
3469: lead to great productivity improvements.
3470:
3471: @menu
3472: * Introducing the Text Interpreter::
3473: * Stacks and Postfix notation::
3474: * Your first definition::
3475: * How does that work?::
3476: * Forth is written in Forth::
3477: * Review - elements of a Forth system::
3478: * Where to go next::
3479: * Exercises::
3480: @end menu
3481:
3482: @comment ----------------------------------------------
3483: @node Introducing the Text Interpreter, Stacks and Postfix notation, Introduction, Introduction
3484: @section Introducing the Text Interpreter
3485: @cindex text interpreter
3486: @cindex outer interpreter
3487:
3488: @c IMO this is too detailed and the pace is too slow for
3489: @c an introduction. If you know German, take a look at
3490: @c http://www.complang.tuwien.ac.at/anton/lvas/skriptum-stack.html
3491: @c to see how I do it - anton
3492:
3493: @c nac-> Where I have accepted your comments 100% and modified the text
3494: @c accordingly, I have deleted your comments. Elsewhere I have added a
3495: @c response like this to attempt to rationalise what I have done. Of
3496: @c course, this is a very clumsy mechanism for something that would be
3497: @c done far more efficiently over a beer. Please delete any dialogue
3498: @c you consider closed.
3499:
3500: When you invoke the Forth image, you will see a startup banner printed
3501: and nothing else (if you have Gforth installed on your system, try
3502: invoking it now, by typing @kbd{gforth@key{RET}}). Forth is now running
3503: its command line interpreter, which is called the @dfn{Text Interpreter}
3504: (also known as the @dfn{Outer Interpreter}). (You will learn a lot
3505: about the text interpreter as you read through this chapter, for more
3506: detail @pxref{The Text Interpreter}).
3507:
3508: Although it's not obvious, Forth is actually waiting for your
3509: input. Type a number and press the @key{RET} key:
3510:
3511: @example
3512: @kbd{45@key{RET}} ok
3513: @end example
3514:
3515: Rather than give you a prompt to invite you to input something, the text
3516: interpreter prints a status message @i{after} it has processed a line
3517: of input. The status message in this case (``@code{ ok}'' followed by
3518: carriage-return) indicates that the text interpreter was able to process
3519: all of your input successfully. Now type something illegal:
3520:
3521: @example
3522: @kbd{qwer341@key{RET}}
3523: :1: Undefined word
3524: qwer341
3525: ^^^^^^^
3526: $400D2BA8 Bounce
3527: $400DBDA8 no.extensions
3528: @end example
3529:
3530: The exact text, other than the ``Undefined word'' may differ slightly on
3531: your system, but the effect is the same; when the text interpreter
3532: detects an error, it discards any remaining text on a line, resets
3533: certain internal state and prints an error message. For a detailed description of error messages see @ref{Error
3534: messages}.
3535:
3536: The text interpreter waits for you to press carriage-return, and then
3537: processes your input line. Starting at the beginning of the line, it
3538: breaks the line into groups of characters separated by spaces. For each
3539: group of characters in turn, it makes two attempts to do something:
3540:
3541: @itemize @bullet
3542: @item
3543: @cindex name dictionary
3544: It tries to treat it as a command. It does this by searching a @dfn{name
3545: dictionary}. If the group of characters matches an entry in the name
3546: dictionary, the name dictionary provides the text interpreter with
3547: information that allows the text interpreter perform some actions. In
3548: Forth jargon, we say that the group
3549: @cindex word
3550: @cindex definition
3551: @cindex execution token
3552: @cindex xt
3553: of characters names a @dfn{word}, that the dictionary search returns an
3554: @dfn{execution token (xt)} corresponding to the @dfn{definition} of the
3555: word, and that the text interpreter executes the xt. Often, the terms
3556: @dfn{word} and @dfn{definition} are used interchangeably.
3557: @item
3558: If the text interpreter fails to find a match in the name dictionary, it
3559: tries to treat the group of characters as a number in the current number
3560: base (when you start up Forth, the current number base is base 10). If
3561: the group of characters legitimately represents a number, the text
3562: interpreter pushes the number onto a stack (we'll learn more about that
3563: in the next section).
3564: @end itemize
3565:
3566: If the text interpreter is unable to do either of these things with any
3567: group of characters, it discards the group of characters and the rest of
3568: the line, then prints an error message. If the text interpreter reaches
3569: the end of the line without error, it prints the status message ``@code{ ok}''
3570: followed by carriage-return.
3571:
3572: This is the simplest command we can give to the text interpreter:
3573:
3574: @example
3575: @key{RET} ok
3576: @end example
3577:
3578: The text interpreter did everything we asked it to do (nothing) without
3579: an error, so it said that everything is ``@code{ ok}''. Try a slightly longer
3580: command:
3581:
3582: @example
3583: @kbd{12 dup fred dup@key{RET}}
3584: :1: Undefined word
3585: 12 dup fred dup
3586: ^^^^
3587: $400D2BA8 Bounce
3588: $400DBDA8 no.extensions
3589: @end example
3590:
3591: When you press the carriage-return key, the text interpreter starts to
3592: work its way along the line:
3593:
3594: @itemize @bullet
3595: @item
3596: When it gets to the space after the @code{2}, it takes the group of
3597: characters @code{12} and looks them up in the name
3598: dictionary@footnote{We can't tell if it found them or not, but assume
3599: for now that it did not}. There is no match for this group of characters
3600: in the name dictionary, so it tries to treat them as a number. It is
3601: able to do this successfully, so it puts the number, 12, ``on the stack''
3602: (whatever that means).
3603: @item
3604: The text interpreter resumes scanning the line and gets the next group
3605: of characters, @code{dup}. It looks it up in the name dictionary and
3606: (you'll have to take my word for this) finds it, and executes the word
3607: @code{dup} (whatever that means).
3608: @item
3609: Once again, the text interpreter resumes scanning the line and gets the
3610: group of characters @code{fred}. It looks them up in the name
3611: dictionary, but can't find them. It tries to treat them as a number, but
3612: they don't represent any legal number.
3613: @end itemize
3614:
3615: At this point, the text interpreter gives up and prints an error
3616: message. The error message shows exactly how far the text interpreter
3617: got in processing the line. In particular, it shows that the text
3618: interpreter made no attempt to do anything with the final character
3619: group, @code{dup}, even though we have good reason to believe that the
3620: text interpreter would have no problem looking that word up and
3621: executing it a second time.
3622:
3623:
3624: @comment ----------------------------------------------
3625: @node Stacks and Postfix notation, Your first definition, Introducing the Text Interpreter, Introduction
3626: @section Stacks, postfix notation and parameter passing
3627: @cindex text interpreter
3628: @cindex outer interpreter
3629:
3630: In procedural programming languages (like C and Pascal), the
3631: building-block of programs is the @dfn{function} or @dfn{procedure}. These
3632: functions or procedures are called with @dfn{explicit parameters}. For
3633: example, in C we might write:
3634:
3635: @example
3636: total = total + new_volume(length,height,depth);
3637: @end example
3638:
3639: @noindent
3640: where new_volume is a function-call to another piece of code, and total,
3641: length, height and depth are all variables. length, height and depth are
3642: parameters to the function-call.
3643:
3644: In Forth, the equivalent of the function or procedure is the
3645: @dfn{definition} and parameters are implicitly passed between
3646: definitions using a shared stack that is visible to the
3647: programmer. Although Forth does support variables, the existence of the
3648: stack means that they are used far less often than in most other
3649: programming languages. When the text interpreter encounters a number, it
3650: will place (@dfn{push}) it on the stack. There are several stacks (the
3651: actual number is implementation-dependent ...) and the particular stack
3652: used for any operation is implied unambiguously by the operation being
3653: performed. The stack used for all integer operations is called the @dfn{data
3654: stack} and, since this is the stack used most commonly, references to
3655: ``the data stack'' are often abbreviated to ``the stack''.
3656:
3657: The stacks have a last-in, first-out (LIFO) organisation. If you type:
3658:
3659: @example
3660: @kbd{1 2 3@key{RET}} ok
3661: @end example
3662:
3663: Then this instructs the text interpreter to placed three numbers on the
3664: (data) stack. An analogy for the behaviour of the stack is to take a
3665: pack of playing cards and deal out the ace (1), 2 and 3 into a pile on
3666: the table. The 3 was the last card onto the pile (``last-in'') and if
3667: you take a card off the pile then, unless you're prepared to fiddle a
3668: bit, the card that you take off will be the 3 (``first-out''). The
3669: number that will be first-out of the stack is called the @dfn{top of
3670: stack}, which
3671: @cindex TOS definition
3672: is often abbreviated to @dfn{TOS}.
3673:
3674: To understand how parameters are passed in Forth, consider the
3675: behaviour of the definition @code{+} (pronounced ``plus''). You will not
3676: be surprised to learn that this definition performs addition. More
3677: precisely, it adds two number together and produces a result. Where does
3678: it get the two numbers from? It takes the top two numbers off the
3679: stack. Where does it place the result? On the stack. You can act-out the
3680: behaviour of @code{+} with your playing cards like this:
3681:
3682: @itemize @bullet
3683: @item
3684: Pick up two cards from the stack on the table
3685: @item
3686: Stare at them intently and ask yourself ``what @i{is} the sum of these two
3687: numbers''
3688: @item
3689: Decide that the answer is 5
3690: @item
3691: Shuffle the two cards back into the pack and find a 5
3692: @item
3693: Put a 5 on the remaining ace that's on the table.
3694: @end itemize
3695:
3696: If you don't have a pack of cards handy but you do have Forth running,
3697: you can use the definition @code{.s} to show the current state of the stack,
3698: without affecting the stack. Type:
3699:
3700: @example
3701: @kbd{clearstack 1 2 3@key{RET}} ok
3702: @kbd{.s@key{RET}} <3> 1 2 3 ok
3703: @end example
3704:
3705: The text interpreter looks up the word @code{clearstack} and executes
3706: it; it tidies up the stack and removes any entries that may have been
3707: left on it by earlier examples. The text interpreter pushes each of the
3708: three numbers in turn onto the stack. Finally, the text interpreter
3709: looks up the word @code{.s} and executes it. The effect of executing
3710: @code{.s} is to print the ``<3>'' (the total number of items on the stack)
3711: followed by a list of all the items on the stack; the item on the far
3712: right-hand side is the TOS.
3713:
3714: You can now type:
3715:
3716: @example
3717: @kbd{+ .s@key{RET}} <2> 1 5 ok
3718: @end example
3719:
3720: @noindent
3721: which is correct; there are now 2 items on the stack and the result of
3722: the addition is 5.
3723:
3724: If you're playing with cards, try doing a second addition: pick up the
3725: two cards, work out that their sum is 6, shuffle them into the pack,
3726: look for a 6 and place that on the table. You now have just one item on
3727: the stack. What happens if you try to do a third addition? Pick up the
3728: first card, pick up the second card -- ah! There is no second card. This
3729: is called a @dfn{stack underflow} and consitutes an error. If you try to
3730: do the same thing with Forth it will report an error (probably a Stack
3731: Underflow or an Invalid Memory Address error).
3732:
3733: The opposite situation to a stack underflow is a @dfn{stack overflow},
3734: which simply accepts that there is a finite amount of storage space
3735: reserved for the stack. To stretch the playing card analogy, if you had
3736: enough packs of cards and you piled the cards up on the table, you would
3737: eventually be unable to add another card; you'd hit the ceiling. Gforth
3738: allows you to set the maximum size of the stacks. In general, the only
3739: time that you will get a stack overflow is because a definition has a
3740: bug in it and is generating data on the stack uncontrollably.
3741:
3742: There's one final use for the playing card analogy. If you model your
3743: stack using a pack of playing cards, the maximum number of items on
3744: your stack will be 52 (I assume you didn't use the Joker). The maximum
3745: @i{value} of any item on the stack is 13 (the King). In fact, the only
3746: possible numbers are positive integer numbers 1 through 13; you can't
3747: have (for example) 0 or 27 or 3.52 or -2. If you change the way you
3748: think about some of the cards, you can accommodate different
3749: numbers. For example, you could think of the Jack as representing 0,
3750: the Queen as representing -1 and the King as representing -2. Your
3751: @i{range} remains unchanged (you can still only represent a total of 13
3752: numbers) but the numbers that you can represent are -2 through 10.
3753:
3754: In that analogy, the limit was the amount of information that a single
3755: stack entry could hold, and Forth has a similar limit. In Forth, the
3756: size of a stack entry is called a @dfn{cell}. The actual size of a cell is
3757: implementation dependent and affects the maximum value that a stack
3758: entry can hold. A Standard Forth provides a cell size of at least
3759: 16-bits, and most desktop systems use a cell size of 32-bits.
3760:
3761: Forth does not do any type checking for you, so you are free to
3762: manipulate and combine stack items in any way you wish. A convenient way
3763: of treating stack items is as 2's complement signed integers, and that
3764: is what Standard words like @code{+} do. Therefore you can type:
3765:
3766: @example
3767: @kbd{-5 12 + .s@key{RET}} <1> 7 ok
3768: @end example
3769:
3770: If you use numbers and definitions like @code{+} in order to turn Forth
3771: into a great big pocket calculator, you will realise that it's rather
3772: different from a normal calculator. Rather than typing 2 + 3 = you had
3773: to type 2 3 + (ignore the fact that you had to use @code{.s} to see the
3774: result). The terminology used to describe this difference is to say that
3775: your calculator uses @dfn{Infix Notation} (parameters and operators are
3776: mixed) whilst Forth uses @dfn{Postfix Notation} (parameters and
3777: operators are separate), also called @dfn{Reverse Polish Notation}.
3778:
3779: Whilst postfix notation might look confusing to begin with, it has
3780: several important advantages:
3781:
3782: @itemize @bullet
3783: @item
3784: it is unambiguous
3785: @item
3786: it is more concise
3787: @item
3788: it fits naturally with a stack-based system
3789: @end itemize
3790:
3791: To examine these claims in more detail, consider these sums:
3792:
3793: @example
3794: 6 + 5 * 4 =
3795: 4 * 5 + 6 =
3796: @end example
3797:
3798: If you're just learning maths or your maths is very rusty, you will
3799: probably come up with the answer 44 for the first and 26 for the
3800: second. If you are a bit of a whizz at maths you will remember the
3801: @i{convention} that multiplication takes precendence over addition, and
3802: you'd come up with the answer 26 both times. To explain the answer 26
3803: to someone who got the answer 44, you'd probably rewrite the first sum
3804: like this:
3805:
3806: @example
3807: 6 + (5 * 4) =
3808: @end example
3809:
3810: If what you really wanted was to perform the addition before the
3811: multiplication, you would have to use parentheses to force it.
3812:
3813: If you did the first two sums on a pocket calculator you would probably
3814: get the right answers, unless you were very cautious and entered them using
3815: these keystroke sequences:
3816:
3817: 6 + 5 = * 4 =
3818: 4 * 5 = + 6 =
3819:
3820: Postfix notation is unambiguous because the order that the operators
3821: are applied is always explicit; that also means that parentheses are
3822: never required. The operators are @i{active} (the act of quoting the
3823: operator makes the operation occur) which removes the need for ``=''.
3824:
3825: The sum 6 + 5 * 4 can be written (in postfix notation) in two
3826: equivalent ways:
3827:
3828: @example
3829: 6 5 4 * + or:
3830: 5 4 * 6 +
3831: @end example
3832:
3833: An important thing that you should notice about this notation is that
3834: the @i{order} of the numbers does not change; if you want to subtract
3835: 2 from 10 you type @code{10 2 -}.
3836:
3837: The reason that Forth uses postfix notation is very simple to explain: it
3838: makes the implementation extremely simple, and it follows naturally from
3839: using the stack as a mechanism for passing parameters. Another way of
3840: thinking about this is to realise that all Forth definitions are
3841: @i{active}; they execute as they are encountered by the text
3842: interpreter. The result of this is that the syntax of Forth is trivially
3843: simple.
3844:
3845:
3846:
3847: @comment ----------------------------------------------
3848: @node Your first definition, How does that work?, Stacks and Postfix notation, Introduction
3849: @section Your first Forth definition
3850: @cindex first definition
3851:
3852: Until now, the examples we've seen have been trivial; we've just been
3853: using Forth as a bigger-than-pocket calculator. Also, each calculation
3854: we've shown has been a ``one-off'' -- to repeat it we'd need to type it in
3855: again@footnote{That's not quite true. If you press the up-arrow key on
3856: your keyboard you should be able to scroll back to any earlier command,
3857: edit it and re-enter it.} In this section we'll see how to add new
3858: words to Forth's vocabulary.
3859:
3860: The easiest way to create a new word is to use a @dfn{colon
3861: definition}. We'll define a few and try them out before worrying too
3862: much about how they work. Try typing in these examples; be careful to
3863: copy the spaces accurately:
3864:
3865: @example
3866: : add-two 2 + . ;
3867: : greet ." Hello and welcome" ;
3868: : demo 5 add-two ;
3869: @end example
3870:
3871: @noindent
3872: Now try them out:
3873:
3874: @example
3875: @kbd{greet@key{RET}} Hello and welcome ok
3876: @kbd{greet greet@key{RET}} Hello and welcomeHello and welcome ok
3877: @kbd{4 add-two@key{RET}} 6 ok
3878: @kbd{demo@key{RET}} 7 ok
3879: @kbd{9 greet demo add-two@key{RET}} Hello and welcome7 11 ok
3880: @end example
3881:
3882: The first new thing that we've introduced here is the pair of words
3883: @code{:} and @code{;}. These are used to start and terminate a new
3884: definition, respectively. The first word after the @code{:} is the name
3885: for the new definition.
3886:
3887: As you can see from the examples, a definition is built up of words that
3888: have already been defined; Forth makes no distinction between
3889: definitions that existed when you started the system up, and those that
3890: you define yourself.
3891:
3892: The examples also introduce the words @code{.} (dot), @code{."}
3893: (dot-quote) and @code{dup} (dewp). Dot takes the value from the top of
3894: the stack and displays it. It's like @code{.s} except that it only
3895: displays the top item of the stack and it is destructive; after it has
3896: executed, the number is no longer on the stack. There is always one
3897: space printed after the number, and no spaces before it. Dot-quote
3898: defines a string (a sequence of characters) that will be printed when
3899: the word is executed. The string can contain any printable characters
3900: except @code{"}. A @code{"} has a special function; it is not a Forth
3901: word but it acts as a delimiter (the way that delimiters work is
3902: described in the next section). Finally, @code{dup} duplicates the value
3903: at the top of the stack. Try typing @code{5 dup .s} to see what it does.
3904:
3905: We already know that the text interpreter searches through the
3906: dictionary to locate names. If you've followed the examples earlier, you
3907: will already have a definition called @code{add-two}. Lets try modifying
3908: it by typing in a new definition:
3909:
3910: @example
3911: @kbd{: add-two dup . ." + 2 =" 2 + . ;@key{RET}} redefined add-two ok
3912: @end example
3913:
3914: Forth recognised that we were defining a word that already exists, and
3915: printed a message to warn us of that fact. Let's try out the new
3916: definition:
3917:
3918: @example
3919: @kbd{9 add-two@key{RET}} 9 + 2 =11 ok
3920: @end example
3921:
3922: @noindent
3923: All that we've actually done here, though, is to create a new
3924: definition, with a particular name. The fact that there was already a
3925: definition with the same name did not make any difference to the way
3926: that the new definition was created (except that Forth printed a warning
3927: message). The old definition of add-two still exists (try @code{demo}
3928: again to see that this is true). Any new definition will use the new
3929: definition of @code{add-two}, but old definitions continue to use the
3930: version that already existed at the time that they were @code{compiled}.
3931:
3932: Before you go on to the next section, try defining and redefining some
3933: words of your own.
3934:
3935: @comment ----------------------------------------------
3936: @node How does that work?, Forth is written in Forth, Your first definition, Introduction
3937: @section How does that work?
3938: @cindex parsing words
3939:
3940: @c That's pretty deep (IMO way too deep) for an introduction. - anton
3941:
3942: @c Is it a good idea to talk about the interpretation semantics of a
3943: @c number? We don't have an xt to go along with it. - anton
3944:
3945: @c Now that I have eliminated execution semantics, I wonder if it would not
3946: @c be better to keep them (or add run-time semantics), to make it easier to
3947: @c explain what compilation semantics usually does. - anton
3948:
3949: @c nac-> I removed the term ``default compilation sematics'' from the
3950: @c introductory chapter. Removing ``execution semantics'' was making
3951: @c everything simpler to explain, then I think the use of this term made
3952: @c everything more complex again. I replaced it with ``default
3953: @c semantics'' (which is used elsewhere in the manual) by which I mean
3954: @c ``a definition that has neither the immediate nor the compile-only
3955: @c flag set''. I reworded big chunks of the ``how does that work''
3956: @c section (and, unusually for me, I think I even made it shorter!). See
3957: @c what you think -- I know I have not addressed your primary concern
3958: @c that it is too heavy-going for an introduction. From what I understood
3959: @c of your course notes it looks as though they might be a good framework.
3960: @c Things that I've tried to capture here are some things that came as a
3961: @c great revelation here when I first understood them. Also, I like the
3962: @c fact that a very simple code example shows up almost all of the issues
3963: @c that you need to understand to see how Forth works. That's unique and
3964: @c worthwhile to emphasise.
3965:
3966: Now we're going to take another look at the definition of @code{add-two}
3967: from the previous section. From our knowledge of the way that the text
3968: interpreter works, we would have expected this result when we tried to
3969: define @code{add-two}:
3970:
3971: @example
3972: @kbd{: add-two 2 + . ;@key{RET}}
3973: ^^^^^^^
3974: Error: Undefined word
3975: @end example
3976:
3977: The reason that this didn't happen is bound up in the way that @code{:}
3978: works. The word @code{:} does two special things. The first special
3979: thing that it does prevents the text interpreter from ever seeing the
3980: characters @code{add-two}. The text interpreter uses a variable called
3981: @cindex modifying >IN
3982: @code{>IN} (pronounced ``to-in'') to keep track of where it is in the
3983: input line. When it encounters the word @code{:} it behaves in exactly
3984: the same way as it does for any other word; it looks it up in the name
3985: dictionary, finds its xt and executes it. When @code{:} executes, it
3986: looks at the input buffer, finds the word @code{add-two} and advances the
3987: value of @code{>IN} to point past it. It then does some other stuff
3988: associated with creating the new definition (including creating an entry
3989: for @code{add-two} in the name dictionary). When the execution of @code{:}
3990: completes, control returns to the text interpreter, which is oblivious
3991: to the fact that it has been tricked into ignoring part of the input
3992: line.
3993:
3994: @cindex parsing words
3995: Words like @code{:} -- words that advance the value of @code{>IN} and so
3996: prevent the text interpreter from acting on the whole of the input line
3997: -- are called @dfn{parsing words}.
3998:
3999: @cindex @code{state} - effect on the text interpreter
4000: @cindex text interpreter - effect of state
4001: The second special thing that @code{:} does is change the value of a
4002: variable called @code{state}, which affects the way that the text
4003: interpreter behaves. When Gforth starts up, @code{state} has the value
4004: 0, and the text interpreter is said to be @dfn{interpreting}. During a
4005: colon definition (started with @code{:}), @code{state} is set to -1 and
4006: the text interpreter is said to be @dfn{compiling}.
4007:
4008: In this example, the text interpreter is compiling when it processes the
4009: string ``@code{2 + . ;}''. It still breaks the string down into
4010: character sequences in the same way. However, instead of pushing the
4011: number @code{2} onto the stack, it lays down (@dfn{compiles}) some magic
4012: into the definition of @code{add-two} that will make the number @code{2} get
4013: pushed onto the stack when @code{add-two} is @dfn{executed}. Similarly,
4014: the behaviours of @code{+} and @code{.} are also compiled into the
4015: definition.
4016:
4017: One category of words don't get compiled. These so-called @dfn{immediate
4018: words} get executed (performed @i{now}) regardless of whether the text
4019: interpreter is interpreting or compiling. The word @code{;} is an
4020: immediate word. Rather than being compiled into the definition, it
4021: executes. Its effect is to terminate the current definition, which
4022: includes changing the value of @code{state} back to 0.
4023:
4024: When you execute @code{add-two}, it has a @dfn{run-time effect} that is
4025: exactly the same as if you had typed @code{2 + . @key{RET}} outside of a
4026: definition.
4027:
4028: In Forth, every word or number can be described in terms of two
4029: properties:
4030:
4031: @itemize @bullet
4032: @item
4033: @cindex interpretation semantics
4034: Its @dfn{interpretation semantics} describe how it will behave when the
4035: text interpreter encounters it in @dfn{interpret} state. The
4036: interpretation semantics of a word are represented by an @dfn{execution
4037: token}.
4038: @item
4039: @cindex compilation semantics
4040: Its @dfn{compilation semantics} describe how it will behave when the
4041: text interpreter encounters it in @dfn{compile} state. The compilation
4042: semantics of a word are represented in an implementation-dependent way;
4043: Gforth uses a @dfn{compilation token}.
4044: @end itemize
4045:
4046: @noindent
4047: Numbers are always treated in a fixed way:
4048:
4049: @itemize @bullet
4050: @item
4051: When the number is @dfn{interpreted}, its behaviour is to push the
4052: number onto the stack.
4053: @item
4054: When the number is @dfn{compiled}, a piece of code is appended to the
4055: current definition that pushes the number when it runs. (In other words,
4056: the compilation semantics of a number are to postpone its interpretation
4057: semantics until the run-time of the definition that it is being compiled
4058: into.)
4059: @end itemize
4060:
4061: Words don't behave in such a regular way, but most have @i{default
4062: semantics} which means that they behave like this:
4063:
4064: @itemize @bullet
4065: @item
4066: The @dfn{interpretation semantics} of the word are to do something useful.
4067: @item
4068: The @dfn{compilation semantics} of the word are to append its
4069: @dfn{interpretation semantics} to the current definition (so that its
4070: run-time behaviour is to do something useful).
4071: @end itemize
4072:
4073: @cindex immediate words
4074: The actual behaviour of any particular word can be controlled by using
4075: the words @code{immediate} and @code{compile-only} when the word is
4076: defined. These words set flags in the name dictionary entry of the most
4077: recently defined word, and these flags are retrieved by the text
4078: interpreter when it finds the word in the name dictionary.
4079:
4080: A word that is marked as @dfn{immediate} has compilation semantics that
4081: are identical to its interpretation semantics. In other words, it
4082: behaves like this:
4083:
4084: @itemize @bullet
4085: @item
4086: The @dfn{interpretation semantics} of the word are to do something useful.
4087: @item
4088: The @dfn{compilation semantics} of the word are to do something useful
4089: (and actually the same thing); i.e., it is executed during compilation.
4090: @end itemize
4091:
4092: Marking a word as @dfn{compile-only} prohibits the text interpreter from
4093: performing the interpretation semantics of the word directly; an attempt
4094: to do so will generate an error. It is never necessary to use
4095: @code{compile-only} (and it is not even part of ANS Forth, though it is
4096: provided by many implementations) but it is good etiquette to apply it
4097: to a word that will not behave correctly (and might have unexpected
4098: side-effects) in interpret state. For example, it is only legal to use
4099: the conditional word @code{IF} within a definition. If you forget this
4100: and try to use it elsewhere, the fact that (in Gforth) it is marked as
4101: @code{compile-only} allows the text interpreter to generate a helpful
4102: error message rather than subjecting you to the consequences of your
4103: folly.
4104:
4105: This example shows the difference between an immediate and a
4106: non-immediate word:
4107:
4108: @example
4109: : show-state state @@ . ;
4110: : show-state-now show-state ; immediate
4111: : word1 show-state ;
4112: : word2 show-state-now ;
4113: @end example
4114:
4115: The word @code{immediate} after the definition of @code{show-state-now}
4116: makes that word an immediate word. These definitions introduce a new
4117: word: @code{@@} (pronounced ``fetch''). This word fetches the value of a
4118: variable, and leaves it on the stack. Therefore, the behaviour of
4119: @code{show-state} is to print a number that represents the current value
4120: of @code{state}.
4121:
4122: When you execute @code{word1}, it prints the number 0, indicating that
4123: the system is interpreting. When the text interpreter compiled the
4124: definition of @code{word1}, it encountered @code{show-state} whose
4125: compilation semantics are to append its interpretation semantics to the
4126: current definition. When you execute @code{word1}, it performs the
4127: interpretation semantics of @code{show-state}. At the time that @code{word1}
4128: (and therefore @code{show-state}) are executed, the system is
4129: interpreting.
4130:
4131: When you pressed @key{RET} after entering the definition of @code{word2},
4132: you should have seen the number -1 printed, followed by ``@code{
4133: ok}''. When the text interpreter compiled the definition of
4134: @code{word2}, it encountered @code{show-state-now}, an immediate word,
4135: whose compilation semantics are therefore to perform its interpretation
4136: semantics. It is executed straight away (even before the text
4137: interpreter has moved on to process another group of characters; the
4138: @code{;} in this example). The effect of executing it are to display the
4139: value of @code{state} @i{at the time that the definition of}
4140: @code{word2} @i{is being defined}. Printing -1 demonstrates that the
4141: system is compiling at this time. If you execute @code{word2} it does
4142: nothing at all.
4143:
4144: @cindex @code{."}, how it works
4145: Before leaving the subject of immediate words, consider the behaviour of
4146: @code{."} in the definition of @code{greet}, in the previous
4147: section. This word is both a parsing word and an immediate word. Notice
4148: that there is a space between @code{."} and the start of the text
4149: @code{Hello and welcome}, but that there is no space between the last
4150: letter of @code{welcome} and the @code{"} character. The reason for this
4151: is that @code{."} is a Forth word; it must have a space after it so that
4152: the text interpreter can identify it. The @code{"} is not a Forth word;
4153: it is a @dfn{delimiter}. The examples earlier show that, when the string
4154: is displayed, there is neither a space before the @code{H} nor after the
4155: @code{e}. Since @code{."} is an immediate word, it executes at the time
4156: that @code{greet} is defined. When it executes, its behaviour is to
4157: search forward in the input line looking for the delimiter. When it
4158: finds the delimiter, it updates @code{>IN} to point past the
4159: delimiter. It also compiles some magic code into the definition of
4160: @code{greet}; the xt of a run-time routine that prints a text string. It
4161: compiles the string @code{Hello and welcome} into memory so that it is
4162: available to be printed later. When the text interpreter gains control,
4163: the next word it finds in the input stream is @code{;} and so it
4164: terminates the definition of @code{greet}.
4165:
4166:
4167: @comment ----------------------------------------------
4168: @node Forth is written in Forth, Review - elements of a Forth system, How does that work?, Introduction
4169: @section Forth is written in Forth
4170: @cindex structure of Forth programs
4171:
4172: When you start up a Forth compiler, a large number of definitions
4173: already exist. In Forth, you develop a new application using bottom-up
4174: programming techniques to create new definitions that are defined in
4175: terms of existing definitions. As you create each definition you can
4176: test and debug it interactively.
4177:
4178: If you have tried out the examples in this section, you will probably
4179: have typed them in by hand; when you leave Gforth, your definitions will
4180: be lost. You can avoid this by using a text editor to enter Forth source
4181: code into a file, and then loading code from the file using
4182: @code{include} (@pxref{Forth source files}). A Forth source file is
4183: processed by the text interpreter, just as though you had typed it in by
4184: hand@footnote{Actually, there are some subtle differences -- see
4185: @ref{The Text Interpreter}.}.
4186:
4187: Gforth also supports the traditional Forth alternative to using text
4188: files for program entry (@pxref{Blocks}).
4189:
4190: In common with many, if not most, Forth compilers, most of Gforth is
4191: actually written in Forth. All of the @file{.fs} files in the
4192: installation directory@footnote{For example,
4193: @file{/usr/local/share/gforth...}} are Forth source files, which you can
4194: study to see examples of Forth programming.
4195:
4196: Gforth maintains a history file that records every line that you type to
4197: the text interpreter. This file is preserved between sessions, and is
4198: used to provide a command-line recall facility. If you enter long
4199: definitions by hand, you can use a text editor to paste them out of the
4200: history file into a Forth source file for reuse at a later time
4201: (for more information @pxref{Command-line editing}).
4202:
4203:
4204: @comment ----------------------------------------------
4205: @node Review - elements of a Forth system, Where to go next, Forth is written in Forth, Introduction
4206: @section Review - elements of a Forth system
4207: @cindex elements of a Forth system
4208:
4209: To summarise this chapter:
4210:
4211: @itemize @bullet
4212: @item
4213: Forth programs use @dfn{factoring} to break a problem down into small
4214: fragments called @dfn{words} or @dfn{definitions}.
4215: @item
4216: Forth program development is an interactive process.
4217: @item
4218: The main command loop that accepts input, and controls both
4219: interpretation and compilation, is called the @dfn{text interpreter}
4220: (also known as the @dfn{outer interpreter}).
4221: @item
4222: Forth has a very simple syntax, consisting of words and numbers
4223: separated by spaces or carriage-return characters. Any additional syntax
4224: is imposed by @dfn{parsing words}.
4225: @item
4226: Forth uses a stack to pass parameters between words. As a result, it
4227: uses postfix notation.
4228: @item
4229: To use a word that has previously been defined, the text interpreter
4230: searches for the word in the @dfn{name dictionary}.
4231: @item
4232: Words have @dfn{interpretation semantics} and @dfn{compilation semantics}.
4233: @item
4234: The text interpreter uses the value of @code{state} to select between
4235: the use of the @dfn{interpretation semantics} and the @dfn{compilation
4236: semantics} of a word that it encounters.
4237: @item
4238: The relationship between the @dfn{interpretation semantics} and
4239: @dfn{compilation semantics} for a word
4240: depend upon the way in which the word was defined (for example, whether
4241: it is an @dfn{immediate} word).
4242: @item
4243: Forth definitions can be implemented in Forth (called @dfn{high-level
4244: definitions}) or in some other way (usually a lower-level language and
4245: as a result often called @dfn{low-level definitions}, @dfn{code
4246: definitions} or @dfn{primitives}).
4247: @item
4248: Many Forth systems are implemented mainly in Forth.
4249: @end itemize
4250:
4251:
4252: @comment ----------------------------------------------
4253: @node Where to go next, Exercises, Review - elements of a Forth system, Introduction
4254: @section Where To Go Next
4255: @cindex where to go next
4256:
4257: Amazing as it may seem, if you have read (and understood) this far, you
4258: know almost all the fundamentals about the inner workings of a Forth
4259: system. You certainly know enough to be able to read and understand the
4260: rest of this manual and the ANS Forth document, to learn more about the
4261: facilities that Forth in general and Gforth in particular provide. Even
4262: scarier, you know almost enough to implement your own Forth system.
4263: However, that's not a good idea just yet... better to try writing some
4264: programs in Gforth.
4265:
4266: Forth has such a rich vocabulary that it can be hard to know where to
4267: start in learning it. This section suggests a few sets of words that are
4268: enough to write small but useful programs. Use the word index in this
4269: document to learn more about each word, then try it out and try to write
4270: small definitions using it. Start by experimenting with these words:
4271:
4272: @itemize @bullet
4273: @item
4274: Arithmetic: @code{+ - * / /MOD */ ABS INVERT}
4275: @item
4276: Comparison: @code{MIN MAX =}
4277: @item
4278: Logic: @code{AND OR XOR NOT}
4279: @item
4280: Stack manipulation: @code{DUP DROP SWAP OVER}
4281: @item
4282: Loops and decisions: @code{IF ELSE ENDIF ?DO I LOOP}
4283: @item
4284: Input/Output: @code{. ." EMIT CR KEY}
4285: @item
4286: Defining words: @code{: ; CREATE}
4287: @item
4288: Memory allocation words: @code{ALLOT ,}
4289: @item
4290: Tools: @code{SEE WORDS .S MARKER}
4291: @end itemize
4292:
4293: When you have mastered those, go on to:
4294:
4295: @itemize @bullet
4296: @item
4297: More defining words: @code{VARIABLE CONSTANT VALUE TO CREATE DOES>}
4298: @item
4299: Memory access: @code{@@ !}
4300: @end itemize
4301:
4302: When you have mastered these, there's nothing for it but to read through
4303: the whole of this manual and find out what you've missed.
4304:
4305: @comment ----------------------------------------------
4306: @node Exercises, , Where to go next, Introduction
4307: @section Exercises
4308: @cindex exercises
4309:
4310: TODO: provide a set of programming excercises linked into the stuff done
4311: already and into other sections of the manual. Provide solutions to all
4312: the exercises in a .fs file in the distribution.
4313:
4314: @c Get some inspiration from Starting Forth and Kelly&Spies.
4315:
4316: @c excercises:
4317: @c 1. take inches and convert to feet and inches.
4318: @c 2. take temperature and convert from fahrenheight to celcius;
4319: @c may need to care about symmetric vs floored??
4320: @c 3. take input line and do character substitution
4321: @c to encipher or decipher
4322: @c 4. as above but work on a file for in and out
4323: @c 5. take input line and convert to pig-latin
4324: @c
4325: @c thing of sets of things to exercise then come up with
4326: @c problems that need those things.
4327:
4328:
4329: @c ******************************************************************
4330: @node Words, Error messages, Introduction, Top
4331: @chapter Forth Words
4332: @cindex words
4333:
4334: @menu
4335: * Notation::
4336: * Case insensitivity::
4337: * Comments::
4338: * Boolean Flags::
4339: * Arithmetic::
4340: * Stack Manipulation::
4341: * Memory::
4342: * Control Structures::
4343: * Defining Words::
4344: * Interpretation and Compilation Semantics::
4345: * Tokens for Words::
4346: * The Text Interpreter::
4347: * Word Lists::
4348: * Environmental Queries::
4349: * Files::
4350: * Blocks::
4351: * Other I/O::
4352: * Locals::
4353: * Structures::
4354: * Object-oriented Forth::
4355: * Programming Tools::
4356: * Assembler and Code Words::
4357: * Threading Words::
4358: * Passing Commands to the OS::
4359: * Keeping track of Time::
4360: * Miscellaneous Words::
4361: @end menu
4362:
4363: @node Notation, Case insensitivity, Words, Words
4364: @section Notation
4365: @cindex notation of glossary entries
4366: @cindex format of glossary entries
4367: @cindex glossary notation format
4368: @cindex word glossary entry format
4369:
4370: The Forth words are described in this section in the glossary notation
4371: that has become a de-facto standard for Forth texts:
4372:
4373: @format
4374: @i{word} @i{Stack effect} @i{wordset} @i{pronunciation}
4375: @end format
4376: @i{Description}
4377:
4378: @table @var
4379: @item word
4380: The name of the word.
4381:
4382: @item Stack effect
4383: @cindex stack effect
4384: The stack effect is written in the notation @code{@i{before} --
4385: @i{after}}, where @i{before} and @i{after} describe the top of
4386: stack entries before and after the execution of the word. The rest of
4387: the stack is not touched by the word. The top of stack is rightmost,
4388: i.e., a stack sequence is written as it is typed in. Note that Gforth
4389: uses a separate floating point stack, but a unified stack
4390: notation. Also, return stack effects are not shown in @i{stack
4391: effect}, but in @i{Description}. The name of a stack item describes
4392: the type and/or the function of the item. See below for a discussion of
4393: the types.
4394:
4395: All words have two stack effects: A compile-time stack effect and a
4396: run-time stack effect. The compile-time stack-effect of most words is
4397: @i{ -- }. If the compile-time stack-effect of a word deviates from
4398: this standard behaviour, or the word does other unusual things at
4399: compile time, both stack effects are shown; otherwise only the run-time
4400: stack effect is shown.
4401:
4402: @cindex pronounciation of words
4403: @item pronunciation
4404: How the word is pronounced.
4405:
4406: @cindex wordset
4407: @cindex environment wordset
4408: @item wordset
4409: The ANS Forth standard is divided into several word sets. A standard
4410: system need not support all of them. Therefore, in theory, the fewer
4411: word sets your program uses the more portable it will be. However, we
4412: suspect that most ANS Forth systems on personal machines will feature
4413: all word sets. Words that are not defined in ANS Forth have
4414: @code{gforth} or @code{gforth-internal} as word set. @code{gforth}
4415: describes words that will work in future releases of Gforth;
4416: @code{gforth-internal} words are more volatile. Environmental query
4417: strings are also displayed like words; you can recognize them by the
4418: @code{environment} in the word set field.
4419:
4420: @item Description
4421: A description of the behaviour of the word.
4422: @end table
4423:
4424: @cindex types of stack items
4425: @cindex stack item types
4426: The type of a stack item is specified by the character(s) the name
4427: starts with:
4428:
4429: @table @code
4430: @item f
4431: @cindex @code{f}, stack item type
4432: Boolean flags, i.e. @code{false} or @code{true}.
4433: @item c
4434: @cindex @code{c}, stack item type
4435: Char
4436: @item w
4437: @cindex @code{w}, stack item type
4438: Cell, can contain an integer or an address
4439: @item n
4440: @cindex @code{n}, stack item type
4441: signed integer
4442: @item u
4443: @cindex @code{u}, stack item type
4444: unsigned integer
4445: @item d
4446: @cindex @code{d}, stack item type
4447: double sized signed integer
4448: @item ud
4449: @cindex @code{ud}, stack item type
4450: double sized unsigned integer
4451: @item r
4452: @cindex @code{r}, stack item type
4453: Float (on the FP stack)
4454: @item a-
4455: @cindex @code{a_}, stack item type
4456: Cell-aligned address
4457: @item c-
4458: @cindex @code{c_}, stack item type
4459: Char-aligned address (note that a Char may have two bytes in Windows NT)
4460: @item f-
4461: @cindex @code{f_}, stack item type
4462: Float-aligned address
4463: @item df-
4464: @cindex @code{df_}, stack item type
4465: Address aligned for IEEE double precision float
4466: @item sf-
4467: @cindex @code{sf_}, stack item type
4468: Address aligned for IEEE single precision float
4469: @item xt
4470: @cindex @code{xt}, stack item type
4471: Execution token, same size as Cell
4472: @item wid
4473: @cindex @code{wid}, stack item type
4474: Word list ID, same size as Cell
4475: @item ior, wior
4476: @cindex ior type description
4477: @cindex wior type description
4478: I/O result code, cell-sized. In Gforth, you can @code{throw} iors.
4479: @item f83name
4480: @cindex @code{f83name}, stack item type
4481: Pointer to a name structure
4482: @item "
4483: @cindex @code{"}, stack item type
4484: string in the input stream (not on the stack). The terminating character
4485: is a blank by default. If it is not a blank, it is shown in @code{<>}
4486: quotes.
4487: @end table
4488:
4489: @comment ----------------------------------------------
4490: @node Case insensitivity, Comments, Notation, Words
4491: @section Case insensitivity
4492: @cindex case sensitivity
4493: @cindex upper and lower case
4494:
4495: Gforth is case-insensitive; you can enter definitions and invoke
4496: Standard words using upper, lower or mixed case (however,
4497: @pxref{core-idef, Implementation-defined options, Implementation-defined
4498: options}).
4499:
4500: ANS Forth only @i{requires} implementations to recognise Standard words
4501: when they are typed entirely in upper case. Therefore, a Standard
4502: program must use upper case for all Standard words. You can use whatever
4503: case you like for words that you define, but in a Standard program you
4504: have to use the words in the same case that you defined them.
4505:
4506: Gforth supports case sensitivity through @code{table}s (case-sensitive
4507: wordlists, @pxref{Word Lists}).
4508:
4509: Two people have asked how to convert Gforth to be case-sensitive; while
4510: we think this is a bad idea, you can change all wordlists into tables
4511: like this:
4512:
4513: @example
4514: ' table-find forth-wordlist wordlist-map @ !
4515: @end example
4516:
4517: Note that you now have to type the predefined words in the same case
4518: that we defined them, which are varying. You may want to convert them
4519: to your favourite case before doing this operation (I won't explain how,
4520: because if you are even contemplating doing this, you'd better have
4521: enough knowledge of Forth systems to know this already).
4522:
4523: @node Comments, Boolean Flags, Case insensitivity, Words
4524: @section Comments
4525: @cindex comments
4526:
4527: Forth supports two styles of comment; the traditional @i{in-line} comment,
4528: @code{(} and its modern cousin, the @i{comment to end of line}; @code{\}.
4529:
4530:
4531: doc-(
4532: doc-\
4533: doc-\G
4534:
4535:
4536: @node Boolean Flags, Arithmetic, Comments, Words
4537: @section Boolean Flags
4538: @cindex Boolean flags
4539:
4540: A Boolean flag is cell-sized. A cell with all bits clear represents the
4541: flag @code{false} and a flag with all bits set represents the flag
4542: @code{true}. Words that check a flag (for example, @code{IF}) will treat
4543: a cell that has @i{any} bit set as @code{true}.
4544: @c on and off to Memory?
4545: @c true and false to "Bitwise operations" or "Numeric comparison"?
4546:
4547: doc-true
4548: doc-false
4549: doc-on
4550: doc-off
4551:
4552:
4553: @node Arithmetic, Stack Manipulation, Boolean Flags, Words
4554: @section Arithmetic
4555: @cindex arithmetic words
4556:
4557: @cindex division with potentially negative operands
4558: Forth arithmetic is not checked, i.e., you will not hear about integer
4559: overflow on addition or multiplication, you may hear about division by
4560: zero if you are lucky. The operator is written after the operands, but
4561: the operands are still in the original order. I.e., the infix @code{2-1}
4562: corresponds to @code{2 1 -}. Forth offers a variety of division
4563: operators. If you perform division with potentially negative operands,
4564: you do not want to use @code{/} or @code{/mod} with its undefined
4565: behaviour, but rather @code{fm/mod} or @code{sm/mod} (probably the
4566: former, @pxref{Mixed precision}).
4567: @comment TODO discuss the different division forms and the std approach
4568:
4569: @menu
4570: * Single precision::
4571: * Double precision:: Double-cell integer arithmetic
4572: * Bitwise operations::
4573: * Numeric comparison::
4574: * Mixed precision:: Operations with single and double-cell integers
4575: * Floating Point::
4576: @end menu
4577:
4578: @node Single precision, Double precision, Arithmetic, Arithmetic
4579: @subsection Single precision
4580: @cindex single precision arithmetic words
4581:
4582: @c !! cell undefined
4583:
4584: By default, numbers in Forth are single-precision integers that are one
4585: cell in size. They can be signed or unsigned, depending upon how you
4586: treat them. For the rules used by the text interpreter for recognising
4587: single-precision integers see @ref{Number Conversion}.
4588:
4589: These words are all defined for signed operands, but some of them also
4590: work for unsigned numbers: @code{+}, @code{1+}, @code{-}, @code{1-},
4591: @code{*}.
4592:
4593: doc-+
4594: doc-1+
4595: doc--
4596: doc-1-
4597: doc-*
4598: doc-/
4599: doc-mod
4600: doc-/mod
4601: doc-negate
4602: doc-abs
4603: doc-min
4604: doc-max
4605: doc-floored
4606:
4607:
4608: @node Double precision, Bitwise operations, Single precision, Arithmetic
4609: @subsection Double precision
4610: @cindex double precision arithmetic words
4611:
4612: For the rules used by the text interpreter for
4613: recognising double-precision integers, see @ref{Number Conversion}.
4614:
4615: A double precision number is represented by a cell pair, with the most
4616: significant cell at the TOS. It is trivial to convert an unsigned single
4617: to a double: simply push a @code{0} onto the TOS. Since numbers are
4618: represented by Gforth using 2's complement arithmetic, converting a
4619: signed single to a (signed) double requires sign-extension across the
4620: most significant cell. This can be achieved using @code{s>d}. The moral
4621: of the story is that you cannot convert a number without knowing whether
4622: it represents an unsigned or a signed number.
4623:
4624: These words are all defined for signed operands, but some of them also
4625: work for unsigned numbers: @code{d+}, @code{d-}.
4626:
4627: doc-s>d
4628: doc-d>s
4629: doc-d+
4630: doc-d-
4631: doc-dnegate
4632: doc-dabs
4633: doc-dmin
4634: doc-dmax
4635:
4636:
4637: @node Bitwise operations, Numeric comparison, Double precision, Arithmetic
4638: @subsection Bitwise operations
4639: @cindex bitwise operation words
4640:
4641:
4642: doc-and
4643: doc-or
4644: doc-xor
4645: doc-invert
4646: doc-lshift
4647: doc-rshift
4648: doc-2*
4649: doc-d2*
4650: doc-2/
4651: doc-d2/
4652:
4653:
4654: @node Numeric comparison, Mixed precision, Bitwise operations, Arithmetic
4655: @subsection Numeric comparison
4656: @cindex numeric comparison words
4657:
4658: Note that the words that compare for equality (@code{= <> 0= 0<> d= d<>
4659: d0= d0<>}) work for for both signed and unsigned numbers.
4660:
4661: doc-<
4662: doc-<=
4663: doc-<>
4664: doc-=
4665: doc->
4666: doc->=
4667:
4668: doc-0<
4669: doc-0<=
4670: doc-0<>
4671: doc-0=
4672: doc-0>
4673: doc-0>=
4674:
4675: doc-u<
4676: doc-u<=
4677: @c u<> and u= exist but are the same as <> and =
4678: @c doc-u<>
4679: @c doc-u=
4680: doc-u>
4681: doc-u>=
4682:
4683: doc-within
4684:
4685: doc-d<
4686: doc-d<=
4687: doc-d<>
4688: doc-d=
4689: doc-d>
4690: doc-d>=
4691:
4692: doc-d0<
4693: doc-d0<=
4694: doc-d0<>
4695: doc-d0=
4696: doc-d0>
4697: doc-d0>=
4698:
4699: doc-du<
4700: doc-du<=
4701: @c du<> and du= exist but are the same as d<> and d=
4702: @c doc-du<>
4703: @c doc-du=
4704: doc-du>
4705: doc-du>=
4706:
4707:
4708: @node Mixed precision, Floating Point, Numeric comparison, Arithmetic
4709: @subsection Mixed precision
4710: @cindex mixed precision arithmetic words
4711:
4712:
4713: doc-m+
4714: doc-*/
4715: doc-*/mod
4716: doc-m*
4717: doc-um*
4718: doc-m*/
4719: doc-um/mod
4720: doc-fm/mod
4721: doc-sm/rem
4722:
4723:
4724: @node Floating Point, , Mixed precision, Arithmetic
4725: @subsection Floating Point
4726: @cindex floating point arithmetic words
4727:
4728: For the rules used by the text interpreter for
4729: recognising floating-point numbers see @ref{Number Conversion}.
4730:
4731: Gforth has a separate floating point stack, but the documentation uses
4732: the unified notation.@footnote{It's easy to generate the separate
4733: notation from that by just separating the floating-point numbers out:
4734: e.g. @code{( n r1 u r2 -- r3 )} becomes @code{( n u -- ) ( F: r1 r2 --
4735: r3 )}.}
4736:
4737: @cindex floating-point arithmetic, pitfalls
4738: Floating point numbers have a number of unpleasant surprises for the
4739: unwary (e.g., floating point addition is not associative) and even a few
4740: for the wary. You should not use them unless you know what you are doing
4741: or you don't care that the results you get are totally bogus. If you
4742: want to learn about the problems of floating point numbers (and how to
4743: avoid them), you might start with @cite{David Goldberg,
4744: @uref{http://www.validgh.com/goldberg/paper.ps,What Every Computer
4745: Scientist Should Know About Floating-Point Arithmetic}, ACM Computing
4746: Surveys 23(1):5@minus{}48, March 1991}.
4747:
4748:
4749: doc-d>f
4750: doc-f>d
4751: doc-f+
4752: doc-f-
4753: doc-f*
4754: doc-f/
4755: doc-fnegate
4756: doc-fabs
4757: doc-fmax
4758: doc-fmin
4759: doc-floor
4760: doc-fround
4761: doc-f**
4762: doc-fsqrt
4763: doc-fexp
4764: doc-fexpm1
4765: doc-fln
4766: doc-flnp1
4767: doc-flog
4768: doc-falog
4769: doc-f2*
4770: doc-f2/
4771: doc-1/f
4772: doc-precision
4773: doc-set-precision
4774:
4775: @cindex angles in trigonometric operations
4776: @cindex trigonometric operations
4777: Angles in floating point operations are given in radians (a full circle
4778: has 2 pi radians).
4779:
4780: doc-fsin
4781: doc-fcos
4782: doc-fsincos
4783: doc-ftan
4784: doc-fasin
4785: doc-facos
4786: doc-fatan
4787: doc-fatan2
4788: doc-fsinh
4789: doc-fcosh
4790: doc-ftanh
4791: doc-fasinh
4792: doc-facosh
4793: doc-fatanh
4794: doc-pi
4795:
4796: @cindex equality of floats
4797: @cindex floating-point comparisons
4798: One particular problem with floating-point arithmetic is that comparison
4799: for equality often fails when you would expect it to succeed. For this
4800: reason approximate equality is often preferred (but you still have to
4801: know what you are doing). Also note that IEEE NaNs may compare
4802: differently from what you might expect. The comparison words are:
4803:
4804: doc-f~rel
4805: doc-f~abs
4806: doc-f~
4807: doc-f=
4808: doc-f<>
4809:
4810: doc-f<
4811: doc-f<=
4812: doc-f>
4813: doc-f>=
4814:
4815: doc-f0<
4816: doc-f0<=
4817: doc-f0<>
4818: doc-f0=
4819: doc-f0>
4820: doc-f0>=
4821:
4822:
4823: @node Stack Manipulation, Memory, Arithmetic, Words
4824: @section Stack Manipulation
4825: @cindex stack manipulation words
4826:
4827: @cindex floating-point stack in the standard
4828: Gforth maintains a number of separate stacks:
4829:
4830: @cindex data stack
4831: @cindex parameter stack
4832: @itemize @bullet
4833: @item
4834: A data stack (also known as the @dfn{parameter stack}) -- for
4835: characters, cells, addresses, and double cells.
4836:
4837: @cindex floating-point stack
4838: @item
4839: A floating point stack -- for holding floating point (FP) numbers.
4840:
4841: @cindex return stack
4842: @item
4843: A return stack -- for holding the return addresses of colon
4844: definitions and other (non-FP) data.
4845:
4846: @cindex locals stack
4847: @item
4848: A locals stack -- for holding local variables.
4849: @end itemize
4850:
4851: @menu
4852: * Data stack::
4853: * Floating point stack::
4854: * Return stack::
4855: * Locals stack::
4856: * Stack pointer manipulation::
4857: @end menu
4858:
4859: @node Data stack, Floating point stack, Stack Manipulation, Stack Manipulation
4860: @subsection Data stack
4861: @cindex data stack manipulation words
4862: @cindex stack manipulations words, data stack
4863:
4864:
4865: doc-drop
4866: doc-nip
4867: doc-dup
4868: doc-over
4869: doc-tuck
4870: doc-swap
4871: doc-pick
4872: doc-rot
4873: doc--rot
4874: doc-?dup
4875: doc-roll
4876: doc-2drop
4877: doc-2nip
4878: doc-2dup
4879: doc-2over
4880: doc-2tuck
4881: doc-2swap
4882: doc-2rot
4883:
4884:
4885: @node Floating point stack, Return stack, Data stack, Stack Manipulation
4886: @subsection Floating point stack
4887: @cindex floating-point stack manipulation words
4888: @cindex stack manipulation words, floating-point stack
4889:
4890: Whilst every sane Forth has a separate floating-point stack, it is not
4891: strictly required; an ANS Forth system could theoretically keep
4892: floating-point numbers on the data stack. As an additional difficulty,
4893: you don't know how many cells a floating-point number takes. It is
4894: reportedly possible to write words in a way that they work also for a
4895: unified stack model, but we do not recommend trying it. Instead, just
4896: say that your program has an environmental dependency on a separate
4897: floating-point stack.
4898:
4899: doc-floating-stack
4900:
4901: doc-fdrop
4902: doc-fnip
4903: doc-fdup
4904: doc-fover
4905: doc-ftuck
4906: doc-fswap
4907: doc-fpick
4908: doc-frot
4909:
4910:
4911: @node Return stack, Locals stack, Floating point stack, Stack Manipulation
4912: @subsection Return stack
4913: @cindex return stack manipulation words
4914: @cindex stack manipulation words, return stack
4915:
4916: @cindex return stack and locals
4917: @cindex locals and return stack
4918: A Forth system is allowed to keep local variables on the
4919: return stack. This is reasonable, as local variables usually eliminate
4920: the need to use the return stack explicitly. So, if you want to produce
4921: a standard compliant program and you are using local variables in a
4922: word, forget about return stack manipulations in that word (refer to the
4923: standard document for the exact rules).
4924:
4925: doc->r
4926: doc-r>
4927: doc-r@
4928: doc-rdrop
4929: doc-2>r
4930: doc-2r>
4931: doc-2r@
4932: doc-2rdrop
4933:
4934:
4935: @node Locals stack, Stack pointer manipulation, Return stack, Stack Manipulation
4936: @subsection Locals stack
4937:
4938: Gforth uses an extra locals stack. It is described, along with the
4939: reasons for its existence, in @ref{Locals implementation}.
4940:
4941: @node Stack pointer manipulation, , Locals stack, Stack Manipulation
4942: @subsection Stack pointer manipulation
4943: @cindex stack pointer manipulation words
4944:
4945: @c removed s0 r0 l0 -- they are obsolete aliases for sp0 rp0 lp0
4946: doc-sp0
4947: doc-sp@
4948: doc-sp!
4949: doc-fp0
4950: doc-fp@
4951: doc-fp!
4952: doc-rp0
4953: doc-rp@
4954: doc-rp!
4955: doc-lp0
4956: doc-lp@
4957: doc-lp!
4958:
4959:
4960: @node Memory, Control Structures, Stack Manipulation, Words
4961: @section Memory
4962: @cindex memory words
4963:
4964: @menu
4965: * Memory model::
4966: * Dictionary allocation::
4967: * Heap Allocation::
4968: * Memory Access::
4969: * Address arithmetic::
4970: * Memory Blocks::
4971: @end menu
4972:
4973: In addition to the standard Forth memory allocation words, there is also
4974: a @uref{http://www.complang.tuwien.ac.at/forth/garbage-collection.zip,
4975: garbage collector}.
4976:
4977: @node Memory model, Dictionary allocation, Memory, Memory
4978: @subsection ANS Forth and Gforth memory models
4979:
4980: @c The ANS Forth description is a mess (e.g., is the heap part of
4981: @c the dictionary?), so let's not stick to closely with it.
4982:
4983: ANS Forth considers a Forth system as consisting of several address
4984: spaces, of which only @dfn{data space} is managed and accessible with
4985: the memory words. Memory not necessarily in data space includes the
4986: stacks, the code (called code space) and the headers (called name
4987: space). In Gforth everything is in data space, but the code for the
4988: primitives is usually read-only.
4989:
4990: Data space is divided into a number of areas: The (data space portion of
4991: the) dictionary@footnote{Sometimes, the term @dfn{dictionary} is used to
4992: refer to the search data structure embodied in word lists and headers,
4993: because it is used for looking up names, just as you would in a
4994: conventional dictionary.}, the heap, and a number of system-allocated
4995: buffers.
4996:
4997: @cindex address arithmetic restrictions, ANS vs. Gforth
4998: @cindex contiguous regions, ANS vs. Gforth
4999: In ANS Forth data space is also divided into contiguous regions. You
5000: can only use address arithmetic within a contiguous region, not between
5001: them. Usually each allocation gives you one contiguous region, but the
5002: dictionary allocation words have additional rules (@pxref{Dictionary
5003: allocation}).
5004:
5005: Gforth provides one big address space, and address arithmetic can be
5006: performed between any addresses. However, in the dictionary headers or
5007: code are interleaved with data, so almost the only contiguous data space
5008: regions there are those described by ANS Forth as contiguous; but you
5009: can be sure that the dictionary is allocated towards increasing
5010: addresses even between contiguous regions. The memory order of
5011: allocations in the heap is platform-dependent (and possibly different
5012: from one run to the next).
5013:
5014:
5015: @node Dictionary allocation, Heap Allocation, Memory model, Memory
5016: @subsection Dictionary allocation
5017: @cindex reserving data space
5018: @cindex data space - reserving some
5019:
5020: Dictionary allocation is a stack-oriented allocation scheme, i.e., if
5021: you want to deallocate X, you also deallocate everything
5022: allocated after X.
5023:
5024: @cindex contiguous regions in dictionary allocation
5025: The allocations using the words below are contiguous and grow the region
5026: towards increasing addresses. Other words that allocate dictionary
5027: memory of any kind (i.e., defining words including @code{:noname}) end
5028: the contiguous region and start a new one.
5029:
5030: In ANS Forth only @code{create}d words are guaranteed to produce an
5031: address that is the start of the following contiguous region. In
5032: particular, the cell allocated by @code{variable} is not guaranteed to
5033: be contiguous with following @code{allot}ed memory.
5034:
5035: You can deallocate memory by using @code{allot} with a negative argument
5036: (with some restrictions, see @code{allot}). For larger deallocations use
5037: @code{marker}.
5038:
5039:
5040: doc-here
5041: doc-unused
5042: doc-allot
5043: doc-c,
5044: doc-f,
5045: doc-,
5046: doc-2,
5047:
5048: Memory accesses have to be aligned (@pxref{Address arithmetic}). So of
5049: course you should allocate memory in an aligned way, too. I.e., before
5050: allocating allocating a cell, @code{here} must be cell-aligned, etc.
5051: The words below align @code{here} if it is not already. Basically it is
5052: only already aligned for a type, if the last allocation was a multiple
5053: of the size of this type and if @code{here} was aligned for this type
5054: before.
5055:
5056: After freshly @code{create}ing a word, @code{here} is @code{align}ed in
5057: ANS Forth (@code{maxalign}ed in Gforth).
5058:
5059: doc-align
5060: doc-falign
5061: doc-sfalign
5062: doc-dfalign
5063: doc-maxalign
5064: doc-cfalign
5065:
5066:
5067: @node Heap Allocation, Memory Access, Dictionary allocation, Memory
5068: @subsection Heap allocation
5069: @cindex heap allocation
5070: @cindex dynamic allocation of memory
5071: @cindex memory-allocation word set
5072:
5073: @cindex contiguous regions and heap allocation
5074: Heap allocation supports deallocation of allocated memory in any
5075: order. Dictionary allocation is not affected by it (i.e., it does not
5076: end a contiguous region). In Gforth, these words are implemented using
5077: the standard C library calls malloc(), free() and resize().
5078:
5079: The memory region produced by one invocation of @code{allocate} or
5080: @code{resize} is internally contiguous. There is no contiguity between
5081: such a region and any other region (including others allocated from the
5082: heap).
5083:
5084: doc-allocate
5085: doc-free
5086: doc-resize
5087:
5088:
5089: @node Memory Access, Address arithmetic, Heap Allocation, Memory
5090: @subsection Memory Access
5091: @cindex memory access words
5092:
5093: doc-@
5094: doc-!
5095: doc-+!
5096: doc-c@
5097: doc-c!
5098: doc-2@
5099: doc-2!
5100: doc-f@
5101: doc-f!
5102: doc-sf@
5103: doc-sf!
5104: doc-df@
5105: doc-df!
5106:
5107:
5108: @node Address arithmetic, Memory Blocks, Memory Access, Memory
5109: @subsection Address arithmetic
5110: @cindex address arithmetic words
5111:
5112: Address arithmetic is the foundation on which you can build data
5113: structures like arrays, records (@pxref{Structures}) and objects
5114: (@pxref{Object-oriented Forth}).
5115:
5116: @cindex address unit
5117: @cindex au (address unit)
5118: ANS Forth does not specify the sizes of the data types. Instead, it
5119: offers a number of words for computing sizes and doing address
5120: arithmetic. Address arithmetic is performed in terms of address units
5121: (aus); on most systems the address unit is one byte. Note that a
5122: character may have more than one au, so @code{chars} is no noop (on
5123: platforms where it is a noop, it compiles to nothing).
5124:
5125: The basic address arithmetic words are @code{+} and @code{-}. E.g., if
5126: you have the address of a cell, perform @code{1 cells +}, and you will
5127: have the address of the next cell.
5128:
5129: @cindex contiguous regions and address arithmetic
5130: In ANS Forth you can perform address arithmetic only within a contiguous
5131: region, i.e., if you have an address into one region, you can only add
5132: and subtract such that the result is still within the region; you can
5133: only subtract or compare addresses from within the same contiguous
5134: region. Reasons: several contiguous regions can be arranged in memory
5135: in any way; on segmented systems addresses may have unusual
5136: representations, such that address arithmetic only works within a
5137: region. Gforth provides a few more guarantees (linear address space,
5138: dictionary grows upwards), but in general I have found it easy to stay
5139: within contiguous regions (exception: computing and comparing to the
5140: address just beyond the end of an array).
5141:
5142: @cindex alignment of addresses for types
5143: ANS Forth also defines words for aligning addresses for specific
5144: types. Many computers require that accesses to specific data types
5145: must only occur at specific addresses; e.g., that cells may only be
5146: accessed at addresses divisible by 4. Even if a machine allows unaligned
5147: accesses, it can usually perform aligned accesses faster.
5148:
5149: For the performance-conscious: alignment operations are usually only
5150: necessary during the definition of a data structure, not during the
5151: (more frequent) accesses to it.
5152:
5153: ANS Forth defines no words for character-aligning addresses. This is not
5154: an oversight, but reflects the fact that addresses that are not
5155: char-aligned have no use in the standard and therefore will not be
5156: created.
5157:
5158: @cindex @code{CREATE} and alignment
5159: ANS Forth guarantees that addresses returned by @code{CREATE}d words
5160: are cell-aligned; in addition, Gforth guarantees that these addresses
5161: are aligned for all purposes.
5162:
5163: Note that the ANS Forth word @code{char} has nothing to do with address
5164: arithmetic.
5165:
5166:
5167: doc-chars
5168: doc-char+
5169: doc-cells
5170: doc-cell+
5171: doc-cell
5172: doc-aligned
5173: doc-floats
5174: doc-float+
5175: doc-float
5176: doc-faligned
5177: doc-sfloats
5178: doc-sfloat+
5179: doc-sfaligned
5180: doc-dfloats
5181: doc-dfloat+
5182: doc-dfaligned
5183: doc-maxaligned
5184: doc-cfaligned
5185: doc-address-unit-bits
5186:
5187:
5188: @node Memory Blocks, , Address arithmetic, Memory
5189: @subsection Memory Blocks
5190: @cindex memory block words
5191: @cindex character strings - moving and copying
5192:
5193: Memory blocks often represent character strings; For ways of storing
5194: character strings in memory see @ref{String Formats}. For other
5195: string-processing words see @ref{Displaying characters and strings}.
5196:
5197: A few of these words work on address unit blocks. In that case, you
5198: usually have to insert @code{CHARS} before the word when working on
5199: character strings. Most words work on character blocks, and expect a
5200: char-aligned address.
5201:
5202: When copying characters between overlapping memory regions, use
5203: @code{chars move} or choose carefully between @code{cmove} and
5204: @code{cmove>}.
5205:
5206: doc-move
5207: doc-erase
5208: doc-cmove
5209: doc-cmove>
5210: doc-fill
5211: doc-blank
5212: doc-compare
5213: doc-search
5214: doc--trailing
5215: doc-/string
5216:
5217:
5218: @comment TODO examples
5219:
5220:
5221: @node Control Structures, Defining Words, Memory, Words
5222: @section Control Structures
5223: @cindex control structures
5224:
5225: Control structures in Forth cannot be used interpretively, only in a
5226: colon definition@footnote{To be precise, they have no interpretation
5227: semantics (@pxref{Interpretation and Compilation Semantics}).}. We do
5228: not like this limitation, but have not seen a satisfying way around it
5229: yet, although many schemes have been proposed.
5230:
5231: @menu
5232: * Selection:: IF ... ELSE ... ENDIF
5233: * Simple Loops:: BEGIN ...
5234: * Counted Loops:: DO
5235: * Arbitrary control structures::
5236: * Calls and returns::
5237: * Exception Handling::
5238: @end menu
5239:
5240: @node Selection, Simple Loops, Control Structures, Control Structures
5241: @subsection Selection
5242: @cindex selection control structures
5243: @cindex control structures for selection
5244:
5245: @cindex @code{IF} control structure
5246: @example
5247: @i{flag}
5248: IF
5249: @i{code}
5250: ENDIF
5251: @end example
5252: @noindent
5253:
5254: If @i{flag} is non-zero (as far as @code{IF} etc. are concerned, a cell
5255: with any bit set represents truth) @i{code} is executed.
5256:
5257: @example
5258: @i{flag}
5259: IF
5260: @i{code1}
5261: ELSE
5262: @i{code2}
5263: ENDIF
5264: @end example
5265:
5266: If @var{flag} is true, @i{code1} is executed, otherwise @i{code2} is
5267: executed.
5268:
5269: You can use @code{THEN} instead of @code{ENDIF}. Indeed, @code{THEN} is
5270: standard, and @code{ENDIF} is not, although it is quite popular. We
5271: recommend using @code{ENDIF}, because it is less confusing for people
5272: who also know other languages (and is not prone to reinforcing negative
5273: prejudices against Forth in these people). Adding @code{ENDIF} to a
5274: system that only supplies @code{THEN} is simple:
5275: @example
5276: : ENDIF POSTPONE THEN ; immediate
5277: @end example
5278:
5279: [According to @cite{Webster's New Encyclopedic Dictionary}, @dfn{then
5280: (adv.)} has the following meanings:
5281: @quotation
5282: ... 2b: following next after in order ... 3d: as a necessary consequence
5283: (if you were there, then you saw them).
5284: @end quotation
5285: Forth's @code{THEN} has the meaning 2b, whereas @code{THEN} in Pascal
5286: and many other programming languages has the meaning 3d.]
5287:
5288: Gforth also provides the words @code{?DUP-IF} and @code{?DUP-0=-IF}, so
5289: you can avoid using @code{?dup}. Using these alternatives is also more
5290: efficient than using @code{?dup}. Definitions in ANS Forth
5291: for @code{ENDIF}, @code{?DUP-IF} and @code{?DUP-0=-IF} are provided in
5292: @file{compat/control.fs}.
5293:
5294: @cindex @code{CASE} control structure
5295: @example
5296: @i{n}
5297: CASE
5298: @i{n1} OF @i{code1} ENDOF
5299: @i{n2} OF @i{code2} ENDOF
5300: @dots{}
5301: ( n ) @i{default-code} ( n )
5302: ENDCASE
5303: @end example
5304:
5305: Executes the first @i{codei}, where the @i{ni} is equal to @i{n}. If no
5306: @i{ni} matches, the optional @i{default-code} is executed. The optional
5307: default case can be added by simply writing the code after the last
5308: @code{ENDOF}. It may use @i{n}, which is on top of the stack, but must
5309: not consume it.
5310:
5311: @progstyle
5312: To keep the code understandable, you should ensure that on all paths
5313: through a selection construct the stack is changed in the same way
5314: (wrt. number and types of stack items consumed and pushed).
5315:
5316: @node Simple Loops, Counted Loops, Selection, Control Structures
5317: @subsection Simple Loops
5318: @cindex simple loops
5319: @cindex loops without count
5320:
5321: @cindex @code{WHILE} loop
5322: @example
5323: BEGIN
5324: @i{code1}
5325: @i{flag}
5326: WHILE
5327: @i{code2}
5328: REPEAT
5329: @end example
5330:
5331: @i{code1} is executed and @i{flag} is computed. If it is true,
5332: @i{code2} is executed and the loop is restarted; If @i{flag} is
5333: false, execution continues after the @code{REPEAT}.
5334:
5335: @cindex @code{UNTIL} loop
5336: @example
5337: BEGIN
5338: @i{code}
5339: @i{flag}
5340: UNTIL
5341: @end example
5342:
5343: @i{code} is executed. The loop is restarted if @code{flag} is false.
5344:
5345: @progstyle
5346: To keep the code understandable, a complete iteration of the loop should
5347: not change the number and types of the items on the stacks.
5348:
5349: @cindex endless loop
5350: @cindex loops, endless
5351: @example
5352: BEGIN
5353: @i{code}
5354: AGAIN
5355: @end example
5356:
5357: This is an endless loop.
5358:
5359: @node Counted Loops, Arbitrary control structures, Simple Loops, Control Structures
5360: @subsection Counted Loops
5361: @cindex counted loops
5362: @cindex loops, counted
5363: @cindex @code{DO} loops
5364:
5365: The basic counted loop is:
5366: @example
5367: @i{limit} @i{start}
5368: ?DO
5369: @i{body}
5370: LOOP
5371: @end example
5372:
5373: This performs one iteration for every integer, starting from @i{start}
5374: and up to, but excluding @i{limit}. The counter, or @i{index}, can be
5375: accessed with @code{i}. For example, the loop:
5376: @example
5377: 10 0 ?DO
5378: i .
5379: LOOP
5380: @end example
5381: @noindent
5382: prints @code{0 1 2 3 4 5 6 7 8 9}
5383:
5384: The index of the innermost loop can be accessed with @code{i}, the index
5385: of the next loop with @code{j}, and the index of the third loop with
5386: @code{k}.
5387:
5388:
5389: doc-i
5390: doc-j
5391: doc-k
5392:
5393:
5394: The loop control data are kept on the return stack, so there are some
5395: restrictions on mixing return stack accesses and counted loop words. In
5396: particuler, if you put values on the return stack outside the loop, you
5397: cannot read them inside the loop@footnote{well, not in a way that is
5398: portable.}. If you put values on the return stack within a loop, you
5399: have to remove them before the end of the loop and before accessing the
5400: index of the loop.
5401:
5402: There are several variations on the counted loop:
5403:
5404: @itemize @bullet
5405: @item
5406: @code{LEAVE} leaves the innermost counted loop immediately; execution
5407: continues after the associated @code{LOOP} or @code{NEXT}. For example:
5408:
5409: @example
5410: 10 0 ?DO i DUP . 3 = IF LEAVE THEN LOOP
5411: @end example
5412: prints @code{0 1 2 3}
5413:
5414:
5415: @item
5416: @code{UNLOOP} prepares for an abnormal loop exit, e.g., via
5417: @code{EXIT}. @code{UNLOOP} removes the loop control parameters from the
5418: return stack so @code{EXIT} can get to its return address. For example:
5419:
5420: @example
5421: : demo 10 0 ?DO i DUP . 3 = IF UNLOOP EXIT THEN LOOP ." Done" ;
5422: @end example
5423: prints @code{0 1 2 3}
5424:
5425:
5426: @item
5427: If @i{start} is greater than @i{limit}, a @code{?DO} loop is entered
5428: (and @code{LOOP} iterates until they become equal by wrap-around
5429: arithmetic). This behaviour is usually not what you want. Therefore,
5430: Gforth offers @code{+DO} and @code{U+DO} (as replacements for
5431: @code{?DO}), which do not enter the loop if @i{start} is greater than
5432: @i{limit}; @code{+DO} is for signed loop parameters, @code{U+DO} for
5433: unsigned loop parameters.
5434:
5435: @item
5436: @code{?DO} can be replaced by @code{DO}. @code{DO} always enters
5437: the loop, independent of the loop parameters. Do not use @code{DO}, even
5438: if you know that the loop is entered in any case. Such knowledge tends
5439: to become invalid during maintenance of a program, and then the
5440: @code{DO} will make trouble.
5441:
5442: @item
5443: @code{LOOP} can be replaced with @code{@i{n} +LOOP}; this updates the
5444: index by @i{n} instead of by 1. The loop is terminated when the border
5445: between @i{limit-1} and @i{limit} is crossed. E.g.:
5446:
5447: @example
5448: 4 0 +DO i . 2 +LOOP
5449: @end example
5450: @noindent
5451: prints @code{0 2}
5452:
5453: @example
5454: 4 1 +DO i . 2 +LOOP
5455: @end example
5456: @noindent
5457: prints @code{1 3}
5458:
5459: @item
5460: @cindex negative increment for counted loops
5461: @cindex counted loops with negative increment
5462: The behaviour of @code{@i{n} +LOOP} is peculiar when @i{n} is negative:
5463:
5464: @example
5465: -1 0 ?DO i . -1 +LOOP
5466: @end example
5467: @noindent
5468: prints @code{0 -1}
5469:
5470: @example
5471: 0 0 ?DO i . -1 +LOOP
5472: @end example
5473: prints nothing.
5474:
5475: Therefore we recommend avoiding @code{@i{n} +LOOP} with negative
5476: @i{n}. One alternative is @code{@i{u} -LOOP}, which reduces the
5477: index by @i{u} each iteration. The loop is terminated when the border
5478: between @i{limit+1} and @i{limit} is crossed. Gforth also provides
5479: @code{-DO} and @code{U-DO} for down-counting loops. E.g.:
5480:
5481: @example
5482: -2 0 -DO i . 1 -LOOP
5483: @end example
5484: @noindent
5485: prints @code{0 -1}
5486:
5487: @example
5488: -1 0 -DO i . 1 -LOOP
5489: @end example
5490: @noindent
5491: prints @code{0}
5492:
5493: @example
5494: 0 0 -DO i . 1 -LOOP
5495: @end example
5496: @noindent
5497: prints nothing.
5498:
5499: @end itemize
5500:
5501: Unfortunately, @code{+DO}, @code{U+DO}, @code{-DO}, @code{U-DO} and
5502: @code{-LOOP} are not defined in ANS Forth. However, an implementation
5503: for these words that uses only standard words is provided in
5504: @file{compat/loops.fs}.
5505:
5506:
5507: @cindex @code{FOR} loops
5508: Another counted loop is:
5509: @example
5510: @i{n}
5511: FOR
5512: @i{body}
5513: NEXT
5514: @end example
5515: This is the preferred loop of native code compiler writers who are too
5516: lazy to optimize @code{?DO} loops properly. This loop structure is not
5517: defined in ANS Forth. In Gforth, this loop iterates @i{n+1} times;
5518: @code{i} produces values starting with @i{n} and ending with 0. Other
5519: Forth systems may behave differently, even if they support @code{FOR}
5520: loops. To avoid problems, don't use @code{FOR} loops.
5521:
5522: @node Arbitrary control structures, Calls and returns, Counted Loops, Control Structures
5523: @subsection Arbitrary control structures
5524: @cindex control structures, user-defined
5525:
5526: @cindex control-flow stack
5527: ANS Forth permits and supports using control structures in a non-nested
5528: way. Information about incomplete control structures is stored on the
5529: control-flow stack. This stack may be implemented on the Forth data
5530: stack, and this is what we have done in Gforth.
5531:
5532: @cindex @code{orig}, control-flow stack item
5533: @cindex @code{dest}, control-flow stack item
5534: An @i{orig} entry represents an unresolved forward branch, a @i{dest}
5535: entry represents a backward branch target. A few words are the basis for
5536: building any control structure possible (except control structures that
5537: need storage, like calls, coroutines, and backtracking).
5538:
5539:
5540: doc-if
5541: doc-ahead
5542: doc-then
5543: doc-begin
5544: doc-until
5545: doc-again
5546: doc-cs-pick
5547: doc-cs-roll
5548:
5549:
5550: The Standard words @code{CS-PICK} and @code{CS-ROLL} allow you to
5551: manipulate the control-flow stack in a portable way. Without them, you
5552: would need to know how many stack items are occupied by a control-flow
5553: entry (many systems use one cell. In Gforth they currently take three,
5554: but this may change in the future).
5555:
5556: Some standard control structure words are built from these words:
5557:
5558:
5559: doc-else
5560: doc-while
5561: doc-repeat
5562:
5563:
5564: @noindent
5565: Gforth adds some more control-structure words:
5566:
5567:
5568: doc-endif
5569: doc-?dup-if
5570: doc-?dup-0=-if
5571:
5572:
5573: @noindent
5574: Counted loop words constitute a separate group of words:
5575:
5576:
5577: doc-?do
5578: doc-+do
5579: doc-u+do
5580: doc--do
5581: doc-u-do
5582: doc-do
5583: doc-for
5584: doc-loop
5585: doc-+loop
5586: doc--loop
5587: doc-next
5588: doc-leave
5589: doc-?leave
5590: doc-unloop
5591: doc-done
5592:
5593:
5594: The standard does not allow using @code{CS-PICK} and @code{CS-ROLL} on
5595: @i{do-sys}. Gforth allows it, but it's your job to ensure that for
5596: every @code{?DO} etc. there is exactly one @code{UNLOOP} on any path
5597: through the definition (@code{LOOP} etc. compile an @code{UNLOOP} on the
5598: fall-through path). Also, you have to ensure that all @code{LEAVE}s are
5599: resolved (by using one of the loop-ending words or @code{DONE}).
5600:
5601: @noindent
5602: Another group of control structure words are:
5603:
5604:
5605: doc-case
5606: doc-endcase
5607: doc-of
5608: doc-endof
5609:
5610:
5611: @i{case-sys} and @i{of-sys} cannot be processed using @code{CS-PICK} and
5612: @code{CS-ROLL}.
5613:
5614: @subsubsection Programming Style
5615: @cindex control structures programming style
5616: @cindex programming style, arbitrary control structures
5617:
5618: In order to ensure readability we recommend that you do not create
5619: arbitrary control structures directly, but define new control structure
5620: words for the control structure you want and use these words in your
5621: program. For example, instead of writing:
5622:
5623: @example
5624: BEGIN
5625: ...
5626: IF [ 1 CS-ROLL ]
5627: ...
5628: AGAIN THEN
5629: @end example
5630:
5631: @noindent
5632: we recommend defining control structure words, e.g.,
5633:
5634: @example
5635: : WHILE ( DEST -- ORIG DEST )
5636: POSTPONE IF
5637: 1 CS-ROLL ; immediate
5638:
5639: : REPEAT ( orig dest -- )
5640: POSTPONE AGAIN
5641: POSTPONE THEN ; immediate
5642: @end example
5643:
5644: @noindent
5645: and then using these to create the control structure:
5646:
5647: @example
5648: BEGIN
5649: ...
5650: WHILE
5651: ...
5652: REPEAT
5653: @end example
5654:
5655: That's much easier to read, isn't it? Of course, @code{REPEAT} and
5656: @code{WHILE} are predefined, so in this example it would not be
5657: necessary to define them.
5658:
5659: @node Calls and returns, Exception Handling, Arbitrary control structures, Control Structures
5660: @subsection Calls and returns
5661: @cindex calling a definition
5662: @cindex returning from a definition
5663:
5664: @cindex recursive definitions
5665: A definition can be called simply be writing the name of the definition
5666: to be called. Normally a definition is invisible during its own
5667: definition. If you want to write a directly recursive definition, you
5668: can use @code{recursive} to make the current definition visible, or
5669: @code{recurse} to call the current definition directly.
5670:
5671:
5672: doc-recursive
5673: doc-recurse
5674:
5675:
5676: @comment TODO add example of the two recursion methods
5677: @quotation
5678: @progstyle
5679: I prefer using @code{recursive} to @code{recurse}, because calling the
5680: definition by name is more descriptive (if the name is well-chosen) than
5681: the somewhat cryptic @code{recurse}. E.g., in a quicksort
5682: implementation, it is much better to read (and think) ``now sort the
5683: partitions'' than to read ``now do a recursive call''.
5684: @end quotation
5685:
5686: For mutual recursion, use @code{Defer}red words, like this:
5687:
5688: @example
5689: Defer foo
5690:
5691: : bar ( ... -- ... )
5692: ... foo ... ;
5693:
5694: :noname ( ... -- ... )
5695: ... bar ... ;
5696: IS foo
5697: @end example
5698:
5699: Deferred words are discussed in more detail in @ref{Deferred words}.
5700:
5701: The current definition returns control to the calling definition when
5702: the end of the definition is reached or @code{EXIT} is encountered.
5703:
5704: doc-exit
5705: doc-;s
5706:
5707:
5708: @node Exception Handling, , Calls and returns, Control Structures
5709: @subsection Exception Handling
5710: @cindex exceptions
5711:
5712: @c quit is a very bad idea for error handling,
5713: @c because it does not translate into a THROW
5714: @c it also does not belong into this chapter
5715:
5716: If a word detects an error condition that it cannot handle, it can
5717: @code{throw} an exception. In the simplest case, this will terminate
5718: your program, and report an appropriate error.
5719:
5720: doc-throw
5721:
5722: @code{Throw} consumes a cell-sized error number on the stack. There are
5723: some predefined error numbers in ANS Forth (see @file{errors.fs}). In
5724: Gforth (and most other systems) you can use the iors produced by various
5725: words as error numbers (e.g., a typical use of @code{allocate} is
5726: @code{allocate throw}). Gforth also provides the word @code{exception}
5727: to define your own error numbers (with decent error reporting); an ANS
5728: Forth version of this word (but without the error messages) is available
5729: in @code{compat/except.fs}. And finally, you can use your own error
5730: numbers (anything outside the range -4095..0), but won't get nice error
5731: messages, only numbers. For example, try:
5732:
5733: @example
5734: -10 throw \ ANS defined
5735: -267 throw \ system defined
5736: s" my error" exception throw \ user defined
5737: 7 throw \ arbitrary number
5738: @end example
5739:
5740: doc---exception-exception
5741:
5742: A common idiom to @code{THROW} a specific error if a flag is true is
5743: this:
5744:
5745: @example
5746: @code{( flag ) 0<> @i{errno} and throw}
5747: @end example
5748:
5749: Your program can provide exception handlers to catch exceptions. An
5750: exception handler can be used to correct the problem, or to clean up
5751: some data structures and just throw the exception to the next exception
5752: handler. Note that @code{throw} jumps to the dynamically innermost
5753: exception handler. The system's exception handler is outermost, and just
5754: prints an error and restarts command-line interpretation (or, in batch
5755: mode (i.e., while processing the shell command line), leaves Gforth).
5756:
5757: The ANS Forth way to catch exceptions is @code{catch}:
5758:
5759: doc-catch
5760:
5761: The most common use of exception handlers is to clean up the state when
5762: an error happens. E.g.,
5763:
5764: @example
5765: base @ >r hex \ actually the hex should be inside foo, or we h
5766: ['] foo catch ( nerror|0 )
5767: r> base !
5768: ( nerror|0 ) throw \ pass it on
5769: @end example
5770:
5771: A use of @code{catch} for handling the error @code{myerror} might look
5772: like this:
5773:
5774: @example
5775: ['] foo catch
5776: CASE
5777: myerror OF ... ( do something about it ) ENDOF
5778: dup throw \ default: pass other errors on, do nothing on non-errors
5779: ENDCASE
5780: @end example
5781:
5782: Having to wrap the code into a separate word is often cumbersome,
5783: therefore Gforth provides an alternative syntax:
5784:
5785: @example
5786: TRY
5787: @i{code1}
5788: RECOVER \ optional
5789: @i{code2} \ optional
5790: ENDTRY
5791: @end example
5792:
5793: This performs @i{Code1}. If @i{code1} completes normally, execution
5794: continues after the @code{endtry}. If @i{Code1} throws, the stacks are
5795: reset to the state during @code{try}, the throw value is pushed on the
5796: data stack, and execution constinues at @i{code2}, and finally falls
5797: through the @code{endtry} into the following code. If there is no
5798: @code{recover} clause, this works like an empty recover clause.
5799:
5800: doc-try
5801: doc-recover
5802: doc-endtry
5803:
5804: The cleanup example from above in this syntax:
5805:
5806: @example
5807: base @ >r TRY
5808: hex foo \ now the hex is placed correctly
5809: 0 \ value for throw
5810: ENDTRY
5811: r> base ! throw
5812: @end example
5813:
5814: And here's the error handling example:
5815:
5816: @example
5817: TRY
5818: foo
5819: RECOVER
5820: CASE
5821: myerror OF ... ( do something about it ) ENDOF
5822: throw \ pass other errors on
5823: ENDCASE
5824: ENDTRY
5825: @end example
5826:
5827: @progstyle
5828: As usual, you should ensure that the stack depth is statically known at
5829: the end: either after the @code{throw} for passing on errors, or after
5830: the @code{ENDTRY} (or, if you use @code{catch}, after the end of the
5831: selection construct for handling the error).
5832:
5833: There are two alternatives to @code{throw}: @code{Abort"} is conditional
5834: and you can provide an error message. @code{Abort} just produces an
5835: ``Aborted'' error.
5836:
5837: The problem with these words is that exception handlers cannot
5838: differentiate between different @code{abort"}s; they just look like
5839: @code{-2 throw} to them (the error message cannot be accessed by
5840: standard programs). Similar @code{abort} looks like @code{-1 throw} to
5841: exception handlers.
5842:
5843: doc-abort"
5844: doc-abort
5845:
5846:
5847:
5848: @c -------------------------------------------------------------
5849: @node Defining Words, Interpretation and Compilation Semantics, Control Structures, Words
5850: @section Defining Words
5851: @cindex defining words
5852:
5853: Defining words are used to extend Forth by creating new entries in the dictionary.
5854:
5855: @menu
5856: * CREATE::
5857: * Variables:: Variables and user variables
5858: * Constants::
5859: * Values:: Initialised variables
5860: * Colon Definitions::
5861: * Anonymous Definitions:: Definitions without names
5862: * Supplying names:: Passing definition names as strings
5863: * User-defined Defining Words::
5864: * Deferred words:: Allow forward references
5865: * Aliases::
5866: @end menu
5867:
5868: @node CREATE, Variables, Defining Words, Defining Words
5869: @subsection @code{CREATE}
5870: @cindex simple defining words
5871: @cindex defining words, simple
5872:
5873: Defining words are used to create new entries in the dictionary. The
5874: simplest defining word is @code{CREATE}. @code{CREATE} is used like
5875: this:
5876:
5877: @example
5878: CREATE new-word1
5879: @end example
5880:
5881: @code{CREATE} is a parsing word, i.e., it takes an argument from the
5882: input stream (@code{new-word1} in our example). It generates a
5883: dictionary entry for @code{new-word1}. When @code{new-word1} is
5884: executed, all that it does is leave an address on the stack. The address
5885: represents the value of the data space pointer (@code{HERE}) at the time
5886: that @code{new-word1} was defined. Therefore, @code{CREATE} is a way of
5887: associating a name with the address of a region of memory.
5888:
5889: doc-create
5890:
5891: Note that in ANS Forth guarantees only for @code{create} that its body
5892: is in dictionary data space (i.e., where @code{here}, @code{allot}
5893: etc. work, @pxref{Dictionary allocation}). Also, in ANS Forth only
5894: @code{create}d words can be modified with @code{does>}
5895: (@pxref{User-defined Defining Words}). And in ANS Forth @code{>body}
5896: can only be applied to @code{create}d words.
5897:
5898: By extending this example to reserve some memory in data space, we end
5899: up with something like a @i{variable}. Here are two different ways to do
5900: it:
5901:
5902: @example
5903: CREATE new-word2 1 cells allot \ reserve 1 cell - initial value undefined
5904: CREATE new-word3 4 , \ reserve 1 cell and initialise it (to 4)
5905: @end example
5906:
5907: The variable can be examined and modified using @code{@@} (``fetch'') and
5908: @code{!} (``store'') like this:
5909:
5910: @example
5911: new-word2 @@ . \ get address, fetch from it and display
5912: 1234 new-word2 ! \ new value, get address, store to it
5913: @end example
5914:
5915: @cindex arrays
5916: A similar mechanism can be used to create arrays. For example, an
5917: 80-character text input buffer:
5918:
5919: @example
5920: CREATE text-buf 80 chars allot
5921:
5922: text-buf 0 chars c@@ \ the 1st character (offset 0)
5923: text-buf 3 chars c@@ \ the 4th character (offset 3)
5924: @end example
5925:
5926: You can build arbitrarily complex data structures by allocating
5927: appropriate areas of memory. For further discussions of this, and to
5928: learn about some Gforth tools that make it easier,
5929: @xref{Structures}.
5930:
5931:
5932: @node Variables, Constants, CREATE, Defining Words
5933: @subsection Variables
5934: @cindex variables
5935:
5936: The previous section showed how a sequence of commands could be used to
5937: generate a variable. As a final refinement, the whole code sequence can
5938: be wrapped up in a defining word (pre-empting the subject of the next
5939: section), making it easier to create new variables:
5940:
5941: @example
5942: : myvariableX ( "name" -- a-addr ) CREATE 1 cells allot ;
5943: : myvariable0 ( "name" -- a-addr ) CREATE 0 , ;
5944:
5945: myvariableX foo \ variable foo starts off with an unknown value
5946: myvariable0 joe \ whilst joe is initialised to 0
5947:
5948: 45 3 * foo ! \ set foo to 135
5949: 1234 joe ! \ set joe to 1234
5950: 3 joe +! \ increment joe by 3.. to 1237
5951: @end example
5952:
5953: Not surprisingly, there is no need to define @code{myvariable}, since
5954: Forth already has a definition @code{Variable}. ANS Forth does not
5955: guarantee that a @code{Variable} is initialised when it is created
5956: (i.e., it may behave like @code{myvariableX}). In contrast, Gforth's
5957: @code{Variable} initialises the variable to 0 (i.e., it behaves exactly
5958: like @code{myvariable0}). Forth also provides @code{2Variable} and
5959: @code{fvariable} for double and floating-point variables, respectively
5960: -- they are initialised to 0. and 0e in Gforth. If you use a @code{Variable} to
5961: store a boolean, you can use @code{on} and @code{off} to toggle its
5962: state.
5963:
5964: doc-variable
5965: doc-2variable
5966: doc-fvariable
5967:
5968: @cindex user variables
5969: @cindex user space
5970: The defining word @code{User} behaves in the same way as @code{Variable}.
5971: The difference is that it reserves space in @i{user (data) space} rather
5972: than normal data space. In a Forth system that has a multi-tasker, each
5973: task has its own set of user variables.
5974:
5975: doc-user
5976: @c doc-udp
5977: @c doc-uallot
5978:
5979: @comment TODO is that stuff about user variables strictly correct? Is it
5980: @comment just terminal tasks that have user variables?
5981: @comment should document tasker.fs (with some examples) elsewhere
5982: @comment in this manual, then expand on user space and user variables.
5983:
5984: @node Constants, Values, Variables, Defining Words
5985: @subsection Constants
5986: @cindex constants
5987:
5988: @code{Constant} allows you to declare a fixed value and refer to it by
5989: name. For example:
5990:
5991: @example
5992: 12 Constant INCHES-PER-FOOT
5993: 3E+08 fconstant SPEED-O-LIGHT
5994: @end example
5995:
5996: A @code{Variable} can be both read and written, so its run-time
5997: behaviour is to supply an address through which its current value can be
5998: manipulated. In contrast, the value of a @code{Constant} cannot be
5999: changed once it has been declared@footnote{Well, often it can be -- but
6000: not in a Standard, portable way. It's safer to use a @code{Value} (read
6001: on).} so it's not necessary to supply the address -- it is more
6002: efficient to return the value of the constant directly. That's exactly
6003: what happens; the run-time effect of a constant is to put its value on
6004: the top of the stack (You can find one
6005: way of implementing @code{Constant} in @ref{User-defined Defining Words}).
6006:
6007: Forth also provides @code{2Constant} and @code{fconstant} for defining
6008: double and floating-point constants, respectively.
6009:
6010: doc-constant
6011: doc-2constant
6012: doc-fconstant
6013:
6014: @c that's too deep, and it's not necessarily true for all ANS Forths. - anton
6015: @c nac-> How could that not be true in an ANS Forth? You can't define a
6016: @c constant, use it and then delete the definition of the constant..
6017:
6018: @c anton->An ANS Forth system can compile a constant to a literal; On
6019: @c decompilation you would see only the number, just as if it had been used
6020: @c in the first place. The word will stay, of course, but it will only be
6021: @c used by the text interpreter (no run-time duties, except when it is
6022: @c POSTPONEd or somesuch).
6023:
6024: @c nac:
6025: @c I agree that it's rather deep, but IMO it is an important difference
6026: @c relative to other programming languages.. often it's annoying: it
6027: @c certainly changes my programming style relative to C.
6028:
6029: @c anton: In what way?
6030:
6031: Constants in Forth behave differently from their equivalents in other
6032: programming languages. In other languages, a constant (such as an EQU in
6033: assembler or a #define in C) only exists at compile-time; in the
6034: executable program the constant has been translated into an absolute
6035: number and, unless you are using a symbolic debugger, it's impossible to
6036: know what abstract thing that number represents. In Forth a constant has
6037: an entry in the header space and remains there after the code that uses
6038: it has been defined. In fact, it must remain in the dictionary since it
6039: has run-time duties to perform. For example:
6040:
6041: @example
6042: 12 Constant INCHES-PER-FOOT
6043: : FEET-TO-INCHES ( n1 -- n2 ) INCHES-PER-FOOT * ;
6044: @end example
6045:
6046: @cindex in-lining of constants
6047: When @code{FEET-TO-INCHES} is executed, it will in turn execute the xt
6048: associated with the constant @code{INCHES-PER-FOOT}. If you use
6049: @code{see} to decompile the definition of @code{FEET-TO-INCHES}, you can
6050: see that it makes a call to @code{INCHES-PER-FOOT}. Some Forth compilers
6051: attempt to optimise constants by in-lining them where they are used. You
6052: can force Gforth to in-line a constant like this:
6053:
6054: @example
6055: : FEET-TO-INCHES ( n1 -- n2 ) [ INCHES-PER-FOOT ] LITERAL * ;
6056: @end example
6057:
6058: If you use @code{see} to decompile @i{this} version of
6059: @code{FEET-TO-INCHES}, you can see that @code{INCHES-PER-FOOT} is no
6060: longer present. To understand how this works, read
6061: @ref{Interpret/Compile states}, and @ref{Literals}.
6062:
6063: In-lining constants in this way might improve execution time
6064: fractionally, and can ensure that a constant is now only referenced at
6065: compile-time. However, the definition of the constant still remains in
6066: the dictionary. Some Forth compilers provide a mechanism for controlling
6067: a second dictionary for holding transient words such that this second
6068: dictionary can be deleted later in order to recover memory
6069: space. However, there is no standard way of doing this.
6070:
6071:
6072: @node Values, Colon Definitions, Constants, Defining Words
6073: @subsection Values
6074: @cindex values
6075:
6076: A @code{Value} behaves like a @code{Constant}, but it can be changed.
6077: @code{TO} is a parsing word that changes a @code{Values}. In Gforth
6078: (not in ANS Forth) you can access (and change) a @code{value} also with
6079: @code{>body}.
6080:
6081: Here are some
6082: examples:
6083:
6084: @example
6085: 12 Value APPLES \ Define APPLES with an initial value of 12
6086: 34 TO APPLES \ Change the value of APPLES. TO is a parsing word
6087: 1 ' APPLES >body +! \ Increment APPLES. Non-standard usage.
6088: APPLES \ puts 35 on the top of the stack.
6089: @end example
6090:
6091: doc-value
6092: doc-to
6093:
6094:
6095:
6096: @node Colon Definitions, Anonymous Definitions, Values, Defining Words
6097: @subsection Colon Definitions
6098: @cindex colon definitions
6099:
6100: @example
6101: : name ( ... -- ... )
6102: word1 word2 word3 ;
6103: @end example
6104:
6105: @noindent
6106: Creates a word called @code{name} that, upon execution, executes
6107: @code{word1 word2 word3}. @code{name} is a @dfn{(colon) definition}.
6108:
6109: The explanation above is somewhat superficial. For simple examples of
6110: colon definitions see @ref{Your first definition}. For an in-depth
6111: discussion of some of the issues involved, @xref{Interpretation and
6112: Compilation Semantics}.
6113:
6114: doc-:
6115: doc-;
6116:
6117:
6118: @node Anonymous Definitions, Supplying names, Colon Definitions, Defining Words
6119: @subsection Anonymous Definitions
6120: @cindex colon definitions
6121: @cindex defining words without name
6122:
6123: Sometimes you want to define an @dfn{anonymous word}; a word without a
6124: name. You can do this with:
6125:
6126: doc-:noname
6127:
6128: This leaves the execution token for the word on the stack after the
6129: closing @code{;}. Here's an example in which a deferred word is
6130: initialised with an @code{xt} from an anonymous colon definition:
6131:
6132: @example
6133: Defer deferred
6134: :noname ( ... -- ... )
6135: ... ;
6136: IS deferred
6137: @end example
6138:
6139: @noindent
6140: Gforth provides an alternative way of doing this, using two separate
6141: words:
6142:
6143: doc-noname
6144: @cindex execution token of last defined word
6145: doc-lastxt
6146:
6147: @noindent
6148: The previous example can be rewritten using @code{noname} and
6149: @code{lastxt}:
6150:
6151: @example
6152: Defer deferred
6153: noname : ( ... -- ... )
6154: ... ;
6155: lastxt IS deferred
6156: @end example
6157:
6158: @noindent
6159: @code{noname} works with any defining word, not just @code{:}.
6160:
6161: @code{lastxt} also works when the last word was not defined as
6162: @code{noname}. It does not work for combined words, though. It also has
6163: the useful property that is is valid as soon as the header for a
6164: definition has been built. Thus:
6165:
6166: @example
6167: lastxt . : foo [ lastxt . ] ; ' foo .
6168: @end example
6169:
6170: @noindent
6171: prints 3 numbers; the last two are the same.
6172:
6173: @node Supplying names, User-defined Defining Words, Anonymous Definitions, Defining Words
6174: @subsection Supplying the name of a defined word
6175: @cindex names for defined words
6176: @cindex defining words, name given in a string
6177:
6178: By default, a defining word takes the name for the defined word from the
6179: input stream. Sometimes you want to supply the name from a string. You
6180: can do this with:
6181:
6182: doc-nextname
6183:
6184: For example:
6185:
6186: @example
6187: s" foo" nextname create
6188: @end example
6189:
6190: @noindent
6191: is equivalent to:
6192:
6193: @example
6194: create foo
6195: @end example
6196:
6197: @noindent
6198: @code{nextname} works with any defining word.
6199:
6200:
6201: @node User-defined Defining Words, Deferred words, Supplying names, Defining Words
6202: @subsection User-defined Defining Words
6203: @cindex user-defined defining words
6204: @cindex defining words, user-defined
6205:
6206: You can create a new defining word by wrapping defining-time code around
6207: an existing defining word and putting the sequence in a colon
6208: definition.
6209:
6210: @c anton: This example is very complex and leads in a quite different
6211: @c direction from the CREATE-DOES> stuff that follows. It should probably
6212: @c be done elsewhere, or as a subsubsection of this subsection (or as a
6213: @c subsection of Defining Words)
6214:
6215: For example, suppose that you have a word @code{stats} that
6216: gathers statistics about colon definitions given the @i{xt} of the
6217: definition, and you want every colon definition in your application to
6218: make a call to @code{stats}. You can define and use a new version of
6219: @code{:} like this:
6220:
6221: @example
6222: : stats ( xt -- ) DUP ." (Gathering statistics for " . ." )"
6223: ... ; \ other code
6224:
6225: : my: : lastxt postpone literal ['] stats compile, ;
6226:
6227: my: foo + - ;
6228: @end example
6229:
6230: When @code{foo} is defined using @code{my:} these steps occur:
6231:
6232: @itemize @bullet
6233: @item
6234: @code{my:} is executed.
6235: @item
6236: The @code{:} within the definition (the one between @code{my:} and
6237: @code{lastxt}) is executed, and does just what it always does; it parses
6238: the input stream for a name, builds a dictionary header for the name
6239: @code{foo} and switches @code{state} from interpret to compile.
6240: @item
6241: The word @code{lastxt} is executed. It puts the @i{xt} for the word that is
6242: being defined -- @code{foo} -- onto the stack.
6243: @item
6244: The code that was produced by @code{postpone literal} is executed; this
6245: causes the value on the stack to be compiled as a literal in the code
6246: area of @code{foo}.
6247: @item
6248: The code @code{['] stats} compiles a literal into the definition of
6249: @code{my:}. When @code{compile,} is executed, that literal -- the
6250: execution token for @code{stats} -- is layed down in the code area of
6251: @code{foo} , following the literal@footnote{Strictly speaking, the
6252: mechanism that @code{compile,} uses to convert an @i{xt} into something
6253: in the code area is implementation-dependent. A threaded implementation
6254: might spit out the execution token directly whilst another
6255: implementation might spit out a native code sequence.}.
6256: @item
6257: At this point, the execution of @code{my:} is complete, and control
6258: returns to the text interpreter. The text interpreter is in compile
6259: state, so subsequent text @code{+ -} is compiled into the definition of
6260: @code{foo} and the @code{;} terminates the definition as always.
6261: @end itemize
6262:
6263: You can use @code{see} to decompile a word that was defined using
6264: @code{my:} and see how it is different from a normal @code{:}
6265: definition. For example:
6266:
6267: @example
6268: : bar + - ; \ like foo but using : rather than my:
6269: see bar
6270: : bar
6271: + - ;
6272: see foo
6273: : foo
6274: 107645672 stats + - ;
6275:
6276: \ use ' stats . to show that 107645672 is the xt for stats
6277: @end example
6278:
6279: You can use techniques like this to make new defining words in terms of
6280: @i{any} existing defining word.
6281:
6282:
6283: @cindex defining defining words
6284: @cindex @code{CREATE} ... @code{DOES>}
6285: If you want the words defined with your defining words to behave
6286: differently from words defined with standard defining words, you can
6287: write your defining word like this:
6288:
6289: @example
6290: : def-word ( "name" -- )
6291: CREATE @i{code1}
6292: DOES> ( ... -- ... )
6293: @i{code2} ;
6294:
6295: def-word name
6296: @end example
6297:
6298: @cindex child words
6299: This fragment defines a @dfn{defining word} @code{def-word} and then
6300: executes it. When @code{def-word} executes, it @code{CREATE}s a new
6301: word, @code{name}, and executes the code @i{code1}. The code @i{code2}
6302: is not executed at this time. The word @code{name} is sometimes called a
6303: @dfn{child} of @code{def-word}.
6304:
6305: When you execute @code{name}, the address of the body of @code{name} is
6306: put on the data stack and @i{code2} is executed (the address of the body
6307: of @code{name} is the address @code{HERE} returns immediately after the
6308: @code{CREATE}, i.e., the address a @code{create}d word returns by
6309: default).
6310:
6311: @c anton:
6312: @c www.dictionary.com says:
6313: @c at·a·vism: 1.The reappearance of a characteristic in an organism after
6314: @c several generations of absence, usually caused by the chance
6315: @c recombination of genes. 2.An individual or a part that exhibits
6316: @c atavism. Also called throwback. 3.The return of a trait or recurrence
6317: @c of previous behavior after a period of absence.
6318: @c
6319: @c Doesn't seem to fit.
6320:
6321: @c @cindex atavism in child words
6322: You can use @code{def-word} to define a set of child words that behave
6323: similarly; they all have a common run-time behaviour determined by
6324: @i{code2}. Typically, the @i{code1} sequence builds a data area in the
6325: body of the child word. The structure of the data is common to all
6326: children of @code{def-word}, but the data values are specific -- and
6327: private -- to each child word. When a child word is executed, the
6328: address of its private data area is passed as a parameter on TOS to be
6329: used and manipulated@footnote{It is legitimate both to read and write to
6330: this data area.} by @i{code2}.
6331:
6332: The two fragments of code that make up the defining words act (are
6333: executed) at two completely separate times:
6334:
6335: @itemize @bullet
6336: @item
6337: At @i{define time}, the defining word executes @i{code1} to generate a
6338: child word
6339: @item
6340: At @i{child execution time}, when a child word is invoked, @i{code2}
6341: is executed, using parameters (data) that are private and specific to
6342: the child word.
6343: @end itemize
6344:
6345: Another way of understanding the behaviour of @code{def-word} and
6346: @code{name} is to say that, if you make the following definitions:
6347: @example
6348: : def-word1 ( "name" -- )
6349: CREATE @i{code1} ;
6350:
6351: : action1 ( ... -- ... )
6352: @i{code2} ;
6353:
6354: def-word1 name1
6355: @end example
6356:
6357: @noindent
6358: Then using @code{name1 action1} is equivalent to using @code{name}.
6359:
6360: The classic example is that you can define @code{CONSTANT} in this way:
6361:
6362: @example
6363: : CONSTANT ( w "name" -- )
6364: CREATE ,
6365: DOES> ( -- w )
6366: @@ ;
6367: @end example
6368:
6369: @comment There is a beautiful description of how this works and what
6370: @comment it does in the Forthwrite 100th edition.. as well as an elegant
6371: @comment commentary on the Counting Fruits problem.
6372:
6373: When you create a constant with @code{5 CONSTANT five}, a set of
6374: define-time actions take place; first a new word @code{five} is created,
6375: then the value 5 is laid down in the body of @code{five} with
6376: @code{,}. When @code{five} is executed, the address of the body is put on
6377: the stack, and @code{@@} retrieves the value 5. The word @code{five} has
6378: no code of its own; it simply contains a data field and a pointer to the
6379: code that follows @code{DOES>} in its defining word. That makes words
6380: created in this way very compact.
6381:
6382: The final example in this section is intended to remind you that space
6383: reserved in @code{CREATE}d words is @i{data} space and therefore can be
6384: both read and written by a Standard program@footnote{Exercise: use this
6385: example as a starting point for your own implementation of @code{Value}
6386: and @code{TO} -- if you get stuck, investigate the behaviour of @code{'} and
6387: @code{[']}.}:
6388:
6389: @example
6390: : foo ( "name" -- )
6391: CREATE -1 ,
6392: DOES> ( -- )
6393: @@ . ;
6394:
6395: foo first-word
6396: foo second-word
6397:
6398: 123 ' first-word >BODY !
6399: @end example
6400:
6401: If @code{first-word} had been a @code{CREATE}d word, we could simply
6402: have executed it to get the address of its data field. However, since it
6403: was defined to have @code{DOES>} actions, its execution semantics are to
6404: perform those @code{DOES>} actions. To get the address of its data field
6405: it's necessary to use @code{'} to get its xt, then @code{>BODY} to
6406: translate the xt into the address of the data field. When you execute
6407: @code{first-word}, it will display @code{123}. When you execute
6408: @code{second-word} it will display @code{-1}.
6409:
6410: @cindex stack effect of @code{DOES>}-parts
6411: @cindex @code{DOES>}-parts, stack effect
6412: In the examples above the stack comment after the @code{DOES>} specifies
6413: the stack effect of the defined words, not the stack effect of the
6414: following code (the following code expects the address of the body on
6415: the top of stack, which is not reflected in the stack comment). This is
6416: the convention that I use and recommend (it clashes a bit with using
6417: locals declarations for stack effect specification, though).
6418:
6419: @menu
6420: * CREATE..DOES> applications::
6421: * CREATE..DOES> details::
6422: * Advanced does> usage example::
6423: @end menu
6424:
6425: @node CREATE..DOES> applications, CREATE..DOES> details, User-defined Defining Words, User-defined Defining Words
6426: @subsubsection Applications of @code{CREATE..DOES>}
6427: @cindex @code{CREATE} ... @code{DOES>}, applications
6428:
6429: You may wonder how to use this feature. Here are some usage patterns:
6430:
6431: @cindex factoring similar colon definitions
6432: When you see a sequence of code occurring several times, and you can
6433: identify a meaning, you will factor it out as a colon definition. When
6434: you see similar colon definitions, you can factor them using
6435: @code{CREATE..DOES>}. E.g., an assembler usually defines several words
6436: that look very similar:
6437: @example
6438: : ori, ( reg-target reg-source n -- )
6439: 0 asm-reg-reg-imm ;
6440: : andi, ( reg-target reg-source n -- )
6441: 1 asm-reg-reg-imm ;
6442: @end example
6443:
6444: @noindent
6445: This could be factored with:
6446: @example
6447: : reg-reg-imm ( op-code -- )
6448: CREATE ,
6449: DOES> ( reg-target reg-source n -- )
6450: @@ asm-reg-reg-imm ;
6451:
6452: 0 reg-reg-imm ori,
6453: 1 reg-reg-imm andi,
6454: @end example
6455:
6456: @cindex currying
6457: Another view of @code{CREATE..DOES>} is to consider it as a crude way to
6458: supply a part of the parameters for a word (known as @dfn{currying} in
6459: the functional language community). E.g., @code{+} needs two
6460: parameters. Creating versions of @code{+} with one parameter fixed can
6461: be done like this:
6462: @example
6463: : curry+ ( n1 -- )
6464: CREATE ,
6465: DOES> ( n2 -- n1+n2 )
6466: @@ + ;
6467:
6468: 3 curry+ 3+
6469: -2 curry+ 2-
6470: @end example
6471:
6472: @node CREATE..DOES> details, Advanced does> usage example, CREATE..DOES> applications, User-defined Defining Words
6473: @subsubsection The gory details of @code{CREATE..DOES>}
6474: @cindex @code{CREATE} ... @code{DOES>}, details
6475:
6476: doc-does>
6477:
6478: @cindex @code{DOES>} in a separate definition
6479: This means that you need not use @code{CREATE} and @code{DOES>} in the
6480: same definition; you can put the @code{DOES>}-part in a separate
6481: definition. This allows us to, e.g., select among different @code{DOES>}-parts:
6482: @example
6483: : does1
6484: DOES> ( ... -- ... )
6485: ... ;
6486:
6487: : does2
6488: DOES> ( ... -- ... )
6489: ... ;
6490:
6491: : def-word ( ... -- ... )
6492: create ...
6493: IF
6494: does1
6495: ELSE
6496: does2
6497: ENDIF ;
6498: @end example
6499:
6500: In this example, the selection of whether to use @code{does1} or
6501: @code{does2} is made at definition-time; at the time that the child word is
6502: @code{CREATE}d.
6503:
6504: @cindex @code{DOES>} in interpretation state
6505: In a standard program you can apply a @code{DOES>}-part only if the last
6506: word was defined with @code{CREATE}. In Gforth, the @code{DOES>}-part
6507: will override the behaviour of the last word defined in any case. In a
6508: standard program, you can use @code{DOES>} only in a colon
6509: definition. In Gforth, you can also use it in interpretation state, in a
6510: kind of one-shot mode; for example:
6511: @example
6512: CREATE name ( ... -- ... )
6513: @i{initialization}
6514: DOES>
6515: @i{code} ;
6516: @end example
6517:
6518: @noindent
6519: is equivalent to the standard:
6520: @example
6521: :noname
6522: DOES>
6523: @i{code} ;
6524: CREATE name EXECUTE ( ... -- ... )
6525: @i{initialization}
6526: @end example
6527:
6528: doc->body
6529:
6530: @node Advanced does> usage example, , CREATE..DOES> details, User-defined Defining Words
6531: @subsubsection Advanced does> usage example
6532:
6533: The MIPS disassembler (@file{arch/mips/disasm.fs}) contains many words
6534: for disassembling instructions, that follow a very repetetive scheme:
6535:
6536: @example
6537: :noname @var{disasm-operands} s" @var{inst-name}" type ;
6538: @var{entry-num} cells @var{table} + !
6539: @end example
6540:
6541: Of course, this inspires the idea to factor out the commonalities to
6542: allow a definition like
6543:
6544: @example
6545: @var{disasm-operands} @var{entry-num} @var{table} define-inst @var{inst-name}
6546: @end example
6547:
6548: The parameters @var{disasm-operands} and @var{table} are usually
6549: correlated. Moreover, before I wrote the disassembler, there already
6550: existed code that defines instructions like this:
6551:
6552: @example
6553: @var{entry-num} @var{inst-format} @var{inst-name}
6554: @end example
6555:
6556: This code comes from the assembler and resides in
6557: @file{arch/mips/insts.fs}.
6558:
6559: So I had to define the @var{inst-format} words that performed the scheme
6560: above when executed. At first I chose to use run-time code-generation:
6561:
6562: @example
6563: : @var{inst-format} ( entry-num "name" -- ; compiled code: addr w -- )
6564: :noname Postpone @var{disasm-operands}
6565: name Postpone sliteral Postpone type Postpone ;
6566: swap cells @var{table} + ! ;
6567: @end example
6568:
6569: Note that this supplies the other two parameters of the scheme above.
6570:
6571: An alternative would have been to write this using
6572: @code{create}/@code{does>}:
6573:
6574: @example
6575: : @var{inst-format} ( entry-num "name" -- )
6576: here name string, ( entry-num c-addr ) \ parse and save "name"
6577: noname create , ( entry-num )
6578: lastxt swap cells @var{table} + !
6579: does> ( addr w -- )
6580: \ disassemble instruction w at addr
6581: @@ >r
6582: @var{disasm-operands}
6583: r> count type ;
6584: @end example
6585:
6586: Somehow the first solution is simpler, mainly because it's simpler to
6587: shift a string from definition-time to use-time with @code{sliteral}
6588: than with @code{string,} and friends.
6589:
6590: I wrote a lot of words following this scheme and soon thought about
6591: factoring out the commonalities among them. Note that this uses a
6592: two-level defining word, i.e., a word that defines ordinary defining
6593: words.
6594:
6595: This time a solution involving @code{postpone} and friends seemed more
6596: difficult (try it as an exercise), so I decided to use a
6597: @code{create}/@code{does>} word; since I was already at it, I also used
6598: @code{create}/@code{does>} for the lower level (try using
6599: @code{postpone} etc. as an exercise), resulting in the following
6600: definition:
6601:
6602: @example
6603: : define-format ( disasm-xt table-xt -- )
6604: \ define an instruction format that uses disasm-xt for
6605: \ disassembling and enters the defined instructions into table
6606: \ table-xt
6607: create 2,
6608: does> ( u "inst" -- )
6609: \ defines an anonymous word for disassembling instruction inst,
6610: \ and enters it as u-th entry into table-xt
6611: 2@@ swap here name string, ( u table-xt disasm-xt c-addr ) \ remember string
6612: noname create 2, \ define anonymous word
6613: execute lastxt swap ! \ enter xt of defined word into table-xt
6614: does> ( addr w -- )
6615: \ disassemble instruction w at addr
6616: 2@@ >r ( addr w disasm-xt R: c-addr )
6617: execute ( R: c-addr ) \ disassemble operands
6618: r> count type ; \ print name
6619: @end example
6620:
6621: Note that the tables here (in contrast to above) do the @code{cells +}
6622: by themselves (that's why you have to pass an xt). This word is used in
6623: the following way:
6624:
6625: @example
6626: ' @var{disasm-operands} ' @var{table} define-format @var{inst-format}
6627: @end example
6628:
6629: As shown above, the defined instruction format is then used like this:
6630:
6631: @example
6632: @var{entry-num} @var{inst-format} @var{inst-name}
6633: @end example
6634:
6635: In terms of currying, this kind of two-level defining word provides the
6636: parameters in three stages: first @var{disasm-operands} and @var{table},
6637: then @var{entry-num} and @var{inst-name}, finally @code{addr w}, i.e.,
6638: the instruction to be disassembled.
6639:
6640: Of course this did not quite fit all the instruction format names used
6641: in @file{insts.fs}, so I had to define a few wrappers that conditioned
6642: the parameters into the right form.
6643:
6644: If you have trouble following this section, don't worry. First, this is
6645: involved and takes time (and probably some playing around) to
6646: understand; second, this is the first two-level
6647: @code{create}/@code{does>} word I have written in seventeen years of
6648: Forth; and if I did not have @file{insts.fs} to start with, I may well
6649: have elected to use just a one-level defining word (with some repeating
6650: of parameters when using the defining word). So it is not necessary to
6651: understand this, but it may improve your understanding of Forth.
6652:
6653:
6654: @node Deferred words, Aliases, User-defined Defining Words, Defining Words
6655: @subsection Deferred words
6656: @cindex deferred words
6657:
6658: The defining word @code{Defer} allows you to define a word by name
6659: without defining its behaviour; the definition of its behaviour is
6660: deferred. Here are two situation where this can be useful:
6661:
6662: @itemize @bullet
6663: @item
6664: Where you want to allow the behaviour of a word to be altered later, and
6665: for all precompiled references to the word to change when its behaviour
6666: is changed.
6667: @item
6668: For mutual recursion; @xref{Calls and returns}.
6669: @end itemize
6670:
6671: In the following example, @code{foo} always invokes the version of
6672: @code{greet} that prints ``@code{Good morning}'' whilst @code{bar}
6673: always invokes the version that prints ``@code{Hello}''. There is no way
6674: of getting @code{foo} to use the later version without re-ordering the
6675: source code and recompiling it.
6676:
6677: @example
6678: : greet ." Good morning" ;
6679: : foo ... greet ... ;
6680: : greet ." Hello" ;
6681: : bar ... greet ... ;
6682: @end example
6683:
6684: This problem can be solved by defining @code{greet} as a @code{Defer}red
6685: word. The behaviour of a @code{Defer}red word can be defined and
6686: redefined at any time by using @code{IS} to associate the xt of a
6687: previously-defined word with it. The previous example becomes:
6688:
6689: @example
6690: Defer greet ( -- )
6691: : foo ... greet ... ;
6692: : bar ... greet ... ;
6693: : greet1 ( -- ) ." Good morning" ;
6694: : greet2 ( -- ) ." Hello" ;
6695: ' greet2 <IS> greet \ make greet behave like greet2
6696: @end example
6697:
6698: @progstyle
6699: You should write a stack comment for every deferred word, and put only
6700: XTs into deferred words that conform to this stack effect. Otherwise
6701: it's too difficult to use the deferred word.
6702:
6703: A deferred word can be used to improve the statistics-gathering example
6704: from @ref{User-defined Defining Words}; rather than edit the
6705: application's source code to change every @code{:} to a @code{my:}, do
6706: this:
6707:
6708: @example
6709: : real: : ; \ retain access to the original
6710: defer : \ redefine as a deferred word
6711: ' my: <IS> : \ use special version of :
6712: \
6713: \ load application here
6714: \
6715: ' real: <IS> : \ go back to the original
6716: @end example
6717:
6718:
6719: One thing to note is that @code{<IS>} consumes its name when it is
6720: executed. If you want to specify the name at compile time, use
6721: @code{[IS]}:
6722:
6723: @example
6724: : set-greet ( xt -- )
6725: [IS] greet ;
6726:
6727: ' greet1 set-greet
6728: @end example
6729:
6730: A deferred word can only inherit execution semantics from the xt
6731: (because that is all that an xt can represent -- for more discussion of
6732: this @pxref{Tokens for Words}); by default it will have default
6733: interpretation and compilation semantics deriving from this execution
6734: semantics. However, you can change the interpretation and compilation
6735: semantics of the deferred word in the usual ways:
6736:
6737: @example
6738: : bar .... ; compile-only
6739: Defer fred immediate
6740: Defer jim
6741:
6742: ' bar <IS> jim \ jim has default semantics
6743: ' bar <IS> fred \ fred is immediate
6744: @end example
6745:
6746: doc-defer
6747: doc-<is>
6748: doc-[is]
6749: doc-is
6750: @comment TODO document these: what's defers [is]
6751: doc-what's
6752: doc-defers
6753:
6754: @c Use @code{words-deferred} to see a list of deferred words.
6755:
6756: Definitions in ANS Forth for @code{defer}, @code{<is>} and @code{[is]}
6757: are provided in @file{compat/defer.fs}.
6758:
6759:
6760: @node Aliases, , Deferred words, Defining Words
6761: @subsection Aliases
6762: @cindex aliases
6763:
6764: The defining word @code{Alias} allows you to define a word by name that
6765: has the same behaviour as some other word. Here are two situation where
6766: this can be useful:
6767:
6768: @itemize @bullet
6769: @item
6770: When you want access to a word's definition from a different word list
6771: (for an example of this, see the definition of the @code{Root} word list
6772: in the Gforth source).
6773: @item
6774: When you want to create a synonym; a definition that can be known by
6775: either of two names (for example, @code{THEN} and @code{ENDIF} are
6776: aliases).
6777: @end itemize
6778:
6779: Like deferred words, an alias has default compilation and interpretation
6780: semantics at the beginning (not the modifications of the other word),
6781: but you can change them in the usual ways (@code{immediate},
6782: @code{compile-only}). For example:
6783:
6784: @example
6785: : foo ... ; immediate
6786:
6787: ' foo Alias bar \ bar is not an immediate word
6788: ' foo Alias fooby immediate \ fooby is an immediate word
6789: @end example
6790:
6791: Words that are aliases have the same xt, different headers in the
6792: dictionary, and consequently different name tokens (@pxref{Tokens for
6793: Words}) and possibly different immediate flags. An alias can only have
6794: default or immediate compilation semantics; you can define aliases for
6795: combined words with @code{interpret/compile:} -- see @ref{Combined words}.
6796:
6797: doc-alias
6798:
6799:
6800: @node Interpretation and Compilation Semantics, Tokens for Words, Defining Words, Words
6801: @section Interpretation and Compilation Semantics
6802: @cindex semantics, interpretation and compilation
6803:
6804: @c !! state and ' are used without explanation
6805: @c example for immediate/compile-only? or is the tutorial enough
6806:
6807: @cindex interpretation semantics
6808: The @dfn{interpretation semantics} of a (named) word are what the text
6809: interpreter does when it encounters the word in interpret state. It also
6810: appears in some other contexts, e.g., the execution token returned by
6811: @code{' @i{word}} identifies the interpretation semantics of @i{word}
6812: (in other words, @code{' @i{word} execute} is equivalent to
6813: interpret-state text interpretation of @code{@i{word}}).
6814:
6815: @cindex compilation semantics
6816: The @dfn{compilation semantics} of a (named) word are what the text
6817: interpreter does when it encounters the word in compile state. It also
6818: appears in other contexts, e.g, @code{POSTPONE @i{word}}
6819: compiles@footnote{In standard terminology, ``appends to the current
6820: definition''.} the compilation semantics of @i{word}.
6821:
6822: @cindex execution semantics
6823: The standard also talks about @dfn{execution semantics}. They are used
6824: only for defining the interpretation and compilation semantics of many
6825: words. By default, the interpretation semantics of a word are to
6826: @code{execute} its execution semantics, and the compilation semantics of
6827: a word are to @code{compile,} its execution semantics.@footnote{In
6828: standard terminology: The default interpretation semantics are its
6829: execution semantics; the default compilation semantics are to append its
6830: execution semantics to the execution semantics of the current
6831: definition.}
6832:
6833: Unnamed words (@pxref{Anonymous Definitions}) cannot be encountered by
6834: the text interpreter, ticked, or @code{postpone}d, so they have no
6835: interpretation or compilation semantics. Their behaviour is represented
6836: by their XT (@pxref{Tokens for Words}), and we call it execution
6837: semantics, too.
6838:
6839: @comment TODO expand, make it co-operate with new sections on text interpreter.
6840:
6841: @cindex immediate words
6842: @cindex compile-only words
6843: You can change the semantics of the most-recently defined word:
6844:
6845:
6846: doc-immediate
6847: doc-compile-only
6848: doc-restrict
6849:
6850:
6851: Note that ticking (@code{'}) a compile-only word gives an error
6852: (``Interpreting a compile-only word'').
6853:
6854: @menu
6855: * Combined words::
6856: @end menu
6857:
6858:
6859: @node Combined words, , Interpretation and Compilation Semantics, Interpretation and Compilation Semantics
6860: @subsection Combined Words
6861: @cindex combined words
6862:
6863: Gforth allows you to define @dfn{combined words} -- words that have an
6864: arbitrary combination of interpretation and compilation semantics.
6865:
6866: doc-interpret/compile:
6867:
6868: This feature was introduced for implementing @code{TO} and @code{S"}. I
6869: recommend that you do not define such words, as cute as they may be:
6870: they make it hard to get at both parts of the word in some contexts.
6871: E.g., assume you want to get an execution token for the compilation
6872: part. Instead, define two words, one that embodies the interpretation
6873: part, and one that embodies the compilation part. Once you have done
6874: that, you can define a combined word with @code{interpret/compile:} for
6875: the convenience of your users.
6876:
6877: You might try to use this feature to provide an optimizing
6878: implementation of the default compilation semantics of a word. For
6879: example, by defining:
6880: @example
6881: :noname
6882: foo bar ;
6883: :noname
6884: POSTPONE foo POSTPONE bar ;
6885: interpret/compile: opti-foobar
6886: @end example
6887:
6888: @noindent
6889: as an optimizing version of:
6890:
6891: @example
6892: : foobar
6893: foo bar ;
6894: @end example
6895:
6896: Unfortunately, this does not work correctly with @code{[compile]},
6897: because @code{[compile]} assumes that the compilation semantics of all
6898: @code{interpret/compile:} words are non-default. I.e., @code{[compile]
6899: opti-foobar} would compile compilation semantics, whereas
6900: @code{[compile] foobar} would compile interpretation semantics.
6901:
6902: @cindex state-smart words (are a bad idea)
6903: Some people try to use @dfn{state-smart} words to emulate the feature provided
6904: by @code{interpret/compile:} (words are state-smart if they check
6905: @code{STATE} during execution). E.g., they would try to code
6906: @code{foobar} like this:
6907:
6908: @example
6909: : foobar
6910: STATE @@
6911: IF ( compilation state )
6912: POSTPONE foo POSTPONE bar
6913: ELSE
6914: foo bar
6915: ENDIF ; immediate
6916: @end example
6917:
6918: Although this works if @code{foobar} is only processed by the text
6919: interpreter, it does not work in other contexts (like @code{'} or
6920: @code{POSTPONE}). E.g., @code{' foobar} will produce an execution token
6921: for a state-smart word, not for the interpretation semantics of the
6922: original @code{foobar}; when you execute this execution token (directly
6923: with @code{EXECUTE} or indirectly through @code{COMPILE,}) in compile
6924: state, the result will not be what you expected (i.e., it will not
6925: perform @code{foo bar}). State-smart words are a bad idea. Simply don't
6926: write them@footnote{For a more detailed discussion of this topic, see
6927: M. Anton Ertl,
6928: @cite{@uref{http://www.complang.tuwien.ac.at/papers/ertl98.ps.gz,@code{State}-smartness---Why
6929: it is Evil and How to Exorcise it}}, EuroForth '98.}!
6930:
6931: @cindex defining words with arbitrary semantics combinations
6932: It is also possible to write defining words that define words with
6933: arbitrary combinations of interpretation and compilation semantics. In
6934: general, they look like this:
6935:
6936: @example
6937: : def-word
6938: create-interpret/compile
6939: @i{code1}
6940: interpretation>
6941: @i{code2}
6942: <interpretation
6943: compilation>
6944: @i{code3}
6945: <compilation ;
6946: @end example
6947:
6948: For a @i{word} defined with @code{def-word}, the interpretation
6949: semantics are to push the address of the body of @i{word} and perform
6950: @i{code2}, and the compilation semantics are to push the address of
6951: the body of @i{word} and perform @i{code3}. E.g., @code{constant}
6952: can also be defined like this (except that the defined constants don't
6953: behave correctly when @code{[compile]}d):
6954:
6955: @example
6956: : constant ( n "name" -- )
6957: create-interpret/compile
6958: ,
6959: interpretation> ( -- n )
6960: @@
6961: <interpretation
6962: compilation> ( compilation. -- ; run-time. -- n )
6963: @@ postpone literal
6964: <compilation ;
6965: @end example
6966:
6967:
6968: doc-create-interpret/compile
6969: doc-interpretation>
6970: doc-<interpretation
6971: doc-compilation>
6972: doc-<compilation
6973:
6974:
6975: Words defined with @code{interpret/compile:} and
6976: @code{create-interpret/compile} have an extended header structure that
6977: differs from other words; however, unless you try to access them with
6978: plain address arithmetic, you should not notice this. Words for
6979: accessing the header structure usually know how to deal with this; e.g.,
6980: @code{'} @i{word} @code{>body} also gives you the body of a word created
6981: with @code{create-interpret/compile}.
6982:
6983:
6984: doc-postpone
6985:
6986: @comment TODO -- expand glossary text for POSTPONE
6987:
6988:
6989: @c -------------------------------------------------------------
6990: @node Tokens for Words, The Text Interpreter, Interpretation and Compilation Semantics, Words
6991: @section Tokens for Words
6992: @cindex tokens for words
6993:
6994: This section describes the creation and use of tokens that represent
6995: words.
6996:
6997: @menu
6998: * Execution token:: represents execution/interpretation semantics
6999: * Compilation token:: represents compilation semantics
7000: * Name token:: represents named words
7001: @end menu
7002:
7003: @node Execution token, Compilation token, Tokens for Words, Tokens for Words
7004: @subsection Execution token
7005:
7006: @cindex xt
7007: @cindex execution token
7008: An @dfn{execution token} (@i{XT}) represents some behaviour of a word.
7009: You can use @code{execute} to invoke this behaviour.
7010:
7011: @cindex tick (')
7012: You can use @code{'} to get an execution token that represents the
7013: interpretation semantics of a named word:
7014:
7015: @example
7016: 5 ' .
7017: execute
7018: @end example
7019:
7020: doc-'
7021:
7022: @code{'} parses at run-time; there is also a word @code{[']} that parses
7023: when it is compiled, and compiles the resulting XT:
7024:
7025: @example
7026: : foo ['] . execute ;
7027: 5 foo
7028: : bar ' execute ; \ by contrast,
7029: 5 bar . \ ' parses "." when bar executes
7030: @end example
7031:
7032: doc-[']
7033:
7034: If you want the execution token of @i{word}, write @code{['] @i{word}}
7035: in compiled code and @code{' @i{word}} in interpreted code. Gforth's
7036: @code{'} and @code{[']} behave somewhat unusually by complaining about
7037: compile-only words (because these words have no interpretation
7038: semantics). You might get what you want by using @code{COMP' @i{word}
7039: DROP} or @code{[COMP'] @i{word} DROP} (for details @pxref{Compilation
7040: token}).
7041:
7042: Another way to get an XT is @code{:noname} or @code{lastxt}
7043: (@pxref{Anonymous Definitions}). For anonymous words this gives an xt
7044: for the only behaviour the word has (the execution semantics). For
7045: named words, @code{lastxt} produces an XT for the same behaviour it
7046: would produce if the word was defined anonymously.
7047:
7048: @example
7049: :noname ." hello" ;
7050: execute
7051: @end example
7052:
7053: An XT occupies one cell and can be manipulated like any other cell.
7054:
7055: @cindex code field address
7056: @cindex CFA
7057: In ANS Forth the XT is just an abstract data type (i.e., defined by the
7058: operations that produce or consume it). For old hands: In Gforth, the
7059: XT is implemented as a code field address (CFA).
7060:
7061: @c !! discuss "compile," some more (or in Macros).
7062:
7063: doc-execute
7064: doc-perform
7065: doc-compile,
7066:
7067: @node Compilation token, Name token, Execution token, Tokens for Words
7068: @subsection Compilation token
7069:
7070: @cindex compilation token
7071: @cindex CT (compilation token)
7072: Gforth represents the compilation semantics of a named word by a
7073: @dfn{compilation token} consisting of two cells: @i{w xt}. The top cell
7074: @i{xt} is an execution token. The compilation semantics represented by
7075: the compilation token can be performed with @code{execute}, which
7076: consumes the whole compilation token, with an additional stack effect
7077: determined by the represented compilation semantics.
7078:
7079: At present, the @i{w} part of a compilation token is an execution token,
7080: and the @i{xt} part represents either @code{execute} or
7081: @code{compile,}@footnote{Depending upon the compilation semantics of the
7082: word. If the word has default compilation semantics, the @i{xt} will
7083: represent @code{compile,}. Otherwise (e.g., for immediate words), the
7084: @i{xt} will represent @code{execute}.}. However, don't rely on that
7085: knowledge, unless necessary; future versions of Gforth may introduce
7086: unusual compilation tokens (e.g., a compilation token that represents
7087: the compilation semantics of a literal).
7088:
7089: You can perform the compilation semantics represented by the compilation
7090: token with @code{execute}. You can compile the compilation semantics
7091: with @code{postpone,}. I.e., @code{COMP' @i{word} postpone,} is
7092: equivalent to @code{postpone @i{word}}.
7093:
7094: doc-[comp']
7095: doc-comp'
7096: doc-postpone,
7097:
7098: @node Name token, , Compilation token, Tokens for Words
7099: @subsection Name token
7100:
7101: @cindex name token
7102: @cindex name field address
7103: @cindex NFA
7104: Gforth represents named words by the @dfn{name token}, (@i{nt}). In
7105: Gforth, the abstract data type @emph{name token} is implemented as a
7106: name field address (NFA).
7107:
7108: doc-find-name
7109: doc-name>int
7110: doc-name?int
7111: doc-name>comp
7112: doc-name>string
7113:
7114:
7115: @c ----------------------------------------------------------
7116: @node The Text Interpreter, Word Lists, Tokens for Words, Words
7117: @section The Text Interpreter
7118: @cindex interpreter - outer
7119: @cindex text interpreter
7120: @cindex outer interpreter
7121:
7122: @c Should we really describe all these ugly details? IMO the text
7123: @c interpreter should be much cleaner, but that may not be possible within
7124: @c ANS Forth. - anton
7125: @c nac-> I wanted to explain how it works to show how you can exploit
7126: @c it in your own programs. When I was writing a cross-compiler, figuring out
7127: @c some of these gory details was very helpful to me. None of the textbooks
7128: @c I've seen cover it, and the most modern Forth textbook -- Forth Inc's,
7129: @c seems to positively avoid going into too much detail for some of
7130: @c the internals.
7131:
7132: @c anton: ok. I wonder, though, if this is the right place; for some stuff
7133: @c it is; for the ugly details, I would prefer another place. I wonder
7134: @c whether we should have a chapter before "Words" that describes some
7135: @c basic concepts referred to in words, and a chapter after "Words" that
7136: @c describes implementation details.
7137:
7138: The text interpreter@footnote{This is an expanded version of the
7139: material in @ref{Introducing the Text Interpreter}.} is an endless loop
7140: that processes input from the current input device. It is also called
7141: the outer interpreter, in contrast to the inner interpreter
7142: (@pxref{Engine}) which executes the compiled Forth code on interpretive
7143: implementations.
7144:
7145: @cindex interpret state
7146: @cindex compile state
7147: The text interpreter operates in one of two states: @dfn{interpret
7148: state} and @dfn{compile state}. The current state is defined by the
7149: aptly-named variable @code{state}.
7150:
7151: This section starts by describing how the text interpreter behaves when
7152: it is in interpret state, processing input from the user input device --
7153: the keyboard. This is the mode that a Forth system is in after it starts
7154: up.
7155:
7156: @cindex input buffer
7157: @cindex terminal input buffer
7158: The text interpreter works from an area of memory called the @dfn{input
7159: buffer}@footnote{When the text interpreter is processing input from the
7160: keyboard, this area of memory is called the @dfn{terminal input buffer}
7161: (TIB) and is addressed by the (obsolescent) words @code{TIB} and
7162: @code{#TIB}.}, which stores your keyboard input when you press the
7163: @key{RET} key. Starting at the beginning of the input buffer, it skips
7164: leading spaces (called @dfn{delimiters}) then parses a string (a
7165: sequence of non-space characters) until it reaches either a space
7166: character or the end of the buffer. Having parsed a string, it makes two
7167: attempts to process it:
7168:
7169: @cindex dictionary
7170: @itemize @bullet
7171: @item
7172: It looks for the string in a @dfn{dictionary} of definitions. If the
7173: string is found, the string names a @dfn{definition} (also known as a
7174: @dfn{word}) and the dictionary search returns information that allows
7175: the text interpreter to perform the word's @dfn{interpretation
7176: semantics}. In most cases, this simply means that the word will be
7177: executed.
7178: @item
7179: If the string is not found in the dictionary, the text interpreter
7180: attempts to treat it as a number, using the rules described in
7181: @ref{Number Conversion}. If the string represents a legal number in the
7182: current radix, the number is pushed onto a parameter stack (the data
7183: stack for integers, the floating-point stack for floating-point
7184: numbers).
7185: @end itemize
7186:
7187: If both attempts fail, or if the word is found in the dictionary but has
7188: no interpretation semantics@footnote{This happens if the word was
7189: defined as @code{COMPILE-ONLY}.} the text interpreter discards the
7190: remainder of the input buffer, issues an error message and waits for
7191: more input. If one of the attempts succeeds, the text interpreter
7192: repeats the parsing process until the whole of the input buffer has been
7193: processed, at which point it prints the status message ``@code{ ok}''
7194: and waits for more input.
7195:
7196: @c anton: this should be in the input stream subsection (or below it)
7197:
7198: @cindex parse area
7199: The text interpreter keeps track of its position in the input buffer by
7200: updating a variable called @code{>IN} (pronounced ``to-in''). The value
7201: of @code{>IN} starts out as 0, indicating an offset of 0 from the start
7202: of the input buffer. The region from offset @code{>IN @@} to the end of
7203: the input buffer is called the @dfn{parse area}@footnote{In other words,
7204: the text interpreter processes the contents of the input buffer by
7205: parsing strings from the parse area until the parse area is empty.}.
7206: This example shows how @code{>IN} changes as the text interpreter parses
7207: the input buffer:
7208:
7209: @example
7210: : remaining >IN @@ SOURCE 2 PICK - -ROT + SWAP
7211: CR ." ->" TYPE ." <-" ; IMMEDIATE
7212:
7213: 1 2 3 remaining + remaining .
7214:
7215: : foo 1 2 3 remaining SWAP remaining ;
7216: @end example
7217:
7218: @noindent
7219: The result is:
7220:
7221: @example
7222: ->+ remaining .<-
7223: ->.<-5 ok
7224:
7225: ->SWAP remaining ;-<
7226: ->;<- ok
7227: @end example
7228:
7229: @cindex parsing words
7230: The value of @code{>IN} can also be modified by a word in the input
7231: buffer that is executed by the text interpreter. This means that a word
7232: can ``trick'' the text interpreter into either skipping a section of the
7233: input buffer@footnote{This is how parsing words work.} or into parsing a
7234: section twice. For example:
7235:
7236: @example
7237: : lat ." <<foo>>" ;
7238: : flat ." <<bar>>" >IN DUP @@ 3 - SWAP ! ;
7239: @end example
7240:
7241: @noindent
7242: When @code{flat} is executed, this output is produced@footnote{Exercise
7243: for the reader: what would happen if the @code{3} were replaced with
7244: @code{4}?}:
7245:
7246: @example
7247: <<bar>><<foo>>
7248: @end example
7249:
7250: This technique can be used to work around some of the interoperability
7251: problems of parsing words. Of course, it's better to avoid parsing
7252: words where possible.
7253:
7254: @noindent
7255: Two important notes about the behaviour of the text interpreter:
7256:
7257: @itemize @bullet
7258: @item
7259: It processes each input string to completion before parsing additional
7260: characters from the input buffer.
7261: @item
7262: It treats the input buffer as a read-only region (and so must your code).
7263: @end itemize
7264:
7265: @noindent
7266: When the text interpreter is in compile state, its behaviour changes in
7267: these ways:
7268:
7269: @itemize @bullet
7270: @item
7271: If a parsed string is found in the dictionary, the text interpreter will
7272: perform the word's @dfn{compilation semantics}. In most cases, this
7273: simply means that the execution semantics of the word will be appended
7274: to the current definition.
7275: @item
7276: When a number is encountered, it is compiled into the current definition
7277: (as a literal) rather than being pushed onto a parameter stack.
7278: @item
7279: If an error occurs, @code{state} is modified to put the text interpreter
7280: back into interpret state.
7281: @item
7282: Each time a line is entered from the keyboard, Gforth prints
7283: ``@code{ compiled}'' rather than `` @code{ok}''.
7284: @end itemize
7285:
7286: @cindex text interpreter - input sources
7287: When the text interpreter is using an input device other than the
7288: keyboard, its behaviour changes in these ways:
7289:
7290: @itemize @bullet
7291: @item
7292: When the parse area is empty, the text interpreter attempts to refill
7293: the input buffer from the input source. When the input source is
7294: exhausted, the input source is set back to the previous input source.
7295: @item
7296: It doesn't print out ``@code{ ok}'' or ``@code{ compiled}'' messages each
7297: time the parse area is emptied.
7298: @item
7299: If an error occurs, the input source is set back to the user input
7300: device.
7301: @end itemize
7302:
7303: You can read about this in more detail in @ref{Input Sources}.
7304:
7305: doc->in
7306: doc-source
7307:
7308: doc-tib
7309: doc-#tib
7310:
7311:
7312: @menu
7313: * Input Sources::
7314: * Number Conversion::
7315: * Interpret/Compile states::
7316: * Literals::
7317: * Interpreter Directives::
7318: @end menu
7319:
7320: @node Input Sources, Number Conversion, The Text Interpreter, The Text Interpreter
7321: @subsection Input Sources
7322: @cindex input sources
7323: @cindex text interpreter - input sources
7324:
7325: By default, the text interpreter processes input from the user input
7326: device (the keyboard) when Forth starts up. The text interpreter can
7327: process input from any of these sources:
7328:
7329: @itemize @bullet
7330: @item
7331: The user input device -- the keyboard.
7332: @item
7333: A file, using the words described in @ref{Forth source files}.
7334: @item
7335: A block, using the words described in @ref{Blocks}.
7336: @item
7337: A text string, using @code{evaluate}.
7338: @end itemize
7339:
7340: A program can identify the current input device from the values of
7341: @code{source-id} and @code{blk}.
7342:
7343:
7344: doc-source-id
7345: doc-blk
7346:
7347: doc-save-input
7348: doc-restore-input
7349:
7350: doc-evaluate
7351:
7352:
7353:
7354: @node Number Conversion, Interpret/Compile states, Input Sources, The Text Interpreter
7355: @subsection Number Conversion
7356: @cindex number conversion
7357: @cindex double-cell numbers, input format
7358: @cindex input format for double-cell numbers
7359: @cindex single-cell numbers, input format
7360: @cindex input format for single-cell numbers
7361: @cindex floating-point numbers, input format
7362: @cindex input format for floating-point numbers
7363:
7364: This section describes the rules that the text interpreter uses when it
7365: tries to convert a string into a number.
7366:
7367: Let <digit> represent any character that is a legal digit in the current
7368: number base@footnote{For example, 0-9 when the number base is decimal or
7369: 0-9, A-F when the number base is hexadecimal.}.
7370:
7371: Let <decimal digit> represent any character in the range 0-9.
7372:
7373: Let @{@i{a b}@} represent the @i{optional} presence of any of the characters
7374: in the braces (@i{a} or @i{b} or neither).
7375:
7376: Let * represent any number of instances of the previous character
7377: (including none).
7378:
7379: Let any other character represent itself.
7380:
7381: @noindent
7382: Now, the conversion rules are:
7383:
7384: @itemize @bullet
7385: @item
7386: A string of the form <digit><digit>* is treated as a single-precision
7387: (cell-sized) positive integer. Examples are 0 123 6784532 32343212343456 42
7388: @item
7389: A string of the form -<digit><digit>* is treated as a single-precision
7390: (cell-sized) negative integer, and is represented using 2's-complement
7391: arithmetic. Examples are -45 -5681 -0
7392: @item
7393: A string of the form <digit><digit>*.<digit>* is treated as a double-precision
7394: (double-cell-sized) positive integer. Examples are 3465. 3.465 34.65
7395: (all three of these represent the same number).
7396: @item
7397: A string of the form -<digit><digit>*.<digit>* is treated as a
7398: double-precision (double-cell-sized) negative integer, and is
7399: represented using 2's-complement arithmetic. Examples are -3465. -3.465
7400: -34.65 (all three of these represent the same number).
7401: @item
7402: A string of the form @{+ -@}<decimal digit>@{.@}<decimal digit>*@{e
7403: E@}@{+ -@}<decimal digit><decimal digit>* is treated as a floating-point
7404: number. Examples are 1e 1e0 1.e 1.e0 +1e+0 (which all represent the same
7405: number) +12.E-4
7406: @end itemize
7407:
7408: By default, the number base used for integer number conversion is given
7409: by the contents of the variable @code{base}. Note that a lot of
7410: confusion can result from unexpected values of @code{base}. If you
7411: change @code{base} anywhere, make sure to save the old value and restore
7412: it afterwards. In general I recommend keeping @code{base} decimal, and
7413: using the prefixes described below for the popular non-decimal bases.
7414:
7415: doc-dpl
7416: doc-base
7417: doc-hex
7418: doc-decimal
7419:
7420:
7421: @cindex '-prefix for character strings
7422: @cindex &-prefix for decimal numbers
7423: @cindex %-prefix for binary numbers
7424: @cindex $-prefix for hexadecimal numbers
7425: Gforth allows you to override the value of @code{base} by using a
7426: prefix@footnote{Some Forth implementations provide a similar scheme by
7427: implementing @code{$} etc. as parsing words that process the subsequent
7428: number in the input stream and push it onto the stack. For example, see
7429: @cite{Number Conversion and Literals}, by Wil Baden; Forth Dimensions
7430: 20(3) pages 26--27. In such implementations, unlike in Gforth, a space
7431: is required between the prefix and the number.} before the first digit
7432: of an (integer) number. Four prefixes are supported:
7433:
7434: @itemize @bullet
7435: @item
7436: @code{&} -- decimal
7437: @item
7438: @code{%} -- binary
7439: @item
7440: @code{$} -- hexadecimal
7441: @item
7442: @code{'} -- base @code{max-char+1}
7443: @end itemize
7444:
7445: Here are some examples, with the equivalent decimal number shown after
7446: in braces:
7447:
7448: -$41 (-65), %1001101 (205), %1001.0001 (145 - a double-precision number),
7449: 'AB (16706; ascii A is 65, ascii B is 66, number is 65*256 + 66),
7450: 'ab (24930; ascii a is 97, ascii B is 98, number is 97*256 + 98),
7451: &905 (905), $abc (2478), $ABC (2478).
7452:
7453: @cindex number conversion - traps for the unwary
7454: @noindent
7455: Number conversion has a number of traps for the unwary:
7456:
7457: @itemize @bullet
7458: @item
7459: You cannot determine the current number base using the code sequence
7460: @code{base @@ .} -- the number base is always 10 in the current number
7461: base. Instead, use something like @code{base @@ dec.}
7462: @item
7463: If the number base is set to a value greater than 14 (for example,
7464: hexadecimal), the number 123E4 is ambiguous; the conversion rules allow
7465: it to be intepreted as either a single-precision integer or a
7466: floating-point number (Gforth treats it as an integer). The ambiguity
7467: can be resolved by explicitly stating the sign of the mantissa and/or
7468: exponent: 123E+4 or +123E4 -- if the number base is decimal, no
7469: ambiguity arises; either representation will be treated as a
7470: floating-point number.
7471: @item
7472: There is a word @code{bin} but it does @i{not} set the number base!
7473: It is used to specify file types.
7474: @item
7475: ANS Forth requires the @code{.} of a double-precision number to be the
7476: final character in the string. Gforth allows the @code{.} to be
7477: anywhere after the first digit.
7478: @item
7479: The number conversion process does not check for overflow.
7480: @item
7481: In an ANS Forth program @code{base} is required to be decimal when
7482: converting floating-point numbers. In Gforth, number conversion to
7483: floating-point numbers always uses base &10, irrespective of the value
7484: of @code{base}.
7485: @end itemize
7486:
7487: You can read numbers into your programs with the words described in
7488: @ref{Input}.
7489:
7490: @node Interpret/Compile states, Literals, Number Conversion, The Text Interpreter
7491: @subsection Interpret/Compile states
7492: @cindex Interpret/Compile states
7493:
7494: A standard program is not permitted to change @code{state}
7495: explicitly. However, it can change @code{state} implicitly, using the
7496: words @code{[} and @code{]}. When @code{[} is executed it switches
7497: @code{state} to interpret state, and therefore the text interpreter
7498: starts interpreting. When @code{]} is executed it switches @code{state}
7499: to compile state and therefore the text interpreter starts
7500: compiling. The most common usage for these words is for switching into
7501: interpret state and back from within a colon definition; this technique
7502: can be used to compile a literal (for an example, @pxref{Literals}) or
7503: for conditional compilation (for an example, @pxref{Interpreter
7504: Directives}).
7505:
7506:
7507: @c This is a bad example: It's non-standard, and it's not necessary.
7508: @c However, I can't think of a good example for switching into compile
7509: @c state when there is no current word (@code{state}-smart words are not a
7510: @c good reason). So maybe we should use an example for switching into
7511: @c interpret @code{state} in a colon def. - anton
7512: @c nac-> I agree. I started out by putting in the example, then realised
7513: @c that it was non-ANS, so wrote more words around it. I hope this
7514: @c re-written version is acceptable to you. I do want to keep the example
7515: @c as it is helpful for showing what is and what is not portable, particularly
7516: @c where it outlaws a style in common use.
7517:
7518: @c anton: it's more important to show what's portable. After we have done
7519: @c that, we can also show what's not. In any case, I intend to write a
7520: @c section Macros (or so) which will also deal with [ ].
7521:
7522: @code{[} and @code{]} also give you the ability to switch into compile
7523: state and back, but we cannot think of any useful Standard application
7524: for this ability. Pre-ANS Forth textbooks have examples like this:
7525:
7526: @example
7527: : AA ." this is A" ;
7528: : BB ." this is B" ;
7529: : CC ." this is C" ;
7530:
7531: create table ] aa bb cc [
7532:
7533: : go ( n -- ) \ n is offset into table.. 0 for 1st entry
7534: cells table + @ execute ;
7535: @end example
7536:
7537: This example builds a jump table; @code{0 go} will display ``@code{this
7538: is A}''. Using @code{[} and @code{]} in this example is equivalent to
7539: defining @code{table} like this:
7540:
7541: @example
7542: create table ' aa COMPILE, ' bb COMPILE, ' cc COMPILE,
7543: @end example
7544:
7545: The problem with this code is that the definition of @code{table} is not
7546: portable -- it @i{compile}s execution tokens into code space. Whilst it
7547: @i{may} work on systems where code space and data space co-incide, the
7548: Standard only allows data space to be assigned for a @code{CREATE}d
7549: word. In addition, the Standard only allows @code{@@} to access data
7550: space, whilst this example is using it to access code space. The only
7551: portable, Standard way to build this table is to build it in data space,
7552: like this:
7553:
7554: @example
7555: create table ' aa , ' bb , ' cc ,
7556: @end example
7557:
7558: doc-state
7559: doc-[
7560: doc-]
7561:
7562:
7563: @node Literals, Interpreter Directives, Interpret/Compile states, The Text Interpreter
7564: @subsection Literals
7565: @cindex Literals
7566:
7567: Often, you want to use a number within a colon definition. When you do
7568: this, the text interpreter automatically compiles the number as a
7569: @i{literal}. A literal is a number whose run-time effect is to be pushed
7570: onto the stack. If you had to do some maths to generate the number, you
7571: might write it like this:
7572:
7573: @example
7574: : HOUR-TO-SEC ( n1 -- n2 )
7575: 60 * \ to minutes
7576: 60 * ; \ to seconds
7577: @end example
7578:
7579: It is very clear what this definition is doing, but it's inefficient
7580: since it is performing 2 multiples at run-time. An alternative would be
7581: to write:
7582:
7583: @example
7584: : HOUR-TO-SEC ( n1 -- n2 )
7585: 3600 * ; \ to seconds
7586: @end example
7587:
7588: Which does the same thing, and has the advantage of using a single
7589: multiply. Ideally, we'd like the efficiency of the second with the
7590: readability of the first.
7591:
7592: @code{Literal} allows us to achieve that. It takes a number from the
7593: stack and lays it down in the current definition just as though the
7594: number had been typed directly into the definition. Our first attempt
7595: might look like this:
7596:
7597: @example
7598: 60 \ mins per hour
7599: 60 * \ seconds per minute
7600: : HOUR-TO-SEC ( n1 -- n2 )
7601: Literal * ; \ to seconds
7602: @end example
7603:
7604: But this produces the error message @code{unstructured}. What happened?
7605: The stack notation for @code{:} is (@i{ -- colon-sys}) and the size of
7606: @i{colon-sys} is implementation-defined. In other words, once we start a
7607: colon definition we can't portably access anything that was on the stack
7608: before the definition began@footnote{@cite{Two Problems in ANS Forth},
7609: by Thomas Worthington; Forth Dimensions 20(2) pages 32--34 describes
7610: some situations where you might want to access stack items above
7611: colon-sys, and provides a solution to the problem.}. The correct way of
7612: solving this problem in this instance is to use @code{[ ]} like this:
7613:
7614: @example
7615: : HOUR-TO-SEC ( n1 -- n2 )
7616: [ 60 \ minutes per hour
7617: 60 * ] \ seconds per minute
7618: LITERAL * ; \ to seconds
7619: @end example
7620:
7621:
7622: doc-literal
7623: doc-]L
7624: doc-2literal
7625: doc-fliteral
7626:
7627:
7628: @node Interpreter Directives, , Literals, The Text Interpreter
7629: @subsection Interpreter Directives
7630: @cindex interpreter directives
7631: @cindex conditional compilation
7632:
7633: These words are usually used in interpret state; typically to control
7634: which parts of a source file are processed by the text
7635: interpreter. There are only a few ANS Forth Standard words, but Gforth
7636: supplements these with a rich set of immediate control structure words
7637: to compensate for the fact that the non-immediate versions can only be
7638: used in compile state (@pxref{Control Structures}). Typical usages:
7639:
7640: @example
7641: FALSE Constant HAVE-ASSEMBLER
7642: .
7643: .
7644: HAVE-ASSEMBLER [IF]
7645: : ASSEMBLER-FEATURE
7646: ...
7647: ;
7648: [ENDIF]
7649: .
7650: .
7651: : SEE
7652: ... \ general-purpose SEE code
7653: [ HAVE-ASSEMBLER [IF] ]
7654: ... \ assembler-specific SEE code
7655: [ [ENDIF] ]
7656: ;
7657: @end example
7658:
7659:
7660: doc-[IF]
7661: doc-[ELSE]
7662: doc-[THEN]
7663: doc-[ENDIF]
7664:
7665: doc-[IFDEF]
7666: doc-[IFUNDEF]
7667:
7668: doc-[?DO]
7669: doc-[DO]
7670: doc-[FOR]
7671: doc-[LOOP]
7672: doc-[+LOOP]
7673: doc-[NEXT]
7674:
7675: doc-[BEGIN]
7676: doc-[UNTIL]
7677: doc-[AGAIN]
7678: doc-[WHILE]
7679: doc-[REPEAT]
7680:
7681:
7682: @c -------------------------------------------------------------
7683: @node Word Lists, Environmental Queries, The Text Interpreter, Words
7684: @section Word Lists
7685: @cindex word lists
7686: @cindex header space
7687:
7688: A wordlist is a list of named words; you can add new words and look up
7689: words by name (and you can remove words in a restricted way with
7690: markers). Every named (and @code{reveal}ed) word is in one wordlist.
7691:
7692: @cindex search order stack
7693: The text interpreter searches the wordlists present in the search order
7694: (a stack of wordlists), from the top to the bottom. Within each
7695: wordlist, the search starts conceptually at the newest word; i.e., if
7696: two words in a wordlist have the same name, the newer word is found.
7697:
7698: @cindex compilation word list
7699: New words are added to the @dfn{compilation wordlist} (aka current
7700: wordlist).
7701:
7702: @cindex wid
7703: A word list is identified by a cell-sized word list identifier (@i{wid})
7704: in much the same way as a file is identified by a file handle. The
7705: numerical value of the wid has no (portable) meaning, and might change
7706: from session to session.
7707:
7708: The ANS Forth ``Search order'' word set is intended to provide a set of
7709: low-level tools that allow various different schemes to be
7710: implemented. Gforth also provides @code{vocabulary}, a traditional Forth
7711: word. @file{compat/vocabulary.fs} provides an implementation in ANS
7712: Forth.
7713:
7714: @comment TODO: locals section refers to here, saying that every word list (aka
7715: @comment vocabulary) has its own methods for searching etc. Need to document that.
7716: @c anton: but better in a separate subsection on wordlist internals
7717:
7718: @comment TODO: document markers, reveal, tables, mappedwordlist
7719:
7720: @comment the gforthman- prefix is used to pick out the true definition of a
7721: @comment word from the source files, rather than some alias.
7722:
7723: doc-forth-wordlist
7724: doc-definitions
7725: doc-get-current
7726: doc-set-current
7727: doc-get-order
7728: doc---gforthman-set-order
7729: doc-wordlist
7730: doc-table
7731: doc-push-order
7732: doc-previous
7733: doc-also
7734: doc---gforthman-forth
7735: doc-only
7736: doc---gforthman-order
7737:
7738: doc-find
7739: doc-search-wordlist
7740:
7741: doc-words
7742: doc-vlist
7743: @c doc-words-deferred
7744:
7745: @c doc-mappedwordlist @c map-structure undefined, implemantation-specific
7746: doc-root
7747: doc-vocabulary
7748: doc-seal
7749: doc-vocs
7750: doc-current
7751: doc-context
7752:
7753:
7754: @menu
7755: * Vocabularies::
7756: * Why use word lists?::
7757: * Word list example::
7758: @end menu
7759:
7760: @node Vocabularies, Why use word lists?, Word Lists, Word Lists
7761: @subsection Vocabularies
7762: @cindex Vocabularies, detailed explanation
7763:
7764: Here is an example of creating and using a new wordlist using ANS
7765: Forth words:
7766:
7767: @example
7768: wordlist constant my-new-words-wordlist
7769: : my-new-words get-order nip my-new-words-wordlist swap set-order ;
7770:
7771: \ add it to the search order
7772: also my-new-words
7773:
7774: \ alternatively, add it to the search order and make it
7775: \ the compilation word list
7776: also my-new-words definitions
7777: \ type "order" to see the problem
7778: @end example
7779:
7780: The problem with this example is that @code{order} has no way to
7781: associate the name @code{my-new-words} with the wid of the word list (in
7782: Gforth, @code{order} and @code{vocs} will display @code{???} for a wid
7783: that has no associated name). There is no Standard way of associating a
7784: name with a wid.
7785:
7786: In Gforth, this example can be re-coded using @code{vocabulary}, which
7787: associates a name with a wid:
7788:
7789: @example
7790: vocabulary my-new-words
7791:
7792: \ add it to the search order
7793: also my-new-words
7794:
7795: \ alternatively, add it to the search order and make it
7796: \ the compilation word list
7797: my-new-words definitions
7798: \ type "order" to see that the problem is solved
7799: @end example
7800:
7801:
7802: @node Why use word lists?, Word list example, Vocabularies, Word Lists
7803: @subsection Why use word lists?
7804: @cindex word lists - why use them?
7805:
7806: Here are some reasons why people use wordlists:
7807:
7808: @itemize @bullet
7809:
7810: @c anton: Gforth's hashing implementation makes the search speed
7811: @c independent from the number of words. But it is linear with the number
7812: @c of wordlists that have to be searched, so in effect using more wordlists
7813: @c actually slows down compilation.
7814:
7815: @c @item
7816: @c To improve compilation speed by reducing the number of header space
7817: @c entries that must be searched. This is achieved by creating a new
7818: @c word list that contains all of the definitions that are used in the
7819: @c definition of a Forth system but which would not usually be used by
7820: @c programs running on that system. That word list would be on the search
7821: @c list when the Forth system was compiled but would be removed from the
7822: @c search list for normal operation. This can be a useful technique for
7823: @c low-performance systems (for example, 8-bit processors in embedded
7824: @c systems) but is unlikely to be necessary in high-performance desktop
7825: @c systems.
7826:
7827: @item
7828: To prevent a set of words from being used outside the context in which
7829: they are valid. Two classic examples of this are an integrated editor
7830: (all of the edit commands are defined in a separate word list; the
7831: search order is set to the editor word list when the editor is invoked;
7832: the old search order is restored when the editor is terminated) and an
7833: integrated assembler (the op-codes for the machine are defined in a
7834: separate word list which is used when a @code{CODE} word is defined).
7835:
7836: @item
7837: To organize the words of an application or library into a user-visible
7838: set (in @code{forth-wordlist} or some other common wordlist) and a set
7839: of helper words used just for the implementation (hidden in a separate
7840: wordlist). This keeps @code{words}' output smaller, separates
7841: implementation and interface, and reduces the chance of name conflicts
7842: within the common wordlist.
7843:
7844: @item
7845: To prevent a name-space clash between multiple definitions with the same
7846: name. For example, when building a cross-compiler you might have a word
7847: @code{IF} that generates conditional code for your target system. By
7848: placing this definition in a different word list you can control whether
7849: the host system's @code{IF} or the target system's @code{IF} get used in
7850: any particular context by controlling the order of the word lists on the
7851: search order stack.
7852:
7853: @end itemize
7854:
7855: The downsides of using wordlists are:
7856:
7857: @itemize
7858:
7859: @item
7860: Debugging becomes more cumbersome.
7861:
7862: @item
7863: Name conflicts worked around with wordlists are still there, and you
7864: have to arrange the search order carefully to get the desired results;
7865: if you forget to do that, you get hard-to-find errors (as in any case
7866: where you read the code differently from the compiler; @code{see} can
7867: help seeing which of several possible words the name resolves to in such
7868: cases). @code{See} displays just the name of the words, not what
7869: wordlist they belong to, so it might be misleading. Using unique names
7870: is a better approach to avoid name conflicts.
7871:
7872: @item
7873: You have to explicitly undo any changes to the search order. In many
7874: cases it would be more convenient if this happened implicitly. Gforth
7875: currently does not provide such a feature, but it may do so in the
7876: future.
7877: @end itemize
7878:
7879:
7880: @node Word list example, , Why use word lists?, Word Lists
7881: @subsection Word list example
7882: @cindex word lists - example
7883:
7884: The following example is from the
7885: @uref{http://www.complang.tuwien.ac.at/forth/garbage-collection.zip,
7886: garbage collector} and uses wordlists to separate public words from
7887: helper words:
7888:
7889: @example
7890: get-current ( wid )
7891: vocabulary garbage-collector also garbage-collector definitions
7892: ... \ define helper words
7893: ( wid ) set-current \ restore original (i.e., public) compilation wordlist
7894: ... \ define the public (i.e., API) words
7895: \ they can refer to the helper words
7896: previous \ restore original search order (helper words become invisible)
7897: @end example
7898:
7899: @c -------------------------------------------------------------
7900: @node Environmental Queries, Files, Word Lists, Words
7901: @section Environmental Queries
7902: @cindex environmental queries
7903:
7904: ANS Forth introduced the idea of ``environmental queries'' as a way
7905: for a program running on a system to determine certain characteristics of the system.
7906: The Standard specifies a number of strings that might be recognised by a system.
7907:
7908: The Standard requires that the header space used for environmental queries
7909: be distinct from the header space used for definitions.
7910:
7911: Typically, environmental queries are supported by creating a set of
7912: definitions in a word list that is @i{only} used during environmental
7913: queries; that is what Gforth does. There is no Standard way of adding
7914: definitions to the set of recognised environmental queries, but any
7915: implementation that supports the loading of optional word sets must have
7916: some mechanism for doing this (after loading the word set, the
7917: associated environmental query string must return @code{true}). In
7918: Gforth, the word list used to honour environmental queries can be
7919: manipulated just like any other word list.
7920:
7921:
7922: doc-environment?
7923: doc-environment-wordlist
7924:
7925: doc-gforth
7926: doc-os-class
7927:
7928:
7929: Note that, whilst the documentation for (e.g.) @code{gforth} shows it
7930: returning two items on the stack, querying it using @code{environment?}
7931: will return an additional item; the @code{true} flag that shows that the
7932: string was recognised.
7933:
7934: @comment TODO Document the standard strings or note where they are documented herein
7935:
7936: Here are some examples of using environmental queries:
7937:
7938: @example
7939: s" address-unit-bits" environment? 0=
7940: [IF]
7941: cr .( environmental attribute address-units-bits unknown... ) cr
7942: [ELSE]
7943: drop \ ensure balanced stack effect
7944: [THEN]
7945:
7946: \ this might occur in the prelude of a standard program that uses THROW
7947: s" exception" environment? [IF]
7948: 0= [IF]
7949: : throw abort" exception thrown" ;
7950: [THEN]
7951: [ELSE] \ we don't know, so make sure
7952: : throw abort" exception thrown" ;
7953: [THEN]
7954:
7955: s" gforth" environment? [IF] .( Gforth version ) TYPE
7956: [ELSE] .( Not Gforth..) [THEN]
7957:
7958: \ a program using v*
7959: s" gforth" environment? [IF]
7960: s" 0.5.0" compare 0< [IF] \ v* is a primitive since 0.5.0
7961: : v* ( f_addr1 nstride1 f_addr2 nstride2 ucount -- r )
7962: >r swap 2swap swap 0e r> 0 ?DO
7963: dup f@ over + 2swap dup f@ f* f+ over + 2swap
7964: LOOP
7965: 2drop 2drop ;
7966: [THEN]
7967: [ELSE] \
7968: : v* ( f_addr1 nstride1 f_addr2 nstride2 ucount -- r )
7969: ...
7970: [THEN]
7971: @end example
7972:
7973: Here is an example of adding a definition to the environment word list:
7974:
7975: @example
7976: get-current environment-wordlist set-current
7977: true constant block
7978: true constant block-ext
7979: set-current
7980: @end example
7981:
7982: You can see what definitions are in the environment word list like this:
7983:
7984: @example
7985: environment-wordlist push-order words previous
7986: @end example
7987:
7988:
7989: @c -------------------------------------------------------------
7990: @node Files, Blocks, Environmental Queries, Words
7991: @section Files
7992: @cindex files
7993: @cindex I/O - file-handling
7994:
7995: Gforth provides facilities for accessing files that are stored in the
7996: host operating system's file-system. Files that are processed by Gforth
7997: can be divided into two categories:
7998:
7999: @itemize @bullet
8000: @item
8001: Files that are processed by the Text Interpreter (@dfn{Forth source files}).
8002: @item
8003: Files that are processed by some other program (@dfn{general files}).
8004: @end itemize
8005:
8006: @menu
8007: * Forth source files::
8008: * General files::
8009: * Search Paths::
8010: @end menu
8011:
8012: @c -------------------------------------------------------------
8013: @node Forth source files, General files, Files, Files
8014: @subsection Forth source files
8015: @cindex including files
8016: @cindex Forth source files
8017:
8018: The simplest way to interpret the contents of a file is to use one of
8019: these two formats:
8020:
8021: @example
8022: include mysource.fs
8023: s" mysource.fs" included
8024: @end example
8025:
8026: You usually want to include a file only if it is not included already
8027: (by, say, another source file). In that case, you can use one of these
8028: three formats:
8029:
8030: @example
8031: require mysource.fs
8032: needs mysource.fs
8033: s" mysource.fs" required
8034: @end example
8035:
8036: @cindex stack effect of included files
8037: @cindex including files, stack effect
8038: It is good practice to write your source files such that interpreting them
8039: does not change the stack. Source files designed in this way can be used with
8040: @code{required} and friends without complications. For example:
8041:
8042: @example
8043: 1024 require foo.fs drop
8044: @end example
8045:
8046: Here you want to pass the argument 1024 (e.g., a buffer size) to
8047: @file{foo.fs}. Interpreting @file{foo.fs} has the stack effect ( n -- n
8048: ), which allows its use with @code{require}. Of course with such
8049: parameters to required files, you have to ensure that the first
8050: @code{require} fits for all uses (i.e., @code{require} it early in the
8051: master load file).
8052:
8053: doc-include-file
8054: doc-included
8055: doc-included?
8056: doc-include
8057: doc-required
8058: doc-require
8059: doc-needs
8060: @c doc-init-included-files @c internal
8061: @c doc-loadfilename @c internal word
8062: doc-sourcefilename
8063: doc-sourceline#
8064:
8065: A definition in ANS Forth for @code{required} is provided in
8066: @file{compat/required.fs}.
8067:
8068: @c -------------------------------------------------------------
8069: @node General files, Search Paths, Forth source files, Files
8070: @subsection General files
8071: @cindex general files
8072: @cindex file-handling
8073:
8074: Files are opened/created by name and type. The following file access
8075: methods (FAMs) are recognised:
8076:
8077: @cindex fam (file access method)
8078: doc-r/o
8079: doc-r/w
8080: doc-w/o
8081: doc-bin
8082:
8083:
8084: When a file is opened/created, it returns a file identifier,
8085: @i{wfileid} that is used for all other file commands. All file
8086: commands also return a status value, @i{wior}, that is 0 for a
8087: successful operation and an implementation-defined non-zero value in the
8088: case of an error.
8089:
8090:
8091: doc-open-file
8092: doc-create-file
8093:
8094: doc-close-file
8095: doc-delete-file
8096: doc-rename-file
8097: doc-read-file
8098: doc-read-line
8099: doc-write-file
8100: doc-write-line
8101: doc-emit-file
8102: doc-flush-file
8103:
8104: doc-file-status
8105: doc-file-position
8106: doc-reposition-file
8107: doc-file-size
8108: doc-resize-file
8109:
8110:
8111: @c ---------------------------------------------------------
8112: @node Search Paths, , General files, Files
8113: @subsection Search Paths
8114: @cindex path for @code{included}
8115: @cindex file search path
8116: @cindex @code{include} search path
8117: @cindex search path for files
8118:
8119: If you specify an absolute filename (i.e., a filename starting with
8120: @file{/} or @file{~}, or with @file{:} in the second position (as in
8121: @samp{C:...})) for @code{included} and friends, that file is included
8122: just as you would expect.
8123:
8124: If the filename starts with @file{./}, this refers to the directory that
8125: the present file was @code{included} from. This allows files to include
8126: other files relative to their own position (irrespective of the current
8127: working directory or the absolute position). This feature is essential
8128: for libraries consisting of several files, where a file may include
8129: other files from the library. It corresponds to @code{#include "..."}
8130: in C. If the current input source is not a file, @file{.} refers to the
8131: directory of the innermost file being included, or, if there is no file
8132: being included, to the current working directory.
8133:
8134: For relative filenames (not starting with @file{./}), Gforth uses a
8135: search path similar to Forth's search order (@pxref{Word Lists}). It
8136: tries to find the given filename in the directories present in the path,
8137: and includes the first one it finds. There are separate search paths for
8138: Forth source files and general files. If the search path contains the
8139: directory @file{.}, this refers to the directory of the current file, or
8140: the working directory, as if the file had been specified with @file{./}.
8141:
8142: Use @file{~+} to refer to the current working directory (as in the
8143: @code{bash}).
8144:
8145: @c anton: fold the following subsubsections into this subsection?
8146:
8147: @menu
8148: * Source Search Paths::
8149: * General Search Paths::
8150: @end menu
8151:
8152: @c ---------------------------------------------------------
8153: @node Source Search Paths, General Search Paths, Search Paths, Search Paths
8154: @subsubsection Source Search Paths
8155: @cindex search path control, source files
8156:
8157: The search path is initialized when you start Gforth (@pxref{Invoking
8158: Gforth}). You can display it and change it using @code{fpath} in
8159: combination with the general path handling words.
8160:
8161: doc-fpath
8162: @c the functionality of the following words is easily available through
8163: @c fpath and the general path words. The may go away.
8164: @c doc-.fpath
8165: @c doc-fpath+
8166: @c doc-fpath=
8167: @c doc-open-fpath-file
8168:
8169: @noindent
8170: Here is an example of using @code{fpath} and @code{require}:
8171:
8172: @example
8173: fpath path= /usr/lib/forth/|./
8174: require timer.fs
8175: @end example
8176:
8177:
8178: @c ---------------------------------------------------------
8179: @node General Search Paths, , Source Search Paths, Search Paths
8180: @subsubsection General Search Paths
8181: @cindex search path control, source files
8182:
8183: Your application may need to search files in several directories, like
8184: @code{included} does. To facilitate this, Gforth allows you to define
8185: and use your own search paths, by providing generic equivalents of the
8186: Forth search path words:
8187:
8188: doc-open-path-file
8189: doc-path-allot
8190: doc-clear-path
8191: doc-also-path
8192: doc-.path
8193: doc-path+
8194: doc-path=
8195:
8196: @c anton: better define a word for it, say "path-allot ( ucount -- path-addr )
8197:
8198: Here's an example of creating an empty search path:
8199: @c
8200: @example
8201: create mypath 500 path-allot \ maximum length 500 chars (is checked)
8202: @end example
8203:
8204: @c -------------------------------------------------------------
8205: @node Blocks, Other I/O, Files, Words
8206: @section Blocks
8207: @cindex I/O - blocks
8208: @cindex blocks
8209:
8210: When you run Gforth on a modern desk-top computer, it runs under the
8211: control of an operating system which provides certain services. One of
8212: these services is @var{file services}, which allows Forth source code
8213: and data to be stored in files and read into Gforth (@pxref{Files}).
8214:
8215: Traditionally, Forth has been an important programming language on
8216: systems where it has interfaced directly to the underlying hardware with
8217: no intervening operating system. Forth provides a mechanism, called
8218: @dfn{blocks}, for accessing mass storage on such systems.
8219:
8220: A block is a 1024-byte data area, which can be used to hold data or
8221: Forth source code. No structure is imposed on the contents of the
8222: block. A block is identified by its number; blocks are numbered
8223: contiguously from 1 to an implementation-defined maximum.
8224:
8225: A typical system that used blocks but no operating system might use a
8226: single floppy-disk drive for mass storage, with the disks formatted to
8227: provide 256-byte sectors. Blocks would be implemented by assigning the
8228: first four sectors of the disk to block 1, the second four sectors to
8229: block 2 and so on, up to the limit of the capacity of the disk. The disk
8230: would not contain any file system information, just the set of blocks.
8231:
8232: @cindex blocks file
8233: On systems that do provide file services, blocks are typically
8234: implemented by storing a sequence of blocks within a single @dfn{blocks
8235: file}. The size of the blocks file will be an exact multiple of 1024
8236: bytes, corresponding to the number of blocks it contains. This is the
8237: mechanism that Gforth uses.
8238:
8239: @cindex @file{blocks.fb}
8240: Only one blocks file can be open at a time. If you use block words without
8241: having specified a blocks file, Gforth defaults to the blocks file
8242: @file{blocks.fb}. Gforth uses the Forth search path when attempting to
8243: locate a blocks file (@pxref{Source Search Paths}).
8244:
8245: @cindex block buffers
8246: When you read and write blocks under program control, Gforth uses a
8247: number of @dfn{block buffers} as intermediate storage. These buffers are
8248: not used when you use @code{load} to interpret the contents of a block.
8249:
8250: The behaviour of the block buffers is analagous to that of a cache.
8251: Each block buffer has three states:
8252:
8253: @itemize @bullet
8254: @item
8255: Unassigned
8256: @item
8257: Assigned-clean
8258: @item
8259: Assigned-dirty
8260: @end itemize
8261:
8262: Initially, all block buffers are @i{unassigned}. In order to access a
8263: block, the block (specified by its block number) must be assigned to a
8264: block buffer.
8265:
8266: The assignment of a block to a block buffer is performed by @code{block}
8267: or @code{buffer}. Use @code{block} when you wish to modify the existing
8268: contents of a block. Use @code{buffer} when you don't care about the
8269: existing contents of the block@footnote{The ANS Forth definition of
8270: @code{buffer} is intended not to cause disk I/O; if the data associated
8271: with the particular block is already stored in a block buffer due to an
8272: earlier @code{block} command, @code{buffer} will return that block
8273: buffer and the existing contents of the block will be
8274: available. Otherwise, @code{buffer} will simply assign a new, empty
8275: block buffer for the block.}.
8276:
8277: Once a block has been assigned to a block buffer using @code{block} or
8278: @code{buffer}, that block buffer becomes the @i{current block
8279: buffer}. Data may only be manipulated (read or written) within the
8280: current block buffer.
8281:
8282: When the contents of the current block buffer has been modified it is
8283: necessary, @emph{before calling @code{block} or @code{buffer} again}, to
8284: either abandon the changes (by doing nothing) or mark the block as
8285: changed (assigned-dirty), using @code{update}. Using @code{update} does
8286: not change the blocks file; it simply changes a block buffer's state to
8287: @i{assigned-dirty}. The block will be written implicitly when it's
8288: buffer is needed for another block, or explicitly by @code{flush} or
8289: @code{save-buffers}.
8290:
8291: word @code{Flush} writes all @i{assigned-dirty} blocks back to the
8292: blocks file on disk. Leaving Gforth with @code{bye} also performs a
8293: @code{flush}.
8294:
8295: In Gforth, @code{block} and @code{buffer} use a @i{direct-mapped}
8296: algorithm to assign a block buffer to a block. That means that any
8297: particular block can only be assigned to one specific block buffer,
8298: called (for the particular operation) the @i{victim buffer}. If the
8299: victim buffer is @i{unassigned} or @i{assigned-clean} it is allocated to
8300: the new block immediately. If it is @i{assigned-dirty} its current
8301: contents are written back to the blocks file on disk before it is
8302: allocated to the new block.
8303:
8304: Although no structure is imposed on the contents of a block, it is
8305: traditional to display the contents as 16 lines each of 64 characters. A
8306: block provides a single, continuous stream of input (for example, it
8307: acts as a single parse area) -- there are no end-of-line characters
8308: within a block, and no end-of-file character at the end of a
8309: block. There are two consequences of this:
8310:
8311: @itemize @bullet
8312: @item
8313: The last character of one line wraps straight into the first character
8314: of the following line
8315: @item
8316: The word @code{\} -- comment to end of line -- requires special
8317: treatment; in the context of a block it causes all characters until the
8318: end of the current 64-character ``line'' to be ignored.
8319: @end itemize
8320:
8321: In Gforth, when you use @code{block} with a non-existent block number,
8322: the current blocks file will be extended to the appropriate size and the
8323: block buffer will be initialised with spaces.
8324:
8325: Gforth includes a simple block editor (type @code{use blocked.fb 0 list}
8326: for details) but doesn't encourage the use of blocks; the mechanism is
8327: only provided for backward compatibility -- ANS Forth requires blocks to
8328: be available when files are.
8329:
8330: Common techniques that are used when working with blocks include:
8331:
8332: @itemize @bullet
8333: @item
8334: A screen editor that allows you to edit blocks without leaving the Forth
8335: environment.
8336: @item
8337: Shadow screens; where every code block has an associated block
8338: containing comments (for example: code in odd block numbers, comments in
8339: even block numbers). Typically, the block editor provides a convenient
8340: mechanism to toggle between code and comments.
8341: @item
8342: Load blocks; a single block (typically block 1) contains a number of
8343: @code{thru} commands which @code{load} the whole of the application.
8344: @end itemize
8345:
8346: See Frank Sergeant's Pygmy Forth to see just how well blocks can be
8347: integrated into a Forth programming environment.
8348:
8349: @comment TODO what about errors on open-blocks?
8350:
8351: doc-open-blocks
8352: doc-use
8353: doc-block-offset
8354: doc-get-block-fid
8355: doc-block-position
8356:
8357: doc-list
8358: doc-scr
8359:
8360: doc---gforthman-block
8361: doc-buffer
8362:
8363: doc-empty-buffers
8364: doc-empty-buffer
8365: doc-update
8366: doc-updated?
8367: doc-save-buffers
8368: doc-save-buffer
8369: doc-flush
8370:
8371: doc-load
8372: doc-thru
8373: doc-+load
8374: doc-+thru
8375: doc---gforthman--->
8376: doc-block-included
8377:
8378:
8379: @c -------------------------------------------------------------
8380: @node Other I/O, Locals, Blocks, Words
8381: @section Other I/O
8382: @cindex I/O - keyboard and display
8383:
8384: @menu
8385: * Simple numeric output:: Predefined formats
8386: * Formatted numeric output:: Formatted (pictured) output
8387: * String Formats:: How Forth stores strings in memory
8388: * Displaying characters and strings:: Other stuff
8389: * Input:: Input
8390: @end menu
8391:
8392: @node Simple numeric output, Formatted numeric output, Other I/O, Other I/O
8393: @subsection Simple numeric output
8394: @cindex numeric output - simple/free-format
8395:
8396: The simplest output functions are those that display numbers from the
8397: data or floating-point stacks. Floating-point output is always displayed
8398: using base 10. Numbers displayed from the data stack use the value stored
8399: in @code{base}.
8400:
8401:
8402: doc-.
8403: doc-dec.
8404: doc-hex.
8405: doc-u.
8406: doc-.r
8407: doc-u.r
8408: doc-d.
8409: doc-ud.
8410: doc-d.r
8411: doc-ud.r
8412: doc-f.
8413: doc-fe.
8414: doc-fs.
8415:
8416:
8417: Examples of printing the number 1234.5678E23 in the different floating-point output
8418: formats are shown below:
8419:
8420: @example
8421: f. 123456779999999000000000000.
8422: fe. 123.456779999999E24
8423: fs. 1.23456779999999E26
8424: @end example
8425:
8426:
8427: @node Formatted numeric output, String Formats, Simple numeric output, Other I/O
8428: @subsection Formatted numeric output
8429: @cindex formatted numeric output
8430: @cindex pictured numeric output
8431: @cindex numeric output - formatted
8432:
8433: Forth traditionally uses a technique called @dfn{pictured numeric
8434: output} for formatted printing of integers. In this technique, digits
8435: are extracted from the number (using the current output radix defined by
8436: @code{base}), converted to ASCII codes and appended to a string that is
8437: built in a scratch-pad area of memory (@pxref{core-idef,
8438: Implementation-defined options, Implementation-defined
8439: options}). Arbitrary characters can be appended to the string during the
8440: extraction process. The completed string is specified by an address
8441: and length and can be manipulated (@code{TYPE}ed, copied, modified)
8442: under program control.
8443:
8444: All of the integer output words described in the previous section
8445: (@pxref{Simple numeric output}) are implemented in Gforth using pictured
8446: numeric output.
8447:
8448: Three important things to remember about pictured numeric output:
8449:
8450: @itemize @bullet
8451: @item
8452: It always operates on double-precision numbers; to display a
8453: single-precision number, convert it first (for ways of doing this
8454: @pxref{Double precision}).
8455: @item
8456: It always treats the double-precision number as though it were
8457: unsigned. The examples below show ways of printing signed numbers.
8458: @item
8459: The string is built up from right to left; least significant digit first.
8460: @end itemize
8461:
8462:
8463: doc-<#
8464: doc-<<#
8465: doc-#
8466: doc-#s
8467: doc-hold
8468: doc-sign
8469: doc-#>
8470: doc-#>>
8471:
8472: doc-represent
8473:
8474:
8475: @noindent
8476: Here are some examples of using pictured numeric output:
8477:
8478: @example
8479: : my-u. ( u -- )
8480: \ Simplest use of pns.. behaves like Standard u.
8481: 0 \ convert to unsigned double
8482: <<# \ start conversion
8483: #s \ convert all digits
8484: #> \ complete conversion
8485: TYPE SPACE \ display, with trailing space
8486: #>> ; \ release hold area
8487:
8488: : cents-only ( u -- )
8489: 0 \ convert to unsigned double
8490: <<# \ start conversion
8491: # # \ convert two least-significant digits
8492: #> \ complete conversion, discard other digits
8493: TYPE SPACE \ display, with trailing space
8494: #>> ; \ release hold area
8495:
8496: : dollars-and-cents ( u -- )
8497: 0 \ convert to unsigned double
8498: <<# \ start conversion
8499: # # \ convert two least-significant digits
8500: [char] . hold \ insert decimal point
8501: #s \ convert remaining digits
8502: [char] $ hold \ append currency symbol
8503: #> \ complete conversion
8504: TYPE SPACE \ display, with trailing space
8505: #>> ; \ release hold area
8506:
8507: : my-. ( n -- )
8508: \ handling negatives.. behaves like Standard .
8509: s>d \ convert to signed double
8510: swap over dabs \ leave sign byte followed by unsigned double
8511: <<# \ start conversion
8512: #s \ convert all digits
8513: rot sign \ get at sign byte, append "-" if needed
8514: #> \ complete conversion
8515: TYPE SPACE \ display, with trailing space
8516: #>> ; \ release hold area
8517:
8518: : account. ( n -- )
8519: \ accountants don't like minus signs, they use parentheses
8520: \ for negative numbers
8521: s>d \ convert to signed double
8522: swap over dabs \ leave sign byte followed by unsigned double
8523: <<# \ start conversion
8524: 2 pick \ get copy of sign byte
8525: 0< IF [char] ) hold THEN \ right-most character of output
8526: #s \ convert all digits
8527: rot \ get at sign byte
8528: 0< IF [char] ( hold THEN
8529: #> \ complete conversion
8530: TYPE SPACE \ display, with trailing space
8531: #>> ; \ release hold area
8532:
8533: @end example
8534:
8535: Here are some examples of using these words:
8536:
8537: @example
8538: 1 my-u. 1
8539: hex -1 my-u. decimal FFFFFFFF
8540: 1 cents-only 01
8541: 1234 cents-only 34
8542: 2 dollars-and-cents $0.02
8543: 1234 dollars-and-cents $12.34
8544: 123 my-. 123
8545: -123 my. -123
8546: 123 account. 123
8547: -456 account. (456)
8548: @end example
8549:
8550:
8551: @node String Formats, Displaying characters and strings, Formatted numeric output, Other I/O
8552: @subsection String Formats
8553: @cindex strings - see character strings
8554: @cindex character strings - formats
8555: @cindex I/O - see character strings
8556: @cindex counted strings
8557:
8558: @c anton: this does not really belong here; maybe the memory section,
8559: @c or the principles chapter
8560:
8561: Forth commonly uses two different methods for representing character
8562: strings:
8563:
8564: @itemize @bullet
8565: @item
8566: @cindex address of counted string
8567: @cindex counted string
8568: As a @dfn{counted string}, represented by a @i{c-addr}. The char
8569: addressed by @i{c-addr} contains a character-count, @i{n}, of the
8570: string and the string occupies the subsequent @i{n} char addresses in
8571: memory.
8572: @item
8573: As cell pair on the stack; @i{c-addr u}, where @i{u} is the length
8574: of the string in characters, and @i{c-addr} is the address of the
8575: first byte of the string.
8576: @end itemize
8577:
8578: ANS Forth encourages the use of the second format when representing
8579: strings.
8580:
8581:
8582: doc-count
8583:
8584:
8585: For words that move, copy and search for strings see @ref{Memory
8586: Blocks}. For words that display characters and strings see
8587: @ref{Displaying characters and strings}.
8588:
8589: @node Displaying characters and strings, Input, String Formats, Other I/O
8590: @subsection Displaying characters and strings
8591: @cindex characters - compiling and displaying
8592: @cindex character strings - compiling and displaying
8593:
8594: This section starts with a glossary of Forth words and ends with a set
8595: of examples.
8596:
8597:
8598: doc-bl
8599: doc-space
8600: doc-spaces
8601: doc-emit
8602: doc-toupper
8603: doc-."
8604: doc-.(
8605: doc-type
8606: doc-typewhite
8607: doc-cr
8608: @cindex cursor control
8609: doc-at-xy
8610: doc-page
8611: doc-s"
8612: doc-c"
8613: doc-char
8614: doc-[char]
8615: doc-sliteral
8616:
8617:
8618: @noindent
8619: As an example, consider the following text, stored in a file @file{test.fs}:
8620:
8621: @example
8622: .( text-1)
8623: : my-word
8624: ." text-2" cr
8625: .( text-3)
8626: ;
8627:
8628: ." text-4"
8629:
8630: : my-char
8631: [char] ALPHABET emit
8632: char emit
8633: ;
8634: @end example
8635:
8636: When you load this code into Gforth, the following output is generated:
8637:
8638: @example
8639: @kbd{include test.fs @key{RET}} text-1text-3text-4 ok
8640: @end example
8641:
8642: @itemize @bullet
8643: @item
8644: Messages @code{text-1} and @code{text-3} are displayed because @code{.(}
8645: is an immediate word; it behaves in the same way whether it is used inside
8646: or outside a colon definition.
8647: @item
8648: Message @code{text-4} is displayed because of Gforth's added interpretation
8649: semantics for @code{."}.
8650: @item
8651: Message @code{text-2} is @i{not} displayed, because the text interpreter
8652: performs the compilation semantics for @code{."} within the definition of
8653: @code{my-word}.
8654: @end itemize
8655:
8656: Here are some examples of executing @code{my-word} and @code{my-char}:
8657:
8658: @example
8659: @kbd{my-word @key{RET}} text-2
8660: ok
8661: @kbd{my-char fred @key{RET}} Af ok
8662: @kbd{my-char jim @key{RET}} Aj ok
8663: @end example
8664:
8665: @itemize @bullet
8666: @item
8667: Message @code{text-2} is displayed because of the run-time behaviour of
8668: @code{."}.
8669: @item
8670: @code{[char]} compiles the ``A'' from ``ALPHABET'' and puts its display code
8671: on the stack at run-time. @code{emit} always displays the character
8672: when @code{my-char} is executed.
8673: @item
8674: @code{char} parses a string at run-time and the second @code{emit} displays
8675: the first character of the string.
8676: @item
8677: If you type @code{see my-char} you can see that @code{[char]} discarded
8678: the text ``LPHABET'' and only compiled the display code for ``A'' into the
8679: definition of @code{my-char}.
8680: @end itemize
8681:
8682:
8683:
8684: @node Input, , Displaying characters and strings, Other I/O
8685: @subsection Input
8686: @cindex input
8687: @cindex I/O - see input
8688: @cindex parsing a string
8689:
8690: For ways of storing character strings in memory see @ref{String Formats}.
8691:
8692: @comment TODO examples for >number >float accept key key? pad parse word refill
8693: @comment then index them
8694:
8695:
8696: doc-key
8697: doc-key?
8698: doc-ekey
8699: doc-ekey?
8700: doc-ekey>char
8701: doc->number
8702: doc->float
8703: doc-accept
8704: doc-pad
8705: @c anton: these belong in the input stream section
8706: doc-parse
8707: doc-word
8708: doc-sword
8709: doc-name
8710: doc-refill
8711: @comment obsolescent words..
8712: doc-convert
8713: doc-query
8714: doc-expect
8715: doc-span
8716:
8717:
8718: @c -------------------------------------------------------------
8719: @node Locals, Structures, Other I/O, Words
8720: @section Locals
8721: @cindex locals
8722:
8723: Local variables can make Forth programming more enjoyable and Forth
8724: programs easier to read. Unfortunately, the locals of ANS Forth are
8725: laden with restrictions. Therefore, we provide not only the ANS Forth
8726: locals wordset, but also our own, more powerful locals wordset (we
8727: implemented the ANS Forth locals wordset through our locals wordset).
8728:
8729: The ideas in this section have also been published in M. Anton Ertl,
8730: @cite{@uref{http://www.complang.tuwien.ac.at/papers/ertl94l.ps.gz,
8731: Automatic Scoping of Local Variables}}, EuroForth '94.
8732:
8733: @menu
8734: * Gforth locals::
8735: * ANS Forth locals::
8736: @end menu
8737:
8738: @node Gforth locals, ANS Forth locals, Locals, Locals
8739: @subsection Gforth locals
8740: @cindex Gforth locals
8741: @cindex locals, Gforth style
8742:
8743: Locals can be defined with
8744:
8745: @example
8746: @{ local1 local2 ... -- comment @}
8747: @end example
8748: or
8749: @example
8750: @{ local1 local2 ... @}
8751: @end example
8752:
8753: E.g.,
8754: @example
8755: : max @{ n1 n2 -- n3 @}
8756: n1 n2 > if
8757: n1
8758: else
8759: n2
8760: endif ;
8761: @end example
8762:
8763: The similarity of locals definitions with stack comments is intended. A
8764: locals definition often replaces the stack comment of a word. The order
8765: of the locals corresponds to the order in a stack comment and everything
8766: after the @code{--} is really a comment.
8767:
8768: This similarity has one disadvantage: It is too easy to confuse locals
8769: declarations with stack comments, causing bugs and making them hard to
8770: find. However, this problem can be avoided by appropriate coding
8771: conventions: Do not use both notations in the same program. If you do,
8772: they should be distinguished using additional means, e.g. by position.
8773:
8774: @cindex types of locals
8775: @cindex locals types
8776: The name of the local may be preceded by a type specifier, e.g.,
8777: @code{F:} for a floating point value:
8778:
8779: @example
8780: : CX* @{ F: Ar F: Ai F: Br F: Bi -- Cr Ci @}
8781: \ complex multiplication
8782: Ar Br f* Ai Bi f* f-
8783: Ar Bi f* Ai Br f* f+ ;
8784: @end example
8785:
8786: @cindex flavours of locals
8787: @cindex locals flavours
8788: @cindex value-flavoured locals
8789: @cindex variable-flavoured locals
8790: Gforth currently supports cells (@code{W:}, @code{W^}), doubles
8791: (@code{D:}, @code{D^}), floats (@code{F:}, @code{F^}) and characters
8792: (@code{C:}, @code{C^}) in two flavours: a value-flavoured local (defined
8793: with @code{W:}, @code{D:} etc.) produces its value and can be changed
8794: with @code{TO}. A variable-flavoured local (defined with @code{W^} etc.)
8795: produces its address (which becomes invalid when the variable's scope is
8796: left). E.g., the standard word @code{emit} can be defined in terms of
8797: @code{type} like this:
8798:
8799: @example
8800: : emit @{ C^ char* -- @}
8801: char* 1 type ;
8802: @end example
8803:
8804: @cindex default type of locals
8805: @cindex locals, default type
8806: A local without type specifier is a @code{W:} local. Both flavours of
8807: locals are initialized with values from the data or FP stack.
8808:
8809: Currently there is no way to define locals with user-defined data
8810: structures, but we are working on it.
8811:
8812: Gforth allows defining locals everywhere in a colon definition. This
8813: poses the following questions:
8814:
8815: @menu
8816: * Where are locals visible by name?::
8817: * How long do locals live?::
8818: * Locals programming style::
8819: * Locals implementation::
8820: @end menu
8821:
8822: @node Where are locals visible by name?, How long do locals live?, Gforth locals, Gforth locals
8823: @subsubsection Where are locals visible by name?
8824: @cindex locals visibility
8825: @cindex visibility of locals
8826: @cindex scope of locals
8827:
8828: Basically, the answer is that locals are visible where you would expect
8829: it in block-structured languages, and sometimes a little longer. If you
8830: want to restrict the scope of a local, enclose its definition in
8831: @code{SCOPE}...@code{ENDSCOPE}.
8832:
8833:
8834: doc-scope
8835: doc-endscope
8836:
8837:
8838: These words behave like control structure words, so you can use them
8839: with @code{CS-PICK} and @code{CS-ROLL} to restrict the scope in
8840: arbitrary ways.
8841:
8842: If you want a more exact answer to the visibility question, here's the
8843: basic principle: A local is visible in all places that can only be
8844: reached through the definition of the local@footnote{In compiler
8845: construction terminology, all places dominated by the definition of the
8846: local.}. In other words, it is not visible in places that can be reached
8847: without going through the definition of the local. E.g., locals defined
8848: in @code{IF}...@code{ENDIF} are visible until the @code{ENDIF}, locals
8849: defined in @code{BEGIN}...@code{UNTIL} are visible after the
8850: @code{UNTIL} (until, e.g., a subsequent @code{ENDSCOPE}).
8851:
8852: The reasoning behind this solution is: We want to have the locals
8853: visible as long as it is meaningful. The user can always make the
8854: visibility shorter by using explicit scoping. In a place that can
8855: only be reached through the definition of a local, the meaning of a
8856: local name is clear. In other places it is not: How is the local
8857: initialized at the control flow path that does not contain the
8858: definition? Which local is meant, if the same name is defined twice in
8859: two independent control flow paths?
8860:
8861: This should be enough detail for nearly all users, so you can skip the
8862: rest of this section. If you really must know all the gory details and
8863: options, read on.
8864:
8865: In order to implement this rule, the compiler has to know which places
8866: are unreachable. It knows this automatically after @code{AHEAD},
8867: @code{AGAIN}, @code{EXIT} and @code{LEAVE}; in other cases (e.g., after
8868: most @code{THROW}s), you can use the word @code{UNREACHABLE} to tell the
8869: compiler that the control flow never reaches that place. If
8870: @code{UNREACHABLE} is not used where it could, the only consequence is
8871: that the visibility of some locals is more limited than the rule above
8872: says. If @code{UNREACHABLE} is used where it should not (i.e., if you
8873: lie to the compiler), buggy code will be produced.
8874:
8875:
8876: doc-unreachable
8877:
8878:
8879: Another problem with this rule is that at @code{BEGIN}, the compiler
8880: does not know which locals will be visible on the incoming
8881: back-edge. All problems discussed in the following are due to this
8882: ignorance of the compiler (we discuss the problems using @code{BEGIN}
8883: loops as examples; the discussion also applies to @code{?DO} and other
8884: loops). Perhaps the most insidious example is:
8885: @example
8886: AHEAD
8887: BEGIN
8888: x
8889: [ 1 CS-ROLL ] THEN
8890: @{ x @}
8891: ...
8892: UNTIL
8893: @end example
8894:
8895: This should be legal according to the visibility rule. The use of
8896: @code{x} can only be reached through the definition; but that appears
8897: textually below the use.
8898:
8899: From this example it is clear that the visibility rules cannot be fully
8900: implemented without major headaches. Our implementation treats common
8901: cases as advertised and the exceptions are treated in a safe way: The
8902: compiler makes a reasonable guess about the locals visible after a
8903: @code{BEGIN}; if it is too pessimistic, the
8904: user will get a spurious error about the local not being defined; if the
8905: compiler is too optimistic, it will notice this later and issue a
8906: warning. In the case above the compiler would complain about @code{x}
8907: being undefined at its use. You can see from the obscure examples in
8908: this section that it takes quite unusual control structures to get the
8909: compiler into trouble, and even then it will often do fine.
8910:
8911: If the @code{BEGIN} is reachable from above, the most optimistic guess
8912: is that all locals visible before the @code{BEGIN} will also be
8913: visible after the @code{BEGIN}. This guess is valid for all loops that
8914: are entered only through the @code{BEGIN}, in particular, for normal
8915: @code{BEGIN}...@code{WHILE}...@code{REPEAT} and
8916: @code{BEGIN}...@code{UNTIL} loops and it is implemented in our
8917: compiler. When the branch to the @code{BEGIN} is finally generated by
8918: @code{AGAIN} or @code{UNTIL}, the compiler checks the guess and
8919: warns the user if it was too optimistic:
8920: @example
8921: IF
8922: @{ x @}
8923: BEGIN
8924: \ x ?
8925: [ 1 cs-roll ] THEN
8926: ...
8927: UNTIL
8928: @end example
8929:
8930: Here, @code{x} lives only until the @code{BEGIN}, but the compiler
8931: optimistically assumes that it lives until the @code{THEN}. It notices
8932: this difference when it compiles the @code{UNTIL} and issues a
8933: warning. The user can avoid the warning, and make sure that @code{x}
8934: is not used in the wrong area by using explicit scoping:
8935: @example
8936: IF
8937: SCOPE
8938: @{ x @}
8939: ENDSCOPE
8940: BEGIN
8941: [ 1 cs-roll ] THEN
8942: ...
8943: UNTIL
8944: @end example
8945:
8946: Since the guess is optimistic, there will be no spurious error messages
8947: about undefined locals.
8948:
8949: If the @code{BEGIN} is not reachable from above (e.g., after
8950: @code{AHEAD} or @code{EXIT}), the compiler cannot even make an
8951: optimistic guess, as the locals visible after the @code{BEGIN} may be
8952: defined later. Therefore, the compiler assumes that no locals are
8953: visible after the @code{BEGIN}. However, the user can use
8954: @code{ASSUME-LIVE} to make the compiler assume that the same locals are
8955: visible at the BEGIN as at the point where the top control-flow stack
8956: item was created.
8957:
8958:
8959: doc-assume-live
8960:
8961:
8962: @noindent
8963: E.g.,
8964: @example
8965: @{ x @}
8966: AHEAD
8967: ASSUME-LIVE
8968: BEGIN
8969: x
8970: [ 1 CS-ROLL ] THEN
8971: ...
8972: UNTIL
8973: @end example
8974:
8975: Other cases where the locals are defined before the @code{BEGIN} can be
8976: handled by inserting an appropriate @code{CS-ROLL} before the
8977: @code{ASSUME-LIVE} (and changing the control-flow stack manipulation
8978: behind the @code{ASSUME-LIVE}).
8979:
8980: Cases where locals are defined after the @code{BEGIN} (but should be
8981: visible immediately after the @code{BEGIN}) can only be handled by
8982: rearranging the loop. E.g., the ``most insidious'' example above can be
8983: arranged into:
8984: @example
8985: BEGIN
8986: @{ x @}
8987: ... 0=
8988: WHILE
8989: x
8990: REPEAT
8991: @end example
8992:
8993: @node How long do locals live?, Locals programming style, Where are locals visible by name?, Gforth locals
8994: @subsubsection How long do locals live?
8995: @cindex locals lifetime
8996: @cindex lifetime of locals
8997:
8998: The right answer for the lifetime question would be: A local lives at
8999: least as long as it can be accessed. For a value-flavoured local this
9000: means: until the end of its visibility. However, a variable-flavoured
9001: local could be accessed through its address far beyond its visibility
9002: scope. Ultimately, this would mean that such locals would have to be
9003: garbage collected. Since this entails un-Forth-like implementation
9004: complexities, I adopted the same cowardly solution as some other
9005: languages (e.g., C): The local lives only as long as it is visible;
9006: afterwards its address is invalid (and programs that access it
9007: afterwards are erroneous).
9008:
9009: @node Locals programming style, Locals implementation, How long do locals live?, Gforth locals
9010: @subsubsection Locals programming style
9011: @cindex locals programming style
9012: @cindex programming style, locals
9013:
9014: The freedom to define locals anywhere has the potential to change
9015: programming styles dramatically. In particular, the need to use the
9016: return stack for intermediate storage vanishes. Moreover, all stack
9017: manipulations (except @code{PICK}s and @code{ROLL}s with run-time
9018: determined arguments) can be eliminated: If the stack items are in the
9019: wrong order, just write a locals definition for all of them; then
9020: write the items in the order you want.
9021:
9022: This seems a little far-fetched and eliminating stack manipulations is
9023: unlikely to become a conscious programming objective. Still, the number
9024: of stack manipulations will be reduced dramatically if local variables
9025: are used liberally (e.g., compare @code{max} (@pxref{Gforth locals}) with
9026: a traditional implementation of @code{max}).
9027:
9028: This shows one potential benefit of locals: making Forth programs more
9029: readable. Of course, this benefit will only be realized if the
9030: programmers continue to honour the principle of factoring instead of
9031: using the added latitude to make the words longer.
9032:
9033: @cindex single-assignment style for locals
9034: Using @code{TO} can and should be avoided. Without @code{TO},
9035: every value-flavoured local has only a single assignment and many
9036: advantages of functional languages apply to Forth. I.e., programs are
9037: easier to analyse, to optimize and to read: It is clear from the
9038: definition what the local stands for, it does not turn into something
9039: different later.
9040:
9041: E.g., a definition using @code{TO} might look like this:
9042: @example
9043: : strcmp @{ addr1 u1 addr2 u2 -- n @}
9044: u1 u2 min 0
9045: ?do
9046: addr1 c@@ addr2 c@@ -
9047: ?dup-if
9048: unloop exit
9049: then
9050: addr1 char+ TO addr1
9051: addr2 char+ TO addr2
9052: loop
9053: u1 u2 - ;
9054: @end example
9055: Here, @code{TO} is used to update @code{addr1} and @code{addr2} at
9056: every loop iteration. @code{strcmp} is a typical example of the
9057: readability problems of using @code{TO}. When you start reading
9058: @code{strcmp}, you think that @code{addr1} refers to the start of the
9059: string. Only near the end of the loop you realize that it is something
9060: else.
9061:
9062: This can be avoided by defining two locals at the start of the loop that
9063: are initialized with the right value for the current iteration.
9064: @example
9065: : strcmp @{ addr1 u1 addr2 u2 -- n @}
9066: addr1 addr2
9067: u1 u2 min 0
9068: ?do @{ s1 s2 @}
9069: s1 c@@ s2 c@@ -
9070: ?dup-if
9071: unloop exit
9072: then
9073: s1 char+ s2 char+
9074: loop
9075: 2drop
9076: u1 u2 - ;
9077: @end example
9078: Here it is clear from the start that @code{s1} has a different value
9079: in every loop iteration.
9080:
9081: @node Locals implementation, , Locals programming style, Gforth locals
9082: @subsubsection Locals implementation
9083: @cindex locals implementation
9084: @cindex implementation of locals
9085:
9086: @cindex locals stack
9087: Gforth uses an extra locals stack. The most compelling reason for
9088: this is that the return stack is not float-aligned; using an extra stack
9089: also eliminates the problems and restrictions of using the return stack
9090: as locals stack. Like the other stacks, the locals stack grows toward
9091: lower addresses. A few primitives allow an efficient implementation:
9092:
9093:
9094: doc-@local#
9095: doc-f@local#
9096: doc-laddr#
9097: doc-lp+!#
9098: doc-lp!
9099: doc->l
9100: doc-f>l
9101:
9102:
9103: In addition to these primitives, some specializations of these
9104: primitives for commonly occurring inline arguments are provided for
9105: efficiency reasons, e.g., @code{@@local0} as specialization of
9106: @code{@@local#} for the inline argument 0. The following compiling words
9107: compile the right specialized version, or the general version, as
9108: appropriate:
9109:
9110:
9111: doc-compile-@local
9112: doc-compile-f@local
9113: doc-compile-lp+!
9114:
9115:
9116: Combinations of conditional branches and @code{lp+!#} like
9117: @code{?branch-lp+!#} (the locals pointer is only changed if the branch
9118: is taken) are provided for efficiency and correctness in loops.
9119:
9120: A special area in the dictionary space is reserved for keeping the
9121: local variable names. @code{@{} switches the dictionary pointer to this
9122: area and @code{@}} switches it back and generates the locals
9123: initializing code. @code{W:} etc.@ are normal defining words. This
9124: special area is cleared at the start of every colon definition.
9125:
9126: @cindex word list for defining locals
9127: A special feature of Gforth's dictionary is used to implement the
9128: definition of locals without type specifiers: every word list (aka
9129: vocabulary) has its own methods for searching
9130: etc. (@pxref{Word Lists}). For the present purpose we defined a word list
9131: with a special search method: When it is searched for a word, it
9132: actually creates that word using @code{W:}. @code{@{} changes the search
9133: order to first search the word list containing @code{@}}, @code{W:} etc.,
9134: and then the word list for defining locals without type specifiers.
9135:
9136: The lifetime rules support a stack discipline within a colon
9137: definition: The lifetime of a local is either nested with other locals
9138: lifetimes or it does not overlap them.
9139:
9140: At @code{BEGIN}, @code{IF}, and @code{AHEAD} no code for locals stack
9141: pointer manipulation is generated. Between control structure words
9142: locals definitions can push locals onto the locals stack. @code{AGAIN}
9143: is the simplest of the other three control flow words. It has to
9144: restore the locals stack depth of the corresponding @code{BEGIN}
9145: before branching. The code looks like this:
9146: @format
9147: @code{lp+!#} current-locals-size @minus{} dest-locals-size
9148: @code{branch} <begin>
9149: @end format
9150:
9151: @code{UNTIL} is a little more complicated: If it branches back, it
9152: must adjust the stack just like @code{AGAIN}. But if it falls through,
9153: the locals stack must not be changed. The compiler generates the
9154: following code:
9155: @format
9156: @code{?branch-lp+!#} <begin> current-locals-size @minus{} dest-locals-size
9157: @end format
9158: The locals stack pointer is only adjusted if the branch is taken.
9159:
9160: @code{THEN} can produce somewhat inefficient code:
9161: @format
9162: @code{lp+!#} current-locals-size @minus{} orig-locals-size
9163: <orig target>:
9164: @code{lp+!#} orig-locals-size @minus{} new-locals-size
9165: @end format
9166: The second @code{lp+!#} adjusts the locals stack pointer from the
9167: level at the @i{orig} point to the level after the @code{THEN}. The
9168: first @code{lp+!#} adjusts the locals stack pointer from the current
9169: level to the level at the orig point, so the complete effect is an
9170: adjustment from the current level to the right level after the
9171: @code{THEN}.
9172:
9173: @cindex locals information on the control-flow stack
9174: @cindex control-flow stack items, locals information
9175: In a conventional Forth implementation a dest control-flow stack entry
9176: is just the target address and an orig entry is just the address to be
9177: patched. Our locals implementation adds a word list to every orig or dest
9178: item. It is the list of locals visible (or assumed visible) at the point
9179: described by the entry. Our implementation also adds a tag to identify
9180: the kind of entry, in particular to differentiate between live and dead
9181: (reachable and unreachable) orig entries.
9182:
9183: A few unusual operations have to be performed on locals word lists:
9184:
9185:
9186: doc-common-list
9187: doc-sub-list?
9188: doc-list-size
9189:
9190:
9191: Several features of our locals word list implementation make these
9192: operations easy to implement: The locals word lists are organised as
9193: linked lists; the tails of these lists are shared, if the lists
9194: contain some of the same locals; and the address of a name is greater
9195: than the address of the names behind it in the list.
9196:
9197: Another important implementation detail is the variable
9198: @code{dead-code}. It is used by @code{BEGIN} and @code{THEN} to
9199: determine if they can be reached directly or only through the branch
9200: that they resolve. @code{dead-code} is set by @code{UNREACHABLE},
9201: @code{AHEAD}, @code{EXIT} etc., and cleared at the start of a colon
9202: definition, by @code{BEGIN} and usually by @code{THEN}.
9203:
9204: Counted loops are similar to other loops in most respects, but
9205: @code{LEAVE} requires special attention: It performs basically the same
9206: service as @code{AHEAD}, but it does not create a control-flow stack
9207: entry. Therefore the information has to be stored elsewhere;
9208: traditionally, the information was stored in the target fields of the
9209: branches created by the @code{LEAVE}s, by organizing these fields into a
9210: linked list. Unfortunately, this clever trick does not provide enough
9211: space for storing our extended control flow information. Therefore, we
9212: introduce another stack, the leave stack. It contains the control-flow
9213: stack entries for all unresolved @code{LEAVE}s.
9214:
9215: Local names are kept until the end of the colon definition, even if
9216: they are no longer visible in any control-flow path. In a few cases
9217: this may lead to increased space needs for the locals name area, but
9218: usually less than reclaiming this space would cost in code size.
9219:
9220:
9221: @node ANS Forth locals, , Gforth locals, Locals
9222: @subsection ANS Forth locals
9223: @cindex locals, ANS Forth style
9224:
9225: The ANS Forth locals wordset does not define a syntax for locals, but
9226: words that make it possible to define various syntaxes. One of the
9227: possible syntaxes is a subset of the syntax we used in the Gforth locals
9228: wordset, i.e.:
9229:
9230: @example
9231: @{ local1 local2 ... -- comment @}
9232: @end example
9233: @noindent
9234: or
9235: @example
9236: @{ local1 local2 ... @}
9237: @end example
9238:
9239: The order of the locals corresponds to the order in a stack comment. The
9240: restrictions are:
9241:
9242: @itemize @bullet
9243: @item
9244: Locals can only be cell-sized values (no type specifiers are allowed).
9245: @item
9246: Locals can be defined only outside control structures.
9247: @item
9248: Locals can interfere with explicit usage of the return stack. For the
9249: exact (and long) rules, see the standard. If you don't use return stack
9250: accessing words in a definition using locals, you will be all right. The
9251: purpose of this rule is to make locals implementation on the return
9252: stack easier.
9253: @item
9254: The whole definition must be in one line.
9255: @end itemize
9256:
9257: Locals defined in ANS Forth behave like @code{VALUE}s
9258: (@pxref{Values}). I.e., they are initialized from the stack. Using their
9259: name produces their value. Their value can be changed using @code{TO}.
9260:
9261: Since the syntax above is supported by Gforth directly, you need not do
9262: anything to use it. If you want to port a program using this syntax to
9263: another ANS Forth system, use @file{compat/anslocal.fs} to implement the
9264: syntax on the other system.
9265:
9266: Note that a syntax shown in the standard, section A.13 looks
9267: similar, but is quite different in having the order of locals
9268: reversed. Beware!
9269:
9270: The ANS Forth locals wordset itself consists of one word:
9271:
9272: doc-(local)
9273:
9274: The ANS Forth locals extension wordset defines a syntax using
9275: @code{locals|}, but it is so awful that we strongly recommend not to use
9276: it. We have implemented this syntax to make porting to Gforth easy, but
9277: do not document it here. The problem with this syntax is that the locals
9278: are defined in an order reversed with respect to the standard stack
9279: comment notation, making programs harder to read, and easier to misread
9280: and miswrite. The only merit of this syntax is that it is easy to
9281: implement using the ANS Forth locals wordset.
9282:
9283:
9284: @c ----------------------------------------------------------
9285: @node Structures, Object-oriented Forth, Locals, Words
9286: @section Structures
9287: @cindex structures
9288: @cindex records
9289:
9290: This section presents the structure package that comes with Gforth. A
9291: version of the package implemented in ANS Forth is available in
9292: @file{compat/struct.fs}. This package was inspired by a posting on
9293: comp.lang.forth in 1989 (unfortunately I don't remember, by whom;
9294: possibly John Hayes). A version of this section has been published in
9295: M. Anton Ertl,
9296: @uref{http://www.complang.tuwien.ac.at/forth/objects/structs.html, Yet
9297: Another Forth Structures Package}, Forth Dimensions 19(3), pages
9298: 13--16. Marcel Hendrix provided helpful comments.
9299:
9300: @menu
9301: * Why explicit structure support?::
9302: * Structure Usage::
9303: * Structure Naming Convention::
9304: * Structure Implementation::
9305: * Structure Glossary::
9306: @end menu
9307:
9308: @node Why explicit structure support?, Structure Usage, Structures, Structures
9309: @subsection Why explicit structure support?
9310:
9311: @cindex address arithmetic for structures
9312: @cindex structures using address arithmetic
9313: If we want to use a structure containing several fields, we could simply
9314: reserve memory for it, and access the fields using address arithmetic
9315: (@pxref{Address arithmetic}). As an example, consider a structure with
9316: the following fields
9317:
9318: @table @code
9319: @item a
9320: is a float
9321: @item b
9322: is a cell
9323: @item c
9324: is a float
9325: @end table
9326:
9327: Given the (float-aligned) base address of the structure we get the
9328: address of the field
9329:
9330: @table @code
9331: @item a
9332: without doing anything further.
9333: @item b
9334: with @code{float+}
9335: @item c
9336: with @code{float+ cell+ faligned}
9337: @end table
9338:
9339: It is easy to see that this can become quite tiring.
9340:
9341: Moreover, it is not very readable, because seeing a
9342: @code{cell+} tells us neither which kind of structure is
9343: accessed nor what field is accessed; we have to somehow infer the kind
9344: of structure, and then look up in the documentation, which field of
9345: that structure corresponds to that offset.
9346:
9347: Finally, this kind of address arithmetic also causes maintenance
9348: troubles: If you add or delete a field somewhere in the middle of the
9349: structure, you have to find and change all computations for the fields
9350: afterwards.
9351:
9352: So, instead of using @code{cell+} and friends directly, how
9353: about storing the offsets in constants:
9354:
9355: @example
9356: 0 constant a-offset
9357: 0 float+ constant b-offset
9358: 0 float+ cell+ faligned c-offset
9359: @end example
9360:
9361: Now we can get the address of field @code{x} with @code{x-offset
9362: +}. This is much better in all respects. Of course, you still
9363: have to change all later offset definitions if you add a field. You can
9364: fix this by declaring the offsets in the following way:
9365:
9366: @example
9367: 0 constant a-offset
9368: a-offset float+ constant b-offset
9369: b-offset cell+ faligned constant c-offset
9370: @end example
9371:
9372: Since we always use the offsets with @code{+}, we could use a defining
9373: word @code{cfield} that includes the @code{+} in the action of the
9374: defined word:
9375:
9376: @example
9377: : cfield ( n "name" -- )
9378: create ,
9379: does> ( name execution: addr1 -- addr2 )
9380: @@ + ;
9381:
9382: 0 cfield a
9383: 0 a float+ cfield b
9384: 0 b cell+ faligned cfield c
9385: @end example
9386:
9387: Instead of @code{x-offset +}, we now simply write @code{x}.
9388:
9389: The structure field words now can be used quite nicely. However,
9390: their definition is still a bit cumbersome: We have to repeat the
9391: name, the information about size and alignment is distributed before
9392: and after the field definitions etc. The structure package presented
9393: here addresses these problems.
9394:
9395: @node Structure Usage, Structure Naming Convention, Why explicit structure support?, Structures
9396: @subsection Structure Usage
9397: @cindex structure usage
9398:
9399: @cindex @code{field} usage
9400: @cindex @code{struct} usage
9401: @cindex @code{end-struct} usage
9402: You can define a structure for a (data-less) linked list with:
9403: @example
9404: struct
9405: cell% field list-next
9406: end-struct list%
9407: @end example
9408:
9409: With the address of the list node on the stack, you can compute the
9410: address of the field that contains the address of the next node with
9411: @code{list-next}. E.g., you can determine the length of a list
9412: with:
9413:
9414: @example
9415: : list-length ( list -- n )
9416: \ "list" is a pointer to the first element of a linked list
9417: \ "n" is the length of the list
9418: 0 BEGIN ( list1 n1 )
9419: over
9420: WHILE ( list1 n1 )
9421: 1+ swap list-next @@ swap
9422: REPEAT
9423: nip ;
9424: @end example
9425:
9426: You can reserve memory for a list node in the dictionary with
9427: @code{list% %allot}, which leaves the address of the list node on the
9428: stack. For the equivalent allocation on the heap you can use @code{list%
9429: %alloc} (or, for an @code{allocate}-like stack effect (i.e., with ior),
9430: use @code{list% %allocate}). You can get the the size of a list
9431: node with @code{list% %size} and its alignment with @code{list%
9432: %alignment}.
9433:
9434: Note that in ANS Forth the body of a @code{create}d word is
9435: @code{aligned} but not necessarily @code{faligned};
9436: therefore, if you do a:
9437:
9438: @example
9439: create @emph{name} foo% %allot drop
9440: @end example
9441:
9442: @noindent
9443: then the memory alloted for @code{foo%} is guaranteed to start at the
9444: body of @code{@emph{name}} only if @code{foo%} contains only character,
9445: cell and double fields. Therefore, if your structure contains floats,
9446: better use
9447:
9448: @example
9449: foo% %allot constant @emph{name}
9450: @end example
9451:
9452: @cindex structures containing structures
9453: You can include a structure @code{foo%} as a field of
9454: another structure, like this:
9455: @example
9456: struct
9457: ...
9458: foo% field ...
9459: ...
9460: end-struct ...
9461: @end example
9462:
9463: @cindex structure extension
9464: @cindex extended records
9465: Instead of starting with an empty structure, you can extend an
9466: existing structure. E.g., a plain linked list without data, as defined
9467: above, is hardly useful; You can extend it to a linked list of integers,
9468: like this:@footnote{This feature is also known as @emph{extended
9469: records}. It is the main innovation in the Oberon language; in other
9470: words, adding this feature to Modula-2 led Wirth to create a new
9471: language, write a new compiler etc. Adding this feature to Forth just
9472: required a few lines of code.}
9473:
9474: @example
9475: list%
9476: cell% field intlist-int
9477: end-struct intlist%
9478: @end example
9479:
9480: @code{intlist%} is a structure with two fields:
9481: @code{list-next} and @code{intlist-int}.
9482:
9483: @cindex structures containing arrays
9484: You can specify an array type containing @emph{n} elements of
9485: type @code{foo%} like this:
9486:
9487: @example
9488: foo% @emph{n} *
9489: @end example
9490:
9491: You can use this array type in any place where you can use a normal
9492: type, e.g., when defining a @code{field}, or with
9493: @code{%allot}.
9494:
9495: @cindex first field optimization
9496: The first field is at the base address of a structure and the word for
9497: this field (e.g., @code{list-next}) actually does not change the address
9498: on the stack. You may be tempted to leave it away in the interest of
9499: run-time and space efficiency. This is not necessary, because the
9500: structure package optimizes this case: If you compile a first-field
9501: words, no code is generated. So, in the interest of readability and
9502: maintainability you should include the word for the field when accessing
9503: the field.
9504:
9505:
9506: @node Structure Naming Convention, Structure Implementation, Structure Usage, Structures
9507: @subsection Structure Naming Convention
9508: @cindex structure naming convention
9509:
9510: The field names that come to (my) mind are often quite generic, and,
9511: if used, would cause frequent name clashes. E.g., many structures
9512: probably contain a @code{counter} field. The structure names
9513: that come to (my) mind are often also the logical choice for the names
9514: of words that create such a structure.
9515:
9516: Therefore, I have adopted the following naming conventions:
9517:
9518: @itemize @bullet
9519: @cindex field naming convention
9520: @item
9521: The names of fields are of the form
9522: @code{@emph{struct}-@emph{field}}, where
9523: @code{@emph{struct}} is the basic name of the structure, and
9524: @code{@emph{field}} is the basic name of the field. You can
9525: think of field words as converting the (address of the)
9526: structure into the (address of the) field.
9527:
9528: @cindex structure naming convention
9529: @item
9530: The names of structures are of the form
9531: @code{@emph{struct}%}, where
9532: @code{@emph{struct}} is the basic name of the structure.
9533: @end itemize
9534:
9535: This naming convention does not work that well for fields of extended
9536: structures; e.g., the integer list structure has a field
9537: @code{intlist-int}, but has @code{list-next}, not
9538: @code{intlist-next}.
9539:
9540: @node Structure Implementation, Structure Glossary, Structure Naming Convention, Structures
9541: @subsection Structure Implementation
9542: @cindex structure implementation
9543: @cindex implementation of structures
9544:
9545: The central idea in the implementation is to pass the data about the
9546: structure being built on the stack, not in some global
9547: variable. Everything else falls into place naturally once this design
9548: decision is made.
9549:
9550: The type description on the stack is of the form @emph{align
9551: size}. Keeping the size on the top-of-stack makes dealing with arrays
9552: very simple.
9553:
9554: @code{field} is a defining word that uses @code{Create}
9555: and @code{DOES>}. The body of the field contains the offset
9556: of the field, and the normal @code{DOES>} action is simply:
9557:
9558: @example
9559: @@ +
9560: @end example
9561:
9562: @noindent
9563: i.e., add the offset to the address, giving the stack effect
9564: @i{addr1 -- addr2} for a field.
9565:
9566: @cindex first field optimization, implementation
9567: This simple structure is slightly complicated by the optimization
9568: for fields with offset 0, which requires a different
9569: @code{DOES>}-part (because we cannot rely on there being
9570: something on the stack if such a field is invoked during
9571: compilation). Therefore, we put the different @code{DOES>}-parts
9572: in separate words, and decide which one to invoke based on the
9573: offset. For a zero offset, the field is basically a noop; it is
9574: immediate, and therefore no code is generated when it is compiled.
9575:
9576: @node Structure Glossary, , Structure Implementation, Structures
9577: @subsection Structure Glossary
9578: @cindex structure glossary
9579:
9580:
9581: doc-%align
9582: doc-%alignment
9583: doc-%alloc
9584: doc-%allocate
9585: doc-%allot
9586: doc-cell%
9587: doc-char%
9588: doc-dfloat%
9589: doc-double%
9590: doc-end-struct
9591: doc-field
9592: doc-float%
9593: doc-naligned
9594: doc-sfloat%
9595: doc-%size
9596: doc-struct
9597:
9598:
9599: @c -------------------------------------------------------------
9600: @node Object-oriented Forth, Programming Tools, Structures, Words
9601: @section Object-oriented Forth
9602:
9603: Gforth comes with three packages for object-oriented programming:
9604: @file{objects.fs}, @file{oof.fs}, and @file{mini-oof.fs}; none of them
9605: is preloaded, so you have to @code{include} them before use. The most
9606: important differences between these packages (and others) are discussed
9607: in @ref{Comparison with other object models}. All packages are written
9608: in ANS Forth and can be used with any other ANS Forth.
9609:
9610: @menu
9611: * Why object-oriented programming?::
9612: * Object-Oriented Terminology::
9613: * Objects::
9614: * OOF::
9615: * Mini-OOF::
9616: * Comparison with other object models::
9617: @end menu
9618:
9619: @c ----------------------------------------------------------------
9620: @node Why object-oriented programming?, Object-Oriented Terminology, Object-oriented Forth, Object-oriented Forth
9621: @subsection Why object-oriented programming?
9622: @cindex object-oriented programming motivation
9623: @cindex motivation for object-oriented programming
9624:
9625: Often we have to deal with several data structures (@emph{objects}),
9626: that have to be treated similarly in some respects, but differently in
9627: others. Graphical objects are the textbook example: circles, triangles,
9628: dinosaurs, icons, and others, and we may want to add more during program
9629: development. We want to apply some operations to any graphical object,
9630: e.g., @code{draw} for displaying it on the screen. However, @code{draw}
9631: has to do something different for every kind of object.
9632: @comment TODO add some other operations eg perimeter, area
9633: @comment and tie in to concrete examples later..
9634:
9635: We could implement @code{draw} as a big @code{CASE}
9636: control structure that executes the appropriate code depending on the
9637: kind of object to be drawn. This would be not be very elegant, and,
9638: moreover, we would have to change @code{draw} every time we add
9639: a new kind of graphical object (say, a spaceship).
9640:
9641: What we would rather do is: When defining spaceships, we would tell
9642: the system: ``Here's how you @code{draw} a spaceship; you figure
9643: out the rest''.
9644:
9645: This is the problem that all systems solve that (rightfully) call
9646: themselves object-oriented; the object-oriented packages presented here
9647: solve this problem (and not much else).
9648: @comment TODO ?list properties of oo systems.. oo vs o-based?
9649:
9650: @c ------------------------------------------------------------------------
9651: @node Object-Oriented Terminology, Objects, Why object-oriented programming?, Object-oriented Forth
9652: @subsection Object-Oriented Terminology
9653: @cindex object-oriented terminology
9654: @cindex terminology for object-oriented programming
9655:
9656: This section is mainly for reference, so you don't have to understand
9657: all of it right away. The terminology is mainly Smalltalk-inspired. In
9658: short:
9659:
9660: @table @emph
9661: @cindex class
9662: @item class
9663: a data structure definition with some extras.
9664:
9665: @cindex object
9666: @item object
9667: an instance of the data structure described by the class definition.
9668:
9669: @cindex instance variables
9670: @item instance variables
9671: fields of the data structure.
9672:
9673: @cindex selector
9674: @cindex method selector
9675: @cindex virtual function
9676: @item selector
9677: (or @emph{method selector}) a word (e.g.,
9678: @code{draw}) that performs an operation on a variety of data
9679: structures (classes). A selector describes @emph{what} operation to
9680: perform. In C++ terminology: a (pure) virtual function.
9681:
9682: @cindex method
9683: @item method
9684: the concrete definition that performs the operation
9685: described by the selector for a specific class. A method specifies
9686: @emph{how} the operation is performed for a specific class.
9687:
9688: @cindex selector invocation
9689: @cindex message send
9690: @cindex invoking a selector
9691: @item selector invocation
9692: a call of a selector. One argument of the call (the TOS (top-of-stack))
9693: is used for determining which method is used. In Smalltalk terminology:
9694: a message (consisting of the selector and the other arguments) is sent
9695: to the object.
9696:
9697: @cindex receiving object
9698: @item receiving object
9699: the object used for determining the method executed by a selector
9700: invocation. In the @file{objects.fs} model, it is the object that is on
9701: the TOS when the selector is invoked. (@emph{Receiving} comes from
9702: the Smalltalk @emph{message} terminology.)
9703:
9704: @cindex child class
9705: @cindex parent class
9706: @cindex inheritance
9707: @item child class
9708: a class that has (@emph{inherits}) all properties (instance variables,
9709: selectors, methods) from a @emph{parent class}. In Smalltalk
9710: terminology: The subclass inherits from the superclass. In C++
9711: terminology: The derived class inherits from the base class.
9712:
9713: @end table
9714:
9715: @c If you wonder about the message sending terminology, it comes from
9716: @c a time when each object had it's own task and objects communicated via
9717: @c message passing; eventually the Smalltalk developers realized that
9718: @c they can do most things through simple (indirect) calls. They kept the
9719: @c terminology.
9720:
9721: @c --------------------------------------------------------------
9722: @node Objects, OOF, Object-Oriented Terminology, Object-oriented Forth
9723: @subsection The @file{objects.fs} model
9724: @cindex objects
9725: @cindex object-oriented programming
9726:
9727: @cindex @file{objects.fs}
9728: @cindex @file{oof.fs}
9729:
9730: This section describes the @file{objects.fs} package. This material also
9731: has been published in M. Anton Ertl,
9732: @cite{@uref{http://www.complang.tuwien.ac.at/forth/objects/objects.html,
9733: Yet Another Forth Objects Package}}, Forth Dimensions 19(2), pages
9734: 37--43.
9735: @c McKewan's and Zsoter's packages
9736:
9737: This section assumes that you have read @ref{Structures}.
9738:
9739: The techniques on which this model is based have been used to implement
9740: the parser generator, Gray, and have also been used in Gforth for
9741: implementing the various flavours of word lists (hashed or not,
9742: case-sensitive or not, special-purpose word lists for locals etc.).
9743:
9744:
9745: @menu
9746: * Properties of the Objects model::
9747: * Basic Objects Usage::
9748: * The Objects base class::
9749: * Creating objects::
9750: * Object-Oriented Programming Style::
9751: * Class Binding::
9752: * Method conveniences::
9753: * Classes and Scoping::
9754: * Dividing classes::
9755: * Object Interfaces::
9756: * Objects Implementation::
9757: * Objects Glossary::
9758: @end menu
9759:
9760: Marcel Hendrix provided helpful comments on this section.
9761:
9762: @node Properties of the Objects model, Basic Objects Usage, Objects, Objects
9763: @subsubsection Properties of the @file{objects.fs} model
9764: @cindex @file{objects.fs} properties
9765:
9766: @itemize @bullet
9767: @item
9768: It is straightforward to pass objects on the stack. Passing
9769: selectors on the stack is a little less convenient, but possible.
9770:
9771: @item
9772: Objects are just data structures in memory, and are referenced by their
9773: address. You can create words for objects with normal defining words
9774: like @code{constant}. Likewise, there is no difference between instance
9775: variables that contain objects and those that contain other data.
9776:
9777: @item
9778: Late binding is efficient and easy to use.
9779:
9780: @item
9781: It avoids parsing, and thus avoids problems with state-smartness
9782: and reduced extensibility; for convenience there are a few parsing
9783: words, but they have non-parsing counterparts. There are also a few
9784: defining words that parse. This is hard to avoid, because all standard
9785: defining words parse (except @code{:noname}); however, such
9786: words are not as bad as many other parsing words, because they are not
9787: state-smart.
9788:
9789: @item
9790: It does not try to incorporate everything. It does a few things and does
9791: them well (IMO). In particular, this model was not designed to support
9792: information hiding (although it has features that may help); you can use
9793: a separate package for achieving this.
9794:
9795: @item
9796: It is layered; you don't have to learn and use all features to use this
9797: model. Only a few features are necessary (@pxref{Basic Objects Usage},
9798: @pxref{The Objects base class}, @pxref{Creating objects}.), the others
9799: are optional and independent of each other.
9800:
9801: @item
9802: An implementation in ANS Forth is available.
9803:
9804: @end itemize
9805:
9806:
9807: @node Basic Objects Usage, The Objects base class, Properties of the Objects model, Objects
9808: @subsubsection Basic @file{objects.fs} Usage
9809: @cindex basic objects usage
9810: @cindex objects, basic usage
9811:
9812: You can define a class for graphical objects like this:
9813:
9814: @cindex @code{class} usage
9815: @cindex @code{end-class} usage
9816: @cindex @code{selector} usage
9817: @example
9818: object class \ "object" is the parent class
9819: selector draw ( x y graphical -- )
9820: end-class graphical
9821: @end example
9822:
9823: This code defines a class @code{graphical} with an
9824: operation @code{draw}. We can perform the operation
9825: @code{draw} on any @code{graphical} object, e.g.:
9826:
9827: @example
9828: 100 100 t-rex draw
9829: @end example
9830:
9831: @noindent
9832: where @code{t-rex} is a word (say, a constant) that produces a
9833: graphical object.
9834:
9835: @comment TODO add a 2nd operation eg perimeter.. and use for
9836: @comment a concrete example
9837:
9838: @cindex abstract class
9839: How do we create a graphical object? With the present definitions,
9840: we cannot create a useful graphical object. The class
9841: @code{graphical} describes graphical objects in general, but not
9842: any concrete graphical object type (C++ users would call it an
9843: @emph{abstract class}); e.g., there is no method for the selector
9844: @code{draw} in the class @code{graphical}.
9845:
9846: For concrete graphical objects, we define child classes of the
9847: class @code{graphical}, e.g.:
9848:
9849: @cindex @code{overrides} usage
9850: @cindex @code{field} usage in class definition
9851: @example
9852: graphical class \ "graphical" is the parent class
9853: cell% field circle-radius
9854:
9855: :noname ( x y circle -- )
9856: circle-radius @@ draw-circle ;
9857: overrides draw
9858:
9859: :noname ( n-radius circle -- )
9860: circle-radius ! ;
9861: overrides construct
9862:
9863: end-class circle
9864: @end example
9865:
9866: Here we define a class @code{circle} as a child of @code{graphical},
9867: with field @code{circle-radius} (which behaves just like a field
9868: (@pxref{Structures}); it defines (using @code{overrides}) new methods
9869: for the selectors @code{draw} and @code{construct} (@code{construct} is
9870: defined in @code{object}, the parent class of @code{graphical}).
9871:
9872: Now we can create a circle on the heap (i.e.,
9873: @code{allocate}d memory) with:
9874:
9875: @cindex @code{heap-new} usage
9876: @example
9877: 50 circle heap-new constant my-circle
9878: @end example
9879:
9880: @noindent
9881: @code{heap-new} invokes @code{construct}, thus
9882: initializing the field @code{circle-radius} with 50. We can draw
9883: this new circle at (100,100) with:
9884:
9885: @example
9886: 100 100 my-circle draw
9887: @end example
9888:
9889: @cindex selector invocation, restrictions
9890: @cindex class definition, restrictions
9891: Note: You can only invoke a selector if the object on the TOS
9892: (the receiving object) belongs to the class where the selector was
9893: defined or one of its descendents; e.g., you can invoke
9894: @code{draw} only for objects belonging to @code{graphical}
9895: or its descendents (e.g., @code{circle}). Immediately before
9896: @code{end-class}, the search order has to be the same as
9897: immediately after @code{class}.
9898:
9899: @node The Objects base class, Creating objects, Basic Objects Usage, Objects
9900: @subsubsection The @file{object.fs} base class
9901: @cindex @code{object} class
9902:
9903: When you define a class, you have to specify a parent class. So how do
9904: you start defining classes? There is one class available from the start:
9905: @code{object}. It is ancestor for all classes and so is the
9906: only class that has no parent. It has two selectors: @code{construct}
9907: and @code{print}.
9908:
9909: @node Creating objects, Object-Oriented Programming Style, The Objects base class, Objects
9910: @subsubsection Creating objects
9911: @cindex creating objects
9912: @cindex object creation
9913: @cindex object allocation options
9914:
9915: @cindex @code{heap-new} discussion
9916: @cindex @code{dict-new} discussion
9917: @cindex @code{construct} discussion
9918: You can create and initialize an object of a class on the heap with
9919: @code{heap-new} ( ... class -- object ) and in the dictionary
9920: (allocation with @code{allot}) with @code{dict-new} (
9921: ... class -- object ). Both words invoke @code{construct}, which
9922: consumes the stack items indicated by "..." above.
9923:
9924: @cindex @code{init-object} discussion
9925: @cindex @code{class-inst-size} discussion
9926: If you want to allocate memory for an object yourself, you can get its
9927: alignment and size with @code{class-inst-size 2@@} ( class --
9928: align size ). Once you have memory for an object, you can initialize
9929: it with @code{init-object} ( ... class object -- );
9930: @code{construct} does only a part of the necessary work.
9931:
9932: @node Object-Oriented Programming Style, Class Binding, Creating objects, Objects
9933: @subsubsection Object-Oriented Programming Style
9934: @cindex object-oriented programming style
9935: @cindex programming style, object-oriented
9936:
9937: This section is not exhaustive.
9938:
9939: @cindex stack effects of selectors
9940: @cindex selectors and stack effects
9941: In general, it is a good idea to ensure that all methods for the
9942: same selector have the same stack effect: when you invoke a selector,
9943: you often have no idea which method will be invoked, so, unless all
9944: methods have the same stack effect, you will not know the stack effect
9945: of the selector invocation.
9946:
9947: One exception to this rule is methods for the selector
9948: @code{construct}. We know which method is invoked, because we
9949: specify the class to be constructed at the same place. Actually, I
9950: defined @code{construct} as a selector only to give the users a
9951: convenient way to specify initialization. The way it is used, a
9952: mechanism different from selector invocation would be more natural
9953: (but probably would take more code and more space to explain).
9954:
9955: @node Class Binding, Method conveniences, Object-Oriented Programming Style, Objects
9956: @subsubsection Class Binding
9957: @cindex class binding
9958: @cindex early binding
9959:
9960: @cindex late binding
9961: Normal selector invocations determine the method at run-time depending
9962: on the class of the receiving object. This run-time selection is called
9963: @i{late binding}.
9964:
9965: Sometimes it's preferable to invoke a different method. For example,
9966: you might want to use the simple method for @code{print}ing
9967: @code{object}s instead of the possibly long-winded @code{print} method
9968: of the receiver class. You can achieve this by replacing the invocation
9969: of @code{print} with:
9970:
9971: @cindex @code{[bind]} usage
9972: @example
9973: [bind] object print
9974: @end example
9975:
9976: @noindent
9977: in compiled code or:
9978:
9979: @cindex @code{bind} usage
9980: @example
9981: bind object print
9982: @end example
9983:
9984: @cindex class binding, alternative to
9985: @noindent
9986: in interpreted code. Alternatively, you can define the method with a
9987: name (e.g., @code{print-object}), and then invoke it through the
9988: name. Class binding is just a (often more convenient) way to achieve
9989: the same effect; it avoids name clutter and allows you to invoke
9990: methods directly without naming them first.
9991:
9992: @cindex superclass binding
9993: @cindex parent class binding
9994: A frequent use of class binding is this: When we define a method
9995: for a selector, we often want the method to do what the selector does
9996: in the parent class, and a little more. There is a special word for
9997: this purpose: @code{[parent]}; @code{[parent]
9998: @emph{selector}} is equivalent to @code{[bind] @emph{parent
9999: selector}}, where @code{@emph{parent}} is the parent
10000: class of the current class. E.g., a method definition might look like:
10001:
10002: @cindex @code{[parent]} usage
10003: @example
10004: :noname
10005: dup [parent] foo \ do parent's foo on the receiving object
10006: ... \ do some more
10007: ; overrides foo
10008: @end example
10009:
10010: @cindex class binding as optimization
10011: In @cite{Object-oriented programming in ANS Forth} (Forth Dimensions,
10012: March 1997), Andrew McKewan presents class binding as an optimization
10013: technique. I recommend not using it for this purpose unless you are in
10014: an emergency. Late binding is pretty fast with this model anyway, so the
10015: benefit of using class binding is small; the cost of using class binding
10016: where it is not appropriate is reduced maintainability.
10017:
10018: While we are at programming style questions: You should bind
10019: selectors only to ancestor classes of the receiving object. E.g., say,
10020: you know that the receiving object is of class @code{foo} or its
10021: descendents; then you should bind only to @code{foo} and its
10022: ancestors.
10023:
10024: @node Method conveniences, Classes and Scoping, Class Binding, Objects
10025: @subsubsection Method conveniences
10026: @cindex method conveniences
10027:
10028: In a method you usually access the receiving object pretty often. If
10029: you define the method as a plain colon definition (e.g., with
10030: @code{:noname}), you may have to do a lot of stack
10031: gymnastics. To avoid this, you can define the method with @code{m:
10032: ... ;m}. E.g., you could define the method for
10033: @code{draw}ing a @code{circle} with
10034:
10035: @cindex @code{this} usage
10036: @cindex @code{m:} usage
10037: @cindex @code{;m} usage
10038: @example
10039: m: ( x y circle -- )
10040: ( x y ) this circle-radius @@ draw-circle ;m
10041: @end example
10042:
10043: @cindex @code{exit} in @code{m: ... ;m}
10044: @cindex @code{exitm} discussion
10045: @cindex @code{catch} in @code{m: ... ;m}
10046: When this method is executed, the receiver object is removed from the
10047: stack; you can access it with @code{this} (admittedly, in this
10048: example the use of @code{m: ... ;m} offers no advantage). Note
10049: that I specify the stack effect for the whole method (i.e. including
10050: the receiver object), not just for the code between @code{m:}
10051: and @code{;m}. You cannot use @code{exit} in
10052: @code{m:...;m}; instead, use
10053: @code{exitm}.@footnote{Moreover, for any word that calls
10054: @code{catch} and was defined before loading
10055: @code{objects.fs}, you have to redefine it like I redefined
10056: @code{catch}: @code{: catch this >r catch r> to-this ;}}
10057:
10058: @cindex @code{inst-var} usage
10059: You will frequently use sequences of the form @code{this
10060: @emph{field}} (in the example above: @code{this
10061: circle-radius}). If you use the field only in this way, you can
10062: define it with @code{inst-var} and eliminate the
10063: @code{this} before the field name. E.g., the @code{circle}
10064: class above could also be defined with:
10065:
10066: @example
10067: graphical class
10068: cell% inst-var radius
10069:
10070: m: ( x y circle -- )
10071: radius @@ draw-circle ;m
10072: overrides draw
10073:
10074: m: ( n-radius circle -- )
10075: radius ! ;m
10076: overrides construct
10077:
10078: end-class circle
10079: @end example
10080:
10081: @code{radius} can only be used in @code{circle} and its
10082: descendent classes and inside @code{m:...;m}.
10083:
10084: @cindex @code{inst-value} usage
10085: You can also define fields with @code{inst-value}, which is
10086: to @code{inst-var} what @code{value} is to
10087: @code{variable}. You can change the value of such a field with
10088: @code{[to-inst]}. E.g., we could also define the class
10089: @code{circle} like this:
10090:
10091: @example
10092: graphical class
10093: inst-value radius
10094:
10095: m: ( x y circle -- )
10096: radius draw-circle ;m
10097: overrides draw
10098:
10099: m: ( n-radius circle -- )
10100: [to-inst] radius ;m
10101: overrides construct
10102:
10103: end-class circle
10104: @end example
10105:
10106: @c !! :m is easy to confuse with m:. Another name would be better.
10107:
10108: @c Finally, you can define named methods with @code{:m}. One use of this
10109: @c feature is the definition of words that occur only in one class and are
10110: @c not intended to be overridden, but which still need method context
10111: @c (e.g., for accessing @code{inst-var}s). Another use is for methods that
10112: @c would be bound frequently, if defined anonymously.
10113:
10114:
10115: @node Classes and Scoping, Dividing classes, Method conveniences, Objects
10116: @subsubsection Classes and Scoping
10117: @cindex classes and scoping
10118: @cindex scoping and classes
10119:
10120: Inheritance is frequent, unlike structure extension. This exacerbates
10121: the problem with the field name convention (@pxref{Structure Naming
10122: Convention}): One always has to remember in which class the field was
10123: originally defined; changing a part of the class structure would require
10124: changes for renaming in otherwise unaffected code.
10125:
10126: @cindex @code{inst-var} visibility
10127: @cindex @code{inst-value} visibility
10128: To solve this problem, I added a scoping mechanism (which was not in my
10129: original charter): A field defined with @code{inst-var} (or
10130: @code{inst-value}) is visible only in the class where it is defined and in
10131: the descendent classes of this class. Using such fields only makes
10132: sense in @code{m:}-defined methods in these classes anyway.
10133:
10134: This scoping mechanism allows us to use the unadorned field name,
10135: because name clashes with unrelated words become much less likely.
10136:
10137: @cindex @code{protected} discussion
10138: @cindex @code{private} discussion
10139: Once we have this mechanism, we can also use it for controlling the
10140: visibility of other words: All words defined after
10141: @code{protected} are visible only in the current class and its
10142: descendents. @code{public} restores the compilation
10143: (i.e. @code{current}) word list that was in effect before. If you
10144: have several @code{protected}s without an intervening
10145: @code{public} or @code{set-current}, @code{public}
10146: will restore the compilation word list in effect before the first of
10147: these @code{protected}s.
10148:
10149: @node Dividing classes, Object Interfaces, Classes and Scoping, Objects
10150: @subsubsection Dividing classes
10151: @cindex Dividing classes
10152: @cindex @code{methods}...@code{end-methods}
10153:
10154: You may want to do the definition of methods separate from the
10155: definition of the class, its selectors, fields, and instance variables,
10156: i.e., separate the implementation from the definition. You can do this
10157: in the following way:
10158:
10159: @example
10160: graphical class
10161: inst-value radius
10162: end-class circle
10163:
10164: ... \ do some other stuff
10165:
10166: circle methods \ now we are ready
10167:
10168: m: ( x y circle -- )
10169: radius draw-circle ;m
10170: overrides draw
10171:
10172: m: ( n-radius circle -- )
10173: [to-inst] radius ;m
10174: overrides construct
10175:
10176: end-methods
10177: @end example
10178:
10179: You can use several @code{methods}...@code{end-methods} sections. The
10180: only things you can do to the class in these sections are: defining
10181: methods, and overriding the class's selectors. You must not define new
10182: selectors or fields.
10183:
10184: Note that you often have to override a selector before using it. In
10185: particular, you usually have to override @code{construct} with a new
10186: method before you can invoke @code{heap-new} and friends. E.g., you
10187: must not create a circle before the @code{overrides construct} sequence
10188: in the example above.
10189:
10190: @node Object Interfaces, Objects Implementation, Dividing classes, Objects
10191: @subsubsection Object Interfaces
10192: @cindex object interfaces
10193: @cindex interfaces for objects
10194:
10195: In this model you can only call selectors defined in the class of the
10196: receiving objects or in one of its ancestors. If you call a selector
10197: with a receiving object that is not in one of these classes, the
10198: result is undefined; if you are lucky, the program crashes
10199: immediately.
10200:
10201: @cindex selectors common to hardly-related classes
10202: Now consider the case when you want to have a selector (or several)
10203: available in two classes: You would have to add the selector to a
10204: common ancestor class, in the worst case to @code{object}. You
10205: may not want to do this, e.g., because someone else is responsible for
10206: this ancestor class.
10207:
10208: The solution for this problem is interfaces. An interface is a
10209: collection of selectors. If a class implements an interface, the
10210: selectors become available to the class and its descendents. A class
10211: can implement an unlimited number of interfaces. For the problem
10212: discussed above, we would define an interface for the selector(s), and
10213: both classes would implement the interface.
10214:
10215: As an example, consider an interface @code{storage} for
10216: writing objects to disk and getting them back, and a class
10217: @code{foo} that implements it. The code would look like this:
10218:
10219: @cindex @code{interface} usage
10220: @cindex @code{end-interface} usage
10221: @cindex @code{implementation} usage
10222: @example
10223: interface
10224: selector write ( file object -- )
10225: selector read1 ( file object -- )
10226: end-interface storage
10227:
10228: bar class
10229: storage implementation
10230:
10231: ... overrides write
10232: ... overrides read1
10233: ...
10234: end-class foo
10235: @end example
10236:
10237: @noindent
10238: (I would add a word @code{read} @i{( file -- object )} that uses
10239: @code{read1} internally, but that's beyond the point illustrated
10240: here.)
10241:
10242: Note that you cannot use @code{protected} in an interface; and
10243: of course you cannot define fields.
10244:
10245: In the Neon model, all selectors are available for all classes;
10246: therefore it does not need interfaces. The price you pay in this model
10247: is slower late binding, and therefore, added complexity to avoid late
10248: binding.
10249:
10250: @node Objects Implementation, Objects Glossary, Object Interfaces, Objects
10251: @subsubsection @file{objects.fs} Implementation
10252: @cindex @file{objects.fs} implementation
10253:
10254: @cindex @code{object-map} discussion
10255: An object is a piece of memory, like one of the data structures
10256: described with @code{struct...end-struct}. It has a field
10257: @code{object-map} that points to the method map for the object's
10258: class.
10259:
10260: @cindex method map
10261: @cindex virtual function table
10262: The @emph{method map}@footnote{This is Self terminology; in C++
10263: terminology: virtual function table.} is an array that contains the
10264: execution tokens (@i{xt}s) of the methods for the object's class. Each
10265: selector contains an offset into a method map.
10266:
10267: @cindex @code{selector} implementation, class
10268: @code{selector} is a defining word that uses
10269: @code{CREATE} and @code{DOES>}. The body of the
10270: selector contains the offset; the @code{DOES>} action for a
10271: class selector is, basically:
10272:
10273: @example
10274: ( object addr ) @@ over object-map @@ + @@ execute
10275: @end example
10276:
10277: Since @code{object-map} is the first field of the object, it
10278: does not generate any code. As you can see, calling a selector has a
10279: small, constant cost.
10280:
10281: @cindex @code{current-interface} discussion
10282: @cindex class implementation and representation
10283: A class is basically a @code{struct} combined with a method
10284: map. During the class definition the alignment and size of the class
10285: are passed on the stack, just as with @code{struct}s, so
10286: @code{field} can also be used for defining class
10287: fields. However, passing more items on the stack would be
10288: inconvenient, so @code{class} builds a data structure in memory,
10289: which is accessed through the variable
10290: @code{current-interface}. After its definition is complete, the
10291: class is represented on the stack by a pointer (e.g., as parameter for
10292: a child class definition).
10293:
10294: A new class starts off with the alignment and size of its parent,
10295: and a copy of the parent's method map. Defining new fields extends the
10296: size and alignment; likewise, defining new selectors extends the
10297: method map. @code{overrides} just stores a new @i{xt} in the method
10298: map at the offset given by the selector.
10299:
10300: @cindex class binding, implementation
10301: Class binding just gets the @i{xt} at the offset given by the selector
10302: from the class's method map and @code{compile,}s (in the case of
10303: @code{[bind]}) it.
10304:
10305: @cindex @code{this} implementation
10306: @cindex @code{catch} and @code{this}
10307: @cindex @code{this} and @code{catch}
10308: I implemented @code{this} as a @code{value}. At the
10309: start of an @code{m:...;m} method the old @code{this} is
10310: stored to the return stack and restored at the end; and the object on
10311: the TOS is stored @code{TO this}. This technique has one
10312: disadvantage: If the user does not leave the method via
10313: @code{;m}, but via @code{throw} or @code{exit},
10314: @code{this} is not restored (and @code{exit} may
10315: crash). To deal with the @code{throw} problem, I have redefined
10316: @code{catch} to save and restore @code{this}; the same
10317: should be done with any word that can catch an exception. As for
10318: @code{exit}, I simply forbid it (as a replacement, there is
10319: @code{exitm}).
10320:
10321: @cindex @code{inst-var} implementation
10322: @code{inst-var} is just the same as @code{field}, with
10323: a different @code{DOES>} action:
10324: @example
10325: @@ this +
10326: @end example
10327: Similar for @code{inst-value}.
10328:
10329: @cindex class scoping implementation
10330: Each class also has a word list that contains the words defined with
10331: @code{inst-var} and @code{inst-value}, and its protected
10332: words. It also has a pointer to its parent. @code{class} pushes
10333: the word lists of the class and all its ancestors onto the search order stack,
10334: and @code{end-class} drops them.
10335:
10336: @cindex interface implementation
10337: An interface is like a class without fields, parent and protected
10338: words; i.e., it just has a method map. If a class implements an
10339: interface, its method map contains a pointer to the method map of the
10340: interface. The positive offsets in the map are reserved for class
10341: methods, therefore interface map pointers have negative
10342: offsets. Interfaces have offsets that are unique throughout the
10343: system, unlike class selectors, whose offsets are only unique for the
10344: classes where the selector is available (invokable).
10345:
10346: This structure means that interface selectors have to perform one
10347: indirection more than class selectors to find their method. Their body
10348: contains the interface map pointer offset in the class method map, and
10349: the method offset in the interface method map. The
10350: @code{does>} action for an interface selector is, basically:
10351:
10352: @example
10353: ( object selector-body )
10354: 2dup selector-interface @@ ( object selector-body object interface-offset )
10355: swap object-map @@ + @@ ( object selector-body map )
10356: swap selector-offset @@ + @@ execute
10357: @end example
10358:
10359: where @code{object-map} and @code{selector-offset} are
10360: first fields and generate no code.
10361:
10362: As a concrete example, consider the following code:
10363:
10364: @example
10365: interface
10366: selector if1sel1
10367: selector if1sel2
10368: end-interface if1
10369:
10370: object class
10371: if1 implementation
10372: selector cl1sel1
10373: cell% inst-var cl1iv1
10374:
10375: ' m1 overrides construct
10376: ' m2 overrides if1sel1
10377: ' m3 overrides if1sel2
10378: ' m4 overrides cl1sel2
10379: end-class cl1
10380:
10381: create obj1 object dict-new drop
10382: create obj2 cl1 dict-new drop
10383: @end example
10384:
10385: The data structure created by this code (including the data structure
10386: for @code{object}) is shown in the
10387: @uref{objects-implementation.eps,figure}, assuming a cell size of 4.
10388: @comment TODO add this diagram..
10389:
10390: @node Objects Glossary, , Objects Implementation, Objects
10391: @subsubsection @file{objects.fs} Glossary
10392: @cindex @file{objects.fs} Glossary
10393:
10394:
10395: doc---objects-bind
10396: doc---objects-<bind>
10397: doc---objects-bind'
10398: doc---objects-[bind]
10399: doc---objects-class
10400: doc---objects-class->map
10401: doc---objects-class-inst-size
10402: doc---objects-class-override!
10403: doc---objects-construct
10404: doc---objects-current'
10405: doc---objects-[current]
10406: doc---objects-current-interface
10407: doc---objects-dict-new
10408: doc---objects-drop-order
10409: doc---objects-end-class
10410: doc---objects-end-class-noname
10411: doc---objects-end-interface
10412: doc---objects-end-interface-noname
10413: doc---objects-end-methods
10414: doc---objects-exitm
10415: doc---objects-heap-new
10416: doc---objects-implementation
10417: doc---objects-init-object
10418: doc---objects-inst-value
10419: doc---objects-inst-var
10420: doc---objects-interface
10421: doc---objects-m:
10422: doc---objects-:m
10423: doc---objects-;m
10424: doc---objects-method
10425: doc---objects-methods
10426: doc---objects-object
10427: doc---objects-overrides
10428: doc---objects-[parent]
10429: doc---objects-print
10430: doc---objects-protected
10431: doc---objects-public
10432: @c !! push-order conflicts
10433: doc---objects-push-order
10434: doc---objects-selector
10435: doc---objects-this
10436: doc---objects-<to-inst>
10437: doc---objects-[to-inst]
10438: doc---objects-to-this
10439: doc---objects-xt-new
10440:
10441:
10442: @c -------------------------------------------------------------
10443: @node OOF, Mini-OOF, Objects, Object-oriented Forth
10444: @subsection The @file{oof.fs} model
10445: @cindex oof
10446: @cindex object-oriented programming
10447:
10448: @cindex @file{objects.fs}
10449: @cindex @file{oof.fs}
10450:
10451: This section describes the @file{oof.fs} package.
10452:
10453: The package described in this section has been used in bigFORTH since 1991, and
10454: used for two large applications: a chromatographic system used to
10455: create new medicaments, and a graphic user interface library (MINOS).
10456:
10457: You can find a description (in German) of @file{oof.fs} in @cite{Object
10458: oriented bigFORTH} by Bernd Paysan, published in @cite{Vierte Dimension}
10459: 10(2), 1994.
10460:
10461: @menu
10462: * Properties of the OOF model::
10463: * Basic OOF Usage::
10464: * The OOF base class::
10465: * Class Declaration::
10466: * Class Implementation::
10467: @end menu
10468:
10469: @node Properties of the OOF model, Basic OOF Usage, OOF, OOF
10470: @subsubsection Properties of the @file{oof.fs} model
10471: @cindex @file{oof.fs} properties
10472:
10473: @itemize @bullet
10474: @item
10475: This model combines object oriented programming with information
10476: hiding. It helps you writing large application, where scoping is
10477: necessary, because it provides class-oriented scoping.
10478:
10479: @item
10480: Named objects, object pointers, and object arrays can be created,
10481: selector invocation uses the ``object selector'' syntax. Selector invocation
10482: to objects and/or selectors on the stack is a bit less convenient, but
10483: possible.
10484:
10485: @item
10486: Selector invocation and instance variable usage of the active object is
10487: straightforward, since both make use of the active object.
10488:
10489: @item
10490: Late binding is efficient and easy to use.
10491:
10492: @item
10493: State-smart objects parse selectors. However, extensibility is provided
10494: using a (parsing) selector @code{postpone} and a selector @code{'}.
10495:
10496: @item
10497: An implementation in ANS Forth is available.
10498:
10499: @end itemize
10500:
10501:
10502: @node Basic OOF Usage, The OOF base class, Properties of the OOF model, OOF
10503: @subsubsection Basic @file{oof.fs} Usage
10504: @cindex @file{oof.fs} usage
10505:
10506: This section uses the same example as for @code{objects} (@pxref{Basic Objects Usage}).
10507:
10508: You can define a class for graphical objects like this:
10509:
10510: @cindex @code{class} usage
10511: @cindex @code{class;} usage
10512: @cindex @code{method} usage
10513: @example
10514: object class graphical \ "object" is the parent class
10515: method draw ( x y graphical -- )
10516: class;
10517: @end example
10518:
10519: This code defines a class @code{graphical} with an
10520: operation @code{draw}. We can perform the operation
10521: @code{draw} on any @code{graphical} object, e.g.:
10522:
10523: @example
10524: 100 100 t-rex draw
10525: @end example
10526:
10527: @noindent
10528: where @code{t-rex} is an object or object pointer, created with e.g.
10529: @code{graphical : t-rex}.
10530:
10531: @cindex abstract class
10532: How do we create a graphical object? With the present definitions,
10533: we cannot create a useful graphical object. The class
10534: @code{graphical} describes graphical objects in general, but not
10535: any concrete graphical object type (C++ users would call it an
10536: @emph{abstract class}); e.g., there is no method for the selector
10537: @code{draw} in the class @code{graphical}.
10538:
10539: For concrete graphical objects, we define child classes of the
10540: class @code{graphical}, e.g.:
10541:
10542: @example
10543: graphical class circle \ "graphical" is the parent class
10544: cell var circle-radius
10545: how:
10546: : draw ( x y -- )
10547: circle-radius @@ draw-circle ;
10548:
10549: : init ( n-radius -- (
10550: circle-radius ! ;
10551: class;
10552: @end example
10553:
10554: Here we define a class @code{circle} as a child of @code{graphical},
10555: with a field @code{circle-radius}; it defines new methods for the
10556: selectors @code{draw} and @code{init} (@code{init} is defined in
10557: @code{object}, the parent class of @code{graphical}).
10558:
10559: Now we can create a circle in the dictionary with:
10560:
10561: @example
10562: 50 circle : my-circle
10563: @end example
10564:
10565: @noindent
10566: @code{:} invokes @code{init}, thus initializing the field
10567: @code{circle-radius} with 50. We can draw this new circle at (100,100)
10568: with:
10569:
10570: @example
10571: 100 100 my-circle draw
10572: @end example
10573:
10574: @cindex selector invocation, restrictions
10575: @cindex class definition, restrictions
10576: Note: You can only invoke a selector if the receiving object belongs to
10577: the class where the selector was defined or one of its descendents;
10578: e.g., you can invoke @code{draw} only for objects belonging to
10579: @code{graphical} or its descendents (e.g., @code{circle}). The scoping
10580: mechanism will check if you try to invoke a selector that is not
10581: defined in this class hierarchy, so you'll get an error at compilation
10582: time.
10583:
10584:
10585: @node The OOF base class, Class Declaration, Basic OOF Usage, OOF
10586: @subsubsection The @file{oof.fs} base class
10587: @cindex @file{oof.fs} base class
10588:
10589: When you define a class, you have to specify a parent class. So how do
10590: you start defining classes? There is one class available from the start:
10591: @code{object}. You have to use it as ancestor for all classes. It is the
10592: only class that has no parent. Classes are also objects, except that
10593: they don't have instance variables; class manipulation such as
10594: inheritance or changing definitions of a class is handled through
10595: selectors of the class @code{object}.
10596:
10597: @code{object} provides a number of selectors:
10598:
10599: @itemize @bullet
10600: @item
10601: @code{class} for subclassing, @code{definitions} to add definitions
10602: later on, and @code{class?} to get type informations (is the class a
10603: subclass of the class passed on the stack?).
10604:
10605: doc---object-class
10606: doc---object-definitions
10607: doc---object-class?
10608:
10609:
10610: @item
10611: @code{init} and @code{dispose} as constructor and destructor of the
10612: object. @code{init} is invocated after the object's memory is allocated,
10613: while @code{dispose} also handles deallocation. Thus if you redefine
10614: @code{dispose}, you have to call the parent's dispose with @code{super
10615: dispose}, too.
10616:
10617: doc---object-init
10618: doc---object-dispose
10619:
10620:
10621: @item
10622: @code{new}, @code{new[]}, @code{:}, @code{ptr}, @code{asptr}, and
10623: @code{[]} to create named and unnamed objects and object arrays or
10624: object pointers.
10625:
10626: doc---object-new
10627: doc---object-new[]
10628: doc---object-:
10629: doc---object-ptr
10630: doc---object-asptr
10631: doc---object-[]
10632:
10633:
10634: @item
10635: @code{::} and @code{super} for explicit scoping. You should use explicit
10636: scoping only for super classes or classes with the same set of instance
10637: variables. Explicitly-scoped selectors use early binding.
10638:
10639: doc---object-::
10640: doc---object-super
10641:
10642:
10643: @item
10644: @code{self} to get the address of the object
10645:
10646: doc---object-self
10647:
10648:
10649: @item
10650: @code{bind}, @code{bound}, @code{link}, and @code{is} to assign object
10651: pointers and instance defers.
10652:
10653: doc---object-bind
10654: doc---object-bound
10655: doc---object-link
10656: doc---object-is
10657:
10658:
10659: @item
10660: @code{'} to obtain selector tokens, @code{send} to invocate selectors
10661: form the stack, and @code{postpone} to generate selector invocation code.
10662:
10663: doc---object-'
10664: doc---object-postpone
10665:
10666:
10667: @item
10668: @code{with} and @code{endwith} to select the active object from the
10669: stack, and enable its scope. Using @code{with} and @code{endwith}
10670: also allows you to create code using selector @code{postpone} without being
10671: trapped by the state-smart objects.
10672:
10673: doc---object-with
10674: doc---object-endwith
10675:
10676:
10677: @end itemize
10678:
10679: @node Class Declaration, Class Implementation, The OOF base class, OOF
10680: @subsubsection Class Declaration
10681: @cindex class declaration
10682:
10683: @itemize @bullet
10684: @item
10685: Instance variables
10686:
10687: doc---oof-var
10688:
10689:
10690: @item
10691: Object pointers
10692:
10693: doc---oof-ptr
10694: doc---oof-asptr
10695:
10696:
10697: @item
10698: Instance defers
10699:
10700: doc---oof-defer
10701:
10702:
10703: @item
10704: Method selectors
10705:
10706: doc---oof-early
10707: doc---oof-method
10708:
10709:
10710: @item
10711: Class-wide variables
10712:
10713: doc---oof-static
10714:
10715:
10716: @item
10717: End declaration
10718:
10719: doc---oof-how:
10720: doc---oof-class;
10721:
10722:
10723: @end itemize
10724:
10725: @c -------------------------------------------------------------
10726: @node Class Implementation, , Class Declaration, OOF
10727: @subsubsection Class Implementation
10728: @cindex class implementation
10729:
10730: @c -------------------------------------------------------------
10731: @node Mini-OOF, Comparison with other object models, OOF, Object-oriented Forth
10732: @subsection The @file{mini-oof.fs} model
10733: @cindex mini-oof
10734:
10735: Gforth's third object oriented Forth package is a 12-liner. It uses a
10736: mixture of the @file{object.fs} and the @file{oof.fs} syntax,
10737: and reduces to the bare minimum of features. This is based on a posting
10738: of Bernd Paysan in comp.lang.forth.
10739:
10740: @menu
10741: * Basic Mini-OOF Usage::
10742: * Mini-OOF Example::
10743: * Mini-OOF Implementation::
10744: @end menu
10745:
10746: @c -------------------------------------------------------------
10747: @node Basic Mini-OOF Usage, Mini-OOF Example, Mini-OOF, Mini-OOF
10748: @subsubsection Basic @file{mini-oof.fs} Usage
10749: @cindex mini-oof usage
10750:
10751: There is a base class (@code{class}, which allocates one cell for the
10752: object pointer) plus seven other words: to define a method, a variable,
10753: a class; to end a class, to resolve binding, to allocate an object and
10754: to compile a class method.
10755: @comment TODO better description of the last one
10756:
10757:
10758: doc-object
10759: doc-method
10760: doc-var
10761: doc-class
10762: doc-end-class
10763: doc-defines
10764: doc-new
10765: doc-::
10766:
10767:
10768:
10769: @c -------------------------------------------------------------
10770: @node Mini-OOF Example, Mini-OOF Implementation, Basic Mini-OOF Usage, Mini-OOF
10771: @subsubsection Mini-OOF Example
10772: @cindex mini-oof example
10773:
10774: A short example shows how to use this package. This example, in slightly
10775: extended form, is supplied as @file{moof-exm.fs}
10776: @comment TODO could flesh this out with some comments from the Forthwrite article
10777:
10778: @example
10779: object class
10780: method init
10781: method draw
10782: end-class graphical
10783: @end example
10784:
10785: This code defines a class @code{graphical} with an
10786: operation @code{draw}. We can perform the operation
10787: @code{draw} on any @code{graphical} object, e.g.:
10788:
10789: @example
10790: 100 100 t-rex draw
10791: @end example
10792:
10793: where @code{t-rex} is an object or object pointer, created with e.g.
10794: @code{graphical new Constant t-rex}.
10795:
10796: For concrete graphical objects, we define child classes of the
10797: class @code{graphical}, e.g.:
10798:
10799: @example
10800: graphical class
10801: cell var circle-radius
10802: end-class circle \ "graphical" is the parent class
10803:
10804: :noname ( x y -- )
10805: circle-radius @@ draw-circle ; circle defines draw
10806: :noname ( r -- )
10807: circle-radius ! ; circle defines init
10808: @end example
10809:
10810: There is no implicit init method, so we have to define one. The creation
10811: code of the object now has to call init explicitely.
10812:
10813: @example
10814: circle new Constant my-circle
10815: 50 my-circle init
10816: @end example
10817:
10818: It is also possible to add a function to create named objects with
10819: automatic call of @code{init}, given that all objects have @code{init}
10820: on the same place:
10821:
10822: @example
10823: : new: ( .. o "name" -- )
10824: new dup Constant init ;
10825: 80 circle new: large-circle
10826: @end example
10827:
10828: We can draw this new circle at (100,100) with:
10829:
10830: @example
10831: 100 100 my-circle draw
10832: @end example
10833:
10834: @node Mini-OOF Implementation, , Mini-OOF Example, Mini-OOF
10835: @subsubsection @file{mini-oof.fs} Implementation
10836:
10837: Object-oriented systems with late binding typically use a
10838: ``vtable''-approach: the first variable in each object is a pointer to a
10839: table, which contains the methods as function pointers. The vtable
10840: may also contain other information.
10841:
10842: So first, let's declare methods:
10843:
10844: @example
10845: : method ( m v -- m' v ) Create over , swap cell+ swap
10846: DOES> ( ... o -- ... ) @@ over @@ + @@ execute ;
10847: @end example
10848:
10849: During method declaration, the number of methods and instance
10850: variables is on the stack (in address units). @code{method} creates
10851: one method and increments the method number. To execute a method, it
10852: takes the object, fetches the vtable pointer, adds the offset, and
10853: executes the @i{xt} stored there. Each method takes the object it is
10854: invoked from as top of stack parameter. The method itself should
10855: consume that object.
10856:
10857: Now, we also have to declare instance variables
10858:
10859: @example
10860: : var ( m v size -- m v' ) Create over , +
10861: DOES> ( o -- addr ) @@ + ;
10862: @end example
10863:
10864: As before, a word is created with the current offset. Instance
10865: variables can have different sizes (cells, floats, doubles, chars), so
10866: all we do is take the size and add it to the offset. If your machine
10867: has alignment restrictions, put the proper @code{aligned} or
10868: @code{faligned} before the variable, to adjust the variable
10869: offset. That's why it is on the top of stack.
10870:
10871: We need a starting point (the base object) and some syntactic sugar:
10872:
10873: @example
10874: Create object 1 cells , 2 cells ,
10875: : class ( class -- class methods vars ) dup 2@@ ;
10876: @end example
10877:
10878: For inheritance, the vtable of the parent object has to be
10879: copied when a new, derived class is declared. This gives all the
10880: methods of the parent class, which can be overridden, though.
10881:
10882: @example
10883: : end-class ( class methods vars -- )
10884: Create here >r , dup , 2 cells ?DO ['] noop , 1 cells +LOOP
10885: cell+ dup cell+ r> rot @@ 2 cells /string move ;
10886: @end example
10887:
10888: The first line creates the vtable, initialized with
10889: @code{noop}s. The second line is the inheritance mechanism, it
10890: copies the xts from the parent vtable.
10891:
10892: We still have no way to define new methods, let's do that now:
10893:
10894: @example
10895: : defines ( xt class -- ) ' >body @@ + ! ;
10896: @end example
10897:
10898: To allocate a new object, we need a word, too:
10899:
10900: @example
10901: : new ( class -- o ) here over @@ allot swap over ! ;
10902: @end example
10903:
10904: Sometimes derived classes want to access the method of the
10905: parent object. There are two ways to achieve this with Mini-OOF:
10906: first, you could use named words, and second, you could look up the
10907: vtable of the parent object.
10908:
10909: @example
10910: : :: ( class "name" -- ) ' >body @@ + @@ compile, ;
10911: @end example
10912:
10913:
10914: Nothing can be more confusing than a good example, so here is
10915: one. First let's declare a text object (called
10916: @code{button}), that stores text and position:
10917:
10918: @example
10919: object class
10920: cell var text
10921: cell var len
10922: cell var x
10923: cell var y
10924: method init
10925: method draw
10926: end-class button
10927: @end example
10928:
10929: @noindent
10930: Now, implement the two methods, @code{draw} and @code{init}:
10931:
10932: @example
10933: :noname ( o -- )
10934: >r r@@ x @@ r@@ y @@ at-xy r@@ text @@ r> len @@ type ;
10935: button defines draw
10936: :noname ( addr u o -- )
10937: >r 0 r@@ x ! 0 r@@ y ! r@@ len ! r> text ! ;
10938: button defines init
10939: @end example
10940:
10941: @noindent
10942: To demonstrate inheritance, we define a class @code{bold-button}, with no
10943: new data and no new methods:
10944:
10945: @example
10946: button class
10947: end-class bold-button
10948:
10949: : bold 27 emit ." [1m" ;
10950: : normal 27 emit ." [0m" ;
10951: @end example
10952:
10953: @noindent
10954: The class @code{bold-button} has a different draw method to
10955: @code{button}, but the new method is defined in terms of the draw method
10956: for @code{button}:
10957:
10958: @example
10959: :noname bold [ button :: draw ] normal ; bold-button defines draw
10960: @end example
10961:
10962: @noindent
10963: Finally, create two objects and apply methods:
10964:
10965: @example
10966: button new Constant foo
10967: s" thin foo" foo init
10968: page
10969: foo draw
10970: bold-button new Constant bar
10971: s" fat bar" bar init
10972: 1 bar y !
10973: bar draw
10974: @end example
10975:
10976:
10977: @node Comparison with other object models, , Mini-OOF, Object-oriented Forth
10978: @subsection Comparison with other object models
10979: @cindex comparison of object models
10980: @cindex object models, comparison
10981:
10982: Many object-oriented Forth extensions have been proposed (@cite{A survey
10983: of object-oriented Forths} (SIGPLAN Notices, April 1996) by Bradford
10984: J. Rodriguez and W. F. S. Poehlman lists 17). This section discusses the
10985: relation of the object models described here to two well-known and two
10986: closely-related (by the use of method maps) models. Andras Zsoter
10987: helped us with this section.
10988:
10989: @cindex Neon model
10990: The most popular model currently seems to be the Neon model (see
10991: @cite{Object-oriented programming in ANS Forth} (Forth Dimensions, March
10992: 1997) by Andrew McKewan) but this model has a number of limitations
10993: @footnote{A longer version of this critique can be
10994: found in @cite{On Standardizing Object-Oriented Forth Extensions} (Forth
10995: Dimensions, May 1997) by Anton Ertl.}:
10996:
10997: @itemize @bullet
10998: @item
10999: It uses a @code{@emph{selector object}} syntax, which makes it unnatural
11000: to pass objects on the stack.
11001:
11002: @item
11003: It requires that the selector parses the input stream (at
11004: compile time); this leads to reduced extensibility and to bugs that are+
11005: hard to find.
11006:
11007: @item
11008: It allows using every selector to every object;
11009: this eliminates the need for classes, but makes it harder to create
11010: efficient implementations.
11011: @end itemize
11012:
11013: @cindex Pountain's object-oriented model
11014: Another well-known publication is @cite{Object-Oriented Forth} (Academic
11015: Press, London, 1987) by Dick Pountain. However, it is not really about
11016: object-oriented programming, because it hardly deals with late
11017: binding. Instead, it focuses on features like information hiding and
11018: overloading that are characteristic of modular languages like Ada (83).
11019:
11020: @cindex Zsoter's object-oriented model
11021: In @cite{Does late binding have to be slow?} (Forth Dimensions 18(1)
11022: 1996, pages 31-35) Andras Zsoter describes a model that makes heavy use
11023: of an active object (like @code{this} in @file{objects.fs}): The active
11024: object is not only used for accessing all fields, but also specifies the
11025: receiving object of every selector invocation; you have to change the
11026: active object explicitly with @code{@{ ... @}}, whereas in
11027: @file{objects.fs} it changes more or less implicitly at @code{m:
11028: ... ;m}. Such a change at the method entry point is unnecessary with the
11029: Zsoter's model, because the receiving object is the active object
11030: already. On the other hand, the explicit change is absolutely necessary
11031: in that model, because otherwise no one could ever change the active
11032: object. An ANS Forth implementation of this model is available at
11033: @uref{http://www.forth.org/fig/oopf.html}.
11034:
11035: @cindex @file{oof.fs}, differences to other models
11036: The @file{oof.fs} model combines information hiding and overloading
11037: resolution (by keeping names in various word lists) with object-oriented
11038: programming. It sets the active object implicitly on method entry, but
11039: also allows explicit changing (with @code{>o...o>} or with
11040: @code{with...endwith}). It uses parsing and state-smart objects and
11041: classes for resolving overloading and for early binding: the object or
11042: class parses the selector and determines the method from this. If the
11043: selector is not parsed by an object or class, it performs a call to the
11044: selector for the active object (late binding), like Zsoter's model.
11045: Fields are always accessed through the active object. The big
11046: disadvantage of this model is the parsing and the state-smartness, which
11047: reduces extensibility and increases the opportunities for subtle bugs;
11048: essentially, you are only safe if you never tick or @code{postpone} an
11049: object or class (Bernd disagrees, but I (Anton) am not convinced).
11050:
11051: @cindex @file{mini-oof.fs}, differences to other models
11052: The @file{mini-oof.fs} model is quite similar to a very stripped-down
11053: version of the @file{objects.fs} model, but syntactically it is a
11054: mixture of the @file{objects.fs} and @file{oof.fs} models.
11055:
11056:
11057: @c -------------------------------------------------------------
11058: @node Programming Tools, Assembler and Code Words, Object-oriented Forth, Words
11059: @section Programming Tools
11060: @cindex programming tools
11061:
11062: @c !! move this and assembler down below OO stuff.
11063:
11064: @menu
11065: * Examining::
11066: * Forgetting words::
11067: * Debugging:: Simple and quick.
11068: * Assertions:: Making your programs self-checking.
11069: * Singlestep Debugger:: Executing your program word by word.
11070: @end menu
11071:
11072: @node Examining, Forgetting words, Programming Tools, Programming Tools
11073: @subsection Examining data and code
11074: @cindex examining data and code
11075: @cindex data examination
11076: @cindex code examination
11077:
11078: The following words inspect the stack non-destructively:
11079:
11080: doc-.s
11081: doc-f.s
11082:
11083: There is a word @code{.r} but it does @i{not} display the return stack!
11084: It is used for formatted numeric output (@pxref{Simple numeric output}).
11085:
11086: doc-depth
11087: doc-fdepth
11088: doc-clearstack
11089:
11090: The following words inspect memory.
11091:
11092: doc-?
11093: doc-dump
11094:
11095: And finally, @code{see} allows to inspect code:
11096:
11097: doc-see
11098: doc-xt-see
11099:
11100: @node Forgetting words, Debugging, Examining, Programming Tools
11101: @subsection Forgetting words
11102: @cindex words, forgetting
11103: @cindex forgeting words
11104:
11105: @c anton: other, maybe better places for this subsection: Defining Words;
11106: @c Dictionary allocation. At least a reference should be there.
11107:
11108: Forth allows you to forget words (and everything that was alloted in the
11109: dictonary after them) in a LIFO manner.
11110:
11111: doc-marker
11112:
11113: The most common use of this feature is during progam development: when
11114: you change a source file, forget all the words it defined and load it
11115: again (since you also forget everything defined after the source file
11116: was loaded, you have to reload that, too). Note that effects like
11117: storing to variables and destroyed system words are not undone when you
11118: forget words. With a system like Gforth, that is fast enough at
11119: starting up and compiling, I find it more convenient to exit and restart
11120: Gforth, as this gives me a clean slate.
11121:
11122: Here's an example of using @code{marker} at the start of a source file
11123: that you are debugging; it ensures that you only ever have one copy of
11124: the file's definitions compiled at any time:
11125:
11126: @example
11127: [IFDEF] my-code
11128: my-code
11129: [ENDIF]
11130:
11131: marker my-code
11132: init-included-files
11133:
11134: \ .. definitions start here
11135: \ .
11136: \ .
11137: \ end
11138: @end example
11139:
11140:
11141: @node Debugging, Assertions, Forgetting words, Programming Tools
11142: @subsection Debugging
11143: @cindex debugging
11144:
11145: Languages with a slow edit/compile/link/test development loop tend to
11146: require sophisticated tracing/stepping debuggers to facilate debugging.
11147:
11148: A much better (faster) way in fast-compiling languages is to add
11149: printing code at well-selected places, let the program run, look at
11150: the output, see where things went wrong, add more printing code, etc.,
11151: until the bug is found.
11152:
11153: The simple debugging aids provided in @file{debugs.fs}
11154: are meant to support this style of debugging.
11155:
11156: The word @code{~~} prints debugging information (by default the source
11157: location and the stack contents). It is easy to insert. If you use Emacs
11158: it is also easy to remove (@kbd{C-x ~} in the Emacs Forth mode to
11159: query-replace them with nothing). The deferred words
11160: @code{printdebugdata} and @code{printdebugline} control the output of
11161: @code{~~}. The default source location output format works well with
11162: Emacs' compilation mode, so you can step through the program at the
11163: source level using @kbd{C-x `} (the advantage over a stepping debugger
11164: is that you can step in any direction and you know where the crash has
11165: happened or where the strange data has occurred).
11166:
11167: doc-~~
11168: doc-printdebugdata
11169: doc-printdebugline
11170:
11171: @node Assertions, Singlestep Debugger, Debugging, Programming Tools
11172: @subsection Assertions
11173: @cindex assertions
11174:
11175: It is a good idea to make your programs self-checking, especially if you
11176: make an assumption that may become invalid during maintenance (for
11177: example, that a certain field of a data structure is never zero). Gforth
11178: supports @dfn{assertions} for this purpose. They are used like this:
11179:
11180: @example
11181: assert( @i{flag} )
11182: @end example
11183:
11184: The code between @code{assert(} and @code{)} should compute a flag, that
11185: should be true if everything is alright and false otherwise. It should
11186: not change anything else on the stack. The overall stack effect of the
11187: assertion is @code{( -- )}. E.g.
11188:
11189: @example
11190: assert( 1 1 + 2 = ) \ what we learn in school
11191: assert( dup 0<> ) \ assert that the top of stack is not zero
11192: assert( false ) \ this code should not be reached
11193: @end example
11194:
11195: The need for assertions is different at different times. During
11196: debugging, we want more checking, in production we sometimes care more
11197: for speed. Therefore, assertions can be turned off, i.e., the assertion
11198: becomes a comment. Depending on the importance of an assertion and the
11199: time it takes to check it, you may want to turn off some assertions and
11200: keep others turned on. Gforth provides several levels of assertions for
11201: this purpose:
11202:
11203:
11204: doc-assert0(
11205: doc-assert1(
11206: doc-assert2(
11207: doc-assert3(
11208: doc-assert(
11209: doc-)
11210:
11211:
11212: The variable @code{assert-level} specifies the highest assertions that
11213: are turned on. I.e., at the default @code{assert-level} of one,
11214: @code{assert0(} and @code{assert1(} assertions perform checking, while
11215: @code{assert2(} and @code{assert3(} assertions are treated as comments.
11216:
11217: The value of @code{assert-level} is evaluated at compile-time, not at
11218: run-time. Therefore you cannot turn assertions on or off at run-time;
11219: you have to set the @code{assert-level} appropriately before compiling a
11220: piece of code. You can compile different pieces of code at different
11221: @code{assert-level}s (e.g., a trusted library at level 1 and
11222: newly-written code at level 3).
11223:
11224:
11225: doc-assert-level
11226:
11227:
11228: If an assertion fails, a message compatible with Emacs' compilation mode
11229: is produced and the execution is aborted (currently with @code{ABORT"}.
11230: If there is interest, we will introduce a special throw code. But if you
11231: intend to @code{catch} a specific condition, using @code{throw} is
11232: probably more appropriate than an assertion).
11233:
11234: Definitions in ANS Forth for these assertion words are provided
11235: in @file{compat/assert.fs}.
11236:
11237:
11238: @node Singlestep Debugger, , Assertions, Programming Tools
11239: @subsection Singlestep Debugger
11240: @cindex singlestep Debugger
11241: @cindex debugging Singlestep
11242:
11243: When you create a new word there's often the need to check whether it
11244: behaves correctly or not. You can do this by typing @code{dbg
11245: badword}. A debug session might look like this:
11246:
11247: @example
11248: : badword 0 DO i . LOOP ; ok
11249: 2 dbg badword
11250: : badword
11251: Scanning code...
11252:
11253: Nesting debugger ready!
11254:
11255: 400D4738 8049BC4 0 -> [ 2 ] 00002 00000
11256: 400D4740 8049F68 DO -> [ 0 ]
11257: 400D4744 804A0C8 i -> [ 1 ] 00000
11258: 400D4748 400C5E60 . -> 0 [ 0 ]
11259: 400D474C 8049D0C LOOP -> [ 0 ]
11260: 400D4744 804A0C8 i -> [ 1 ] 00001
11261: 400D4748 400C5E60 . -> 1 [ 0 ]
11262: 400D474C 8049D0C LOOP -> [ 0 ]
11263: 400D4758 804B384 ; -> ok
11264: @end example
11265:
11266: Each line displayed is one step. You always have to hit return to
11267: execute the next word that is displayed. If you don't want to execute
11268: the next word in a whole, you have to type @kbd{n} for @code{nest}. Here is
11269: an overview what keys are available:
11270:
11271: @table @i
11272:
11273: @item @key{RET}
11274: Next; Execute the next word.
11275:
11276: @item n
11277: Nest; Single step through next word.
11278:
11279: @item u
11280: Unnest; Stop debugging and execute rest of word. If we got to this word
11281: with nest, continue debugging with the calling word.
11282:
11283: @item d
11284: Done; Stop debugging and execute rest.
11285:
11286: @item s
11287: Stop; Abort immediately.
11288:
11289: @end table
11290:
11291: Debugging large application with this mechanism is very difficult, because
11292: you have to nest very deeply into the program before the interesting part
11293: begins. This takes a lot of time.
11294:
11295: To do it more directly put a @code{BREAK:} command into your source code.
11296: When program execution reaches @code{BREAK:} the single step debugger is
11297: invoked and you have all the features described above.
11298:
11299: If you have more than one part to debug it is useful to know where the
11300: program has stopped at the moment. You can do this by the
11301: @code{BREAK" string"} command. This behaves like @code{BREAK:} except that
11302: string is typed out when the ``breakpoint'' is reached.
11303:
11304:
11305: doc-dbg
11306: doc-break:
11307: doc-break"
11308:
11309:
11310:
11311: @c -------------------------------------------------------------
11312: @node Assembler and Code Words, Threading Words, Programming Tools, Words
11313: @section Assembler and Code Words
11314: @cindex assembler
11315: @cindex code words
11316:
11317: @menu
11318: * Code and ;code::
11319: * Common Assembler:: Assembler Syntax
11320: * Common Disassembler::
11321: * 386 Assembler:: Deviations and special cases
11322: * Alpha Assembler:: Deviations and special cases
11323: * MIPS assembler:: Deviations and special cases
11324: * Other assemblers:: How to write them
11325: @end menu
11326:
11327: @node Code and ;code, Common Assembler, Assembler and Code Words, Assembler and Code Words
11328: @subsection @code{Code} and @code{;code}
11329:
11330: Gforth provides some words for defining primitives (words written in
11331: machine code), and for defining the machine-code equivalent of
11332: @code{DOES>}-based defining words. However, the machine-independent
11333: nature of Gforth poses a few problems: First of all, Gforth runs on
11334: several architectures, so it can provide no standard assembler. What's
11335: worse is that the register allocation not only depends on the processor,
11336: but also on the @code{gcc} version and options used.
11337:
11338: The words that Gforth offers encapsulate some system dependences (e.g.,
11339: the header structure), so a system-independent assembler may be used in
11340: Gforth. If you do not have an assembler, you can compile machine code
11341: directly with @code{,} and @code{c,}@footnote{This isn't portable,
11342: because these words emit stuff in @i{data} space; it works because
11343: Gforth has unified code/data spaces. Assembler isn't likely to be
11344: portable anyway.}.
11345:
11346:
11347: doc-assembler
11348: doc-init-asm
11349: doc-code
11350: doc-end-code
11351: doc-;code
11352: doc-flush-icache
11353:
11354:
11355: If @code{flush-icache} does not work correctly, @code{code} words
11356: etc. will not work (reliably), either.
11357:
11358: The typical usage of these @code{code} words can be shown most easily by
11359: analogy to the equivalent high-level defining words:
11360:
11361: @example
11362: : foo code foo
11363: <high-level Forth words> <assembler>
11364: ; end-code
11365:
11366: : bar : bar
11367: <high-level Forth words> <high-level Forth words>
11368: CREATE CREATE
11369: <high-level Forth words> <high-level Forth words>
11370: DOES> ;code
11371: <high-level Forth words> <assembler>
11372: ; end-code
11373: @end example
11374:
11375: @c anton: the following stuff is also in "Common Assembler", in less detail.
11376:
11377: @cindex registers of the inner interpreter
11378: In the assembly code you will want to refer to the inner interpreter's
11379: registers (e.g., the data stack pointer) and you may want to use other
11380: registers for temporary storage. Unfortunately, the register allocation
11381: is installation-dependent.
11382:
11383: In particular, @code{ip} (Forth instruction pointer) and @code{rp}
11384: (return stack pointer) are in different places in @code{gforth} and
11385: @code{gforth-fast}. This means that you cannot write a @code{NEXT}
11386: routine that works on both versions; so for doing @code{NEXT}, I
11387: recomment jumping to @code{' noop >code-address}, which contains nothing
11388: but a @code{NEXT}.
11389:
11390: For general accesses to the inner interpreter's registers, the easiest
11391: solution is to use explicit register declarations (@pxref{Explicit Reg
11392: Vars, , Variables in Specified Registers, gcc.info, GNU C Manual}) for
11393: all of the inner interpreter's registers: You have to compile Gforth
11394: with @code{-DFORCE_REG} (configure option @code{--enable-force-reg}) and
11395: the appropriate declarations must be present in the @code{machine.h}
11396: file (see @code{mips.h} for an example; you can find a full list of all
11397: declarable register symbols with @code{grep register engine.c}). If you
11398: give explicit registers to all variables that are declared at the
11399: beginning of @code{engine()}, you should be able to use the other
11400: caller-saved registers for temporary storage. Alternatively, you can use
11401: the @code{gcc} option @code{-ffixed-REG} (@pxref{Code Gen Options, ,
11402: Options for Code Generation Conventions, gcc.info, GNU C Manual}) to
11403: reserve a register (however, this restriction on register allocation may
11404: slow Gforth significantly).
11405:
11406: If this solution is not viable (e.g., because @code{gcc} does not allow
11407: you to explicitly declare all the registers you need), you have to find
11408: out by looking at the code where the inner interpreter's registers
11409: reside and which registers can be used for temporary storage. You can
11410: get an assembly listing of the engine's code with @code{make engine.s}.
11411:
11412: In any case, it is good practice to abstract your assembly code from the
11413: actual register allocation. E.g., if the data stack pointer resides in
11414: register @code{$17}, create an alias for this register called @code{sp},
11415: and use that in your assembly code.
11416:
11417: @cindex code words, portable
11418: Another option for implementing normal and defining words efficiently
11419: is to add the desired functionality to the source of Gforth. For normal
11420: words you just have to edit @file{primitives} (@pxref{Automatic
11421: Generation}). Defining words (equivalent to @code{;CODE} words, for fast
11422: defined words) may require changes in @file{engine.c}, @file{kernel.fs},
11423: @file{prims2x.fs}, and possibly @file{cross.fs}.
11424:
11425: @node Common Assembler, Common Disassembler, Code and ;code, Assembler and Code Words
11426: @subsection Common Assembler
11427:
11428: The assemblers in Gforth generally use a postfix syntax, i.e., the
11429: instruction name follows the operands.
11430:
11431: The operands are passed in the usual order (the same that is used in the
11432: manual of the architecture). Since they all are Forth words, they have
11433: to be separated by spaces; you can also use Forth words to compute the
11434: operands.
11435:
11436: The instruction names usually end with a @code{,}. This makes it easier
11437: to visually separate instructions if you put several of them on one
11438: line; it also avoids shadowing other Forth words (e.g., @code{and}).
11439:
11440: Registers are usually specified by number; e.g., (decimal) @code{11}
11441: specifies registers R11 and F11 on the Alpha architecture (which one,
11442: depends on the instruction). The usual names are also available, e.g.,
11443: @code{s2} for R11 on Alpha.
11444:
11445: Control flow is specified similar to normal Forth code (@pxref{Arbitrary
11446: control structures}), with @code{if,}, @code{ahead,}, @code{then,},
11447: @code{begin,}, @code{until,}, @code{again,}, @code{cs-roll},
11448: @code{cs-pick}, @code{else,}, @code{while,}, and @code{repeat,}. The
11449: conditions are specified in a way specific to each assembler.
11450:
11451: Note that the register assignments of the Gforth engine can change
11452: between Gforth versions, or even between different compilations of the
11453: same Gforth version (e.g., if you use a different GCC version). So if
11454: you want to refer to Gforth's registers (e.g., the stack pointer or
11455: TOS), I recommend defining your own words for refering to these
11456: registers, and using them later on; then you can easily adapt to a
11457: changed register assignment. The stability of the register assignment
11458: is usually better if you build Gforth with @code{--enable-force-reg}.
11459:
11460: In particular, the return stack pointer and the instruction pointer are
11461: in memory in @code{gforth}, and usually in registers in
11462: @code{gforth-fast}. The most common use of these registers is to
11463: dispatch to the next word (the @code{next} routine). A portable way to
11464: do this is to jump to @code{' noop >code-address} (of course, this is
11465: less efficient than integrating the @code{next} code and scheduling it
11466: well).
11467:
11468: @node Common Disassembler, 386 Assembler, Common Assembler, Assembler and Code Words
11469: @subsection Common Disassembler
11470:
11471: You can disassemble a @code{code} word with @code{see}
11472: (@pxref{Debugging}). You can disassemble a section of memory with
11473:
11474: doc-disasm
11475:
11476: The disassembler generally produces output that can be fed into the
11477: assembler (i.e., same syntax, etc.). It also includes additional
11478: information in comments. In particular, the address of the instruction
11479: is given in a comment before the instruction.
11480:
11481: @code{See} may display more or less than the actual code of the word,
11482: because the recognition of the end of the code is unreliable. You can
11483: use @code{disasm} if it did not display enough. It may display more, if
11484: the code word is not immediately followed by a named word. If you have
11485: something else there, you can follow the word with @code{align last @ ,}
11486: to ensure that the end is recognized.
11487:
11488: @node 386 Assembler, Alpha Assembler, Common Disassembler, Assembler and Code Words
11489: @subsection 386 Assembler
11490:
11491: The 386 assembler included in Gforth was written by Bernd Paysan, it's
11492: available under GPL, and originally part of bigFORTH.
11493:
11494: The 386 disassembler included in Gforth was written by Andrew McKewan
11495: and is in the public domain.
11496:
11497: The disassembler displays code in prefix Intel syntax.
11498:
11499: The assembler uses a postfix syntax with reversed parameters.
11500:
11501: The assembler includes all instruction of the Athlon, i.e. 486 core
11502: instructions, Pentium and PPro extensions, floating point, MMX, 3Dnow!,
11503: but not ISSE. It's an integrated 16- and 32-bit assembler. Default is 32
11504: bit, you can switch to 16 bit with .86 and back to 32 bit with .386.
11505:
11506: There are several prefixes to switch between different operation sizes,
11507: @code{.b} for byte accesses, @code{.w} for word accesses, @code{.d} for
11508: double-word accesses. Addressing modes can be switched with @code{.wa}
11509: for 16 bit addresses, and @code{.da} for 32 bit addresses. You don't
11510: need a prefix for byte register names (@code{AL} et al).
11511:
11512: For floating point operations, the prefixes are @code{.fs} (IEEE
11513: single), @code{.fl} (IEEE double), @code{.fx} (extended), @code{.fw}
11514: (word), @code{.fd} (double-word), and @code{.fq} (quad-word).
11515:
11516: The MMX opcodes don't have size prefixes, they are spelled out like in
11517: the Intel assembler. Instead of move from and to memory, there are
11518: PLDQ/PLDD and PSTQ/PSTD.
11519:
11520: The registers lack the 'e' prefix; even in 32 bit mode, eax is called
11521: ax. Immediate values are indicated by postfixing them with @code{#},
11522: e.g., @code{3 #}. Here are some examples of addressing modes:
11523:
11524: @example
11525: 3 # \ immediate
11526: ax \ register
11527: 100 di d) \ 100[edi]
11528: 4 bx cx di) \ 4[ebx][ecx]
11529: di ax *4 i) \ [edi][eax*4]
11530: 20 ax *4 i#) \ 20[eax*4]
11531: @end example
11532:
11533: Some example of instructions are:
11534:
11535: @example
11536: ax bx mov \ move ebx,eax
11537: 3 # ax mov \ mov eax,3
11538: 100 di ) ax mov \ mov eax,100[edi]
11539: 4 bx cx di) ax mov \ mov eax,4[ebx][ecx]
11540: .w ax bx mov \ mov bx,ax
11541: @end example
11542:
11543: The following forms are supported for binary instructions:
11544:
11545: @example
11546: <reg> <reg> <inst>
11547: <n> # <reg> <inst>
11548: <mem> <reg> <inst>
11549: <reg> <mem> <inst>
11550: @end example
11551:
11552: Immediate to memory is not supported. The shift/rotate syntax is:
11553:
11554: @example
11555: <reg/mem> 1 # shl \ shortens to shift without immediate
11556: <reg/mem> 4 # shl
11557: <reg/mem> cl shl
11558: @end example
11559:
11560: Precede string instructions (@code{movs} etc.) with @code{.b} to get
11561: the byte version.
11562:
11563: The control structure words @code{IF} @code{UNTIL} etc. must be preceded
11564: by one of these conditions: @code{vs vc u< u>= 0= 0<> u<= u> 0< 0>= ps
11565: pc < >= <= >}. (Note that most of these words shadow some Forth words
11566: when @code{assembler} is in front of @code{forth} in the search path,
11567: e.g., in @code{code} words). Currently the control structure words use
11568: one stack item, so you have to use @code{roll} instead of @code{cs-roll}
11569: to shuffle them (you can also use @code{swap} etc.).
11570:
11571: Here is an example of a @code{code} word (assumes that the stack pointer
11572: is in esi and the TOS is in ebx):
11573:
11574: @example
11575: code my+ ( n1 n2 -- n )
11576: 4 si D) bx add
11577: 4 # si add
11578: Next
11579: end-code
11580: @end example
11581:
11582: @node Alpha Assembler, MIPS assembler, 386 Assembler, Assembler and Code Words
11583: @subsection Alpha Assembler
11584:
11585: The Alpha assembler and disassembler were originally written by Bernd
11586: Thallner.
11587:
11588: The register names @code{a0}--@code{a5} are not available to avoid
11589: shadowing hex numbers.
11590:
11591: Immediate forms of arithmetic instructions are distinguished by a
11592: @code{#} just before the @code{,}, e.g., @code{and#,} (note: @code{lda,}
11593: does not count as arithmetic instruction).
11594:
11595: You have to specify all operands to an instruction, even those that
11596: other assemblers consider optional, e.g., the destination register for
11597: @code{br,}, or the destination register and hint for @code{jmp,}.
11598:
11599: You can specify conditions for @code{if,} by removing the first @code{b}
11600: and the trailing @code{,} from a branch with a corresponding name; e.g.,
11601:
11602: @example
11603: 11 fgt if, \ if F11>0e
11604: ...
11605: endif,
11606: @end example
11607:
11608: @code{fbgt,} gives @code{fgt}.
11609:
11610: @node MIPS assembler, Other assemblers, Alpha Assembler, Assembler and Code Words
11611: @subsection MIPS assembler
11612:
11613: The MIPS assembler was originally written by Christian Pirker.
11614:
11615: Currently the assembler and disassembler only cover the MIPS-I
11616: architecture (R3000), and don't support FP instructions.
11617:
11618: The register names @code{$a0}--@code{$a3} are not available to avoid
11619: shadowing hex numbers.
11620:
11621: Because there is no way to distinguish registers from immediate values,
11622: you have to explicitly use the immediate forms of instructions, i.e.,
11623: @code{addiu,}, not just @code{addu,} (@command{as} does this
11624: implicitly).
11625:
11626: If the architecture manual specifies several formats for the instruction
11627: (e.g., for @code{jalr,}), you usually have to use the one with more
11628: arguments (i.e., two for @code{jalr,}). When in doubt, see
11629: @code{arch/mips/testasm.fs} for an example of correct use.
11630:
11631: Branches and jumps in the MIPS architecture have a delay slot. You have
11632: to fill it yourself (the simplest way is to use @code{nop,}), the
11633: assembler does not do it for you (unlike @command{as}). Even
11634: @code{if,}, @code{ahead,}, @code{until,}, @code{again,}, @code{while,},
11635: @code{else,} and @code{repeat,} need a delay slot. Since @code{begin,}
11636: and @code{then,} just specify branch targets, they are not affected.
11637:
11638: Note that you must not put branches, jumps, or @code{li,} into the delay
11639: slot: @code{li,} may expand to several instructions, and control flow
11640: instructions may not be put into the branch delay slot in any case.
11641:
11642: For branches the argument specifying the target is a relative address;
11643: You have to add the address of the delay slot to get the absolute
11644: address.
11645:
11646: The MIPS architecture also has load delay slots and restrictions on
11647: using @code{mfhi,} and @code{mflo,}; you have to order the instructions
11648: yourself to satisfy these restrictions, the assembler does not do it for
11649: you.
11650:
11651: You can specify the conditions for @code{if,} etc. by taking a
11652: conditional branch and leaving away the @code{b} at the start and the
11653: @code{,} at the end. E.g.,
11654:
11655: @example
11656: 4 5 eq if,
11657: ... \ do something if $4 equals $5
11658: then,
11659: @end example
11660:
11661: @node Other assemblers, , MIPS assembler, Assembler and Code Words
11662: @subsection Other assemblers
11663:
11664: If you want to contribute another assembler/disassembler, please contact
11665: us (@email{bug-gforth@@gnu.org}) to check if we have such an assembler
11666: already. If you are writing them from scratch, please use a similar
11667: syntax style as the one we use (i.e., postfix, commas at the end of the
11668: instruction names, @pxref{Common Assembler}); make the output of the
11669: disassembler be valid input for the assembler, and keep the style
11670: similar to the style we used.
11671:
11672: Hints on implementation: The most important part is to have a good test
11673: suite that contains all instructions. Once you have that, the rest is
11674: easy. For actual coding you can take a look at
11675: @file{arch/mips/disasm.fs} to get some ideas on how to use data for both
11676: the assembler and disassembler, avoiding redundancy and some potential
11677: bugs. You can also look at that file (and @pxref{Advanced does> usage
11678: example}) to get ideas how to factor a disassembler.
11679:
11680: Start with the disassembler, because it's easier to reuse data from the
11681: disassembler for the assembler than the other way round.
11682:
11683: For the assembler, take a look at @file{arch/alpha/asm.fs}, which shows
11684: how simple it can be.
11685:
11686: @c -------------------------------------------------------------
11687: @node Threading Words, Passing Commands to the OS, Assembler and Code Words, Words
11688: @section Threading Words
11689: @cindex threading words
11690:
11691: @cindex code address
11692: These words provide access to code addresses and other threading stuff
11693: in Gforth (and, possibly, other interpretive Forths). It more or less
11694: abstracts away the differences between direct and indirect threading
11695: (and, for direct threading, the machine dependences). However, at
11696: present this wordset is still incomplete. It is also pretty low-level;
11697: some day it will hopefully be made unnecessary by an internals wordset
11698: that abstracts implementation details away completely.
11699:
11700: The terminology used here stems from indirect threaded Forth systems; in
11701: such a system, the XT of a word is represented by the CFA (code field
11702: address) of a word; the CFA points to a cell that contains the code
11703: address. The code address is the address of some machine code that
11704: performs the run-time action of invoking the word (e.g., the
11705: @code{dovar:} routine pushes the address of the body of the word (a
11706: variable) on the stack
11707: ).
11708:
11709: @cindex code address
11710: @cindex code field address
11711: In an indirect threaded Forth, you can get the code address of @i{name}
11712: with @code{' @i{name} @@}; in Gforth you can get it with @code{' @i{name}
11713: >code-address}, independent of the threading method.
11714:
11715: doc-threading-method
11716: doc->code-address
11717: doc-code-address!
11718:
11719: @cindex @code{does>}-handler
11720: @cindex @code{does>}-code
11721: For a word defined with @code{DOES>}, the code address usually points to
11722: a jump instruction (the @dfn{does-handler}) that jumps to the dodoes
11723: routine (in Gforth on some platforms, it can also point to the dodoes
11724: routine itself). What you are typically interested in, though, is
11725: whether a word is a @code{DOES>}-defined word, and what Forth code it
11726: executes; @code{>does-code} tells you that.
11727:
11728: doc->does-code
11729:
11730: To create a @code{DOES>}-defined word with the following basic words,
11731: you have to set up a @code{DOES>}-handler with @code{does-handler!};
11732: @code{/does-handler} aus behind you have to place your executable Forth
11733: code. Finally you have to create a word and modify its behaviour with
11734: @code{does-handler!}.
11735:
11736: doc-does-code!
11737: doc-does-handler!
11738: doc-/does-handler
11739:
11740: The code addresses produced by various defining words are produced by
11741: the following words:
11742:
11743: doc-docol:
11744: doc-docon:
11745: doc-dovar:
11746: doc-douser:
11747: doc-dodefer:
11748: doc-dofield:
11749:
11750: @c -------------------------------------------------------------
11751: @node Passing Commands to the OS, Keeping track of Time, Threading Words, Words
11752: @section Passing Commands to the Operating System
11753: @cindex operating system - passing commands
11754: @cindex shell commands
11755:
11756: Gforth allows you to pass an arbitrary string to the host operating
11757: system shell (if such a thing exists) for execution.
11758:
11759:
11760: doc-sh
11761: doc-system
11762: doc-$?
11763: doc-getenv
11764:
11765:
11766: @c -------------------------------------------------------------
11767: @node Keeping track of Time, Miscellaneous Words, Passing Commands to the OS, Words
11768: @section Keeping track of Time
11769: @cindex time-related words
11770:
11771: Gforth implements time-related operations by making calls to the C
11772: library function, @code{gettimeofday}.
11773:
11774: doc-ms
11775: doc-time&date
11776:
11777:
11778:
11779: @c -------------------------------------------------------------
11780: @node Miscellaneous Words, , Keeping track of Time, Words
11781: @section Miscellaneous Words
11782: @cindex miscellaneous words
11783:
11784: @comment TODO find homes for these
11785:
11786: These section lists the ANS Forth words that are not documented
11787: elsewhere in this manual. Ultimately, they all need proper homes.
11788:
11789: doc-[compile]
11790: doc-quit
11791:
11792: The following ANS Forth words are not currently supported by Gforth
11793: (@pxref{ANS conformance}):
11794:
11795: @code{EDITOR}
11796: @code{EMIT?}
11797: @code{FORGET}
11798:
11799: @c ******************************************************************
11800: @node Error messages, Tools, Words, Top
11801: @chapter Error messages
11802: @cindex error messages
11803: @cindex backtrace
11804:
11805: A typical Gforth error message looks like this:
11806:
11807: @example
11808: in file included from :-1
11809: in file included from ./yyy.fs:1
11810: ./xxx.fs:4: Invalid memory address
11811: bar
11812: ^^^
11813: $400E664C @@
11814: $400E6664 foo
11815: @end example
11816:
11817: The message identifying the error is @code{Invalid memory address}. The
11818: error happened when text-interpreting line 4 of the file
11819: @file{./xxx.fs}. This line is given (it contains @code{bar}), and the
11820: word on the line where the error happened, is pointed out (with
11821: @code{^^^}).
11822:
11823: The file containing the error was included in line 1 of @file{./yyy.fs},
11824: and @file{yyy.fs} was included from a non-file (in this case, by giving
11825: @file{yyy.fs} as command-line parameter to Gforth).
11826:
11827: At the end of the error message you find a return stack dump that can be
11828: interpreted as a backtrace (possibly empty). On top you find the top of
11829: the return stack when the @code{throw} happened, and at the bottom you
11830: find the return stack entry just above the return stack of the topmost
11831: text interpreter.
11832:
11833: To the right of most return stack entries you see a guess for the word
11834: that pushed that return stack entry as its return address. This gives a
11835: backtrace. In our case we see that @code{bar} called @code{foo}, and
11836: @code{foo} called @code{@@} (and @code{@@} had an @emph{Invalid memory
11837: address} exception).
11838:
11839: Note that the backtrace is not perfect: We don't know which return stack
11840: entries are return addresses (so we may get false positives); and in
11841: some cases (e.g., for @code{abort"}) we cannot determine from the return
11842: address the word that pushed the return address, so for some return
11843: addresses you see no names in the return stack dump.
11844:
11845: @cindex @code{catch} and backtraces
11846: The return stack dump represents the return stack at the time when a
11847: specific @code{throw} was executed. In programs that make use of
11848: @code{catch}, it is not necessarily clear which @code{throw} should be
11849: used for the return stack dump (e.g., consider one @code{throw} that
11850: indicates an error, which is caught, and during recovery another error
11851: happens; which @code{throw} should be used for the stack dump?). Gforth
11852: presents the return stack dump for the first @code{throw} after the last
11853: executed (not returned-to) @code{catch}; this works well in the usual
11854: case.
11855:
11856: @cindex @code{gforth-fast} and backtraces
11857: @cindex @code{gforth-fast}, difference from @code{gforth}
11858: @cindex backtraces with @code{gforth-fast}
11859: @cindex return stack dump with @code{gforth-fast}
11860: @code{gforth} is able to do a return stack dump for throws generated
11861: from primitives (e.g., invalid memory address, stack empty etc.);
11862: @code{gforth-fast} is only able to do a return stack dump from a
11863: directly called @code{throw} (including @code{abort} etc.). This is the
11864: only difference (apart from a speed factor of between 1.15 (K6-2) and
11865: 2 (21264)) between @code{gforth} and @code{gforth-fast}. Given an
11866: exception caused by a primitive in @code{gforth-fast}, you will
11867: typically see no return stack dump at all; however, if the exception is
11868: caught by @code{catch} (e.g., for restoring some state), and then
11869: @code{throw}n again, the return stack dump will be for the first such
11870: @code{throw}.
11871:
11872: @c ******************************************************************
11873: @node Tools, ANS conformance, Error messages, Top
11874: @chapter Tools
11875:
11876: @menu
11877: * ANS Report:: Report the words used, sorted by wordset.
11878: @end menu
11879:
11880: See also @ref{Emacs and Gforth}.
11881:
11882: @node ANS Report, , Tools, Tools
11883: @section @file{ans-report.fs}: Report the words used, sorted by wordset
11884: @cindex @file{ans-report.fs}
11885: @cindex report the words used in your program
11886: @cindex words used in your program
11887:
11888: If you want to label a Forth program as ANS Forth Program, you must
11889: document which wordsets the program uses; for extension wordsets, it is
11890: helpful to list the words the program requires from these wordsets
11891: (because Forth systems are allowed to provide only some words of them).
11892:
11893: The @file{ans-report.fs} tool makes it easy for you to determine which
11894: words from which wordset and which non-ANS words your application
11895: uses. You simply have to include @file{ans-report.fs} before loading the
11896: program you want to check. After loading your program, you can get the
11897: report with @code{print-ans-report}. A typical use is to run this as
11898: batch job like this:
11899: @example
11900: gforth ans-report.fs myprog.fs -e "print-ans-report bye"
11901: @end example
11902:
11903: The output looks like this (for @file{compat/control.fs}):
11904: @example
11905: The program uses the following words
11906: from CORE :
11907: : POSTPONE THEN ; immediate ?dup IF 0=
11908: from BLOCK-EXT :
11909: \
11910: from FILE :
11911: (
11912: @end example
11913:
11914: @subsection Caveats
11915:
11916: Note that @file{ans-report.fs} just checks which words are used, not whether
11917: they are used in an ANS Forth conforming way!
11918:
11919: Some words are defined in several wordsets in the
11920: standard. @file{ans-report.fs} reports them for only one of the
11921: wordsets, and not necessarily the one you expect. It depends on usage
11922: which wordset is the right one to specify. E.g., if you only use the
11923: compilation semantics of @code{S"}, it is a Core word; if you also use
11924: its interpretation semantics, it is a File word.
11925:
11926: @c ******************************************************************
11927: @node ANS conformance, Standard vs Extensions, Tools, Top
11928: @chapter ANS conformance
11929: @cindex ANS conformance of Gforth
11930:
11931: To the best of our knowledge, Gforth is an
11932:
11933: ANS Forth System
11934: @itemize @bullet
11935: @item providing the Core Extensions word set
11936: @item providing the Block word set
11937: @item providing the Block Extensions word set
11938: @item providing the Double-Number word set
11939: @item providing the Double-Number Extensions word set
11940: @item providing the Exception word set
11941: @item providing the Exception Extensions word set
11942: @item providing the Facility word set
11943: @item providing @code{EKEY}, @code{EKEY>CHAR}, @code{EKEY?}, @code{MS} and @code{TIME&DATE} from the Facility Extensions word set
11944: @item providing the File Access word set
11945: @item providing the File Access Extensions word set
11946: @item providing the Floating-Point word set
11947: @item providing the Floating-Point Extensions word set
11948: @item providing the Locals word set
11949: @item providing the Locals Extensions word set
11950: @item providing the Memory-Allocation word set
11951: @item providing the Memory-Allocation Extensions word set (that one's easy)
11952: @item providing the Programming-Tools word set
11953: @item providing @code{;CODE}, @code{AHEAD}, @code{ASSEMBLER}, @code{BYE}, @code{CODE}, @code{CS-PICK}, @code{CS-ROLL}, @code{STATE}, @code{[ELSE]}, @code{[IF]}, @code{[THEN]} from the Programming-Tools Extensions word set
11954: @item providing the Search-Order word set
11955: @item providing the Search-Order Extensions word set
11956: @item providing the String word set
11957: @item providing the String Extensions word set (another easy one)
11958: @end itemize
11959:
11960: @cindex system documentation
11961: In addition, ANS Forth systems are required to document certain
11962: implementation choices. This chapter tries to meet these
11963: requirements. In many cases it gives a way to ask the system for the
11964: information instead of providing the information directly, in
11965: particular, if the information depends on the processor, the operating
11966: system or the installation options chosen, or if they are likely to
11967: change during the maintenance of Gforth.
11968:
11969: @comment The framework for the rest has been taken from pfe.
11970:
11971: @menu
11972: * The Core Words::
11973: * The optional Block word set::
11974: * The optional Double Number word set::
11975: * The optional Exception word set::
11976: * The optional Facility word set::
11977: * The optional File-Access word set::
11978: * The optional Floating-Point word set::
11979: * The optional Locals word set::
11980: * The optional Memory-Allocation word set::
11981: * The optional Programming-Tools word set::
11982: * The optional Search-Order word set::
11983: @end menu
11984:
11985:
11986: @c =====================================================================
11987: @node The Core Words, The optional Block word set, ANS conformance, ANS conformance
11988: @comment node-name, next, previous, up
11989: @section The Core Words
11990: @c =====================================================================
11991: @cindex core words, system documentation
11992: @cindex system documentation, core words
11993:
11994: @menu
11995: * core-idef:: Implementation Defined Options
11996: * core-ambcond:: Ambiguous Conditions
11997: * core-other:: Other System Documentation
11998: @end menu
11999:
12000: @c ---------------------------------------------------------------------
12001: @node core-idef, core-ambcond, The Core Words, The Core Words
12002: @subsection Implementation Defined Options
12003: @c ---------------------------------------------------------------------
12004: @cindex core words, implementation-defined options
12005: @cindex implementation-defined options, core words
12006:
12007:
12008: @table @i
12009: @item (Cell) aligned addresses:
12010: @cindex cell-aligned addresses
12011: @cindex aligned addresses
12012: processor-dependent. Gforth's alignment words perform natural alignment
12013: (e.g., an address aligned for a datum of size 8 is divisible by
12014: 8). Unaligned accesses usually result in a @code{-23 THROW}.
12015:
12016: @item @code{EMIT} and non-graphic characters:
12017: @cindex @code{EMIT} and non-graphic characters
12018: @cindex non-graphic characters and @code{EMIT}
12019: The character is output using the C library function (actually, macro)
12020: @code{putc}.
12021:
12022: @item character editing of @code{ACCEPT} and @code{EXPECT}:
12023: @cindex character editing of @code{ACCEPT} and @code{EXPECT}
12024: @cindex editing in @code{ACCEPT} and @code{EXPECT}
12025: @cindex @code{ACCEPT}, editing
12026: @cindex @code{EXPECT}, editing
12027: This is modeled on the GNU readline library (@pxref{Readline
12028: Interaction, , Command Line Editing, readline, The GNU Readline
12029: Library}) with Emacs-like key bindings. @kbd{Tab} deviates a little by
12030: producing a full word completion every time you type it (instead of
12031: producing the common prefix of all completions). @xref{Command-line editing}.
12032:
12033: @item character set:
12034: @cindex character set
12035: The character set of your computer and display device. Gforth is
12036: 8-bit-clean (but some other component in your system may make trouble).
12037:
12038: @item Character-aligned address requirements:
12039: @cindex character-aligned address requirements
12040: installation-dependent. Currently a character is represented by a C
12041: @code{unsigned char}; in the future we might switch to @code{wchar_t}
12042: (Comments on that requested).
12043:
12044: @item character-set extensions and matching of names:
12045: @cindex character-set extensions and matching of names
12046: @cindex case-sensitivity for name lookup
12047: @cindex name lookup, case-sensitivity
12048: @cindex locale and case-sensitivity
12049: Any character except the ASCII NUL character can be used in a
12050: name. Matching is case-insensitive (except in @code{TABLE}s). The
12051: matching is performed using the C library function @code{strncasecmp}, whose
12052: function is probably influenced by the locale. E.g., the @code{C} locale
12053: does not know about accents and umlauts, so they are matched
12054: case-sensitively in that locale. For portability reasons it is best to
12055: write programs such that they work in the @code{C} locale. Then one can
12056: use libraries written by a Polish programmer (who might use words
12057: containing ISO Latin-2 encoded characters) and by a French programmer
12058: (ISO Latin-1) in the same program (of course, @code{WORDS} will produce
12059: funny results for some of the words (which ones, depends on the font you
12060: are using)). Also, the locale you prefer may not be available in other
12061: operating systems. Hopefully, Unicode will solve these problems one day.
12062:
12063: @item conditions under which control characters match a space delimiter:
12064: @cindex space delimiters
12065: @cindex control characters as delimiters
12066: If @code{WORD} is called with the space character as a delimiter, all
12067: white-space characters (as identified by the C macro @code{isspace()})
12068: are delimiters. @code{PARSE}, on the other hand, treats space like other
12069: delimiters. @code{SWORD} treats space like @code{WORD}, but behaves
12070: like @code{PARSE} otherwise. @code{(NAME)}, which is used by the outer
12071: interpreter (aka text interpreter) by default, treats all white-space
12072: characters as delimiters.
12073:
12074: @item format of the control-flow stack:
12075: @cindex control-flow stack, format
12076: The data stack is used as control-flow stack. The size of a control-flow
12077: stack item in cells is given by the constant @code{cs-item-size}. At the
12078: time of this writing, an item consists of a (pointer to a) locals list
12079: (third), an address in the code (second), and a tag for identifying the
12080: item (TOS). The following tags are used: @code{defstart},
12081: @code{live-orig}, @code{dead-orig}, @code{dest}, @code{do-dest},
12082: @code{scopestart}.
12083:
12084: @item conversion of digits > 35
12085: @cindex digits > 35
12086: The characters @code{[\]^_'} are the digits with the decimal value
12087: 36@minus{}41. There is no way to input many of the larger digits.
12088:
12089: @item display after input terminates in @code{ACCEPT} and @code{EXPECT}:
12090: @cindex @code{EXPECT}, display after end of input
12091: @cindex @code{ACCEPT}, display after end of input
12092: The cursor is moved to the end of the entered string. If the input is
12093: terminated using the @kbd{Return} key, a space is typed.
12094:
12095: @item exception abort sequence of @code{ABORT"}:
12096: @cindex exception abort sequence of @code{ABORT"}
12097: @cindex @code{ABORT"}, exception abort sequence
12098: The error string is stored into the variable @code{"error} and a
12099: @code{-2 throw} is performed.
12100:
12101: @item input line terminator:
12102: @cindex input line terminator
12103: @cindex line terminator on input
12104: @cindex newline character on input
12105: For interactive input, @kbd{C-m} (CR) and @kbd{C-j} (LF) terminate
12106: lines. One of these characters is typically produced when you type the
12107: @kbd{Enter} or @kbd{Return} key.
12108:
12109: @item maximum size of a counted string:
12110: @cindex maximum size of a counted string
12111: @cindex counted string, maximum size
12112: @code{s" /counted-string" environment? drop .}. Currently 255 characters
12113: on all ports, but this may change.
12114:
12115: @item maximum size of a parsed string:
12116: @cindex maximum size of a parsed string
12117: @cindex parsed string, maximum size
12118: Given by the constant @code{/line}. Currently 255 characters.
12119:
12120: @item maximum size of a definition name, in characters:
12121: @cindex maximum size of a definition name, in characters
12122: @cindex name, maximum length
12123: 31
12124:
12125: @item maximum string length for @code{ENVIRONMENT?}, in characters:
12126: @cindex maximum string length for @code{ENVIRONMENT?}, in characters
12127: @cindex @code{ENVIRONMENT?} string length, maximum
12128: 31
12129:
12130: @item method of selecting the user input device:
12131: @cindex user input device, method of selecting
12132: The user input device is the standard input. There is currently no way to
12133: change it from within Gforth. However, the input can typically be
12134: redirected in the command line that starts Gforth.
12135:
12136: @item method of selecting the user output device:
12137: @cindex user output device, method of selecting
12138: @code{EMIT} and @code{TYPE} output to the file-id stored in the value
12139: @code{outfile-id} (@code{stdout} by default). Gforth uses unbuffered
12140: output when the user output device is a terminal, otherwise the output
12141: is buffered.
12142:
12143: @item methods of dictionary compilation:
12144: What are we expected to document here?
12145:
12146: @item number of bits in one address unit:
12147: @cindex number of bits in one address unit
12148: @cindex address unit, size in bits
12149: @code{s" address-units-bits" environment? drop .}. 8 in all current
12150: ports.
12151:
12152: @item number representation and arithmetic:
12153: @cindex number representation and arithmetic
12154: Processor-dependent. Binary two's complement on all current ports.
12155:
12156: @item ranges for integer types:
12157: @cindex ranges for integer types
12158: @cindex integer types, ranges
12159: Installation-dependent. Make environmental queries for @code{MAX-N},
12160: @code{MAX-U}, @code{MAX-D} and @code{MAX-UD}. The lower bounds for
12161: unsigned (and positive) types is 0. The lower bound for signed types on
12162: two's complement and one's complement machines machines can be computed
12163: by adding 1 to the upper bound.
12164:
12165: @item read-only data space regions:
12166: @cindex read-only data space regions
12167: @cindex data-space, read-only regions
12168: The whole Forth data space is writable.
12169:
12170: @item size of buffer at @code{WORD}:
12171: @cindex size of buffer at @code{WORD}
12172: @cindex @code{WORD} buffer size
12173: @code{PAD HERE - .}. 104 characters on 32-bit machines. The buffer is
12174: shared with the pictured numeric output string. If overwriting
12175: @code{PAD} is acceptable, it is as large as the remaining dictionary
12176: space, although only as much can be sensibly used as fits in a counted
12177: string.
12178:
12179: @item size of one cell in address units:
12180: @cindex cell size
12181: @code{1 cells .}.
12182:
12183: @item size of one character in address units:
12184: @cindex char size
12185: @code{1 chars .}. 1 on all current ports.
12186:
12187: @item size of the keyboard terminal buffer:
12188: @cindex size of the keyboard terminal buffer
12189: @cindex terminal buffer, size
12190: Varies. You can determine the size at a specific time using @code{lp@@
12191: tib - .}. It is shared with the locals stack and TIBs of files that
12192: include the current file. You can change the amount of space for TIBs
12193: and locals stack at Gforth startup with the command line option
12194: @code{-l}.
12195:
12196: @item size of the pictured numeric output buffer:
12197: @cindex size of the pictured numeric output buffer
12198: @cindex pictured numeric output buffer, size
12199: @code{PAD HERE - .}. 104 characters on 32-bit machines. The buffer is
12200: shared with @code{WORD}.
12201:
12202: @item size of the scratch area returned by @code{PAD}:
12203: @cindex size of the scratch area returned by @code{PAD}
12204: @cindex @code{PAD} size
12205: The remainder of dictionary space. @code{unused pad here - - .}.
12206:
12207: @item system case-sensitivity characteristics:
12208: @cindex case-sensitivity characteristics
12209: Dictionary searches are case-insensitive (except in
12210: @code{TABLE}s). However, as explained above under @i{character-set
12211: extensions}, the matching for non-ASCII characters is determined by the
12212: locale you are using. In the default @code{C} locale all non-ASCII
12213: characters are matched case-sensitively.
12214:
12215: @item system prompt:
12216: @cindex system prompt
12217: @cindex prompt
12218: @code{ ok} in interpret state, @code{ compiled} in compile state.
12219:
12220: @item division rounding:
12221: @cindex division rounding
12222: installation dependent. @code{s" floored" environment? drop .}. We leave
12223: the choice to @code{gcc} (what to use for @code{/}) and to you (whether
12224: to use @code{fm/mod}, @code{sm/rem} or simply @code{/}).
12225:
12226: @item values of @code{STATE} when true:
12227: @cindex @code{STATE} values
12228: -1.
12229:
12230: @item values returned after arithmetic overflow:
12231: On two's complement machines, arithmetic is performed modulo
12232: 2**bits-per-cell for single arithmetic and 4**bits-per-cell for double
12233: arithmetic (with appropriate mapping for signed types). Division by zero
12234: typically results in a @code{-55 throw} (Floating-point unidentified
12235: fault), although a @code{-10 throw} (divide by zero) would be more
12236: appropriate.
12237:
12238: @item whether the current definition can be found after @t{DOES>}:
12239: @cindex @t{DOES>}, visibility of current definition
12240: No.
12241:
12242: @end table
12243:
12244: @c ---------------------------------------------------------------------
12245: @node core-ambcond, core-other, core-idef, The Core Words
12246: @subsection Ambiguous conditions
12247: @c ---------------------------------------------------------------------
12248: @cindex core words, ambiguous conditions
12249: @cindex ambiguous conditions, core words
12250:
12251: @table @i
12252:
12253: @item a name is neither a word nor a number:
12254: @cindex name not found
12255: @cindex undefined word
12256: @code{-13 throw} (Undefined word). Actually, @code{-13 bounce}, which
12257: preserves the data and FP stack, so you don't lose more work than
12258: necessary.
12259:
12260: @item a definition name exceeds the maximum length allowed:
12261: @cindex word name too long
12262: @code{-19 throw} (Word name too long)
12263:
12264: @item addressing a region not inside the various data spaces of the forth system:
12265: @cindex Invalid memory address
12266: The stacks, code space and header space are accessible. Machine code space is
12267: typically readable. Accessing other addresses gives results dependent on
12268: the operating system. On decent systems: @code{-9 throw} (Invalid memory
12269: address).
12270:
12271: @item argument type incompatible with parameter:
12272: @cindex argument type mismatch
12273: This is usually not caught. Some words perform checks, e.g., the control
12274: flow words, and issue a @code{ABORT"} or @code{-12 THROW} (Argument type
12275: mismatch).
12276:
12277: @item attempting to obtain the execution token of a word with undefined execution semantics:
12278: @cindex Interpreting a compile-only word, for @code{'} etc.
12279: @cindex execution token of words with undefined execution semantics
12280: @code{-14 throw} (Interpreting a compile-only word). In some cases, you
12281: get an execution token for @code{compile-only-error} (which performs a
12282: @code{-14 throw} when executed).
12283:
12284: @item dividing by zero:
12285: @cindex dividing by zero
12286: @cindex floating point unidentified fault, integer division
12287: On better platforms, this produces a @code{-10 throw} (Division by
12288: zero); on other systems, this typically results in a @code{-55 throw}
12289: (Floating-point unidentified fault).
12290:
12291: @item insufficient data stack or return stack space:
12292: @cindex insufficient data stack or return stack space
12293: @cindex stack overflow
12294: @cindex address alignment exception, stack overflow
12295: @cindex Invalid memory address, stack overflow
12296: Depending on the operating system, the installation, and the invocation
12297: of Gforth, this is either checked by the memory management hardware, or
12298: it is not checked. If it is checked, you typically get a @code{-3 throw}
12299: (Stack overflow), @code{-5 throw} (Return stack overflow), or @code{-9
12300: throw} (Invalid memory address) (depending on the platform and how you
12301: achieved the overflow) as soon as the overflow happens. If it is not
12302: checked, overflows typically result in mysterious illegal memory
12303: accesses, producing @code{-9 throw} (Invalid memory address) or
12304: @code{-23 throw} (Address alignment exception); they might also destroy
12305: the internal data structure of @code{ALLOCATE} and friends, resulting in
12306: various errors in these words.
12307:
12308: @item insufficient space for loop control parameters:
12309: @cindex insufficient space for loop control parameters
12310: like other return stack overflows.
12311:
12312: @item insufficient space in the dictionary:
12313: @cindex insufficient space in the dictionary
12314: @cindex dictionary overflow
12315: If you try to allot (either directly with @code{allot}, or indirectly
12316: with @code{,}, @code{create} etc.) more memory than available in the
12317: dictionary, you get a @code{-8 throw} (Dictionary overflow). If you try
12318: to access memory beyond the end of the dictionary, the results are
12319: similar to stack overflows.
12320:
12321: @item interpreting a word with undefined interpretation semantics:
12322: @cindex interpreting a word with undefined interpretation semantics
12323: @cindex Interpreting a compile-only word
12324: For some words, we have defined interpretation semantics. For the
12325: others: @code{-14 throw} (Interpreting a compile-only word).
12326:
12327: @item modifying the contents of the input buffer or a string literal:
12328: @cindex modifying the contents of the input buffer or a string literal
12329: These are located in writable memory and can be modified.
12330:
12331: @item overflow of the pictured numeric output string:
12332: @cindex overflow of the pictured numeric output string
12333: @cindex pictured numeric output string, overflow
12334: @code{-17 throw} (Pictured numeric ouput string overflow).
12335:
12336: @item parsed string overflow:
12337: @cindex parsed string overflow
12338: @code{PARSE} cannot overflow. @code{WORD} does not check for overflow.
12339:
12340: @item producing a result out of range:
12341: @cindex result out of range
12342: On two's complement machines, arithmetic is performed modulo
12343: 2**bits-per-cell for single arithmetic and 4**bits-per-cell for double
12344: arithmetic (with appropriate mapping for signed types). Division by zero
12345: typically results in a @code{-10 throw} (divide by zero) or @code{-55
12346: throw} (floating point unidentified fault). @code{convert} and
12347: @code{>number} currently overflow silently.
12348:
12349: @item reading from an empty data or return stack:
12350: @cindex stack empty
12351: @cindex stack underflow
12352: @cindex return stack underflow
12353: The data stack is checked by the outer (aka text) interpreter after
12354: every word executed. If it has underflowed, a @code{-4 throw} (Stack
12355: underflow) is performed. Apart from that, stacks may be checked or not,
12356: depending on operating system, installation, and invocation. If they are
12357: caught by a check, they typically result in @code{-4 throw} (Stack
12358: underflow), @code{-6 throw} (Return stack underflow) or @code{-9 throw}
12359: (Invalid memory address), depending on the platform and which stack
12360: underflows and by how much. Note that even if the system uses checking
12361: (through the MMU), your program may have to underflow by a significant
12362: number of stack items to trigger the reaction (the reason for this is
12363: that the MMU, and therefore the checking, works with a page-size
12364: granularity). If there is no checking, the symptoms resulting from an
12365: underflow are similar to those from an overflow. Unbalanced return
12366: stack errors result in a variaty of symptoms, including @code{-9 throw}
12367: (Invalid memory address) and Illegal Instruction (typically @code{-260
12368: throw}).
12369:
12370: @item unexpected end of the input buffer, resulting in an attempt to use a zero-length string as a name:
12371: @cindex unexpected end of the input buffer
12372: @cindex zero-length string as a name
12373: @cindex Attempt to use zero-length string as a name
12374: @code{Create} and its descendants perform a @code{-16 throw} (Attempt to
12375: use zero-length string as a name). Words like @code{'} probably will not
12376: find what they search. Note that it is possible to create zero-length
12377: names with @code{nextname} (should it not?).
12378:
12379: @item @code{>IN} greater than input buffer:
12380: @cindex @code{>IN} greater than input buffer
12381: The next invocation of a parsing word returns a string with length 0.
12382:
12383: @item @code{RECURSE} appears after @code{DOES>}:
12384: @cindex @code{RECURSE} appears after @code{DOES>}
12385: Compiles a recursive call to the defining word, not to the defined word.
12386:
12387: @item argument input source different than current input source for @code{RESTORE-INPUT}:
12388: @cindex argument input source different than current input source for @code{RESTORE-INPUT}
12389: @cindex argument type mismatch, @code{RESTORE-INPUT}
12390: @cindex @code{RESTORE-INPUT}, Argument type mismatch
12391: @code{-12 THROW}. Note that, once an input file is closed (e.g., because
12392: the end of the file was reached), its source-id may be
12393: reused. Therefore, restoring an input source specification referencing a
12394: closed file may lead to unpredictable results instead of a @code{-12
12395: THROW}.
12396:
12397: In the future, Gforth may be able to restore input source specifications
12398: from other than the current input source.
12399:
12400: @item data space containing definitions gets de-allocated:
12401: @cindex data space containing definitions gets de-allocated
12402: Deallocation with @code{allot} is not checked. This typically results in
12403: memory access faults or execution of illegal instructions.
12404:
12405: @item data space read/write with incorrect alignment:
12406: @cindex data space read/write with incorrect alignment
12407: @cindex alignment faults
12408: @cindex address alignment exception
12409: Processor-dependent. Typically results in a @code{-23 throw} (Address
12410: alignment exception). Under Linux-Intel on a 486 or later processor with
12411: alignment turned on, incorrect alignment results in a @code{-9 throw}
12412: (Invalid memory address). There are reportedly some processors with
12413: alignment restrictions that do not report violations.
12414:
12415: @item data space pointer not properly aligned, @code{,}, @code{C,}:
12416: @cindex data space pointer not properly aligned, @code{,}, @code{C,}
12417: Like other alignment errors.
12418:
12419: @item less than u+2 stack items (@code{PICK} and @code{ROLL}):
12420: Like other stack underflows.
12421:
12422: @item loop control parameters not available:
12423: @cindex loop control parameters not available
12424: Not checked. The counted loop words simply assume that the top of return
12425: stack items are loop control parameters and behave accordingly.
12426:
12427: @item most recent definition does not have a name (@code{IMMEDIATE}):
12428: @cindex most recent definition does not have a name (@code{IMMEDIATE})
12429: @cindex last word was headerless
12430: @code{abort" last word was headerless"}.
12431:
12432: @item name not defined by @code{VALUE} used by @code{TO}:
12433: @cindex name not defined by @code{VALUE} used by @code{TO}
12434: @cindex @code{TO} on non-@code{VALUE}s
12435: @cindex Invalid name argument, @code{TO}
12436: @code{-32 throw} (Invalid name argument) (unless name is a local or was
12437: defined by @code{CONSTANT}; in the latter case it just changes the constant).
12438:
12439: @item name not found (@code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]}):
12440: @cindex name not found (@code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]})
12441: @cindex undefined word, @code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]}
12442: @code{-13 throw} (Undefined word)
12443:
12444: @item parameters are not of the same type (@code{DO}, @code{?DO}, @code{WITHIN}):
12445: @cindex parameters are not of the same type (@code{DO}, @code{?DO}, @code{WITHIN})
12446: Gforth behaves as if they were of the same type. I.e., you can predict
12447: the behaviour by interpreting all parameters as, e.g., signed.
12448:
12449: @item @code{POSTPONE} or @code{[COMPILE]} applied to @code{TO}:
12450: @cindex @code{POSTPONE} or @code{[COMPILE]} applied to @code{TO}
12451: Assume @code{: X POSTPONE TO ; IMMEDIATE}. @code{X} performs the
12452: compilation semantics of @code{TO}.
12453:
12454: @item String longer than a counted string returned by @code{WORD}:
12455: @cindex string longer than a counted string returned by @code{WORD}
12456: @cindex @code{WORD}, string overflow
12457: Not checked. The string will be ok, but the count will, of course,
12458: contain only the least significant bits of the length.
12459:
12460: @item u greater than or equal to the number of bits in a cell (@code{LSHIFT}, @code{RSHIFT}):
12461: @cindex @code{LSHIFT}, large shift counts
12462: @cindex @code{RSHIFT}, large shift counts
12463: Processor-dependent. Typical behaviours are returning 0 and using only
12464: the low bits of the shift count.
12465:
12466: @item word not defined via @code{CREATE}:
12467: @cindex @code{>BODY} of non-@code{CREATE}d words
12468: @code{>BODY} produces the PFA of the word no matter how it was defined.
12469:
12470: @cindex @code{DOES>} of non-@code{CREATE}d words
12471: @code{DOES>} changes the execution semantics of the last defined word no
12472: matter how it was defined. E.g., @code{CONSTANT DOES>} is equivalent to
12473: @code{CREATE , DOES>}.
12474:
12475: @item words improperly used outside @code{<#} and @code{#>}:
12476: Not checked. As usual, you can expect memory faults.
12477:
12478: @end table
12479:
12480:
12481: @c ---------------------------------------------------------------------
12482: @node core-other, , core-ambcond, The Core Words
12483: @subsection Other system documentation
12484: @c ---------------------------------------------------------------------
12485: @cindex other system documentation, core words
12486: @cindex core words, other system documentation
12487:
12488: @table @i
12489: @item nonstandard words using @code{PAD}:
12490: @cindex @code{PAD} use by nonstandard words
12491: None.
12492:
12493: @item operator's terminal facilities available:
12494: @cindex operator's terminal facilities available
12495: After processing the command line, Gforth goes into interactive mode,
12496: and you can give commands to Gforth interactively. The actual facilities
12497: available depend on how you invoke Gforth.
12498:
12499: @item program data space available:
12500: @cindex program data space available
12501: @cindex data space available
12502: @code{UNUSED .} gives the remaining dictionary space. The total
12503: dictionary space can be specified with the @code{-m} switch
12504: (@pxref{Invoking Gforth}) when Gforth starts up.
12505:
12506: @item return stack space available:
12507: @cindex return stack space available
12508: You can compute the total return stack space in cells with
12509: @code{s" RETURN-STACK-CELLS" environment? drop .}. You can specify it at
12510: startup time with the @code{-r} switch (@pxref{Invoking Gforth}).
12511:
12512: @item stack space available:
12513: @cindex stack space available
12514: You can compute the total data stack space in cells with
12515: @code{s" STACK-CELLS" environment? drop .}. You can specify it at
12516: startup time with the @code{-d} switch (@pxref{Invoking Gforth}).
12517:
12518: @item system dictionary space required, in address units:
12519: @cindex system dictionary space required, in address units
12520: Type @code{here forthstart - .} after startup. At the time of this
12521: writing, this gives 80080 (bytes) on a 32-bit system.
12522: @end table
12523:
12524:
12525: @c =====================================================================
12526: @node The optional Block word set, The optional Double Number word set, The Core Words, ANS conformance
12527: @section The optional Block word set
12528: @c =====================================================================
12529: @cindex system documentation, block words
12530: @cindex block words, system documentation
12531:
12532: @menu
12533: * block-idef:: Implementation Defined Options
12534: * block-ambcond:: Ambiguous Conditions
12535: * block-other:: Other System Documentation
12536: @end menu
12537:
12538:
12539: @c ---------------------------------------------------------------------
12540: @node block-idef, block-ambcond, The optional Block word set, The optional Block word set
12541: @subsection Implementation Defined Options
12542: @c ---------------------------------------------------------------------
12543: @cindex implementation-defined options, block words
12544: @cindex block words, implementation-defined options
12545:
12546: @table @i
12547: @item the format for display by @code{LIST}:
12548: @cindex @code{LIST} display format
12549: First the screen number is displayed, then 16 lines of 64 characters,
12550: each line preceded by the line number.
12551:
12552: @item the length of a line affected by @code{\}:
12553: @cindex length of a line affected by @code{\}
12554: @cindex @code{\}, line length in blocks
12555: 64 characters.
12556: @end table
12557:
12558:
12559: @c ---------------------------------------------------------------------
12560: @node block-ambcond, block-other, block-idef, The optional Block word set
12561: @subsection Ambiguous conditions
12562: @c ---------------------------------------------------------------------
12563: @cindex block words, ambiguous conditions
12564: @cindex ambiguous conditions, block words
12565:
12566: @table @i
12567: @item correct block read was not possible:
12568: @cindex block read not possible
12569: Typically results in a @code{throw} of some OS-derived value (between
12570: -512 and -2048). If the blocks file was just not long enough, blanks are
12571: supplied for the missing portion.
12572:
12573: @item I/O exception in block transfer:
12574: @cindex I/O exception in block transfer
12575: @cindex block transfer, I/O exception
12576: Typically results in a @code{throw} of some OS-derived value (between
12577: -512 and -2048).
12578:
12579: @item invalid block number:
12580: @cindex invalid block number
12581: @cindex block number invalid
12582: @code{-35 throw} (Invalid block number)
12583:
12584: @item a program directly alters the contents of @code{BLK}:
12585: @cindex @code{BLK}, altering @code{BLK}
12586: The input stream is switched to that other block, at the same
12587: position. If the storing to @code{BLK} happens when interpreting
12588: non-block input, the system will get quite confused when the block ends.
12589:
12590: @item no current block buffer for @code{UPDATE}:
12591: @cindex @code{UPDATE}, no current block buffer
12592: @code{UPDATE} has no effect.
12593:
12594: @end table
12595:
12596: @c ---------------------------------------------------------------------
12597: @node block-other, , block-ambcond, The optional Block word set
12598: @subsection Other system documentation
12599: @c ---------------------------------------------------------------------
12600: @cindex other system documentation, block words
12601: @cindex block words, other system documentation
12602:
12603: @table @i
12604: @item any restrictions a multiprogramming system places on the use of buffer addresses:
12605: No restrictions (yet).
12606:
12607: @item the number of blocks available for source and data:
12608: depends on your disk space.
12609:
12610: @end table
12611:
12612:
12613: @c =====================================================================
12614: @node The optional Double Number word set, The optional Exception word set, The optional Block word set, ANS conformance
12615: @section The optional Double Number word set
12616: @c =====================================================================
12617: @cindex system documentation, double words
12618: @cindex double words, system documentation
12619:
12620: @menu
12621: * double-ambcond:: Ambiguous Conditions
12622: @end menu
12623:
12624:
12625: @c ---------------------------------------------------------------------
12626: @node double-ambcond, , The optional Double Number word set, The optional Double Number word set
12627: @subsection Ambiguous conditions
12628: @c ---------------------------------------------------------------------
12629: @cindex double words, ambiguous conditions
12630: @cindex ambiguous conditions, double words
12631:
12632: @table @i
12633: @item @i{d} outside of range of @i{n} in @code{D>S}:
12634: @cindex @code{D>S}, @i{d} out of range of @i{n}
12635: The least significant cell of @i{d} is produced.
12636:
12637: @end table
12638:
12639:
12640: @c =====================================================================
12641: @node The optional Exception word set, The optional Facility word set, The optional Double Number word set, ANS conformance
12642: @section The optional Exception word set
12643: @c =====================================================================
12644: @cindex system documentation, exception words
12645: @cindex exception words, system documentation
12646:
12647: @menu
12648: * exception-idef:: Implementation Defined Options
12649: @end menu
12650:
12651:
12652: @c ---------------------------------------------------------------------
12653: @node exception-idef, , The optional Exception word set, The optional Exception word set
12654: @subsection Implementation Defined Options
12655: @c ---------------------------------------------------------------------
12656: @cindex implementation-defined options, exception words
12657: @cindex exception words, implementation-defined options
12658:
12659: @table @i
12660: @item @code{THROW}-codes used in the system:
12661: @cindex @code{THROW}-codes used in the system
12662: The codes -256@minus{}-511 are used for reporting signals. The mapping
12663: from OS signal numbers to throw codes is -256@minus{}@i{signal}. The
12664: codes -512@minus{}-2047 are used for OS errors (for file and memory
12665: allocation operations). The mapping from OS error numbers to throw codes
12666: is -512@minus{}@code{errno}. One side effect of this mapping is that
12667: undefined OS errors produce a message with a strange number; e.g.,
12668: @code{-1000 THROW} results in @code{Unknown error 488} on my system.
12669: @end table
12670:
12671: @c =====================================================================
12672: @node The optional Facility word set, The optional File-Access word set, The optional Exception word set, ANS conformance
12673: @section The optional Facility word set
12674: @c =====================================================================
12675: @cindex system documentation, facility words
12676: @cindex facility words, system documentation
12677:
12678: @menu
12679: * facility-idef:: Implementation Defined Options
12680: * facility-ambcond:: Ambiguous Conditions
12681: @end menu
12682:
12683:
12684: @c ---------------------------------------------------------------------
12685: @node facility-idef, facility-ambcond, The optional Facility word set, The optional Facility word set
12686: @subsection Implementation Defined Options
12687: @c ---------------------------------------------------------------------
12688: @cindex implementation-defined options, facility words
12689: @cindex facility words, implementation-defined options
12690:
12691: @table @i
12692: @item encoding of keyboard events (@code{EKEY}):
12693: @cindex keyboard events, encoding in @code{EKEY}
12694: @cindex @code{EKEY}, encoding of keyboard events
12695: Keys corresponding to ASCII characters are encoded as ASCII characters.
12696: Other keys are encoded with the constants @code{k-left}, @code{k-right},
12697: @code{k-up}, @code{k-down}, @code{k-home}, @code{k-end}, @code{k1},
12698: @code{k2}, @code{k3}, @code{k4}, @code{k5}, @code{k6}, @code{k7},
12699: @code{k8}, @code{k9}, @code{k10}, @code{k11}, @code{k12}.
12700:
12701:
12702: @item duration of a system clock tick:
12703: @cindex duration of a system clock tick
12704: @cindex clock tick duration
12705: System dependent. With respect to @code{MS}, the time is specified in
12706: microseconds. How well the OS and the hardware implement this, is
12707: another question.
12708:
12709: @item repeatability to be expected from the execution of @code{MS}:
12710: @cindex repeatability to be expected from the execution of @code{MS}
12711: @cindex @code{MS}, repeatability to be expected
12712: System dependent. On Unix, a lot depends on load. If the system is
12713: lightly loaded, and the delay is short enough that Gforth does not get
12714: swapped out, the performance should be acceptable. Under MS-DOS and
12715: other single-tasking systems, it should be good.
12716:
12717: @end table
12718:
12719:
12720: @c ---------------------------------------------------------------------
12721: @node facility-ambcond, , facility-idef, The optional Facility word set
12722: @subsection Ambiguous conditions
12723: @c ---------------------------------------------------------------------
12724: @cindex facility words, ambiguous conditions
12725: @cindex ambiguous conditions, facility words
12726:
12727: @table @i
12728: @item @code{AT-XY} can't be performed on user output device:
12729: @cindex @code{AT-XY} can't be performed on user output device
12730: Largely terminal dependent. No range checks are done on the arguments.
12731: No errors are reported. You may see some garbage appearing, you may see
12732: simply nothing happen.
12733:
12734: @end table
12735:
12736:
12737: @c =====================================================================
12738: @node The optional File-Access word set, The optional Floating-Point word set, The optional Facility word set, ANS conformance
12739: @section The optional File-Access word set
12740: @c =====================================================================
12741: @cindex system documentation, file words
12742: @cindex file words, system documentation
12743:
12744: @menu
12745: * file-idef:: Implementation Defined Options
12746: * file-ambcond:: Ambiguous Conditions
12747: @end menu
12748:
12749: @c ---------------------------------------------------------------------
12750: @node file-idef, file-ambcond, The optional File-Access word set, The optional File-Access word set
12751: @subsection Implementation Defined Options
12752: @c ---------------------------------------------------------------------
12753: @cindex implementation-defined options, file words
12754: @cindex file words, implementation-defined options
12755:
12756: @table @i
12757: @item file access methods used:
12758: @cindex file access methods used
12759: @code{R/O}, @code{R/W} and @code{BIN} work as you would
12760: expect. @code{W/O} translates into the C file opening mode @code{w} (or
12761: @code{wb}): The file is cleared, if it exists, and created, if it does
12762: not (with both @code{open-file} and @code{create-file}). Under Unix
12763: @code{create-file} creates a file with 666 permissions modified by your
12764: umask.
12765:
12766: @item file exceptions:
12767: @cindex file exceptions
12768: The file words do not raise exceptions (except, perhaps, memory access
12769: faults when you pass illegal addresses or file-ids).
12770:
12771: @item file line terminator:
12772: @cindex file line terminator
12773: System-dependent. Gforth uses C's newline character as line
12774: terminator. What the actual character code(s) of this are is
12775: system-dependent.
12776:
12777: @item file name format:
12778: @cindex file name format
12779: System dependent. Gforth just uses the file name format of your OS.
12780:
12781: @item information returned by @code{FILE-STATUS}:
12782: @cindex @code{FILE-STATUS}, returned information
12783: @code{FILE-STATUS} returns the most powerful file access mode allowed
12784: for the file: Either @code{R/O}, @code{W/O} or @code{R/W}. If the file
12785: cannot be accessed, @code{R/O BIN} is returned. @code{BIN} is applicable
12786: along with the returned mode.
12787:
12788: @item input file state after an exception when including source:
12789: @cindex exception when including source
12790: All files that are left via the exception are closed.
12791:
12792: @item @i{ior} values and meaning:
12793: @cindex @i{ior} values and meaning
12794: @cindex @i{wior} values and meaning
12795: The @i{ior}s returned by the file and memory allocation words are
12796: intended as throw codes. They typically are in the range
12797: -512@minus{}-2047 of OS errors. The mapping from OS error numbers to
12798: @i{ior}s is -512@minus{}@i{errno}.
12799:
12800: @item maximum depth of file input nesting:
12801: @cindex maximum depth of file input nesting
12802: @cindex file input nesting, maximum depth
12803: limited by the amount of return stack, locals/TIB stack, and the number
12804: of open files available. This should not give you troubles.
12805:
12806: @item maximum size of input line:
12807: @cindex maximum size of input line
12808: @cindex input line size, maximum
12809: @code{/line}. Currently 255.
12810:
12811: @item methods of mapping block ranges to files:
12812: @cindex mapping block ranges to files
12813: @cindex files containing blocks
12814: @cindex blocks in files
12815: By default, blocks are accessed in the file @file{blocks.fb} in the
12816: current working directory. The file can be switched with @code{USE}.
12817:
12818: @item number of string buffers provided by @code{S"}:
12819: @cindex @code{S"}, number of string buffers
12820: 1
12821:
12822: @item size of string buffer used by @code{S"}:
12823: @cindex @code{S"}, size of string buffer
12824: @code{/line}. currently 255.
12825:
12826: @end table
12827:
12828: @c ---------------------------------------------------------------------
12829: @node file-ambcond, , file-idef, The optional File-Access word set
12830: @subsection Ambiguous conditions
12831: @c ---------------------------------------------------------------------
12832: @cindex file words, ambiguous conditions
12833: @cindex ambiguous conditions, file words
12834:
12835: @table @i
12836: @item attempting to position a file outside its boundaries:
12837: @cindex @code{REPOSITION-FILE}, outside the file's boundaries
12838: @code{REPOSITION-FILE} is performed as usual: Afterwards,
12839: @code{FILE-POSITION} returns the value given to @code{REPOSITION-FILE}.
12840:
12841: @item attempting to read from file positions not yet written:
12842: @cindex reading from file positions not yet written
12843: End-of-file, i.e., zero characters are read and no error is reported.
12844:
12845: @item @i{file-id} is invalid (@code{INCLUDE-FILE}):
12846: @cindex @code{INCLUDE-FILE}, @i{file-id} is invalid
12847: An appropriate exception may be thrown, but a memory fault or other
12848: problem is more probable.
12849:
12850: @item I/O exception reading or closing @i{file-id} (@code{INCLUDE-FILE}, @code{INCLUDED}):
12851: @cindex @code{INCLUDE-FILE}, I/O exception reading or closing @i{file-id}
12852: @cindex @code{INCLUDED}, I/O exception reading or closing @i{file-id}
12853: The @i{ior} produced by the operation, that discovered the problem, is
12854: thrown.
12855:
12856: @item named file cannot be opened (@code{INCLUDED}):
12857: @cindex @code{INCLUDED}, named file cannot be opened
12858: The @i{ior} produced by @code{open-file} is thrown.
12859:
12860: @item requesting an unmapped block number:
12861: @cindex unmapped block numbers
12862: There are no unmapped legal block numbers. On some operating systems,
12863: writing a block with a large number may overflow the file system and
12864: have an error message as consequence.
12865:
12866: @item using @code{source-id} when @code{blk} is non-zero:
12867: @cindex @code{SOURCE-ID}, behaviour when @code{BLK} is non-zero
12868: @code{source-id} performs its function. Typically it will give the id of
12869: the source which loaded the block. (Better ideas?)
12870:
12871: @end table
12872:
12873:
12874: @c =====================================================================
12875: @node The optional Floating-Point word set, The optional Locals word set, The optional File-Access word set, ANS conformance
12876: @section The optional Floating-Point word set
12877: @c =====================================================================
12878: @cindex system documentation, floating-point words
12879: @cindex floating-point words, system documentation
12880:
12881: @menu
12882: * floating-idef:: Implementation Defined Options
12883: * floating-ambcond:: Ambiguous Conditions
12884: @end menu
12885:
12886:
12887: @c ---------------------------------------------------------------------
12888: @node floating-idef, floating-ambcond, The optional Floating-Point word set, The optional Floating-Point word set
12889: @subsection Implementation Defined Options
12890: @c ---------------------------------------------------------------------
12891: @cindex implementation-defined options, floating-point words
12892: @cindex floating-point words, implementation-defined options
12893:
12894: @table @i
12895: @item format and range of floating point numbers:
12896: @cindex format and range of floating point numbers
12897: @cindex floating point numbers, format and range
12898: System-dependent; the @code{double} type of C.
12899:
12900: @item results of @code{REPRESENT} when @i{float} is out of range:
12901: @cindex @code{REPRESENT}, results when @i{float} is out of range
12902: System dependent; @code{REPRESENT} is implemented using the C library
12903: function @code{ecvt()} and inherits its behaviour in this respect.
12904:
12905: @item rounding or truncation of floating-point numbers:
12906: @cindex rounding of floating-point numbers
12907: @cindex truncation of floating-point numbers
12908: @cindex floating-point numbers, rounding or truncation
12909: System dependent; the rounding behaviour is inherited from the hosting C
12910: compiler. IEEE-FP-based (i.e., most) systems by default round to
12911: nearest, and break ties by rounding to even (i.e., such that the last
12912: bit of the mantissa is 0).
12913:
12914: @item size of floating-point stack:
12915: @cindex floating-point stack size
12916: @code{s" FLOATING-STACK" environment? drop .} gives the total size of
12917: the floating-point stack (in floats). You can specify this on startup
12918: with the command-line option @code{-f} (@pxref{Invoking Gforth}).
12919:
12920: @item width of floating-point stack:
12921: @cindex floating-point stack width
12922: @code{1 floats}.
12923:
12924: @end table
12925:
12926:
12927: @c ---------------------------------------------------------------------
12928: @node floating-ambcond, , floating-idef, The optional Floating-Point word set
12929: @subsection Ambiguous conditions
12930: @c ---------------------------------------------------------------------
12931: @cindex floating-point words, ambiguous conditions
12932: @cindex ambiguous conditions, floating-point words
12933:
12934: @table @i
12935: @item @code{df@@} or @code{df!} used with an address that is not double-float aligned:
12936: @cindex @code{df@@} or @code{df!} used with an address that is not double-float aligned
12937: System-dependent. Typically results in a @code{-23 THROW} like other
12938: alignment violations.
12939:
12940: @item @code{f@@} or @code{f!} used with an address that is not float aligned:
12941: @cindex @code{f@@} used with an address that is not float aligned
12942: @cindex @code{f!} used with an address that is not float aligned
12943: System-dependent. Typically results in a @code{-23 THROW} like other
12944: alignment violations.
12945:
12946: @item floating-point result out of range:
12947: @cindex floating-point result out of range
12948: System-dependent. Can result in a @code{-55 THROW} (Floating-point
12949: unidentified fault), or can produce a special value representing, e.g.,
12950: Infinity.
12951:
12952: @item @code{sf@@} or @code{sf!} used with an address that is not single-float aligned:
12953: @cindex @code{sf@@} or @code{sf!} used with an address that is not single-float aligned
12954: System-dependent. Typically results in an alignment fault like other
12955: alignment violations.
12956:
12957: @item @code{base} is not decimal (@code{REPRESENT}, @code{F.}, @code{FE.}, @code{FS.}):
12958: @cindex @code{base} is not decimal (@code{REPRESENT}, @code{F.}, @code{FE.}, @code{FS.})
12959: The floating-point number is converted into decimal nonetheless.
12960:
12961: @item Both arguments are equal to zero (@code{FATAN2}):
12962: @cindex @code{FATAN2}, both arguments are equal to zero
12963: System-dependent. @code{FATAN2} is implemented using the C library
12964: function @code{atan2()}.
12965:
12966: @item Using @code{FTAN} on an argument @i{r1} where cos(@i{r1}) is zero:
12967: @cindex @code{FTAN} on an argument @i{r1} where cos(@i{r1}) is zero
12968: System-dependent. Anyway, typically the cos of @i{r1} will not be zero
12969: because of small errors and the tan will be a very large (or very small)
12970: but finite number.
12971:
12972: @item @i{d} cannot be presented precisely as a float in @code{D>F}:
12973: @cindex @code{D>F}, @i{d} cannot be presented precisely as a float
12974: The result is rounded to the nearest float.
12975:
12976: @item dividing by zero:
12977: @cindex dividing by zero, floating-point
12978: @cindex floating-point dividing by zero
12979: @cindex floating-point unidentified fault, FP divide-by-zero
12980: @code{-55 throw} (Floating-point unidentified fault)
12981:
12982: @item exponent too big for conversion (@code{DF!}, @code{DF@@}, @code{SF!}, @code{SF@@}):
12983: @cindex exponent too big for conversion (@code{DF!}, @code{DF@@}, @code{SF!}, @code{SF@@})
12984: System dependent. On IEEE-FP based systems the number is converted into
12985: an infinity.
12986:
12987: @item @i{float}<1 (@code{FACOSH}):
12988: @cindex @code{FACOSH}, @i{float}<1
12989: @cindex floating-point unidentified fault, @code{FACOSH}
12990: @code{-55 throw} (Floating-point unidentified fault)
12991:
12992: @item @i{float}=<-1 (@code{FLNP1}):
12993: @cindex @code{FLNP1}, @i{float}=<-1
12994: @cindex floating-point unidentified fault, @code{FLNP1}
12995: @code{-55 throw} (Floating-point unidentified fault). On IEEE-FP systems
12996: negative infinity is typically produced for @i{float}=-1.
12997:
12998: @item @i{float}=<0 (@code{FLN}, @code{FLOG}):
12999: @cindex @code{FLN}, @i{float}=<0
13000: @cindex @code{FLOG}, @i{float}=<0
13001: @cindex floating-point unidentified fault, @code{FLN} or @code{FLOG}
13002: @code{-55 throw} (Floating-point unidentified fault). On IEEE-FP systems
13003: negative infinity is typically produced for @i{float}=0.
13004:
13005: @item @i{float}<0 (@code{FASINH}, @code{FSQRT}):
13006: @cindex @code{FASINH}, @i{float}<0
13007: @cindex @code{FSQRT}, @i{float}<0
13008: @cindex floating-point unidentified fault, @code{FASINH} or @code{FSQRT}
13009: @code{-55 throw} (Floating-point unidentified fault). @code{fasinh}
13010: produces values for these inputs on my Linux box (Bug in the C library?)
13011:
13012: @item |@i{float}|>1 (@code{FACOS}, @code{FASIN}, @code{FATANH}):
13013: @cindex @code{FACOS}, |@i{float}|>1
13014: @cindex @code{FASIN}, |@i{float}|>1
13015: @cindex @code{FATANH}, |@i{float}|>1
13016: @cindex floating-point unidentified fault, @code{FACOS}, @code{FASIN} or @code{FATANH}
13017: @code{-55 throw} (Floating-point unidentified fault).
13018:
13019: @item integer part of float cannot be represented by @i{d} in @code{F>D}:
13020: @cindex @code{F>D}, integer part of float cannot be represented by @i{d}
13021: @cindex floating-point unidentified fault, @code{F>D}
13022: @code{-55 throw} (Floating-point unidentified fault).
13023:
13024: @item string larger than pictured numeric output area (@code{f.}, @code{fe.}, @code{fs.}):
13025: @cindex string larger than pictured numeric output area (@code{f.}, @code{fe.}, @code{fs.})
13026: This does not happen.
13027: @end table
13028:
13029: @c =====================================================================
13030: @node The optional Locals word set, The optional Memory-Allocation word set, The optional Floating-Point word set, ANS conformance
13031: @section The optional Locals word set
13032: @c =====================================================================
13033: @cindex system documentation, locals words
13034: @cindex locals words, system documentation
13035:
13036: @menu
13037: * locals-idef:: Implementation Defined Options
13038: * locals-ambcond:: Ambiguous Conditions
13039: @end menu
13040:
13041:
13042: @c ---------------------------------------------------------------------
13043: @node locals-idef, locals-ambcond, The optional Locals word set, The optional Locals word set
13044: @subsection Implementation Defined Options
13045: @c ---------------------------------------------------------------------
13046: @cindex implementation-defined options, locals words
13047: @cindex locals words, implementation-defined options
13048:
13049: @table @i
13050: @item maximum number of locals in a definition:
13051: @cindex maximum number of locals in a definition
13052: @cindex locals, maximum number in a definition
13053: @code{s" #locals" environment? drop .}. Currently 15. This is a lower
13054: bound, e.g., on a 32-bit machine there can be 41 locals of up to 8
13055: characters. The number of locals in a definition is bounded by the size
13056: of locals-buffer, which contains the names of the locals.
13057:
13058: @end table
13059:
13060:
13061: @c ---------------------------------------------------------------------
13062: @node locals-ambcond, , locals-idef, The optional Locals word set
13063: @subsection Ambiguous conditions
13064: @c ---------------------------------------------------------------------
13065: @cindex locals words, ambiguous conditions
13066: @cindex ambiguous conditions, locals words
13067:
13068: @table @i
13069: @item executing a named local in interpretation state:
13070: @cindex local in interpretation state
13071: @cindex Interpreting a compile-only word, for a local
13072: Locals have no interpretation semantics. If you try to perform the
13073: interpretation semantics, you will get a @code{-14 throw} somewhere
13074: (Interpreting a compile-only word). If you perform the compilation
13075: semantics, the locals access will be compiled (irrespective of state).
13076:
13077: @item @i{name} not defined by @code{VALUE} or @code{(LOCAL)} (@code{TO}):
13078: @cindex name not defined by @code{VALUE} or @code{(LOCAL)} used by @code{TO}
13079: @cindex @code{TO} on non-@code{VALUE}s and non-locals
13080: @cindex Invalid name argument, @code{TO}
13081: @code{-32 throw} (Invalid name argument)
13082:
13083: @end table
13084:
13085:
13086: @c =====================================================================
13087: @node The optional Memory-Allocation word set, The optional Programming-Tools word set, The optional Locals word set, ANS conformance
13088: @section The optional Memory-Allocation word set
13089: @c =====================================================================
13090: @cindex system documentation, memory-allocation words
13091: @cindex memory-allocation words, system documentation
13092:
13093: @menu
13094: * memory-idef:: Implementation Defined Options
13095: @end menu
13096:
13097:
13098: @c ---------------------------------------------------------------------
13099: @node memory-idef, , The optional Memory-Allocation word set, The optional Memory-Allocation word set
13100: @subsection Implementation Defined Options
13101: @c ---------------------------------------------------------------------
13102: @cindex implementation-defined options, memory-allocation words
13103: @cindex memory-allocation words, implementation-defined options
13104:
13105: @table @i
13106: @item values and meaning of @i{ior}:
13107: @cindex @i{ior} values and meaning
13108: The @i{ior}s returned by the file and memory allocation words are
13109: intended as throw codes. They typically are in the range
13110: -512@minus{}-2047 of OS errors. The mapping from OS error numbers to
13111: @i{ior}s is -512@minus{}@i{errno}.
13112:
13113: @end table
13114:
13115: @c =====================================================================
13116: @node The optional Programming-Tools word set, The optional Search-Order word set, The optional Memory-Allocation word set, ANS conformance
13117: @section The optional Programming-Tools word set
13118: @c =====================================================================
13119: @cindex system documentation, programming-tools words
13120: @cindex programming-tools words, system documentation
13121:
13122: @menu
13123: * programming-idef:: Implementation Defined Options
13124: * programming-ambcond:: Ambiguous Conditions
13125: @end menu
13126:
13127:
13128: @c ---------------------------------------------------------------------
13129: @node programming-idef, programming-ambcond, The optional Programming-Tools word set, The optional Programming-Tools word set
13130: @subsection Implementation Defined Options
13131: @c ---------------------------------------------------------------------
13132: @cindex implementation-defined options, programming-tools words
13133: @cindex programming-tools words, implementation-defined options
13134:
13135: @table @i
13136: @item ending sequence for input following @code{;CODE} and @code{CODE}:
13137: @cindex @code{;CODE} ending sequence
13138: @cindex @code{CODE} ending sequence
13139: @code{END-CODE}
13140:
13141: @item manner of processing input following @code{;CODE} and @code{CODE}:
13142: @cindex @code{;CODE}, processing input
13143: @cindex @code{CODE}, processing input
13144: The @code{ASSEMBLER} vocabulary is pushed on the search order stack, and
13145: the input is processed by the text interpreter, (starting) in interpret
13146: state.
13147:
13148: @item search order capability for @code{EDITOR} and @code{ASSEMBLER}:
13149: @cindex @code{ASSEMBLER}, search order capability
13150: The ANS Forth search order word set.
13151:
13152: @item source and format of display by @code{SEE}:
13153: @cindex @code{SEE}, source and format of output
13154: The source for @code{see} is the intermediate code used by the inner
13155: interpreter. The current @code{see} tries to output Forth source code
13156: as well as possible.
13157:
13158: @end table
13159:
13160: @c ---------------------------------------------------------------------
13161: @node programming-ambcond, , programming-idef, The optional Programming-Tools word set
13162: @subsection Ambiguous conditions
13163: @c ---------------------------------------------------------------------
13164: @cindex programming-tools words, ambiguous conditions
13165: @cindex ambiguous conditions, programming-tools words
13166:
13167: @table @i
13168:
13169: @item deleting the compilation word list (@code{FORGET}):
13170: @cindex @code{FORGET}, deleting the compilation word list
13171: Not implemented (yet).
13172:
13173: @item fewer than @i{u}+1 items on the control-flow stack (@code{CS-PICK}, @code{CS-ROLL}):
13174: @cindex @code{CS-PICK}, fewer than @i{u}+1 items on the control flow-stack
13175: @cindex @code{CS-ROLL}, fewer than @i{u}+1 items on the control flow-stack
13176: @cindex control-flow stack underflow
13177: This typically results in an @code{abort"} with a descriptive error
13178: message (may change into a @code{-22 throw} (Control structure mismatch)
13179: in the future). You may also get a memory access error. If you are
13180: unlucky, this ambiguous condition is not caught.
13181:
13182: @item @i{name} can't be found (@code{FORGET}):
13183: @cindex @code{FORGET}, @i{name} can't be found
13184: Not implemented (yet).
13185:
13186: @item @i{name} not defined via @code{CREATE}:
13187: @cindex @code{;CODE}, @i{name} not defined via @code{CREATE}
13188: @code{;CODE} behaves like @code{DOES>} in this respect, i.e., it changes
13189: the execution semantics of the last defined word no matter how it was
13190: defined.
13191:
13192: @item @code{POSTPONE} applied to @code{[IF]}:
13193: @cindex @code{POSTPONE} applied to @code{[IF]}
13194: @cindex @code{[IF]} and @code{POSTPONE}
13195: After defining @code{: X POSTPONE [IF] ; IMMEDIATE}. @code{X} is
13196: equivalent to @code{[IF]}.
13197:
13198: @item reaching the end of the input source before matching @code{[ELSE]} or @code{[THEN]}:
13199: @cindex @code{[IF]}, end of the input source before matching @code{[ELSE]} or @code{[THEN]}
13200: Continue in the same state of conditional compilation in the next outer
13201: input source. Currently there is no warning to the user about this.
13202:
13203: @item removing a needed definition (@code{FORGET}):
13204: @cindex @code{FORGET}, removing a needed definition
13205: Not implemented (yet).
13206:
13207: @end table
13208:
13209:
13210: @c =====================================================================
13211: @node The optional Search-Order word set, , The optional Programming-Tools word set, ANS conformance
13212: @section The optional Search-Order word set
13213: @c =====================================================================
13214: @cindex system documentation, search-order words
13215: @cindex search-order words, system documentation
13216:
13217: @menu
13218: * search-idef:: Implementation Defined Options
13219: * search-ambcond:: Ambiguous Conditions
13220: @end menu
13221:
13222:
13223: @c ---------------------------------------------------------------------
13224: @node search-idef, search-ambcond, The optional Search-Order word set, The optional Search-Order word set
13225: @subsection Implementation Defined Options
13226: @c ---------------------------------------------------------------------
13227: @cindex implementation-defined options, search-order words
13228: @cindex search-order words, implementation-defined options
13229:
13230: @table @i
13231: @item maximum number of word lists in search order:
13232: @cindex maximum number of word lists in search order
13233: @cindex search order, maximum depth
13234: @code{s" wordlists" environment? drop .}. Currently 16.
13235:
13236: @item minimum search order:
13237: @cindex minimum search order
13238: @cindex search order, minimum
13239: @code{root root}.
13240:
13241: @end table
13242:
13243: @c ---------------------------------------------------------------------
13244: @node search-ambcond, , search-idef, The optional Search-Order word set
13245: @subsection Ambiguous conditions
13246: @c ---------------------------------------------------------------------
13247: @cindex search-order words, ambiguous conditions
13248: @cindex ambiguous conditions, search-order words
13249:
13250: @table @i
13251: @item changing the compilation word list (during compilation):
13252: @cindex changing the compilation word list (during compilation)
13253: @cindex compilation word list, change before definition ends
13254: The word is entered into the word list that was the compilation word list
13255: at the start of the definition. Any changes to the name field (e.g.,
13256: @code{immediate}) or the code field (e.g., when executing @code{DOES>})
13257: are applied to the latest defined word (as reported by @code{last} or
13258: @code{lastxt}), if possible, irrespective of the compilation word list.
13259:
13260: @item search order empty (@code{previous}):
13261: @cindex @code{previous}, search order empty
13262: @cindex vocstack empty, @code{previous}
13263: @code{abort" Vocstack empty"}.
13264:
13265: @item too many word lists in search order (@code{also}):
13266: @cindex @code{also}, too many word lists in search order
13267: @cindex vocstack full, @code{also}
13268: @code{abort" Vocstack full"}.
13269:
13270: @end table
13271:
13272: @c ***************************************************************
13273: @node Standard vs Extensions, Model, ANS conformance, Top
13274: @chapter Should I use Gforth extensions?
13275: @cindex Gforth extensions
13276:
13277: As you read through the rest of this manual, you will see documentation
13278: for @i{Standard} words, and documentation for some appealing Gforth
13279: @i{extensions}. You might ask yourself the question: @i{``Should I
13280: restrict myself to the standard, or should I use the extensions?''}
13281:
13282: The answer depends on the goals you have for the program you are working
13283: on:
13284:
13285: @itemize @bullet
13286:
13287: @item Is it just for yourself or do you want to share it with others?
13288:
13289: @item
13290: If you want to share it, do the others all use Gforth?
13291:
13292: @item
13293: If it is just for yourself, do you want to restrict yourself to Gforth?
13294:
13295: @end itemize
13296:
13297: If restricting the program to Gforth is ok, then there is no reason not
13298: to use extensions. It is still a good idea to keep to the standard
13299: where it is easy, in case you want to reuse these parts in another
13300: program that you want to be portable.
13301:
13302: If you want to be able to port the program to other Forth systems, there
13303: are the following points to consider:
13304:
13305: @itemize @bullet
13306:
13307: @item
13308: Most Forth systems that are being maintained support the ANS Forth
13309: standard. So if your program complies with the standard, it will be
13310: portable among many systems.
13311:
13312: @item
13313: A number of the Gforth extensions can be implemented in ANS Forth using
13314: public-domain files provided in the @file{compat/} directory. These are
13315: mentioned in the text in passing. There is no reason not to use these
13316: extensions, your program will still be ANS Forth compliant; just include
13317: the appropriate compat files with your program.
13318:
13319: @item
13320: The tool @file{ans-report.fs} (@pxref{ANS Report}) makes it easy to
13321: analyse your program and determine what non-Standard words it relies
13322: upon. However, it does not check whether you use standard words in a
13323: non-standard way.
13324:
13325: @item
13326: Some techniques are not standardized by ANS Forth, and are hard or
13327: impossible to implement in a standard way, but can be implemented in
13328: most Forth systems easily, and usually in similar ways (e.g., accessing
13329: word headers). Forth has a rich historical precedent for programmers
13330: taking advantage of implementation-dependent features of their tools
13331: (for example, relying on a knowledge of the dictionary
13332: structure). Sometimes these techniques are necessary to extract every
13333: last bit of performance from the hardware, sometimes they are just a
13334: programming shorthand.
13335:
13336: @item
13337: Does using a Gforth extension save more work than the porting this part
13338: to other Forth systems (if any) will cost?
13339:
13340: @item
13341: Is the additional functionality worth the reduction in portability and
13342: the additional porting problems?
13343:
13344: @end itemize
13345:
13346: In order to perform these consideratios, you need to know what's
13347: standard and what's not. This manual generally states if something is
13348: non-standard, but the authoritative source is the standard document.
13349: Appendix A of the Standard (@var{Rationale}) provides a valuable insight
13350: into the thought processes of the technical committee.
13351:
13352: Note also that portability between Forth systems is not the only
13353: portability issue; there is also the issue of portability between
13354: different platforms (processor/OS combinations).
13355:
13356: @c ***************************************************************
13357: @node Model, Integrating Gforth, Standard vs Extensions, Top
13358: @chapter Model
13359:
13360: This chapter has yet to be written. It will contain information, on
13361: which internal structures you can rely.
13362:
13363: @c ***************************************************************
13364: @node Integrating Gforth, Emacs and Gforth, Model, Top
13365: @chapter Integrating Gforth into C programs
13366:
13367: This is not yet implemented.
13368:
13369: Several people like to use Forth as scripting language for applications
13370: that are otherwise written in C, C++, or some other language.
13371:
13372: The Forth system ATLAST provides facilities for embedding it into
13373: applications; unfortunately it has several disadvantages: most
13374: importantly, it is not based on ANS Forth, and it is apparently dead
13375: (i.e., not developed further and not supported). The facilities
13376: provided by Gforth in this area are inspired by ATLAST's facilities, so
13377: making the switch should not be hard.
13378:
13379: We also tried to design the interface such that it can easily be
13380: implemented by other Forth systems, so that we may one day arrive at a
13381: standardized interface. Such a standard interface would allow you to
13382: replace the Forth system without having to rewrite C code.
13383:
13384: You embed the Gforth interpreter by linking with the library
13385: @code{libgforth.a} (give the compiler the option @code{-lgforth}). All
13386: global symbols in this library that belong to the interface, have the
13387: prefix @code{forth_}. (Global symbols that are used internally have the
13388: prefix @code{gforth_}).
13389:
13390: You can include the declarations of Forth types and the functions and
13391: variables of the interface with @code{#include <forth.h>}.
13392:
13393: Types.
13394:
13395: Variables.
13396:
13397: Data and FP Stack pointer. Area sizes.
13398:
13399: functions.
13400:
13401: forth_init(imagefile)
13402: forth_evaluate(string) exceptions?
13403: forth_goto(address) (or forth_execute(xt)?)
13404: forth_continue() (a corountining mechanism)
13405:
13406: Adding primitives.
13407:
13408: No checking.
13409:
13410: Signals?
13411:
13412: Accessing the Stacks
13413:
13414: @c ******************************************************************
13415: @node Emacs and Gforth, Image Files, Integrating Gforth, Top
13416: @chapter Emacs and Gforth
13417: @cindex Emacs and Gforth
13418:
13419: @cindex @file{gforth.el}
13420: @cindex @file{forth.el}
13421: @cindex Rydqvist, Goran
13422: @cindex comment editing commands
13423: @cindex @code{\}, editing with Emacs
13424: @cindex debug tracer editing commands
13425: @cindex @code{~~}, removal with Emacs
13426: @cindex Forth mode in Emacs
13427: Gforth comes with @file{gforth.el}, an improved version of
13428: @file{forth.el} by Goran Rydqvist (included in the TILE package). The
13429: improvements are:
13430:
13431: @itemize @bullet
13432: @item
13433: A better (but still not perfect) handling of indentation.
13434: @item
13435: Comment paragraph filling (@kbd{M-q})
13436: @item
13437: Commenting (@kbd{C-x \}) and uncommenting (@kbd{C-u C-x \}) of regions
13438: @item
13439: Removal of debugging tracers (@kbd{C-x ~}, @pxref{Debugging}).
13440: @item
13441: Support of the @code{info-lookup} feature for looking up the
13442: documentation of a word.
13443: @end itemize
13444:
13445: I left the stuff I do not use alone, even though some of it only makes
13446: sense for TILE. To get a description of these features, enter Forth mode
13447: and type @kbd{C-h m}.
13448:
13449: @cindex source location of error or debugging output in Emacs
13450: @cindex error output, finding the source location in Emacs
13451: @cindex debugging output, finding the source location in Emacs
13452: In addition, Gforth supports Emacs quite well: The source code locations
13453: given in error messages, debugging output (from @code{~~}) and failed
13454: assertion messages are in the right format for Emacs' compilation mode
13455: (@pxref{Compilation, , Running Compilations under Emacs, emacs, Emacs
13456: Manual}) so the source location corresponding to an error or other
13457: message is only a few keystrokes away (@kbd{C-x `} for the next error,
13458: @kbd{C-c C-c} for the error under the cursor).
13459:
13460: @cindex @file{TAGS} file
13461: @cindex @file{etags.fs}
13462: @cindex viewing the source of a word in Emacs
13463: @cindex @code{require}, placement in files
13464: @cindex @code{include}, placement in files
13465: Also, if you @code{require} @file{etags.fs}, a new @file{TAGS} file will
13466: be produced (@pxref{Tags, , Tags Tables, emacs, Emacs Manual}) that
13467: contains the definitions of all words defined afterwards. You can then
13468: find the source for a word using @kbd{M-.}. Note that emacs can use
13469: several tags files at the same time (e.g., one for the Gforth sources
13470: and one for your program, @pxref{Select Tags Table,,Selecting a Tags
13471: Table,emacs, Emacs Manual}). The TAGS file for the preloaded words is
13472: @file{$(datadir)/gforth/$(VERSION)/TAGS} (e.g.,
13473: @file{/usr/local/share/gforth/0.2.0/TAGS}). To get the best behaviour
13474: with @file{etags.fs}, you should avoid putting definitions both before
13475: and after @code{require} etc., otherwise you will see the same file
13476: visited several times by commands like @code{tags-search}.
13477:
13478: @cindex viewing the documentation of a word in Emacs
13479: @cindex context-sensitive help
13480: Moreover, for words documented in this manual, you can look up the
13481: glossary entry quickly by using @kbd{C-h TAB}
13482: (@code{info-lookup-symbol}, see @pxref{Documentation, ,Documentation
13483: Commands, emacs, Emacs Manual}). This feature requires Emacs 20.3 or
13484: later and does not work for words containing @code{:}.
13485:
13486:
13487: @cindex @file{.emacs}
13488: To get all these benefits, add the following lines to your @file{.emacs}
13489: file:
13490:
13491: @example
13492: (autoload 'forth-mode "gforth.el")
13493: (setq auto-mode-alist (cons '("\\.fs\\'" . forth-mode) auto-mode-alist))
13494: @end example
13495:
13496: @c ******************************************************************
13497: @node Image Files, Engine, Emacs and Gforth, Top
13498: @chapter Image Files
13499: @cindex image file
13500: @cindex @file{.fi} files
13501: @cindex precompiled Forth code
13502: @cindex dictionary in persistent form
13503: @cindex persistent form of dictionary
13504:
13505: An image file is a file containing an image of the Forth dictionary,
13506: i.e., compiled Forth code and data residing in the dictionary. By
13507: convention, we use the extension @code{.fi} for image files.
13508:
13509: @menu
13510: * Image Licensing Issues:: Distribution terms for images.
13511: * Image File Background:: Why have image files?
13512: * Non-Relocatable Image Files:: don't always work.
13513: * Data-Relocatable Image Files:: are better.
13514: * Fully Relocatable Image Files:: better yet.
13515: * Stack and Dictionary Sizes:: Setting the default sizes for an image.
13516: * Running Image Files:: @code{gforth -i @i{file}} or @i{file}.
13517: * Modifying the Startup Sequence:: and turnkey applications.
13518: @end menu
13519:
13520: @node Image Licensing Issues, Image File Background, Image Files, Image Files
13521: @section Image Licensing Issues
13522: @cindex license for images
13523: @cindex image license
13524:
13525: An image created with @code{gforthmi} (@pxref{gforthmi}) or
13526: @code{savesystem} (@pxref{Non-Relocatable Image Files}) includes the
13527: original image; i.e., according to copyright law it is a derived work of
13528: the original image.
13529:
13530: Since Gforth is distributed under the GNU GPL, the newly created image
13531: falls under the GNU GPL, too. In particular, this means that if you
13532: distribute the image, you have to make all of the sources for the image
13533: available, including those you wrote. For details see @ref{License, ,
13534: GNU General Public License (Section 3)}.
13535:
13536: If you create an image with @code{cross} (@pxref{cross.fs}), the image
13537: contains only code compiled from the sources you gave it; if none of
13538: these sources is under the GPL, the terms discussed above do not apply
13539: to the image. However, if your image needs an engine (a gforth binary)
13540: that is under the GPL, you should make sure that you distribute both in
13541: a way that is at most a @emph{mere aggregation}, if you don't want the
13542: terms of the GPL to apply to the image.
13543:
13544: @node Image File Background, Non-Relocatable Image Files, Image Licensing Issues, Image Files
13545: @section Image File Background
13546: @cindex image file background
13547:
13548: Our Forth system consists not only of primitives, but also of
13549: definitions written in Forth. Since the Forth compiler itself belongs to
13550: those definitions, it is not possible to start the system with the
13551: primitives and the Forth source alone. Therefore we provide the Forth
13552: code as an image file in nearly executable form. When Gforth starts up,
13553: a C routine loads the image file into memory, optionally relocates the
13554: addresses, then sets up the memory (stacks etc.) according to
13555: information in the image file, and (finally) starts executing Forth
13556: code.
13557:
13558: The image file variants represent different compromises between the
13559: goals of making it easy to generate image files and making them
13560: portable.
13561:
13562: @cindex relocation at run-time
13563: Win32Forth 3.4 and Mitch Bradley's @code{cforth} use relocation at
13564: run-time. This avoids many of the complications discussed below (image
13565: files are data relocatable without further ado), but costs performance
13566: (one addition per memory access).
13567:
13568: @cindex relocation at load-time
13569: By contrast, the Gforth loader performs relocation at image load time. The
13570: loader also has to replace tokens that represent primitive calls with the
13571: appropriate code-field addresses (or code addresses in the case of
13572: direct threading).
13573:
13574: There are three kinds of image files, with different degrees of
13575: relocatability: non-relocatable, data-relocatable, and fully relocatable
13576: image files.
13577:
13578: @cindex image file loader
13579: @cindex relocating loader
13580: @cindex loader for image files
13581: These image file variants have several restrictions in common; they are
13582: caused by the design of the image file loader:
13583:
13584: @itemize @bullet
13585: @item
13586: There is only one segment; in particular, this means, that an image file
13587: cannot represent @code{ALLOCATE}d memory chunks (and pointers to
13588: them). The contents of the stacks are not represented, either.
13589:
13590: @item
13591: The only kinds of relocation supported are: adding the same offset to
13592: all cells that represent data addresses; and replacing special tokens
13593: with code addresses or with pieces of machine code.
13594:
13595: If any complex computations involving addresses are performed, the
13596: results cannot be represented in the image file. Several applications that
13597: use such computations come to mind:
13598: @itemize @minus
13599: @item
13600: Hashing addresses (or data structures which contain addresses) for table
13601: lookup. If you use Gforth's @code{table}s or @code{wordlist}s for this
13602: purpose, you will have no problem, because the hash tables are
13603: recomputed automatically when the system is started. If you use your own
13604: hash tables, you will have to do something similar.
13605:
13606: @item
13607: There's a cute implementation of doubly-linked lists that uses
13608: @code{XOR}ed addresses. You could represent such lists as singly-linked
13609: in the image file, and restore the doubly-linked representation on
13610: startup.@footnote{In my opinion, though, you should think thrice before
13611: using a doubly-linked list (whatever implementation).}
13612:
13613: @item
13614: The code addresses of run-time routines like @code{docol:} cannot be
13615: represented in the image file (because their tokens would be replaced by
13616: machine code in direct threaded implementations). As a workaround,
13617: compute these addresses at run-time with @code{>code-address} from the
13618: executions tokens of appropriate words (see the definitions of
13619: @code{docol:} and friends in @file{kernel.fs}).
13620:
13621: @item
13622: On many architectures addresses are represented in machine code in some
13623: shifted or mangled form. You cannot put @code{CODE} words that contain
13624: absolute addresses in this form in a relocatable image file. Workarounds
13625: are representing the address in some relative form (e.g., relative to
13626: the CFA, which is present in some register), or loading the address from
13627: a place where it is stored in a non-mangled form.
13628: @end itemize
13629: @end itemize
13630:
13631: @node Non-Relocatable Image Files, Data-Relocatable Image Files, Image File Background, Image Files
13632: @section Non-Relocatable Image Files
13633: @cindex non-relocatable image files
13634: @cindex image file, non-relocatable
13635:
13636: These files are simple memory dumps of the dictionary. They are specific
13637: to the executable (i.e., @file{gforth} file) they were created
13638: with. What's worse, they are specific to the place on which the
13639: dictionary resided when the image was created. Now, there is no
13640: guarantee that the dictionary will reside at the same place the next
13641: time you start Gforth, so there's no guarantee that a non-relocatable
13642: image will work the next time (Gforth will complain instead of crashing,
13643: though).
13644:
13645: You can create a non-relocatable image file with
13646:
13647:
13648: doc-savesystem
13649:
13650:
13651: @node Data-Relocatable Image Files, Fully Relocatable Image Files, Non-Relocatable Image Files, Image Files
13652: @section Data-Relocatable Image Files
13653: @cindex data-relocatable image files
13654: @cindex image file, data-relocatable
13655:
13656: These files contain relocatable data addresses, but fixed code addresses
13657: (instead of tokens). They are specific to the executable (i.e.,
13658: @file{gforth} file) they were created with. For direct threading on some
13659: architectures (e.g., the i386), data-relocatable images do not work. You
13660: get a data-relocatable image, if you use @file{gforthmi} with a
13661: Gforth binary that is not doubly indirect threaded (@pxref{Fully
13662: Relocatable Image Files}).
13663:
13664: @node Fully Relocatable Image Files, Stack and Dictionary Sizes, Data-Relocatable Image Files, Image Files
13665: @section Fully Relocatable Image Files
13666: @cindex fully relocatable image files
13667: @cindex image file, fully relocatable
13668:
13669: @cindex @file{kern*.fi}, relocatability
13670: @cindex @file{gforth.fi}, relocatability
13671: These image files have relocatable data addresses, and tokens for code
13672: addresses. They can be used with different binaries (e.g., with and
13673: without debugging) on the same machine, and even across machines with
13674: the same data formats (byte order, cell size, floating point
13675: format). However, they are usually specific to the version of Gforth
13676: they were created with. The files @file{gforth.fi} and @file{kernl*.fi}
13677: are fully relocatable.
13678:
13679: There are two ways to create a fully relocatable image file:
13680:
13681: @menu
13682: * gforthmi:: The normal way
13683: * cross.fs:: The hard way
13684: @end menu
13685:
13686: @node gforthmi, cross.fs, Fully Relocatable Image Files, Fully Relocatable Image Files
13687: @subsection @file{gforthmi}
13688: @cindex @file{comp-i.fs}
13689: @cindex @file{gforthmi}
13690:
13691: You will usually use @file{gforthmi}. If you want to create an
13692: image @i{file} that contains everything you would load by invoking
13693: Gforth with @code{gforth @i{options}}, you simply say:
13694: @example
13695: gforthmi @i{file} @i{options}
13696: @end example
13697:
13698: E.g., if you want to create an image @file{asm.fi} that has the file
13699: @file{asm.fs} loaded in addition to the usual stuff, you could do it
13700: like this:
13701:
13702: @example
13703: gforthmi asm.fi asm.fs
13704: @end example
13705:
13706: @file{gforthmi} is implemented as a sh script and works like this: It
13707: produces two non-relocatable images for different addresses and then
13708: compares them. Its output reflects this: first you see the output (if
13709: any) of the two Gforth invocations that produce the non-relocatable image
13710: files, then you see the output of the comparing program: It displays the
13711: offset used for data addresses and the offset used for code addresses;
13712: moreover, for each cell that cannot be represented correctly in the
13713: image files, it displays a line like this:
13714:
13715: @example
13716: 78DC BFFFFA50 BFFFFA40
13717: @end example
13718:
13719: This means that at offset $78dc from @code{forthstart}, one input image
13720: contains $bffffa50, and the other contains $bffffa40. Since these cells
13721: cannot be represented correctly in the output image, you should examine
13722: these places in the dictionary and verify that these cells are dead
13723: (i.e., not read before they are written).
13724:
13725: @cindex --application, @code{gforthmi} option
13726: If you insert the option @code{--application} in front of the image file
13727: name, you will get an image that uses the @code{--appl-image} option
13728: instead of the @code{--image-file} option (@pxref{Invoking
13729: Gforth}). When you execute such an image on Unix (by typing the image
13730: name as command), the Gforth engine will pass all options to the image
13731: instead of trying to interpret them as engine options.
13732:
13733: If you type @file{gforthmi} with no arguments, it prints some usage
13734: instructions.
13735:
13736: @cindex @code{savesystem} during @file{gforthmi}
13737: @cindex @code{bye} during @file{gforthmi}
13738: @cindex doubly indirect threaded code
13739: @cindex environment variables
13740: @cindex @code{GFORTHD} -- environment variable
13741: @cindex @code{GFORTH} -- environment variable
13742: @cindex @code{gforth-ditc}
13743: There are a few wrinkles: After processing the passed @i{options}, the
13744: words @code{savesystem} and @code{bye} must be visible. A special doubly
13745: indirect threaded version of the @file{gforth} executable is used for
13746: creating the non-relocatable images; you can pass the exact filename of
13747: this executable through the environment variable @code{GFORTHD}
13748: (default: @file{gforth-ditc}); if you pass a version that is not doubly
13749: indirect threaded, you will not get a fully relocatable image, but a
13750: data-relocatable image (because there is no code address offset). The
13751: normal @file{gforth} executable is used for creating the relocatable
13752: image; you can pass the exact filename of this executable through the
13753: environment variable @code{GFORTH}.
13754:
13755: @node cross.fs, , gforthmi, Fully Relocatable Image Files
13756: @subsection @file{cross.fs}
13757: @cindex @file{cross.fs}
13758: @cindex cross-compiler
13759: @cindex metacompiler
13760: @cindex target compiler
13761:
13762: You can also use @code{cross}, a batch compiler that accepts a Forth-like
13763: programming language (@pxref{Cross Compiler}).
13764:
13765: @code{cross} allows you to create image files for machines with
13766: different data sizes and data formats than the one used for generating
13767: the image file. You can also use it to create an application image that
13768: does not contain a Forth compiler. These features are bought with
13769: restrictions and inconveniences in programming. E.g., addresses have to
13770: be stored in memory with special words (@code{A!}, @code{A,}, etc.) in
13771: order to make the code relocatable.
13772:
13773:
13774: @node Stack and Dictionary Sizes, Running Image Files, Fully Relocatable Image Files, Image Files
13775: @section Stack and Dictionary Sizes
13776: @cindex image file, stack and dictionary sizes
13777: @cindex dictionary size default
13778: @cindex stack size default
13779:
13780: If you invoke Gforth with a command line flag for the size
13781: (@pxref{Invoking Gforth}), the size you specify is stored in the
13782: dictionary. If you save the dictionary with @code{savesystem} or create
13783: an image with @file{gforthmi}, this size will become the default
13784: for the resulting image file. E.g., the following will create a
13785: fully relocatable version of @file{gforth.fi} with a 1MB dictionary:
13786:
13787: @example
13788: gforthmi gforth.fi -m 1M
13789: @end example
13790:
13791: In other words, if you want to set the default size for the dictionary
13792: and the stacks of an image, just invoke @file{gforthmi} with the
13793: appropriate options when creating the image.
13794:
13795: @cindex stack size, cache-friendly
13796: Note: For cache-friendly behaviour (i.e., good performance), you should
13797: make the sizes of the stacks modulo, say, 2K, somewhat different. E.g.,
13798: the default stack sizes are: data: 16k (mod 2k=0); fp: 15.5k (mod
13799: 2k=1.5k); return: 15k(mod 2k=1k); locals: 14.5k (mod 2k=0.5k).
13800:
13801: @node Running Image Files, Modifying the Startup Sequence, Stack and Dictionary Sizes, Image Files
13802: @section Running Image Files
13803: @cindex running image files
13804: @cindex invoking image files
13805: @cindex image file invocation
13806:
13807: @cindex -i, invoke image file
13808: @cindex --image file, invoke image file
13809: You can invoke Gforth with an image file @i{image} instead of the
13810: default @file{gforth.fi} with the @code{-i} flag (@pxref{Invoking Gforth}):
13811: @example
13812: gforth -i @i{image}
13813: @end example
13814:
13815: @cindex executable image file
13816: @cindex image file, executable
13817: If your operating system supports starting scripts with a line of the
13818: form @code{#! ...}, you just have to type the image file name to start
13819: Gforth with this image file (note that the file extension @code{.fi} is
13820: just a convention). I.e., to run Gforth with the image file @i{image},
13821: you can just type @i{image} instead of @code{gforth -i @i{image}}.
13822: This works because every @code{.fi} file starts with a line of this
13823: format:
13824:
13825: @example
13826: #! /usr/local/bin/gforth-0.4.0 -i
13827: @end example
13828:
13829: The file and pathname for the Gforth engine specified on this line is
13830: the specific Gforth executable that it was built against; i.e. the value
13831: of the environment variable @code{GFORTH} at the time that
13832: @file{gforthmi} was executed.
13833:
13834: You can make use of the same shell capability to make a Forth source
13835: file into an executable. For example, if you place this text in a file:
13836:
13837: @example
13838: #! /usr/local/bin/gforth
13839:
13840: ." Hello, world" CR
13841: bye
13842: @end example
13843:
13844: @noindent
13845: and then make the file executable (chmod +x in Unix), you can run it
13846: directly from the command line. The sequence @code{#!} is used in two
13847: ways; firstly, it is recognised as a ``magic sequence'' by the operating
13848: system@footnote{The Unix kernel actually recognises two types of files:
13849: executable files and files of data, where the data is processed by an
13850: interpreter that is specified on the ``interpreter line'' -- the first
13851: line of the file, starting with the sequence #!. There may be a small
13852: limit (e.g., 32) on the number of characters that may be specified on
13853: the interpreter line.} secondly it is treated as a comment character by
13854: Gforth. Because of the second usage, a space is required between
13855: @code{#!} and the path to the executable.
13856:
13857: The disadvantage of this latter technique, compared with using
13858: @file{gforthmi}, is that it is slower; the Forth source code is compiled
13859: on-the-fly, each time the program is invoked.
13860:
13861:
13862: doc-#!
13863:
13864:
13865: @node Modifying the Startup Sequence, , Running Image Files, Image Files
13866: @section Modifying the Startup Sequence
13867: @cindex startup sequence for image file
13868: @cindex image file initialization sequence
13869: @cindex initialization sequence of image file
13870:
13871: You can add your own initialization to the startup sequence through the
13872: deferred word @code{'cold}. @code{'cold} is invoked just before the
13873: image-specific command line processing (by default, loading files and
13874: evaluating (@code{-e}) strings) starts.
13875:
13876: A sequence for adding your initialization usually looks like this:
13877:
13878: @example
13879: :noname
13880: Defers 'cold \ do other initialization stuff (e.g., rehashing wordlists)
13881: ... \ your stuff
13882: ; IS 'cold
13883: @end example
13884:
13885: @cindex turnkey image files
13886: @cindex image file, turnkey applications
13887: You can make a turnkey image by letting @code{'cold} execute a word
13888: (your turnkey application) that never returns; instead, it exits Gforth
13889: via @code{bye} or @code{throw}.
13890:
13891: @cindex command-line arguments, access
13892: @cindex arguments on the command line, access
13893: You can access the (image-specific) command-line arguments through the
13894: variables @code{argc} and @code{argv}. @code{arg} provides convenient
13895: access to @code{argv}.
13896:
13897: If @code{'cold} exits normally, Gforth processes the command-line
13898: arguments as files to be loaded and strings to be evaluated. Therefore,
13899: @code{'cold} should remove the arguments it has used in this case.
13900:
13901:
13902:
13903: doc-'cold
13904: doc-argc
13905: doc-argv
13906: doc-arg
13907:
13908:
13909:
13910: @c ******************************************************************
13911: @node Engine, Binding to System Library, Image Files, Top
13912: @chapter Engine
13913: @cindex engine
13914: @cindex virtual machine
13915:
13916: Reading this chapter is not necessary for programming with Gforth. It
13917: may be helpful for finding your way in the Gforth sources.
13918:
13919: The ideas in this section have also been published in Bernd Paysan,
13920: @cite{ANS fig/GNU/??? Forth} (in German), Forth-Tagung '93 and M. Anton
13921: Ertl, @cite{@uref{http://www.complang.tuwien.ac.at/papers/ertl93.ps.Z, A
13922: Portable Forth Engine}}, EuroForth '93.
13923:
13924: @menu
13925: * Portability::
13926: * Threading::
13927: * Primitives::
13928: * Performance::
13929: @end menu
13930:
13931: @node Portability, Threading, Engine, Engine
13932: @section Portability
13933: @cindex engine portability
13934:
13935: An important goal of the Gforth Project is availability across a wide
13936: range of personal machines. fig-Forth, and, to a lesser extent, F83,
13937: achieved this goal by manually coding the engine in assembly language
13938: for several then-popular processors. This approach is very
13939: labor-intensive and the results are short-lived due to progress in
13940: computer architecture.
13941:
13942: @cindex C, using C for the engine
13943: Others have avoided this problem by coding in C, e.g., Mitch Bradley
13944: (cforth), Mikael Patel (TILE) and Dirk Zoller (pfe). This approach is
13945: particularly popular for UNIX-based Forths due to the large variety of
13946: architectures of UNIX machines. Unfortunately an implementation in C
13947: does not mix well with the goals of efficiency and with using
13948: traditional techniques: Indirect or direct threading cannot be expressed
13949: in C, and switch threading, the fastest technique available in C, is
13950: significantly slower. Another problem with C is that it is very
13951: cumbersome to express double integer arithmetic.
13952:
13953: @cindex GNU C for the engine
13954: @cindex long long
13955: Fortunately, there is a portable language that does not have these
13956: limitations: GNU C, the version of C processed by the GNU C compiler
13957: (@pxref{C Extensions, , Extensions to the C Language Family, gcc.info,
13958: GNU C Manual}). Its labels as values feature (@pxref{Labels as Values, ,
13959: Labels as Values, gcc.info, GNU C Manual}) makes direct and indirect
13960: threading possible, its @code{long long} type (@pxref{Long Long, ,
13961: Double-Word Integers, gcc.info, GNU C Manual}) corresponds to Forth's
13962: double numbers@footnote{Unfortunately, long longs are not implemented
13963: properly on all machines (e.g., on alpha-osf1, long longs are only 64
13964: bits, the same size as longs (and pointers), but they should be twice as
13965: long according to @pxref{Long Long, , Double-Word Integers, gcc.info, GNU
13966: C Manual}). So, we had to implement doubles in C after all. Still, on
13967: most machines we can use long longs and achieve better performance than
13968: with the emulation package.}. GNU C is available for free on all
13969: important (and many unimportant) UNIX machines, VMS, 80386s running
13970: MS-DOS, the Amiga, and the Atari ST, so a Forth written in GNU C can run
13971: on all these machines.
13972:
13973: Writing in a portable language has the reputation of producing code that
13974: is slower than assembly. For our Forth engine we repeatedly looked at
13975: the code produced by the compiler and eliminated most compiler-induced
13976: inefficiencies by appropriate changes in the source code.
13977:
13978: @cindex explicit register declarations
13979: @cindex --enable-force-reg, configuration flag
13980: @cindex -DFORCE_REG
13981: However, register allocation cannot be portably influenced by the
13982: programmer, leading to some inefficiencies on register-starved
13983: machines. We use explicit register declarations (@pxref{Explicit Reg
13984: Vars, , Variables in Specified Registers, gcc.info, GNU C Manual}) to
13985: improve the speed on some machines. They are turned on by using the
13986: configuration flag @code{--enable-force-reg} (@code{gcc} switch
13987: @code{-DFORCE_REG}). Unfortunately, this feature not only depends on the
13988: machine, but also on the compiler version: On some machines some
13989: compiler versions produce incorrect code when certain explicit register
13990: declarations are used. So by default @code{-DFORCE_REG} is not used.
13991:
13992: @node Threading, Primitives, Portability, Engine
13993: @section Threading
13994: @cindex inner interpreter implementation
13995: @cindex threaded code implementation
13996:
13997: @cindex labels as values
13998: GNU C's labels as values extension (available since @code{gcc-2.0},
13999: @pxref{Labels as Values, , Labels as Values, gcc.info, GNU C Manual})
14000: makes it possible to take the address of @i{label} by writing
14001: @code{&&@i{label}}. This address can then be used in a statement like
14002: @code{goto *@i{address}}. I.e., @code{goto *&&x} is the same as
14003: @code{goto x}.
14004:
14005: @cindex @code{NEXT}, indirect threaded
14006: @cindex indirect threaded inner interpreter
14007: @cindex inner interpreter, indirect threaded
14008: With this feature an indirect threaded @code{NEXT} looks like:
14009: @example
14010: cfa = *ip++;
14011: ca = *cfa;
14012: goto *ca;
14013: @end example
14014: @cindex instruction pointer
14015: For those unfamiliar with the names: @code{ip} is the Forth instruction
14016: pointer; the @code{cfa} (code-field address) corresponds to ANS Forths
14017: execution token and points to the code field of the next word to be
14018: executed; The @code{ca} (code address) fetched from there points to some
14019: executable code, e.g., a primitive or the colon definition handler
14020: @code{docol}.
14021:
14022: @cindex @code{NEXT}, direct threaded
14023: @cindex direct threaded inner interpreter
14024: @cindex inner interpreter, direct threaded
14025: Direct threading is even simpler:
14026: @example
14027: ca = *ip++;
14028: goto *ca;
14029: @end example
14030:
14031: Of course we have packaged the whole thing neatly in macros called
14032: @code{NEXT} and @code{NEXT1} (the part of @code{NEXT} after fetching the cfa).
14033:
14034: @menu
14035: * Scheduling::
14036: * Direct or Indirect Threaded?::
14037: * DOES>::
14038: @end menu
14039:
14040: @node Scheduling, Direct or Indirect Threaded?, Threading, Threading
14041: @subsection Scheduling
14042: @cindex inner interpreter optimization
14043:
14044: There is a little complication: Pipelined and superscalar processors,
14045: i.e., RISC and some modern CISC machines can process independent
14046: instructions while waiting for the results of an instruction. The
14047: compiler usually reorders (schedules) the instructions in a way that
14048: achieves good usage of these delay slots. However, on our first tries
14049: the compiler did not do well on scheduling primitives. E.g., for
14050: @code{+} implemented as
14051: @example
14052: n=sp[0]+sp[1];
14053: sp++;
14054: sp[0]=n;
14055: NEXT;
14056: @end example
14057: the @code{NEXT} comes strictly after the other code, i.e., there is nearly no
14058: scheduling. After a little thought the problem becomes clear: The
14059: compiler cannot know that @code{sp} and @code{ip} point to different
14060: addresses (and the version of @code{gcc} we used would not know it even
14061: if it was possible), so it could not move the load of the cfa above the
14062: store to the TOS. Indeed the pointers could be the same, if code on or
14063: very near the top of stack were executed. In the interest of speed we
14064: chose to forbid this probably unused ``feature'' and helped the compiler
14065: in scheduling: @code{NEXT} is divided into the loading part (@code{NEXT_P1})
14066: and the goto part (@code{NEXT_P2}). @code{+} now looks like:
14067: @example
14068: n=sp[0]+sp[1];
14069: sp++;
14070: NEXT_P1;
14071: sp[0]=n;
14072: NEXT_P2;
14073: @end example
14074: This can be scheduled optimally by the compiler.
14075:
14076: This division can be turned off with the switch @code{-DCISC_NEXT}. This
14077: switch is on by default on machines that do not profit from scheduling
14078: (e.g., the 80386), in order to preserve registers.
14079:
14080: @node Direct or Indirect Threaded?, DOES>, Scheduling, Threading
14081: @subsection Direct or Indirect Threaded?
14082: @cindex threading, direct or indirect?
14083:
14084: @cindex -DDIRECT_THREADED
14085: Both! After packaging the nasty details in macro definitions we
14086: realized that we could switch between direct and indirect threading by
14087: simply setting a compilation flag (@code{-DDIRECT_THREADED}) and
14088: defining a few machine-specific macros for the direct-threading case.
14089: On the Forth level we also offer access words that hide the
14090: differences between the threading methods (@pxref{Threading Words}).
14091:
14092: Indirect threading is implemented completely machine-independently.
14093: Direct threading needs routines for creating jumps to the executable
14094: code (e.g. to @code{docol} or @code{dodoes}). These routines are inherently
14095: machine-dependent, but they do not amount to many source lines. Therefore,
14096: even porting direct threading to a new machine requires little effort.
14097:
14098: @cindex --enable-indirect-threaded, configuration flag
14099: @cindex --enable-direct-threaded, configuration flag
14100: The default threading method is machine-dependent. You can enforce a
14101: specific threading method when building Gforth with the configuration
14102: flag @code{--enable-direct-threaded} or
14103: @code{--enable-indirect-threaded}. Note that direct threading is not
14104: supported on all machines.
14105:
14106: @node DOES>, , Direct or Indirect Threaded?, Threading
14107: @subsection DOES>
14108: @cindex @code{DOES>} implementation
14109:
14110: @cindex @code{dodoes} routine
14111: @cindex @code{DOES>}-code
14112: One of the most complex parts of a Forth engine is @code{dodoes}, i.e.,
14113: the chunk of code executed by every word defined by a
14114: @code{CREATE}...@code{DOES>} pair. The main problem here is: How to find
14115: the Forth code to be executed, i.e. the code after the
14116: @code{DOES>} (the @code{DOES>}-code)? There are two solutions:
14117:
14118: In fig-Forth the code field points directly to the @code{dodoes} and the
14119: @code{DOES>}-code address is stored in the cell after the code address (i.e. at
14120: @code{@i{CFA} cell+}). It may seem that this solution is illegal in
14121: the Forth-79 and all later standards, because in fig-Forth this address
14122: lies in the body (which is illegal in these standards). However, by
14123: making the code field larger for all words this solution becomes legal
14124: again. We use this approach for the indirect threaded version and for
14125: direct threading on some machines. Leaving a cell unused in most words
14126: is a bit wasteful, but on the machines we are targeting this is hardly a
14127: problem. The other reason for having a code field size of two cells is
14128: to avoid having different image files for direct and indirect threaded
14129: systems (direct threaded systems require two-cell code fields on many
14130: machines).
14131:
14132: @cindex @code{DOES>}-handler
14133: The other approach is that the code field points or jumps to the cell
14134: after @code{DOES>}. In this variant there is a jump to @code{dodoes} at
14135: this address (the @code{DOES>}-handler). @code{dodoes} can then get the
14136: @code{DOES>}-code address by computing the code address, i.e., the address of
14137: the jump to @code{dodoes}, and add the length of that jump field. A variant of
14138: this is to have a call to @code{dodoes} after the @code{DOES>}; then the
14139: return address (which can be found in the return register on RISCs) is
14140: the @code{DOES>}-code address. Since the two cells available in the code field
14141: are used up by the jump to the code address in direct threading on many
14142: architectures, we use this approach for direct threading on these
14143: architectures. We did not want to add another cell to the code field.
14144:
14145: @node Primitives, Performance, Threading, Engine
14146: @section Primitives
14147: @cindex primitives, implementation
14148: @cindex virtual machine instructions, implementation
14149:
14150: @menu
14151: * Automatic Generation::
14152: * TOS Optimization::
14153: * Produced code::
14154: @end menu
14155:
14156: @node Automatic Generation, TOS Optimization, Primitives, Primitives
14157: @subsection Automatic Generation
14158: @cindex primitives, automatic generation
14159:
14160: @cindex @file{prims2x.fs}
14161: Since the primitives are implemented in a portable language, there is no
14162: longer any need to minimize the number of primitives. On the contrary,
14163: having many primitives has an advantage: speed. In order to reduce the
14164: number of errors in primitives and to make programming them easier, we
14165: provide a tool, the primitive generator (@file{prims2x.fs}), that
14166: automatically generates most (and sometimes all) of the C code for a
14167: primitive from the stack effect notation. The source for a primitive
14168: has the following form:
14169:
14170: @cindex primitive source format
14171: @format
14172: @i{Forth-name} ( @i{stack-effect} ) @i{category} [@i{pronounc.}]
14173: [@code{""}@i{glossary entry}@code{""}]
14174: @i{C code}
14175: [@code{:}
14176: @i{Forth code}]
14177: @end format
14178:
14179: The items in brackets are optional. The category and glossary fields
14180: are there for generating the documentation, the Forth code is there
14181: for manual implementations on machines without GNU C. E.g., the source
14182: for the primitive @code{+} is:
14183: @example
14184: + ( n1 n2 -- n ) core plus
14185: n = n1+n2;
14186: @end example
14187:
14188: This looks like a specification, but in fact @code{n = n1+n2} is C
14189: code. Our primitive generation tool extracts a lot of information from
14190: the stack effect notations@footnote{We use a one-stack notation, even
14191: though we have separate data and floating-point stacks; The separate
14192: notation can be generated easily from the unified notation.}: The number
14193: of items popped from and pushed on the stack, their type, and by what
14194: name they are referred to in the C code. It then generates a C code
14195: prelude and postlude for each primitive. The final C code for @code{+}
14196: looks like this:
14197:
14198: @example
14199: I_plus: /* + ( n1 n2 -- n ) */ /* label, stack effect */
14200: /* */ /* documentation */
14201: @{
14202: DEF_CA /* definition of variable ca (indirect threading) */
14203: Cell n1; /* definitions of variables */
14204: Cell n2;
14205: Cell n;
14206: n1 = (Cell) sp[1]; /* input */
14207: n2 = (Cell) TOS;
14208: sp += 1; /* stack adjustment */
14209: NAME("+") /* debugging output (with -DDEBUG) */
14210: @{
14211: n = n1+n2; /* C code taken from the source */
14212: @}
14213: NEXT_P1; /* NEXT part 1 */
14214: TOS = (Cell)n; /* output */
14215: NEXT_P2; /* NEXT part 2 */
14216: @}
14217: @end example
14218:
14219: This looks long and inefficient, but the GNU C compiler optimizes quite
14220: well and produces optimal code for @code{+} on, e.g., the R3000 and the
14221: HP RISC machines: Defining the @code{n}s does not produce any code, and
14222: using them as intermediate storage also adds no cost.
14223:
14224: There are also other optimizations that are not illustrated by this
14225: example: assignments between simple variables are usually for free (copy
14226: propagation). If one of the stack items is not used by the primitive
14227: (e.g. in @code{drop}), the compiler eliminates the load from the stack
14228: (dead code elimination). On the other hand, there are some things that
14229: the compiler does not do, therefore they are performed by
14230: @file{prims2x.fs}: The compiler does not optimize code away that stores
14231: a stack item to the place where it just came from (e.g., @code{over}).
14232:
14233: While programming a primitive is usually easy, there are a few cases
14234: where the programmer has to take the actions of the generator into
14235: account, most notably @code{?dup}, but also words that do not (always)
14236: fall through to @code{NEXT}.
14237:
14238: @node TOS Optimization, Produced code, Automatic Generation, Primitives
14239: @subsection TOS Optimization
14240: @cindex TOS optimization for primitives
14241: @cindex primitives, keeping the TOS in a register
14242:
14243: An important optimization for stack machine emulators, e.g., Forth
14244: engines, is keeping one or more of the top stack items in
14245: registers. If a word has the stack effect @i{in1}...@i{inx} @code{--}
14246: @i{out1}...@i{outy}, keeping the top @i{n} items in registers
14247: @itemize @bullet
14248: @item
14249: is better than keeping @i{n-1} items, if @i{x>=n} and @i{y>=n},
14250: due to fewer loads from and stores to the stack.
14251: @item is slower than keeping @i{n-1} items, if @i{x<>y} and @i{x<n} and
14252: @i{y<n}, due to additional moves between registers.
14253: @end itemize
14254:
14255: @cindex -DUSE_TOS
14256: @cindex -DUSE_NO_TOS
14257: In particular, keeping one item in a register is never a disadvantage,
14258: if there are enough registers. Keeping two items in registers is a
14259: disadvantage for frequent words like @code{?branch}, constants,
14260: variables, literals and @code{i}. Therefore our generator only produces
14261: code that keeps zero or one items in registers. The generated C code
14262: covers both cases; the selection between these alternatives is made at
14263: C-compile time using the switch @code{-DUSE_TOS}. @code{TOS} in the C
14264: code for @code{+} is just a simple variable name in the one-item case,
14265: otherwise it is a macro that expands into @code{sp[0]}. Note that the
14266: GNU C compiler tries to keep simple variables like @code{TOS} in
14267: registers, and it usually succeeds, if there are enough registers.
14268:
14269: @cindex -DUSE_FTOS
14270: @cindex -DUSE_NO_FTOS
14271: The primitive generator performs the TOS optimization for the
14272: floating-point stack, too (@code{-DUSE_FTOS}). For floating-point
14273: operations the benefit of this optimization is even larger:
14274: floating-point operations take quite long on most processors, but can be
14275: performed in parallel with other operations as long as their results are
14276: not used. If the FP-TOS is kept in a register, this works. If
14277: it is kept on the stack, i.e., in memory, the store into memory has to
14278: wait for the result of the floating-point operation, lengthening the
14279: execution time of the primitive considerably.
14280:
14281: The TOS optimization makes the automatic generation of primitives a
14282: bit more complicated. Just replacing all occurrences of @code{sp[0]} by
14283: @code{TOS} is not sufficient. There are some special cases to
14284: consider:
14285: @itemize @bullet
14286: @item In the case of @code{dup ( w -- w w )} the generator must not
14287: eliminate the store to the original location of the item on the stack,
14288: if the TOS optimization is turned on.
14289: @item Primitives with stack effects of the form @code{--}
14290: @i{out1}...@i{outy} must store the TOS to the stack at the start.
14291: Likewise, primitives with the stack effect @i{in1}...@i{inx} @code{--}
14292: must load the TOS from the stack at the end. But for the null stack
14293: effect @code{--} no stores or loads should be generated.
14294: @end itemize
14295:
14296: @node Produced code, , TOS Optimization, Primitives
14297: @subsection Produced code
14298: @cindex primitives, assembly code listing
14299:
14300: @cindex @file{engine.s}
14301: To see what assembly code is produced for the primitives on your machine
14302: with your compiler and your flag settings, type @code{make engine.s} and
14303: look at the resulting file @file{engine.s}.
14304:
14305: @node Performance, , Primitives, Engine
14306: @section Performance
14307: @cindex performance of some Forth interpreters
14308: @cindex engine performance
14309: @cindex benchmarking Forth systems
14310: @cindex Gforth performance
14311:
14312: On RISCs the Gforth engine is very close to optimal; i.e., it is usually
14313: impossible to write a significantly faster engine.
14314:
14315: On register-starved machines like the 386 architecture processors
14316: improvements are possible, because @code{gcc} does not utilize the
14317: registers as well as a human, even with explicit register declarations;
14318: e.g., Bernd Beuster wrote a Forth system fragment in assembly language
14319: and hand-tuned it for the 486; this system is 1.19 times faster on the
14320: Sieve benchmark on a 486DX2/66 than Gforth compiled with
14321: @code{gcc-2.6.3} with @code{-DFORCE_REG}. The situation has improved
14322: with gcc-2.95 and gforth-0.4.9; now the most important virtual machine
14323: registers fit in real registers (and we can even afford to use the TOS
14324: optimization), resulting in a speedup of 1.14 on the sieve over the
14325: earlier results.
14326:
14327: @cindex Win32Forth performance
14328: @cindex NT Forth performance
14329: @cindex eforth performance
14330: @cindex ThisForth performance
14331: @cindex PFE performance
14332: @cindex TILE performance
14333: The potential advantage of assembly language implementations
14334: is not necessarily realized in complete Forth systems: We compared
14335: Gforth-0.4.9 (direct threaded, compiled with @code{gcc-2.95.1} and
14336: @code{-DFORCE_REG}) with Win32Forth 1.2093, LMI's NT Forth (Beta, May
14337: 1994) and Eforth (with and without peephole (aka pinhole) optimization
14338: of the threaded code); all these systems were written in assembly
14339: language. We also compared Gforth with three systems written in C:
14340: PFE-0.9.14 (compiled with @code{gcc-2.6.3} with the default
14341: configuration for Linux: @code{-O2 -fomit-frame-pointer -DUSE_REGS
14342: -DUNROLL_NEXT}), ThisForth Beta (compiled with @code{gcc-2.6.3 -O3
14343: -fomit-frame-pointer}; ThisForth employs peephole optimization of the
14344: threaded code) and TILE (compiled with @code{make opt}). We benchmarked
14345: Gforth, PFE, ThisForth and TILE on a 486DX2/66 under Linux. Kenneth
14346: O'Heskin kindly provided the results for Win32Forth and NT Forth on a
14347: 486DX2/66 with similar memory performance under Windows NT. Marcel
14348: Hendrix ported Eforth to Linux, then extended it to run the benchmarks,
14349: added the peephole optimizer, ran the benchmarks and reported the
14350: results.
14351:
14352: We used four small benchmarks: the ubiquitous Sieve; bubble-sorting and
14353: matrix multiplication come from the Stanford integer benchmarks and have
14354: been translated into Forth by Martin Fraeman; we used the versions
14355: included in the TILE Forth package, but with bigger data set sizes; and
14356: a recursive Fibonacci number computation for benchmarking calling
14357: performance. The following table shows the time taken for the benchmarks
14358: scaled by the time taken by Gforth (in other words, it shows the speedup
14359: factor that Gforth achieved over the other systems).
14360:
14361: @example
14362: relative Win32- NT eforth This-
14363: time Gforth Forth Forth eforth +opt PFE Forth TILE
14364: sieve 1.00 1.58 1.30 1.58 0.97 1.80 3.63 9.79
14365: bubble 1.00 1.55 1.67 1.75 1.04 1.78 4.59
14366: matmul 1.00 1.67 1.53 1.66 0.84 1.79 4.63
14367: fib 1.00 1.75 1.53 1.40 0.99 1.99 3.43 4.93
14368: @end example
14369:
14370: You may be quite surprised by the good performance of Gforth when
14371: compared with systems written in assembly language. One important reason
14372: for the disappointing performance of these other systems is probably
14373: that they are not written optimally for the 486 (e.g., they use the
14374: @code{lods} instruction). In addition, Win32Forth uses a comfortable,
14375: but costly method for relocating the Forth image: like @code{cforth}, it
14376: computes the actual addresses at run time, resulting in two address
14377: computations per @code{NEXT} (@pxref{Image File Background}).
14378:
14379: Only Eforth with the peephole optimizer performs comparable to
14380: Gforth. The speedups achieved with peephole optimization of threaded
14381: code are quite remarkable. Adding a peephole optimizer to Gforth should
14382: cause similar speedups.
14383:
14384: The speedup of Gforth over PFE, ThisForth and TILE can be easily
14385: explained with the self-imposed restriction of the latter systems to
14386: standard C, which makes efficient threading impossible (however, the
14387: measured implementation of PFE uses a GNU C extension: @pxref{Global Reg
14388: Vars, , Defining Global Register Variables, gcc.info, GNU C Manual}).
14389: Moreover, current C compilers have a hard time optimizing other aspects
14390: of the ThisForth and the TILE source.
14391:
14392: The performance of Gforth on 386 architecture processors varies widely
14393: with the version of @code{gcc} used. E.g., @code{gcc-2.5.8} failed to
14394: allocate any of the virtual machine registers into real machine
14395: registers by itself and would not work correctly with explicit register
14396: declarations, giving a 1.5 times slower engine (on a 486DX2/66 running
14397: the Sieve) than the one measured above.
14398:
14399: Note that there have been several releases of Win32Forth since the
14400: release presented here, so the results presented above may have little
14401: predictive value for the performance of Win32Forth today (results for
14402: the current release on an i486DX2/66 are welcome).
14403:
14404: @cindex @file{Benchres}
14405: In
14406: @cite{@uref{http://www.complang.tuwien.ac.at/papers/ertl&maierhofer95.ps.gz,
14407: Translating Forth to Efficient C}} by M. Anton Ertl and Martin
14408: Maierhofer (presented at EuroForth '95), an indirect threaded version of
14409: Gforth is compared with Win32Forth, NT Forth, PFE, ThisForth, and
14410: several native code systems; that version of Gforth is slower on a 486
14411: than the direct threaded version used here. You can find a newer version
14412: of these measurements at
14413: @uref{http://www.complang.tuwien.ac.at/forth/performance.html}. You can
14414: find numbers for Gforth on various machines in @file{Benchres}.
14415:
14416: @c ******************************************************************
14417: @node Binding to System Library, Cross Compiler, Engine, Top
14418: @chapter Binding to System Library
14419:
14420: @node Cross Compiler, Bugs, Binding to System Library, Top
14421: @chapter Cross Compiler
14422: @cindex @file{cross.fs}
14423: @cindex cross-compiler
14424: @cindex metacompiler
14425: @cindex target compiler
14426:
14427: The cross compiler is used to bootstrap a Forth kernel. Since Gforth is
14428: mostly written in Forth, including crucial parts like the outer
14429: interpreter and compiler, it needs compiled Forth code to get
14430: started. The cross compiler allows to create new images for other
14431: architectures, even running under another Forth system.
14432:
14433: @menu
14434: * Using the Cross Compiler::
14435: * How the Cross Compiler Works::
14436: @end menu
14437:
14438: @node Using the Cross Compiler, How the Cross Compiler Works, Cross Compiler, Cross Compiler
14439: @section Using the Cross Compiler
14440:
14441: The cross compiler uses a language that resembles Forth, but isn't. The
14442: main difference is that you can execute Forth code after definition,
14443: while you usually can't execute the code compiled by cross, because the
14444: code you are compiling is typically for a different computer than the
14445: one you are compiling on.
14446:
14447: The Makefile is already set up to allow you to create kernels for new
14448: architectures with a simple make command. The generic kernels using the
14449: GCC compiled virtual machine are created in the normal build process
14450: with @code{make}. To create a embedded Gforth executable for e.g. the
14451: 8086 processor (running on a DOS machine), type
14452:
14453: @example
14454: make kernl-8086.fi
14455: @end example
14456:
14457: This will use the machine description from the @file{arch/8086}
14458: directory to create a new kernel. A machine file may look like that:
14459:
14460: @example
14461: \ Parameter for target systems 06oct92py
14462:
14463: 4 Constant cell \ cell size in bytes
14464: 2 Constant cell<< \ cell shift to bytes
14465: 5 Constant cell>bit \ cell shift to bits
14466: 8 Constant bits/char \ bits per character
14467: 8 Constant bits/byte \ bits per byte [default: 8]
14468: 8 Constant float \ bytes per float
14469: 8 Constant /maxalign \ maximum alignment in bytes
14470: false Constant bigendian \ byte order
14471: ( true=big, false=little )
14472:
14473: include machpc.fs \ feature list
14474: @end example
14475:
14476: This part is obligatory for the cross compiler itself, the feature list
14477: is used by the kernel to conditionally compile some features in and out,
14478: depending on whether the target supports these features.
14479:
14480: There are some optional features, if you define your own primitives,
14481: have an assembler, or need special, nonstandard preparation to make the
14482: boot process work. @code{asm-include} include an assembler,
14483: @code{prims-include} includes primitives, and @code{>boot} prepares for
14484: booting.
14485:
14486: @example
14487: : asm-include ." Include assembler" cr
14488: s" arch/8086/asm.fs" included ;
14489:
14490: : prims-include ." Include primitives" cr
14491: s" arch/8086/prim.fs" included ;
14492:
14493: : >boot ." Prepare booting" cr
14494: s" ' boot >body into-forth 1+ !" evaluate ;
14495: @end example
14496:
14497: These words are used as sort of macro during the cross compilation in
14498: the file @file{kernel/main.fs}. Instead of using this macros, it would
14499: be possible --- but more complicated --- to write a new kernel project
14500: file, too.
14501:
14502: @file{kernel/main.fs} expects the machine description file name on the
14503: stack; the cross compiler itself (@file{cross.fs}) assumes that either
14504: @code{mach-file} leaves a counted string on the stack, or
14505: @code{machine-file} leaves an address, count pair of the filename on the
14506: stack.
14507:
14508: The feature list is typically controlled using @code{SetValue}, generic
14509: files that are used by several projects can use @code{DefaultValue}
14510: instead. Both functions work like @code{Value}, when the value isn't
14511: defined, but @code{SetValue} works like @code{to} if the value is
14512: defined, and @code{DefaultValue} doesn't set anything, if the value is
14513: defined.
14514:
14515: @example
14516: \ generic mach file for pc gforth 03sep97jaw
14517:
14518: true DefaultValue NIL \ relocating
14519:
14520: >ENVIRON
14521:
14522: true DefaultValue file \ controls the presence of the
14523: \ file access wordset
14524: true DefaultValue OS \ flag to indicate a operating system
14525:
14526: true DefaultValue prims \ true: primitives are c-code
14527:
14528: true DefaultValue floating \ floating point wordset is present
14529:
14530: true DefaultValue glocals \ gforth locals are present
14531: \ will be loaded
14532: true DefaultValue dcomps \ double number comparisons
14533:
14534: true DefaultValue hash \ hashing primitives are loaded/present
14535:
14536: true DefaultValue xconds \ used together with glocals,
14537: \ special conditionals supporting gforths'
14538: \ local variables
14539: true DefaultValue header \ save a header information
14540:
14541: true DefaultValue backtrace \ enables backtrace code
14542:
14543: false DefaultValue ec
14544: false DefaultValue crlf
14545:
14546: cell 2 = [IF] &32 [ELSE] &256 [THEN] KB DefaultValue kernel-size
14547:
14548: &16 KB DefaultValue stack-size
14549: &15 KB &512 + DefaultValue fstack-size
14550: &15 KB DefaultValue rstack-size
14551: &14 KB &512 + DefaultValue lstack-size
14552: @end example
14553:
14554: @node How the Cross Compiler Works, , Using the Cross Compiler, Cross Compiler
14555: @section How the Cross Compiler Works
14556:
14557: @node Bugs, Origin, Cross Compiler, Top
14558: @appendix Bugs
14559: @cindex bug reporting
14560:
14561: Known bugs are described in the file @file{BUGS} in the Gforth distribution.
14562:
14563: If you find a bug, please send a bug report to
14564: @email{bug-gforth@@gnu.org}. A bug report should include this
14565: information:
14566:
14567: @itemize @bullet
14568: @item
14569: The Gforth version used (it is announced at the start of an
14570: interactive Gforth session).
14571: @item
14572: The machine and operating system (on Unix
14573: systems @code{uname -a} will report this information).
14574: @item
14575: The installation options (send the file @file{config.status}).
14576: @item
14577: A complete list of changes (if any) you (or your installer) have made to the
14578: Gforth sources.
14579: @item
14580: A program (or a sequence of keyboard commands) that reproduces the bug.
14581: @item
14582: A description of what you think constitutes the buggy behaviour.
14583: @end itemize
14584:
14585: For a thorough guide on reporting bugs read @ref{Bug Reporting, , How
14586: to Report Bugs, gcc.info, GNU C Manual}.
14587:
14588:
14589: @node Origin, Forth-related information, Bugs, Top
14590: @appendix Authors and Ancestors of Gforth
14591:
14592: @section Authors and Contributors
14593: @cindex authors of Gforth
14594: @cindex contributors to Gforth
14595:
14596: The Gforth project was started in mid-1992 by Bernd Paysan and Anton
14597: Ertl. The third major author was Jens Wilke. Lennart Benschop (who was
14598: one of Gforth's first users, in mid-1993) and Stuart Ramsden inspired us
14599: with their continuous feedback. Lennart Benshop contributed
14600: @file{glosgen.fs}, while Stuart Ramsden has been working on automatic
14601: support for calling C libraries. Helpful comments also came from Paul
14602: Kleinrubatscher, Christian Pirker, Dirk Zoller, Marcel Hendrix, John
14603: Wavrik, Barrie Stott, Marc de Groot, Jorge Acerada, Bruce Hoyt, and
14604: Robert Epprecht. Since the release of Gforth-0.2.1 there were also
14605: helpful comments from many others; thank you all, sorry for not listing
14606: you here (but digging through my mailbox to extract your names is on my
14607: to-do list). Since the release of Gforth-0.4.0 Neal Crook worked on the
14608: manual.
14609:
14610: Gforth also owes a lot to the authors of the tools we used (GCC, CVS,
14611: and autoconf, among others), and to the creators of the Internet: Gforth
14612: was developed across the Internet, and its authors did not meet
14613: physically for the first 4 years of development.
14614:
14615: @section Pedigree
14616: @cindex pedigree of Gforth
14617:
14618: Gforth descends from bigFORTH (1993) and fig-Forth. Gforth and PFE (by
14619: Dirk Zoller) will cross-fertilize each other. Of course, a significant
14620: part of the design of Gforth was prescribed by ANS Forth.
14621:
14622: Bernd Paysan wrote bigFORTH, a descendent from TurboForth, an unreleased
14623: 32 bit native code version of VolksForth for the Atari ST, written
14624: mostly by Dietrich Weineck.
14625:
14626: VolksForth descends from F83. It was written by Klaus Schleisiek, Bernd
14627: Pennemann, Georg Rehfeld and Dietrich Weineck for the C64 (called
14628: UltraForth there) in the mid-80s and ported to the Atari ST in 1986.
14629:
14630: Henry Laxen and Mike Perry wrote F83 as a model implementation of the
14631: Forth-83 standard. !! Pedigree? When?
14632:
14633: A team led by Bill Ragsdale implemented fig-Forth on many processors in
14634: 1979. Robert Selzer and Bill Ragsdale developed the original
14635: implementation of fig-Forth for the 6502 based on microForth.
14636:
14637: The principal architect of microForth was Dean Sanderson. microForth was
14638: FORTH, Inc.'s first off-the-shelf product. It was developed in 1976 for
14639: the 1802, and subsequently implemented on the 8080, the 6800 and the
14640: Z80.
14641:
14642: All earlier Forth systems were custom-made, usually by Charles Moore,
14643: who discovered (as he puts it) Forth during the late 60s. The first full
14644: Forth existed in 1971.
14645:
14646: A part of the information in this section comes from @cite{The Evolution
14647: of Forth} by Elizabeth D. Rather, Donald R. Colburn and Charles
14648: H. Moore, presented at the HOPL-II conference and preprinted in SIGPLAN
14649: Notices 28(3), 1993. You can find more historical and genealogical
14650: information about Forth there.
14651:
14652: @node Forth-related information, Word Index, Origin, Top
14653: @appendix Other Forth-related information
14654: @cindex Forth-related information
14655:
14656: @menu
14657: * Internet resources::
14658: * Books::
14659: * The Forth Interest Group::
14660: * Conferences::
14661: @end menu
14662:
14663:
14664: @node Internet resources, Books, Forth-related information, Forth-related information
14665: @section Internet resources
14666: @cindex internet resources
14667:
14668: @cindex comp.lang.forth
14669: @cindex frequently asked questions
14670: There is an active news group (comp.lang.forth) discussing Forth and
14671: Forth-related issues. A frequently-asked-questions (FAQ) list
14672: is posted to the news group regularly, and archived at these sites:
14673:
14674: @itemize @bullet
14675: @item
14676: @uref{ftp://rtfm.mit.edu/pub/usenet-by-group/comp.lang.forth/}
14677: @item
14678: @uref{ftp://ftp.forth.org/pub/Forth/FAQ/}
14679: @end itemize
14680:
14681: The FAQ list should be considered mandatory reading before posting to
14682: the news group.
14683:
14684: Here are some other web sites holding Forth-related material:
14685:
14686: @itemize @bullet
14687: @item
14688: @uref{http://www.taygeta.com/forth.html} -- Skip Carter's Forth pages.
14689: @item
14690: @uref{http://www.jwdt.com/~paysan/gforth.html} -- the Gforth home page.
14691: @item
14692: @uref{http://www.minerva.com/uathena.htm} -- home of ANS Forth Standard.
14693: @item
14694: @uref{http://dec.bournemouth.ac.uk/forth/index.html} -- the Forth
14695: Research page, including links to the Journal of Forth Application and
14696: Research (JFAR) and a searchable Forth bibliography.
14697: @end itemize
14698:
14699:
14700: @node Books, The Forth Interest Group, Internet resources, Forth-related information
14701: @section Books
14702: @cindex books on Forth
14703:
14704: As the Standard is relatively new, there are not many books out yet. It
14705: is not recommended to learn Forth by using Gforth and a book that is not
14706: written for ANS Forth, as you will not know your mistakes from the
14707: deviations of the book. However, books based on the Forth-83 standard
14708: should be ok, because ANS Forth is primarily an extension of Forth-83.
14709: Refer to the Forth FAQ for details of Forth-related books.
14710:
14711: @cindex standard document for ANS Forth
14712: @cindex ANS Forth document
14713: The definite reference if you want to write ANS Forth programs is, of
14714: course, the ANS Forth document. It is available in printed form from the
14715: National Standards Institute Sales Department (Tel.: USA (212) 642-4900;
14716: Fax.: USA (212) 302-1286) as document @cite{X3.215-1994} for about
14717: $200. You can also get it from Global Engineering Documents (Tel.: USA
14718: (800) 854-7179; Fax.: (303) 843-9880) for about $300.
14719:
14720: @cite{dpANS6}, the last draft of the standard, which was then submitted
14721: to ANSI for publication is available electronically and for free in some
14722: MS Word format, and it has been converted to HTML
14723: (@uref{http://www.taygeta.com/forth/dpans.html}; this HTML version also
14724: includes the answers to Requests for Interpretation (RFIs). Some
14725: pointers to these versions can be found through
14726: @*@uref{http://www.complang.tuwien.ac.at/projects/forth.html}.
14727:
14728:
14729: @node The Forth Interest Group, Conferences, Books, Forth-related information
14730: @section The Forth Interest Group
14731: @cindex Forth interest group (FIG)
14732:
14733: The Forth Interest Group (FIG) is a world-wide, non-profit,
14734: member-supported organisation. It publishes a regular magazine,
14735: @var{FORTH Dimensions}, and offers other benefits of membership. You can
14736: contact the FIG through their office email address:
14737: @email{office@@forth.org} or by visiting their web site at
14738: @uref{http://www.forth.org/}. This web site also includes links to FIG
14739: chapters in other countries and American cities
14740: (@uref{http://www.forth.org/chapters.html}).
14741:
14742: @node Conferences, , The Forth Interest Group, Forth-related information
14743: @section Conferences
14744: @cindex Conferences
14745:
14746: There are several regular conferences related to Forth. They are all
14747: well-publicised in @var{FORTH Dimensions} and on the comp.lang.forth
14748: news group:
14749:
14750: @itemize @bullet
14751: @item
14752: FORML -- the Forth modification laboratory convenes every year near
14753: Monterey, California.
14754: @item
14755: The Rochester Forth Conference -- an annual conference traditionally
14756: held in Rochester, New York.
14757: @item
14758: EuroForth -- this European conference takes place annually.
14759: @end itemize
14760:
14761:
14762: @node Word Index, Name Index, Forth-related information, Top
14763: @unnumbered Word Index
14764:
14765: This index is a list of Forth words that have ``glossary'' entries
14766: within this manual. Each word is listed with its stack effect and
14767: wordset.
14768:
14769: @printindex fn
14770:
14771: @node Name Index, Concept Index, Word Index, Top
14772: @unnumbered Name Index
14773:
14774: This index is a list of Forth words that have ``glossary'' entries
14775: within this manual.
14776:
14777: @printindex ky
14778:
14779: @node Concept Index, , Name Index, Top
14780: @unnumbered Concept and Word Index
14781:
14782: Not all entries listed in this index are present verbatim in the
14783: text. This index also duplicates, in abbreviated form, all of the words
14784: listed in the Word Index (only the names are listed for the words here).
14785:
14786: @printindex cp
14787:
14788: @contents
14789: @bye
14790:
FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>