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:
18: @comment %**start of header (This is for running Texinfo on a region.)
19: @setfilename gforth.info
20: @settitle Gforth Manual
21: @dircategory GNU programming tools
22: @direntry
23: * Gforth: (gforth). A fast interpreter for the Forth language.
24: @end direntry
25: @comment @setchapternewpage odd
26: @comment TODO this gets left in by HTML converter
27: @macro progstyle {}
28: Programming style note:
29: @end macro
30: @comment %**end of header (This is for running Texinfo on a region.)
31:
32:
33: @comment ----------------------------------------------------------
34: @comment macros for beautifying glossary entries
35: @comment if these are used, need to strip them out for HTML converter
36: @comment else they get repeated verbatim in HTML output.
37: @comment .. not working yet.
38:
39: @macro GLOSS-START {}
40: @iftex
41: @ninerm
42: @end iftex
43: @end macro
44:
45: @macro GLOSS-END {}
46: @iftex
47: @rm
48: @end iftex
49: @end macro
50:
51: @comment ----------------------------------------------------------
52:
53:
54: @include version.texi
55:
56: @ifinfo
57: This file documents Gforth @value{VERSION}
58:
59: Copyright @copyright{} 1995-1999 Free Software Foundation, Inc.
60:
61: Permission is granted to make and distribute verbatim copies of
62: this manual provided the copyright notice and this permission notice
63: are preserved on all copies.
64:
65: @ignore
66: Permission is granted to process this file through TeX and print the
67: results, provided the printed document carries a copying permission
68: notice identical to this one except for the removal of this paragraph
69: (this paragraph not being relevant to the printed manual).
70:
71: @end ignore
72: Permission is granted to copy and distribute modified versions of this
73: manual under the conditions for verbatim copying, provided also that the
74: sections entitled "Distribution" and "General Public License" are
75: included exactly as in the original, and provided that the entire
76: resulting derived work is distributed under the terms of a permission
77: notice identical to this one.
78:
79: Permission is granted to copy and distribute translations of this manual
80: into another language, under the above conditions for modified versions,
81: except that the sections entitled "Distribution" and "General Public
82: License" may be included in a translation approved by the author instead
83: of in the original English.
84: @end ifinfo
85:
86: @finalout
87: @titlepage
88: @sp 10
89: @center @titlefont{Gforth Manual}
90: @sp 2
91: @center for version @value{VERSION}
92: @sp 2
93: @center Neal Crook
94: @center Anton Ertl
95: @center Bernd Paysan
96: @center Jens Wilke
97: @sp 3
98: @center This manual is permanently under construction and was last updated on 05-Jun-1999
99:
100: @comment The following two commands start the copyright page.
101: @page
102: @vskip 0pt plus 1filll
103: Copyright @copyright{} 1995--1999 Free Software Foundation, Inc.
104:
105: @comment !! Published by ... or You can get a copy of this manual ...
106:
107: Permission is granted to make and distribute verbatim copies of
108: this manual provided the copyright notice and this permission notice
109: are preserved on all copies.
110:
111: Permission is granted to copy and distribute modified versions of this
112: manual under the conditions for verbatim copying, provided also that the
113: sections entitled "Distribution" and "General Public License" are
114: included exactly as in the original, and provided that the entire
115: resulting derived work is distributed under the terms of a permission
116: notice identical to this one.
117:
118: Permission is granted to copy and distribute translations of this manual
119: into another language, under the above conditions for modified versions,
120: except that the sections entitled "Distribution" and "General Public
121: License" may be included in a translation approved by the author instead
122: of in the original English.
123: @end titlepage
124:
125: @node Top, License, (dir), (dir)
126: @ifinfo
127: Gforth is a free implementation of ANS Forth available on many
128: personal machines. This manual corresponds to version @value{VERSION}.
129: @end ifinfo
130:
131: @menu
132: * License:: The GPL
133: * Goals:: About the Gforth Project
134: * Gforth Environment:: Starting (and exiting) Gforth
135: * Introduction:: An introduction to ANS Forth
136: * Words:: Forth words available in Gforth
137: * Error messages:: How to interpret them
138: * Tools:: Programming tools
139: * ANS conformance:: Implementation-defined options etc.
140: * Model:: The abstract machine of Gforth
141: * Integrating Gforth:: Forth as scripting language for applications
142: * Emacs and Gforth:: The Gforth Mode
143: * Image Files:: @code{.fi} files contain compiled code
144: * Engine:: The inner interpreter and the primitives
145: * Binding to System Library::
146: * Cross Compiler:: The Cross Compiler
147: * Bugs:: How to report them
148: * Origin:: Authors and ancestors of Gforth
149: * Forth-related information:: Books and places to look on the WWW
150: * Word Index:: An item for each Forth word
151: * Name Index:: Forth words, only names listed
152: * Concept Index:: A menu covering many topics
153:
154: @detailmenu
155: --- The Detailed Node Listing ---
156:
157: Goals of Gforth
158:
159: * Gforth Extensions Sinful?::
160:
161: Gforth Environment
162:
163: * Invoking Gforth:: Getting in
164: * Leaving Gforth:: Getting out
165: * Command-line editing::
166: * Upper and lower case::
167: * Environment variables:: ..that affect how Gforth starts up
168: * Gforth Files:: What gets installed and where
169:
170: An Introduction to ANS Forth
171:
172: * Introducing the Text Interpreter::
173: * Stacks and Postfix notation::
174: * Your first definition::
175: * How does that work?::
176: * Forth is written in Forth::
177: * Review - elements of a Forth system::
178: * Where to go next::
179: * Exercises::
180:
181: Forth Words
182:
183: * Notation::
184: * Comments::
185: * Boolean Flags::
186: * Arithmetic::
187: * Stack Manipulation::
188: * Memory::
189: * Control Structures::
190: * Defining Words::
191: * The Text Interpreter::
192: * Tokens for Words::
193: * Word Lists::
194: * Environmental Queries::
195: * Files::
196: * Blocks::
197: * Other I/O::
198: * Programming Tools::
199: * Assembler and Code Words::
200: * Threading Words::
201: * Locals::
202: * Structures::
203: * Object-oriented Forth::
204: * Passing Commands to the OS::
205: * Miscellaneous Words::
206:
207: Arithmetic
208:
209: * Single precision::
210: * Bitwise operations::
211: * Double precision:: Double-cell integer arithmetic
212: * Numeric comparison::
213: * Mixed precision:: Operations with single and double-cell integers
214: * Floating Point::
215:
216: Stack Manipulation
217:
218: * Data stack::
219: * Floating point stack::
220: * Return stack::
221: * Locals stack::
222: * Stack pointer manipulation::
223:
224: Memory
225:
226: * Memory model::
227: * Dictionary allocation::
228: * Heap Allocation::
229: * Memory Access::
230: * Address arithmetic::
231: * Memory Blocks::
232:
233: Control Structures
234:
235: * Selection:: IF ... ELSE ... ENDIF
236: * Simple Loops:: BEGIN ...
237: * Counted Loops:: DO
238: * Arbitrary control structures::
239: * Calls and returns::
240: * Exception Handling::
241:
242: Defining Words
243:
244: * CREATE::
245: * Variables:: Variables and user variables
246: * Constants::
247: * Values:: Initialised variables
248: * Colon Definitions::
249: * Anonymous Definitions:: Definitions without names
250: * User-defined Defining Words::
251: * Deferred words:: Allow forward references
252: * Aliases::
253: * Supplying names::
254: * Interpretation and Compilation Semantics::
255: * Combined words::
256:
257: The Text Interpreter
258:
259: * Input Sources::
260: * Number Conversion::
261: * Interpret/Compile states::
262: * Literals::
263: * Interpreter Directives::
264:
265: Word Lists
266:
267: * Why use word lists?::
268: * Word list examples::
269:
270: Files
271:
272: * Forth source files::
273: * General files::
274: * Search Paths::
275: * Forth Search Paths::
276: * General Search Paths::
277:
278: Other I/O
279:
280: * Simple numeric output:: Predefined formats
281: * Formatted numeric output:: Formatted (pictured) output
282: * String Formats:: How Forth stores strings in memory
283: * Displaying characters and strings:: Other stuff
284: * Input:: Input
285:
286: Programming Tools
287:
288: * Debugging:: Simple and quick.
289: * Assertions:: Making your programs self-checking.
290: * Singlestep Debugger:: Executing your program word by word.
291:
292: Locals
293:
294: * Gforth locals::
295: * ANS Forth locals::
296:
297: Gforth locals
298:
299: * Where are locals visible by name?::
300: * How long do locals live?::
301: * Programming Style::
302: * Implementation::
303:
304: Structures
305:
306: * Why explicit structure support?::
307: * Structure Usage::
308: * Structure Naming Convention::
309: * Structure Implementation::
310: * Structure Glossary::
311:
312: Object-oriented Forth
313:
314: * Why object-oriented programming?::
315: * Object-Oriented Terminology::
316: * Objects::
317: * OOF::
318: * Mini-OOF::
319: * Comparison with other object models::
320:
321: The @file{objects.fs} model
322:
323: * Properties of the Objects model::
324: * Basic Objects Usage::
325: * The Objects base class::
326: * Creating objects::
327: * Object-Oriented Programming Style::
328: * Class Binding::
329: * Method conveniences::
330: * Classes and Scoping::
331: * Dividing classes::
332: * Object Interfaces::
333: * Objects Implementation::
334: * Objects Glossary::
335:
336: The @file{oof.fs} model
337:
338: * Properties of the OOF model::
339: * Basic OOF Usage::
340: * The OOF base class::
341: * Class Declaration::
342: * Class Implementation::
343:
344: The @file{mini-oof.fs} model
345:
346: * Basic Mini-OOF Usage::
347: * Mini-OOF Example::
348: * Mini-OOF Implementation::
349:
350: Tools
351:
352: * ANS Report:: Report the words used, sorted by wordset.
353:
354: ANS conformance
355:
356: * The Core Words::
357: * The optional Block word set::
358: * The optional Double Number word set::
359: * The optional Exception word set::
360: * The optional Facility word set::
361: * The optional File-Access word set::
362: * The optional Floating-Point word set::
363: * The optional Locals word set::
364: * The optional Memory-Allocation word set::
365: * The optional Programming-Tools word set::
366: * The optional Search-Order word set::
367:
368: The Core Words
369:
370: * core-idef:: Implementation Defined Options
371: * core-ambcond:: Ambiguous Conditions
372: * core-other:: Other System Documentation
373:
374: The optional Block word set
375:
376: * block-idef:: Implementation Defined Options
377: * block-ambcond:: Ambiguous Conditions
378: * block-other:: Other System Documentation
379:
380: The optional Double Number word set
381:
382: * double-ambcond:: Ambiguous Conditions
383:
384: The optional Exception word set
385:
386: * exception-idef:: Implementation Defined Options
387:
388: The optional Facility word set
389:
390: * facility-idef:: Implementation Defined Options
391: * facility-ambcond:: Ambiguous Conditions
392:
393: The optional File-Access word set
394:
395: * file-idef:: Implementation Defined Options
396: * file-ambcond:: Ambiguous Conditions
397:
398: The optional Floating-Point word set
399:
400: * floating-idef:: Implementation Defined Options
401: * floating-ambcond:: Ambiguous Conditions
402:
403: The optional Locals word set
404:
405: * locals-idef:: Implementation Defined Options
406: * locals-ambcond:: Ambiguous Conditions
407:
408: The optional Memory-Allocation word set
409:
410: * memory-idef:: Implementation Defined Options
411:
412: The optional Programming-Tools word set
413:
414: * programming-idef:: Implementation Defined Options
415: * programming-ambcond:: Ambiguous Conditions
416:
417: The optional Search-Order word set
418:
419: * search-idef:: Implementation Defined Options
420: * search-ambcond:: Ambiguous Conditions
421:
422: Image Files
423:
424: * Image Licensing Issues:: Distribution terms for images.
425: * Image File Background:: Why have image files?
426: * Non-Relocatable Image Files:: don't always work.
427: * Data-Relocatable Image Files:: are better.
428: * Fully Relocatable Image Files:: better yet.
429: * Stack and Dictionary Sizes:: Setting the default sizes for an image.
430: * Running Image Files:: @code{gforth -i @i{file}} or @i{file}.
431: * Modifying the Startup Sequence:: and turnkey applications.
432:
433: Fully Relocatable Image Files
434:
435: * gforthmi:: The normal way
436: * cross.fs:: The hard way
437:
438: Engine
439:
440: * Portability::
441: * Threading::
442: * Primitives::
443: * Performance::
444:
445: Threading
446:
447: * Scheduling::
448: * Direct or Indirect Threaded?::
449: * DOES>::
450:
451: Primitives
452:
453: * Automatic Generation::
454: * TOS Optimization::
455: * Produced code::
456:
457: Cross Compiler
458:
459: * Using the Cross Compiler::
460: * How the Cross Compiler Works::
461:
462: Other Forth-related information
463:
464: * Internet resources::
465: * Books::
466: * The Forth Interest Group::
467: * Conferences::
468:
469: @end detailmenu
470: @end menu
471:
472: @node License, Goals, Top, Top
473: @unnumbered GNU GENERAL PUBLIC LICENSE
474: @center Version 2, June 1991
475:
476: @display
477: Copyright @copyright{} 1989, 1991 Free Software Foundation, Inc.
478: 675 Mass Ave, Cambridge, MA 02139, USA
479:
480: Everyone is permitted to copy and distribute verbatim copies
481: of this license document, but changing it is not allowed.
482: @end display
483:
484: @unnumberedsec Preamble
485:
486: The licenses for most software are designed to take away your
487: freedom to share and change it. By contrast, the GNU General Public
488: License is intended to guarantee your freedom to share and change free
489: software---to make sure the software is free for all its users. This
490: General Public License applies to most of the Free Software
491: Foundation's software and to any other program whose authors commit to
492: using it. (Some other Free Software Foundation software is covered by
493: the GNU Library General Public License instead.) You can apply it to
494: your programs, too.
495:
496: When we speak of free software, we are referring to freedom, not
497: price. Our General Public Licenses are designed to make sure that you
498: have the freedom to distribute copies of free software (and charge for
499: this service if you wish), that you receive source code or can get it
500: if you want it, that you can change the software or use pieces of it
501: in new free programs; and that you know you can do these things.
502:
503: To protect your rights, we need to make restrictions that forbid
504: anyone to deny you these rights or to ask you to surrender the rights.
505: These restrictions translate to certain responsibilities for you if you
506: distribute copies of the software, or if you modify it.
507:
508: For example, if you distribute copies of such a program, whether
509: gratis or for a fee, you must give the recipients all the rights that
510: you have. You must make sure that they, too, receive or can get the
511: source code. And you must show them these terms so they know their
512: rights.
513:
514: We protect your rights with two steps: (1) copyright the software, and
515: (2) offer you this license which gives you legal permission to copy,
516: distribute and/or modify the software.
517:
518: Also, for each author's protection and ours, we want to make certain
519: that everyone understands that there is no warranty for this free
520: software. If the software is modified by someone else and passed on, we
521: want its recipients to know that what they have is not the original, so
522: that any problems introduced by others will not reflect on the original
523: authors' reputations.
524:
525: Finally, any free program is threatened constantly by software
526: patents. We wish to avoid the danger that redistributors of a free
527: program will individually obtain patent licenses, in effect making the
528: program proprietary. To prevent this, we have made it clear that any
529: patent must be licensed for everyone's free use or not licensed at all.
530:
531: The precise terms and conditions for copying, distribution and
532: modification follow.
533:
534: @iftex
535: @unnumberedsec TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION
536: @end iftex
537: @ifinfo
538: @center TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION
539: @end ifinfo
540:
541: @enumerate 0
542: @item
543: This License applies to any program or other work which contains
544: a notice placed by the copyright holder saying it may be distributed
545: under the terms of this General Public License. The ``Program'', below,
546: refers to any such program or work, and a ``work based on the Program''
547: means either the Program or any derivative work under copyright law:
548: that is to say, a work containing the Program or a portion of it,
549: either verbatim or with modifications and/or translated into another
550: language. (Hereinafter, translation is included without limitation in
551: the term ``modification''.) Each licensee is addressed as ``you''.
552:
553: Activities other than copying, distribution and modification are not
554: covered by this License; they are outside its scope. The act of
555: running the Program is not restricted, and the output from the Program
556: is covered only if its contents constitute a work based on the
557: Program (independent of having been made by running the Program).
558: Whether that is true depends on what the Program does.
559:
560: @item
561: You may copy and distribute verbatim copies of the Program's
562: source code as you receive it, in any medium, provided that you
563: conspicuously and appropriately publish on each copy an appropriate
564: copyright notice and disclaimer of warranty; keep intact all the
565: notices that refer to this License and to the absence of any warranty;
566: and give any other recipients of the Program a copy of this License
567: along with the Program.
568:
569: You may charge a fee for the physical act of transferring a copy, and
570: you may at your option offer warranty protection in exchange for a fee.
571:
572: @item
573: You may modify your copy or copies of the Program or any portion
574: of it, thus forming a work based on the Program, and copy and
575: distribute such modifications or work under the terms of Section 1
576: above, provided that you also meet all of these conditions:
577:
578: @enumerate a
579: @item
580: You must cause the modified files to carry prominent notices
581: stating that you changed the files and the date of any change.
582:
583: @item
584: You must cause any work that you distribute or publish, that in
585: whole or in part contains or is derived from the Program or any
586: part thereof, to be licensed as a whole at no charge to all third
587: parties under the terms of this License.
588:
589: @item
590: If the modified program normally reads commands interactively
591: when run, you must cause it, when started running for such
592: interactive use in the most ordinary way, to print or display an
593: announcement including an appropriate copyright notice and a
594: notice that there is no warranty (or else, saying that you provide
595: a warranty) and that users may redistribute the program under
596: these conditions, and telling the user how to view a copy of this
597: License. (Exception: if the Program itself is interactive but
598: does not normally print such an announcement, your work based on
599: the Program is not required to print an announcement.)
600: @end enumerate
601:
602: These requirements apply to the modified work as a whole. If
603: identifiable sections of that work are not derived from the Program,
604: and can be reasonably considered independent and separate works in
605: themselves, then this License, and its terms, do not apply to those
606: sections when you distribute them as separate works. But when you
607: distribute the same sections as part of a whole which is a work based
608: on the Program, the distribution of the whole must be on the terms of
609: this License, whose permissions for other licensees extend to the
610: entire whole, and thus to each and every part regardless of who wrote it.
611:
612: Thus, it is not the intent of this section to claim rights or contest
613: your rights to work written entirely by you; rather, the intent is to
614: exercise the right to control the distribution of derivative or
615: collective works based on the Program.
616:
617: In addition, mere aggregation of another work not based on the Program
618: with the Program (or with a work based on the Program) on a volume of
619: a storage or distribution medium does not bring the other work under
620: the scope of this License.
621:
622: @item
623: You may copy and distribute the Program (or a work based on it,
624: under Section 2) in object code or executable form under the terms of
625: Sections 1 and 2 above provided that you also do one of the following:
626:
627: @enumerate a
628: @item
629: Accompany it with the complete corresponding machine-readable
630: source code, which must be distributed under the terms of Sections
631: 1 and 2 above on a medium customarily used for software interchange; or,
632:
633: @item
634: Accompany it with a written offer, valid for at least three
635: years, to give any third party, for a charge no more than your
636: cost of physically performing source distribution, a complete
637: machine-readable copy of the corresponding source code, to be
638: distributed under the terms of Sections 1 and 2 above on a medium
639: customarily used for software interchange; or,
640:
641: @item
642: Accompany it with the information you received as to the offer
643: to distribute corresponding source code. (This alternative is
644: allowed only for noncommercial distribution and only if you
645: received the program in object code or executable form with such
646: an offer, in accord with Subsection b above.)
647: @end enumerate
648:
649: The source code for a work means the preferred form of the work for
650: making modifications to it. For an executable work, complete source
651: code means all the source code for all modules it contains, plus any
652: associated interface definition files, plus the scripts used to
653: control compilation and installation of the executable. However, as a
654: special exception, the source code distributed need not include
655: anything that is normally distributed (in either source or binary
656: form) with the major components (compiler, kernel, and so on) of the
657: operating system on which the executable runs, unless that component
658: itself accompanies the executable.
659:
660: If distribution of executable or object code is made by offering
661: access to copy from a designated place, then offering equivalent
662: access to copy the source code from the same place counts as
663: distribution of the source code, even though third parties are not
664: compelled to copy the source along with the object code.
665:
666: @item
667: You may not copy, modify, sublicense, or distribute the Program
668: except as expressly provided under this License. Any attempt
669: otherwise to copy, modify, sublicense or distribute the Program is
670: void, and will automatically terminate your rights under this License.
671: However, parties who have received copies, or rights, from you under
672: this License will not have their licenses terminated so long as such
673: parties remain in full compliance.
674:
675: @item
676: You are not required to accept this License, since you have not
677: signed it. However, nothing else grants you permission to modify or
678: distribute the Program or its derivative works. These actions are
679: prohibited by law if you do not accept this License. Therefore, by
680: modifying or distributing the Program (or any work based on the
681: Program), you indicate your acceptance of this License to do so, and
682: all its terms and conditions for copying, distributing or modifying
683: the Program or works based on it.
684:
685: @item
686: Each time you redistribute the Program (or any work based on the
687: Program), the recipient automatically receives a license from the
688: original licensor to copy, distribute or modify the Program subject to
689: these terms and conditions. You may not impose any further
690: restrictions on the recipients' exercise of the rights granted herein.
691: You are not responsible for enforcing compliance by third parties to
692: this License.
693:
694: @item
695: If, as a consequence of a court judgment or allegation of patent
696: infringement or for any other reason (not limited to patent issues),
697: conditions are imposed on you (whether by court order, agreement or
698: otherwise) that contradict the conditions of this License, they do not
699: excuse you from the conditions of this License. If you cannot
700: distribute so as to satisfy simultaneously your obligations under this
701: License and any other pertinent obligations, then as a consequence you
702: may not distribute the Program at all. For example, if a patent
703: license would not permit royalty-free redistribution of the Program by
704: all those who receive copies directly or indirectly through you, then
705: the only way you could satisfy both it and this License would be to
706: refrain entirely from distribution of the Program.
707:
708: If any portion of this section is held invalid or unenforceable under
709: any particular circumstance, the balance of the section is intended to
710: apply and the section as a whole is intended to apply in other
711: circumstances.
712:
713: It is not the purpose of this section to induce you to infringe any
714: patents or other property right claims or to contest validity of any
715: such claims; this section has the sole purpose of protecting the
716: integrity of the free software distribution system, which is
717: implemented by public license practices. Many people have made
718: generous contributions to the wide range of software distributed
719: through that system in reliance on consistent application of that
720: system; it is up to the author/donor to decide if he or she is willing
721: to distribute software through any other system and a licensee cannot
722: impose that choice.
723:
724: This section is intended to make thoroughly clear what is believed to
725: be a consequence of the rest of this License.
726:
727: @item
728: If the distribution and/or use of the Program is restricted in
729: certain countries either by patents or by copyrighted interfaces, the
730: original copyright holder who places the Program under this License
731: may add an explicit geographical distribution limitation excluding
732: those countries, so that distribution is permitted only in or among
733: countries not thus excluded. In such case, this License incorporates
734: the limitation as if written in the body of this License.
735:
736: @item
737: The Free Software Foundation may publish revised and/or new versions
738: of the General Public License from time to time. Such new versions will
739: be similar in spirit to the present version, but may differ in detail to
740: address new problems or concerns.
741:
742: Each version is given a distinguishing version number. If the Program
743: specifies a version number of this License which applies to it and ``any
744: later version'', you have the option of following the terms and conditions
745: either of that version or of any later version published by the Free
746: Software Foundation. If the Program does not specify a version number of
747: this License, you may choose any version ever published by the Free Software
748: Foundation.
749:
750: @item
751: If you wish to incorporate parts of the Program into other free
752: programs whose distribution conditions are different, write to the author
753: to ask for permission. For software which is copyrighted by the Free
754: Software Foundation, write to the Free Software Foundation; we sometimes
755: make exceptions for this. Our decision will be guided by the two goals
756: of preserving the free status of all derivatives of our free software and
757: of promoting the sharing and reuse of software generally.
758:
759: @iftex
760: @heading NO WARRANTY
761: @end iftex
762: @ifinfo
763: @center NO WARRANTY
764: @end ifinfo
765:
766: @item
767: BECAUSE THE PROGRAM IS LICENSED FREE OF CHARGE, THERE IS NO WARRANTY
768: FOR THE PROGRAM, TO THE EXTENT PERMITTED BY APPLICABLE LAW. EXCEPT WHEN
769: OTHERWISE STATED IN WRITING THE COPYRIGHT HOLDERS AND/OR OTHER PARTIES
770: PROVIDE THE PROGRAM ``AS IS'' WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESSED
771: OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
772: MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. THE ENTIRE RISK AS
773: TO THE QUALITY AND PERFORMANCE OF THE PROGRAM IS WITH YOU. SHOULD THE
774: PROGRAM PROVE DEFECTIVE, YOU ASSUME THE COST OF ALL NECESSARY SERVICING,
775: REPAIR OR CORRECTION.
776:
777: @item
778: IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING
779: WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MAY MODIFY AND/OR
780: REDISTRIBUTE THE PROGRAM AS PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES,
781: INCLUDING ANY GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING
782: OUT OF THE USE OR INABILITY TO USE THE PROGRAM (INCLUDING BUT NOT LIMITED
783: TO LOSS OF DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY
784: YOU OR THIRD PARTIES OR A FAILURE OF THE PROGRAM TO OPERATE WITH ANY OTHER
785: PROGRAMS), EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE
786: POSSIBILITY OF SUCH DAMAGES.
787: @end enumerate
788:
789: @iftex
790: @heading END OF TERMS AND CONDITIONS
791: @end iftex
792: @ifinfo
793: @center END OF TERMS AND CONDITIONS
794: @end ifinfo
795:
796: @page
797: @unnumberedsec How to Apply These Terms to Your New Programs
798:
799: If you develop a new program, and you want it to be of the greatest
800: possible use to the public, the best way to achieve this is to make it
801: free software which everyone can redistribute and change under these terms.
802:
803: To do so, attach the following notices to the program. It is safest
804: to attach them to the start of each source file to most effectively
805: convey the exclusion of warranty; and each file should have at least
806: the ``copyright'' line and a pointer to where the full notice is found.
807:
808: @smallexample
809: @var{one line to give the program's name and a brief idea of what it does.}
810: Copyright (C) 19@var{yy} @var{name of author}
811:
812: This program is free software; you can redistribute it and/or modify
813: it under the terms of the GNU General Public License as published by
814: the Free Software Foundation; either version 2 of the License, or
815: (at your option) any later version.
816:
817: This program is distributed in the hope that it will be useful,
818: but WITHOUT ANY WARRANTY; without even the implied warranty of
819: MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
820: GNU General Public License for more details.
821:
822: You should have received a copy of the GNU General Public License
823: along with this program; if not, write to the Free Software
824: Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
825: @end smallexample
826:
827: Also add information on how to contact you by electronic and paper mail.
828:
829: If the program is interactive, make it output a short notice like this
830: when it starts in an interactive mode:
831:
832: @smallexample
833: Gnomovision version 69, Copyright (C) 19@var{yy} @var{name of author}
834: Gnomovision comes with ABSOLUTELY NO WARRANTY; for details
835: type `show w'.
836: This is free software, and you are welcome to redistribute it
837: under certain conditions; type `show c' for details.
838: @end smallexample
839:
840: The hypothetical commands @samp{show w} and @samp{show c} should show
841: the appropriate parts of the General Public License. Of course, the
842: commands you use may be called something other than @samp{show w} and
843: @samp{show c}; they could even be mouse-clicks or menu items---whatever
844: suits your program.
845:
846: You should also get your employer (if you work as a programmer) or your
847: school, if any, to sign a ``copyright disclaimer'' for the program, if
848: necessary. Here is a sample; alter the names:
849:
850: @smallexample
851: Yoyodyne, Inc., hereby disclaims all copyright interest in the program
852: `Gnomovision' (which makes passes at compilers) written by James Hacker.
853:
854: @var{signature of Ty Coon}, 1 April 1989
855: Ty Coon, President of Vice
856: @end smallexample
857:
858: This General Public License does not permit incorporating your program into
859: proprietary programs. If your program is a subroutine library, you may
860: consider it more useful to permit linking proprietary applications with the
861: library. If this is what you want to do, use the GNU Library General
862: Public License instead of this License.
863:
864: @iftex
865: @unnumbered Preface
866: @cindex Preface
867: This manual documents Gforth. Some introductory material is provided for
868: readers who are unfamiliar with Forth or who are migrating to Gforth
869: from other Forth compilers. However, this manual is primarily a
870: reference manual.
871: @end iftex
872:
873: @comment TODO much more blurb here.
874:
875: @c ******************************************************************
876: @node Goals, Gforth Environment, License, Top
877: @comment node-name, next, previous, up
878: @chapter Goals of Gforth
879: @cindex goals of the Gforth project
880: The goal of the Gforth Project is to develop a standard model for
881: ANS Forth. This can be split into several subgoals:
882:
883: @itemize @bullet
884: @item
885: Gforth should conform to the ANS Forth Standard.
886: @item
887: It should be a model, i.e. it should define all the
888: implementation-dependent things.
889: @item
890: It should become standard, i.e. widely accepted and used. This goal
891: is the most difficult one.
892: @end itemize
893:
894: To achieve these goals Gforth should be
895: @itemize @bullet
896: @item
897: Similar to previous models (fig-Forth, F83)
898: @item
899: Powerful. It should provide for all the things that are considered
900: necessary today and even some that are not yet considered necessary.
901: @item
902: Efficient. It should not get the reputation of being exceptionally
903: slow.
904: @item
905: Free.
906: @item
907: Available on many machines/easy to port.
908: @end itemize
909:
910: Have we achieved these goals? Gforth conforms to the ANS Forth
911: standard. It may be considered a model, but we have not yet documented
912: which parts of the model are stable and which parts we are likely to
913: change. It certainly has not yet become a de facto standard, but it
914: appears to be quite popular. It has some similarities to and some
915: differences from previous models. It has some powerful features, but not
916: yet everything that we envisioned. We certainly have achieved our
917: execution speed goals (@pxref{Performance}). It is free and available
918: on many machines.
919:
920: @menu
921: * Gforth Extensions Sinful?::
922: @end menu
923:
924: @node Gforth Extensions Sinful?, , Goals, Goals
925: @comment node-name, next, previous, up
926: @section Is it a Sin to use Gforth Extensions?
927: @cindex Gforth extensions
928:
929: If you've been paying attention, you will have realised that there is an
930: ANS (American National Standard) for Forth. As you read through the rest
931: of this manual, you will see documentation for @i{Standard} words, and
932: documentation for some appealing Gforth @i{extensions}. You might ask
933: yourself the question: @i{``Given that there is a standard, would I be
934: committing a sin if I use (non-Standard) Gforth extensions?''}
935:
936: The answer to that question is somewhat pragmatic and somewhat
937: philosophical. Consider these points:
938:
939: @itemize @bullet
940: @item
941: A number of the Gforth extensions can be implemented in ANS Forth using
942: files provided in the @file{compat/} directory. These are mentioned in
943: the text in passing.
944: @item
945: Forth has a rich historical precedent for programmers taking advantage
946: of implementation-dependent features of their tools (for example,
947: relying on a knowledge of the dictionary structure). Sometimes these
948: techniques are necessary to extract every last bit of performance from
949: the hardware, sometimes they are just a programming shorthand.
950: @item
951: The best way to break the rules is to know what the rules are. To learn
952: the rules, there is no substitute for studying the text of the Standard
953: itself. In particular, Appendix A of the Standard (@var{Rationale})
954: provides a valuable insight into the thought processes of the technical
955: committee.
956: @item
957: The best reason to break a rule is because you have to; because it's
958: more productive to do that, because it makes your code run fast enough
959: or because you can see no Standard way to achieve what you want to
960: achieve.
961: @end itemize
962:
963: The tool @file{ans-report.fs} (@pxref{ANS Report}) makes it easy to
964: analyse your program and determine what non-Standard definitions it
965: relies upon.
966:
967:
968: @c ******************************************************************
969: @node Gforth Environment, Introduction, Goals, Top
970: @chapter Gforth Environment
971: @cindex Gforth environment
972:
973: Note: ultimately, the Gforth man page will be auto-generated from the
974: material in this chapter.
975:
976: @menu
977: * Invoking Gforth:: Getting in
978: * Leaving Gforth:: Getting out
979: * Command-line editing::
980: * Upper and lower case::
981: * Environment variables:: ..that affect how Gforth starts up
982: * Gforth Files:: What gets installed and where
983: @end menu
984:
985: @xref{Image Files} for related information about the creation of images.
986:
987: @comment ----------------------------------------------
988: @node Invoking Gforth, Leaving Gforth, ,Gforth Environment
989: @section Invoking Gforth
990: @cindex invoking Gforth
991: @cindex running Gforth
992: @cindex command-line options
993: @cindex options on the command line
994: @cindex flags on the command line
995:
996: Gforth is made up of two parts; an executable ``engine'' (named
997: @file{gforth} or @file{gforth-fast}) and an image file. To start it, you
998: will usually just say @code{gforth} -- this automatically loads the
999: default image file @file{gforth.fi}. In many other cases the default
1000: Gforth image will be invoked like this:
1001: @example
1002: gforth [file | -e forth-code] ...
1003: @end example
1004: @noindent
1005: This interprets the contents of the files and the Forth code in the order they
1006: are given.
1007:
1008: In addition to the @file{gforth} engine, there is also an engine called
1009: @file{gforth-fast}, which is faster, but gives less informative error
1010: messages (@pxref{Error messages}).
1011:
1012: In general, the command line looks like this:
1013:
1014: @example
1015: gforth[-fast] [engine options] [image options]
1016: @end example
1017:
1018: The engine options must come before the rest of the command
1019: line. They are:
1020:
1021: @table @code
1022: @cindex -i, command-line option
1023: @cindex --image-file, command-line option
1024: @item --image-file @i{file}
1025: @itemx -i @i{file}
1026: Loads the Forth image @i{file} instead of the default
1027: @file{gforth.fi} (@pxref{Image Files}).
1028:
1029: @cindex --appl-image, command-line option
1030: @item --appl-image @i{file}
1031: Loads the image @i{file} and leaves all further command-line arguments
1032: to the image (instead of processing them as options). This is useful
1033: for building executable application images on Unix, built with
1034: @code{gforthmi --application ...}.
1035:
1036: @cindex --path, command-line option
1037: @cindex -p, command-line option
1038: @item --path @i{path}
1039: @itemx -p @i{path}
1040: Uses @i{path} for searching the image file and Forth source code files
1041: instead of the default in the environment variable @code{GFORTHPATH} or
1042: the path specified at installation time (e.g.,
1043: @file{/usr/local/share/gforth/0.2.0:.}). A path is given as a list of
1044: directories, separated by @samp{:} (on Unix) or @samp{;} (on other OSs).
1045:
1046: @cindex --dictionary-size, command-line option
1047: @cindex -m, command-line option
1048: @cindex @i{size} parameters for command-line options
1049: @cindex size of the dictionary and the stacks
1050: @item --dictionary-size @i{size}
1051: @itemx -m @i{size}
1052: Allocate @i{size} space for the Forth dictionary space instead of
1053: using the default specified in the image (typically 256K). The
1054: @i{size} specification for this and subsequent options consists of
1055: an integer and a unit (e.g.,
1056: @code{4M}). The unit can be one of @code{b} (bytes), @code{e} (element
1057: size, in this case Cells), @code{k} (kilobytes), @code{M} (Megabytes),
1058: @code{G} (Gigabytes), and @code{T} (Terabytes). If no unit is specified,
1059: @code{e} is used.
1060:
1061: @cindex --data-stack-size, command-line option
1062: @cindex -d, command-line option
1063: @item --data-stack-size @i{size}
1064: @itemx -d @i{size}
1065: Allocate @i{size} space for the data stack instead of using the
1066: default specified in the image (typically 16K).
1067:
1068: @cindex --return-stack-size, command-line option
1069: @cindex -r, command-line option
1070: @item --return-stack-size @i{size}
1071: @itemx -r @i{size}
1072: Allocate @i{size} space for the return stack instead of using the
1073: default specified in the image (typically 15K).
1074:
1075: @cindex --fp-stack-size, command-line option
1076: @cindex -f, command-line option
1077: @item --fp-stack-size @i{size}
1078: @itemx -f @i{size}
1079: Allocate @i{size} space for the floating point stack instead of
1080: using the default specified in the image (typically 15.5K). In this case
1081: the unit specifier @code{e} refers to floating point numbers.
1082:
1083: @cindex --locals-stack-size, command-line option
1084: @cindex -l, command-line option
1085: @item --locals-stack-size @i{size}
1086: @itemx -l @i{size}
1087: Allocate @i{size} space for the locals stack instead of using the
1088: default specified in the image (typically 14.5K).
1089:
1090: @cindex -h, command-line option
1091: @cindex --help, command-line option
1092: @item --help
1093: @itemx -h
1094: Print a message about the command-line options
1095:
1096: @cindex -v, command-line option
1097: @cindex --version, command-line option
1098: @item --version
1099: @itemx -v
1100: Print version and exit
1101:
1102: @cindex --debug, command-line option
1103: @item --debug
1104: Print some information useful for debugging on startup.
1105:
1106: @cindex --offset-image, command-line option
1107: @item --offset-image
1108: Start the dictionary at a slightly different position than would be used
1109: otherwise (useful for creating data-relocatable images,
1110: @pxref{Data-Relocatable Image Files}).
1111:
1112: @cindex --no-offset-im, command-line option
1113: @item --no-offset-im
1114: Start the dictionary at the normal position.
1115:
1116: @cindex --clear-dictionary, command-line option
1117: @item --clear-dictionary
1118: Initialize all bytes in the dictionary to 0 before loading the image
1119: (@pxref{Data-Relocatable Image Files}).
1120:
1121: @cindex --die-on-signal, command-line-option
1122: @item --die-on-signal
1123: Normally Gforth handles most signals (e.g., the user interrupt SIGINT,
1124: or the segmentation violation SIGSEGV) by translating it into a Forth
1125: @code{THROW}. With this option, Gforth exits if it receives such a
1126: signal. This option is useful when the engine and/or the image might be
1127: severely broken (such that it causes another signal before recovering
1128: from the first); this option avoids endless loops in such cases.
1129: @end table
1130:
1131: @cindex loading files at startup
1132: @cindex executing code on startup
1133: @cindex batch processing with Gforth
1134: As explained above, the image-specific command-line arguments for the
1135: default image @file{gforth.fi} consist of a sequence of filenames and
1136: @code{-e @var{forth-code}} options that are interpreted in the sequence
1137: in which they are given. The @code{-e @var{forth-code}} or
1138: @code{--evaluate @var{forth-code}} option evaluates the Forth
1139: code. This option takes only one argument; if you want to evaluate more
1140: Forth words, you have to quote them or use @code{-e} several times. To exit
1141: after processing the command line (instead of entering interactive mode)
1142: append @code{-e bye} to the command line.
1143:
1144: @cindex versions, invoking other versions of Gforth
1145: If you have several versions of Gforth installed, @code{gforth} will
1146: invoke the version that was installed last. @code{gforth-@i{version}}
1147: invokes a specific version. If your environment contains the variable
1148: @code{GFORTHPATH}, you may want to override it by using the
1149: @code{--path} option.
1150:
1151: Not yet implemented:
1152: On startup the system first executes the system initialization file
1153: (unless the option @code{--no-init-file} is given; note that the system
1154: resulting from using this option may not be ANS Forth conformant). Then
1155: the user initialization file @file{.gforth.fs} is executed, unless the
1156: option @code{--no-rc} is given; this file is first searched in @file{.},
1157: then in @file{~}, then in the normal path (see above).
1158:
1159:
1160:
1161: @comment ----------------------------------------------
1162: @node Leaving Gforth, Command-line editing, Invoking Gforth, Gforth Environment
1163: @section Leaving Gforth
1164: @cindex Gforth - leaving
1165: @cindex leaving Gforth
1166:
1167: You can leave Gforth by typing @code{bye} or @kbd{Ctrl-d} (at the start
1168: of a line) or (if you invoked Gforth with the @code{--die-on-signal}
1169: option) @kbd{Ctrl-c}. When you leave Gforth, all of your definitions and
1170: data are discarded. @xref{Image Files} for ways of saving the state of
1171: the system before leaving Gforth.
1172:
1173: doc-bye
1174:
1175:
1176: @comment ----------------------------------------------
1177: @node Command-line editing, Upper and lower case,Leaving Gforth,Gforth Environment
1178: @section Command-line editing
1179: @cindex command-line editing
1180:
1181: Gforth maintains a history file that records every line that you type to
1182: the text interpreter. This file is preserved between sessions, and is
1183: used to provide a command-line recall facility; if you type @kbd{Ctrl-P}
1184: repeatedly you can recall successively older commands from this (or
1185: previous) session(s). The full list of command-line editing facilities is:
1186:
1187: @itemize @bullet
1188: @item
1189: @kbd{Ctrl-p} (``previous'') (or up-arrow) to recall successively older
1190: commands from the history buffer.
1191: @item
1192: @kbd{Ctrl-n} (``next'') (or down-arrow) to recall successively newer commands
1193: from the history buffer.
1194: @item
1195: @kbd{Ctrl-f} (or right-arrow) to move the cursor right, non-destructively.
1196: @item
1197: @kbd{Ctrl-b} (or left-arrow) to move the cursor left, non-destructively.
1198: @item
1199: @kbd{Ctrl-h} (backspace) to delete the character to the left of the cursor,
1200: closing up the line.
1201: @item
1202: @kbd{Ctrl-k} to delete (``kill'') from the cursor to the end of the line.
1203: @item
1204: @kbd{Ctrl-a} to move the cursor to the start of the line.
1205: @item
1206: @kbd{Ctrl-e} to move the cursor to the end of the line.
1207: @item
1208: @key{RET} (@kbd{Ctrl-m}) or @key{LFD} (@kbd{Ctrl-j}) to submit the current
1209: line.
1210: @item
1211: @key{TAB} to step through all possible full-word completions of the word
1212: currently being typed.
1213: @item
1214: @kbd{Ctrl-d} at the start of the line to terminate Gforth (gracefully,
1215: using @code{bye}).
1216: @end itemize
1217:
1218: When editing, displayable characters are inserted to the left of the
1219: cursor position; the line is always in ``insert'' (as opposed to
1220: ``overstrike'') mode.
1221:
1222: @cindex history file
1223: @cindex @file{.gforth-history}
1224: On Unix systems, the history file is @file{~/.gforth-history} by
1225: default@footnote{i.e. it is stored in the user's home directory.}. You
1226: can find out the name and location of your history file using:
1227:
1228: @example
1229: history-file type \ Unix-class systems
1230:
1231: history-file type \ Other systems
1232: history-dir type
1233: @end example
1234:
1235: If you enter long definitions by hand, you can use a text editor to
1236: paste them out of the history file into a Forth source file for reuse at
1237: a later time.
1238:
1239: Gforth never trims the size of the history file, so you should do this
1240: periodically, if necessary.
1241:
1242: @comment this is all defined in history.fs
1243: @comment NAC TODO the ctrl-D behaviour can either do a bye or a beep.. how is that option
1244: @comment chosen?
1245:
1246:
1247:
1248: @comment ----------------------------------------------
1249: @node Upper and lower case, Environment variables,Command-line editing,Gforth Environment
1250: @section Upper and lower case
1251: @cindex case-sensitivity
1252: @cindex upper and lower case
1253:
1254: Gforth is case-insensitive; you can enter definitions and invoke
1255: Standard words using upper, lower or mixed case (however,
1256: @pxref{core-idef, Implementation-defined options, Implementation-defined
1257: options}).
1258:
1259: ANS Forth only @i{requires} implementations to recognise Standard words
1260: when they are typed entirely in upper case. Therefore, a Standard
1261: program must use upper case for all Standard words. You can use whatever
1262: case you like for words that you define, but in a standard program you
1263: have to use the words in the same case that you defined them.
1264:
1265: Gforth supports case sensitivity through @code{table}s (case-sensitive
1266: wordlists, @pxref{Word Lists}).
1267:
1268: Two people have asked how to convert Gforth to case sensitivity; while
1269: we think this is a bad idea, you can change all wordlists into tables
1270: like this:
1271:
1272: @example
1273: ' table-find forth-wordlist wordlist-map @ !
1274: @end example
1275:
1276: Note that you now have to type the predefined words in the same case
1277: that we defined them, which are varying. You may want to convert them
1278: to your favourite case before doing this operation (I won't explain how,
1279: because if you are even contemplating to do this, you'd better have
1280: enough knowledge of Forth systems to know this already).
1281:
1282: @comment ----------------------------------------------
1283: @node Environment variables, Gforth Files, Upper and lower case,Gforth Environment
1284: @section Environment variables
1285: @cindex environment variables
1286:
1287: Gforth uses these environment variables:
1288:
1289: @itemize @bullet
1290: @item
1291: @cindex @code{GFORTHHIST} -- environment variable
1292: @code{GFORTHHIST} -- (Unix systems only) specifies the directory in which to
1293: open/create the history file, @file{.gforth-history}. Default:
1294: @code{$HOME}.
1295:
1296: @item
1297: @cindex @code{GFORTHPATH} -- environment variable
1298: @code{GFORTHPATH} -- specifies the path used when searching for the gforth image file and
1299: for Forth source-code files.
1300:
1301: @item
1302: @cindex @code{GFORTH} -- environment variable
1303: @code{GFORTH} -- used by @file{gforthmi} @xref{gforthmi}.
1304:
1305: @item
1306: @cindex @code{GFORTHD} -- environment variable
1307: @code{GFORTHD} -- used by @file{gforthmi} @xref{gforthmi}.
1308:
1309: @item
1310: @cindex @code{TMP}, @code{TEMP} - environment variable
1311: @code{TMP}, @code{TEMP} - (non-Unix systems only) used as a potential
1312: location for the history file.
1313: @end itemize
1314:
1315: @comment also POSIXELY_CORRECT LINES COLUMNS HOME but no interest in
1316: @comment mentioning these.
1317:
1318: All the Gforth environment variables default to sensible values if they
1319: are not set.
1320:
1321:
1322: @comment ----------------------------------------------
1323: @node Gforth Files, ,Environment variables,Gforth Environment
1324: @section Gforth files
1325: @cindex Gforth files
1326:
1327: When you install Gforth on a Unix system, it installs files in these
1328: locations by default:
1329:
1330: @itemize @bullet
1331: @item
1332: @file{/usr/local/bin/gforth}
1333: @item
1334: @file{/usr/local/bin/gforthmi}
1335: @item
1336: @file{/usr/local/man/man1/gforth.1} - man page.
1337: @item
1338: @file{/usr/local/info} - the Info version of this manual.
1339: @item
1340: @file{/usr/local/lib/gforth/<version>/...} - Gforth @file{.fi} files.
1341: @item
1342: @file{/usr/local/share/gforth/<version>/TAGS} - Emacs TAGS file.
1343: @item
1344: @file{/usr/local/share/gforth/<version>/...} - Gforth source files.
1345: @item
1346: @file{.../emacs/site-lisp/gforth.el} - Emacs gforth mode.
1347: @end itemize
1348:
1349: You can select different places for installation by using
1350: @code{configure} options (listed with @code{configure --help}).
1351:
1352: @c ******************************************************************
1353: @node Introduction, Words, Gforth Environment, Top
1354: @comment node-name, next, previous, up
1355: @chapter An Introduction to ANS Forth
1356: @cindex Forth - an introduction
1357:
1358: The primary purpose of this manual is to document Gforth. However, since
1359: Forth is not a widely-known language and there is a lack of up-to-date
1360: teaching material, it seems worthwhile to provide some introductory
1361: material. @xref{Forth-related information} for other sources of Forth-related
1362: information.
1363:
1364: The examples in this section should work on any ANS Forth; the
1365: output shown was produced using Gforth. Each example attempts to
1366: reproduce the exact output that Gforth produces. If you try out the
1367: examples (and you should), what you should type is shown @kbd{like this}
1368: and Gforth's response is shown @code{like this}. The single exception is
1369: that, where the example shows @key{RET} it means that you should
1370: press the ``carriage return'' key. Unfortunately, some output formats for
1371: this manual cannot show the difference between @kbd{this} and
1372: @code{this} which will make trying out the examples harder (but not
1373: impossible).
1374:
1375: Forth is an unusual language. It provides an interactive development
1376: environment which includes both an interpreter and compiler. Forth
1377: programming style encourages you to break a problem down into many
1378: @cindex factoring
1379: small fragments (@dfn{factoring}), and then to develop and test each
1380: fragment interactively. Forth advocates assert that breaking the
1381: edit-compile-test cycle used by conventional programming languages can
1382: lead to great productivity improvements.
1383:
1384: @menu
1385: * Introducing the Text Interpreter::
1386: * Stacks and Postfix notation::
1387: * Your first definition::
1388: * How does that work?::
1389: * Forth is written in Forth::
1390: * Review - elements of a Forth system::
1391: * Where to go next::
1392: * Exercises::
1393: @end menu
1394:
1395: @comment ----------------------------------------------
1396: @node Introducing the Text Interpreter, Stacks and Postfix notation, Introduction, Introduction
1397: @section Introducing the Text Interpreter
1398: @cindex text interpreter
1399: @cindex outer interpreter
1400:
1401: @c IMO this is too detailed and the pace is too slow for
1402: @c an introduction. If you know German, take a look at
1403: @c http://www.complang.tuwien.ac.at/anton/lvas/skriptum-stack.html
1404: @c to see how I do it - anton
1405:
1406: @c nac-> Where I have accepted your comments 100% and modified the text
1407: @c accordingly, I have deleted your comments. Elsewhere I have added a
1408: @c response like this to attempt to rationalise what I have done. Of
1409: @c course, this is a very clumsy mechanism for something that would be
1410: @c done far more efficiently over a beer. Please delete any dialogue
1411: @c you consider closed.
1412:
1413: When you invoke the Forth image, you will see a startup banner printed
1414: and nothing else (if you have Gforth installed on your system, try
1415: invoking it now, by typing @kbd{gforth@key{RET}}). Forth is now running
1416: its command line interpreter, which is called the @dfn{Text Interpreter}
1417: (also known as the @dfn{Outer Interpreter}). (You will learn a lot
1418: about the text interpreter as you read through this chapter, but
1419: @pxref{The Text Interpreter} for more detail).
1420:
1421: Although it's not obvious, Forth is actually waiting for your
1422: input. Type a number and press the @key{RET} key:
1423:
1424: @example
1425: @kbd{45@key{RET}} ok
1426: @end example
1427:
1428: Rather than give you a prompt to invite you to input something, the text
1429: interpreter prints a status message @i{after} it has processed a line
1430: of input. The status message in this case (``@code{ ok}'' followed by
1431: carriage-return) indicates that the text interpreter was able to process
1432: all of your input successfully. Now type something illegal:
1433:
1434: @example
1435: @kbd{qwer341@key{RET}}
1436: :1: Undefined word
1437: qwer341
1438: ^^^^^^^
1439: $400D2BA8 Bounce
1440: $400DBDA8 no.extensions
1441: @end example
1442:
1443: The exact text, other than the ``Undefined word'' may differ slightly on
1444: your system, but the effect is the same; when the text interpreter
1445: detects an error, it discards any remaining text on a line, resets
1446: certain internal state and prints an error message. @xref{Error
1447: messages} for a detailed description of error messages.
1448:
1449: The text interpreter waits for you to press carriage-return, and then
1450: processes your input line. Starting at the beginning of the line, it
1451: breaks the line into groups of characters separated by spaces. For each
1452: group of characters in turn, it makes two attempts to do something:
1453:
1454: @itemize @bullet
1455: @item
1456: @cindex name dictionary
1457: It tries to treat it as a command. It does this by searching a @dfn{name
1458: dictionary}. If the group of characters matches an entry in the name
1459: dictionary, the name dictionary provides the text interpreter with
1460: information that allows the text interpreter perform some actions. In
1461: Forth jargon, we say that the group
1462: @cindex word
1463: @cindex definition
1464: @cindex execution token
1465: @cindex xt
1466: of characters names a @dfn{word}, that the dictionary search returns an
1467: @dfn{execution token (xt)} corresponding to the @dfn{definition} of the
1468: word, and that the text interpreter executes the xt. Often, the terms
1469: @dfn{word} and @dfn{definition} are used interchangeably.
1470: @item
1471: If the text interpreter fails to find a match in the name dictionary, it
1472: tries to treat the group of characters as a number in the current number
1473: base (when you start up Forth, the current number base is base 10). If
1474: the group of characters legitimately represents a number, the text
1475: interpreter pushes the number onto a stack (we'll learn more about that
1476: in the next section).
1477: @end itemize
1478:
1479: If the text interpreter is unable to do either of these things with any
1480: group of characters, it discards the group of characters and the rest of
1481: the line, then prints an error message. If the text interpreter reaches
1482: the end of the line without error, it prints the status message ``@code{ ok}''
1483: followed by carriage-return.
1484:
1485: This is the simplest command we can give to the text interpreter:
1486:
1487: @example
1488: @key{RET} ok
1489: @end example
1490:
1491: The text interpreter did everything we asked it to do (nothing) without
1492: an error, so it said that everything is ``@code{ ok}''. Try a slightly longer
1493: command:
1494:
1495: @example
1496: @kbd{12 dup fred dup@key{RET}}
1497: :1: Undefined word
1498: 12 dup fred dup
1499: ^^^^
1500: $400D2BA8 Bounce
1501: $400DBDA8 no.extensions
1502: @end example
1503:
1504: When you press the carriage-return key, the text interpreter starts to
1505: work its way along the line:
1506:
1507: @itemize @bullet
1508: @item
1509: When it gets to the space after the @code{2}, it takes the group of
1510: characters @code{12} and looks them up in the name
1511: dictionary@footnote{We can't tell if it found them or not, but assume
1512: for now that it did not}. There is no match for this group of characters
1513: in the name dictionary, so it tries to treat them as a number. It is
1514: able to do this successfully, so it puts the number, 12, ``on the stack''
1515: (whatever that means).
1516: @item
1517: The text interpreter resumes scanning the line and gets the next group
1518: of characters, @code{dup}. It looks it up in the name dictionary and
1519: (you'll have to take my word for this) finds it, and executes the word
1520: @code{dup} (whatever that means).
1521: @item
1522: Once again, the text interpreter resumes scanning the line and gets the
1523: group of characters @code{fred}. It looks them up in the name
1524: dictionary, but can't find them. It tries to treat them as a number, but
1525: they don't represent any legal number.
1526: @end itemize
1527:
1528: At this point, the text interpreter gives up and prints an error
1529: message. The error message shows exactly how far the text interpreter
1530: got in processing the line. In particular, it shows that the text
1531: interpreter made no attempt to do anything with the final character
1532: group, @code{dup}, even though we have good reason to believe that the
1533: text interpreter would have no problem looking that word up and
1534: executing it a second time.
1535:
1536:
1537: @comment ----------------------------------------------
1538: @node Stacks and Postfix notation, Your first definition, Introducing the Text Interpreter, Introduction
1539: @section Stacks, postfix notation and parameter passing
1540: @cindex text interpreter
1541: @cindex outer interpreter
1542:
1543: In procedural programming languages (like C and Pascal), the
1544: building-block of programs is the @dfn{function} or @dfn{procedure}. These
1545: functions or procedures are called with @dfn{explicit parameters}. For
1546: example, in C we might write:
1547:
1548: @example
1549: total = total + new_volume(length,height,depth);
1550: @end example
1551:
1552: @noindent
1553: where new_volume is a function-call to another piece of code, and total,
1554: length, height and depth are all variables. length, height and depth are
1555: parameters to the function-call.
1556:
1557: In Forth, the equivalent of the function or procedure is the
1558: @dfn{definition} and parameters are implicitly passed between
1559: definitions using a shared stack that is visible to the
1560: programmer. Although Forth does support variables, the existence of the
1561: stack means that they are used far less often than in most other
1562: programming languages. When the text interpreter encounters a number, it
1563: will place (@dfn{push}) it on the stack. There are several stacks (the
1564: actual number is implementation-dependent ...) and the particular stack
1565: used for any operation is implied unambiguously by the operation being
1566: performed. The stack used for all integer operations is called the @dfn{data
1567: stack} and, since this is the stack used most commonly, references to
1568: ``the data stack'' are often abbreviated to ``the stack''.
1569:
1570: The stacks have a last-in, first-out (LIFO) organisation. If you type:
1571:
1572: @example
1573: @kbd{1 2 3@key{RET}} ok
1574: @end example
1575:
1576: Then this instructs the text interpreter to placed three numbers on the
1577: (data) stack. An analogy for the behaviour of the stack is to take a
1578: pack of playing cards and deal out the ace (1), 2 and 3 into a pile on
1579: the table. The 3 was the last card onto the pile (``last-in'') and if
1580: you take a card off the pile then, unless you're prepared to fiddle a
1581: bit, the card that you take off will be the 3 (``first-out''). The
1582: number that will be first-out of the stack is called the @dfn{top of
1583: stack}, which
1584: @cindex TOS definition
1585: is often abbreviated to @dfn{TOS}.
1586:
1587: To understand how parameters are passed in Forth, consider the
1588: behaviour of the definition @code{+} (pronounced ``plus''). You will not
1589: be surprised to learn that this definition performs addition. More
1590: precisely, it adds two number together and produces a result. Where does
1591: it get the two numbers from? It takes the top two numbers off the
1592: stack. Where does it place the result? On the stack. You can act-out the
1593: behaviour of @code{+} with your playing cards like this:
1594:
1595: @itemize @bullet
1596: @item
1597: Pick up two cards from the stack on the table
1598: @item
1599: Stare at them intently and ask yourself ``what @i{is} the sum of these two
1600: numbers''
1601: @item
1602: Decide that the answer is 5
1603: @item
1604: Shuffle the two cards back into the pack and find a 5
1605: @item
1606: Put a 5 on the remaining ace that's on the table.
1607: @end itemize
1608:
1609: If you don't have a pack of cards handy but you do have Forth running,
1610: you can use the definition @code{.s} to show the current state of the stack,
1611: without affecting the stack. Type:
1612:
1613: @example
1614: @kbd{clearstack 1 2 3@key{RET}} ok
1615: @kbd{.s@key{RET}} <3> 1 2 3 ok
1616: @end example
1617:
1618: The text interpreter looks up the word @code{clearstack} and executes
1619: it; it tidies up the stack and removes any entries that may have been
1620: left on it by earlier examples. The text interpreter pushes each of the
1621: three numbers in turn onto the stack. Finally, the text interpreter
1622: looks up the word @code{.s} and executes it. The effect of executing
1623: @code{.s} is to print the ``<3>'' (the total number of items on the stack)
1624: followed by a list of all the items on the stack; the item on the far
1625: right-hand side is the TOS.
1626:
1627: You can now type:
1628:
1629: @example
1630: @kbd{+ .s@key{RET}} <2> 1 5 ok
1631: @end example
1632:
1633: @noindent
1634: which is correct; there are now 2 items on the stack and the result of
1635: the addition is 5.
1636:
1637: If you're playing with cards, try doing a second addition: pick up the
1638: two cards, work out that their sum is 6, shuffle them into the pack,
1639: look for a 6 and place that on the table. You now have just one item on
1640: the stack. What happens if you try to do a third addition? Pick up the
1641: first card, pick up the second card -- ah! There is no second card. This
1642: is called a @dfn{stack underflow} and consitutes an error. If you try to
1643: do the same thing with Forth it will report an error (probably a Stack
1644: Underflow or an Invalid Memory Address error).
1645:
1646: The opposite situation to a stack underflow is a @dfn{stack overflow},
1647: which simply accepts that there is a finite amount of storage space
1648: reserved for the stack. To stretch the playing card analogy, if you had
1649: enough packs of cards and you piled the cards up on the table, you would
1650: eventually be unable to add another card; you'd hit the ceiling. Gforth
1651: allows you to set the maximum size of the stacks. In general, the only
1652: time that you will get a stack overflow is because a definition has a
1653: bug in it and is generating data on the stack uncontrollably.
1654:
1655: There's one final use for the playing card analogy. If you model your
1656: stack using a pack of playing cards, the maximum number of items on
1657: your stack will be 52 (I assume you didn't use the Joker). The maximum
1658: @i{value} of any item on the stack is 13 (the King). In fact, the only
1659: possible numbers are positive integer numbers 1 through 13; you can't
1660: have (for example) 0 or 27 or 3.52 or -2. If you change the way you
1661: think about some of the cards, you can accommodate different
1662: numbers. For example, you could think of the Jack as representing 0,
1663: the Queen as representing -1 and the King as representing -2. Your
1664: @i{range} remains unchanged (you can still only represent a total of 13
1665: numbers) but the numbers that you can represent are -2 through 10.
1666:
1667: In that analogy, the limit was the amount of information that a single
1668: stack entry could hold, and Forth has a similar limit. In Forth, the
1669: size of a stack entry is called a @dfn{cell}. The actual size of a cell is
1670: implementation dependent and affects the maximum value that a stack
1671: entry can hold. A Standard Forth provides a cell size of at least
1672: 16-bits, and most desktop systems use a cell size of 32-bits.
1673:
1674: Forth does not do any type checking for you, so you are free to
1675: manipulate and combine stack items in any way you wish. A convenient way
1676: of treating stack items is as 2's complement signed integers, and that
1677: is what Standard words like @code{+} do. Therefore you can type:
1678:
1679: @example
1680: @kbd{-5 12 + .s@key{RET}} <1> 7 ok
1681: @end example
1682:
1683: If you use numbers and definitions like @code{+} in order to turn Forth
1684: into a great big pocket calculator, you will realise that it's rather
1685: different from a normal calculator. Rather than typing 2 + 3 = you had
1686: to type 2 3 + (ignore the fact that you had to use @code{.s} to see the
1687: result). The terminology used to describe this difference is to say that
1688: your calculator uses @dfn{Infix Notation} (parameters and operators are
1689: mixed) whilst Forth uses @dfn{Postfix Notation} (parameters and
1690: operators are separate), also called @dfn{Reverse Polish Notation}.
1691:
1692: Whilst postfix notation might look confusing to begin with, it has
1693: several important advantages:
1694:
1695: @itemize @bullet
1696: @item
1697: it is unambiguous
1698: @item
1699: it is more concise
1700: @item
1701: it fits naturally with a stack-based system
1702: @end itemize
1703:
1704: To examine these claims in more detail, consider these sums:
1705:
1706: @example
1707: 6 + 5 * 4 =
1708: 4 * 5 + 6 =
1709: @end example
1710:
1711: If you're just learning maths or your maths is very rusty, you will
1712: probably come up with the answer 44 for the first and 26 for the
1713: second. If you are a bit of a whizz at maths you will remember the
1714: @i{convention} that multiplication takes precendence over addition, and
1715: you'd come up with the answer 26 both times. To explain the answer 26
1716: to someone who got the answer 44, you'd probably rewrite the first sum
1717: like this:
1718:
1719: @example
1720: 6 + (5 * 4) =
1721: @end example
1722:
1723: If what you really wanted was to perform the addition before the
1724: multiplication, you would have to use parentheses to force it.
1725:
1726: If you did the first two sums on a pocket calculator you would probably
1727: get the right answers, unless you were very cautious and entered them using
1728: these keystroke sequences:
1729:
1730: 6 + 5 = * 4 =
1731: 4 * 5 = + 6 =
1732:
1733: Postfix notation is unambiguous because the order that the operators
1734: are applied is always explicit; that also means that parentheses are
1735: never required. The operators are @i{active} (the act of quoting the
1736: operator makes the operation occur) which removes the need for ``=''.
1737:
1738: The sum 6 + 5 * 4 can be written (in postfix notation) in two
1739: equivalent ways:
1740:
1741: @example
1742: 6 5 4 * + or:
1743: 5 4 * 6 +
1744: @end example
1745:
1746: An important thing that you should notice about this notation is that
1747: the @i{order} of the numbers does not change; if you want to subtract
1748: 2 from 10 you type @code{10 2 -}.
1749:
1750: The reason that Forth uses postfix notation is very simple to explain: it
1751: makes the implementation extremely simple, and it follows naturally from
1752: using the stack as a mechanism for passing parameters. Another way of
1753: thinking about this is to realise that all Forth definitions are
1754: @i{active}; they execute as they are encountered by the text
1755: interpreter. The result of this is that the syntax of Forth is trivially
1756: simple.
1757:
1758:
1759:
1760: @comment ----------------------------------------------
1761: @node Your first definition, How does that work?, Stacks and Postfix notation, Introduction
1762: @section Your first Forth definition
1763: @cindex first definition
1764:
1765: Until now, the examples we've seen have been trivial; we've just been
1766: using Forth as a bigger-than-pocket calculator. Also, each calculation
1767: we've shown has been a ``one-off'' -- to repeat it we'd need to type it in
1768: again@footnote{That's not quite true. If you press the up-arrow key on
1769: your keyboard you should be able to scroll back to any earlier command,
1770: edit it and re-enter it.} In this section we'll see how to add new
1771: words to Forth's vocabulary.
1772:
1773: The easiest way to create a new word is to use a @dfn{colon
1774: definition}. We'll define a few and try them out before worrying too
1775: much about how they work. Try typing in these examples; be careful to
1776: copy the spaces accurately:
1777:
1778: @example
1779: : add-two 2 + . ;
1780: : greet ." Hello and welcome" ;
1781: : demo 5 add-two ;
1782: @end example
1783:
1784: @noindent
1785: Now try them out:
1786:
1787: @example
1788: @kbd{greet@key{RET}} Hello and welcome ok
1789: @kbd{greet greet@key{RET}} Hello and welcomeHello and welcome ok
1790: @kbd{4 add-two@key{RET}} 6 ok
1791: @kbd{demo@key{RET}} 7 ok
1792: @kbd{9 greet demo add-two@key{RET}} Hello and welcome7 11 ok
1793: @end example
1794:
1795: The first new thing that we've introduced here is the pair of words
1796: @code{:} and @code{;}. These are used to start and terminate a new
1797: definition, respectively. The first word after the @code{:} is the name
1798: for the new definition.
1799:
1800: As you can see from the examples, a definition is built up of words that
1801: have already been defined; Forth makes no distinction between
1802: definitions that existed when you started the system up, and those that
1803: you define yourself.
1804:
1805: The examples also introduce the words @code{.} (dot), @code{."}
1806: (dot-quote) and @code{dup} (dewp). Dot takes the value from the top of
1807: the stack and displays it. It's like @code{.s} except that it only
1808: displays the top item of the stack and it is destructive; after it has
1809: executed, the number is no longer on the stack. There is always one
1810: space printed after the number, and no spaces before it. Dot-quote
1811: defines a string (a sequence of characters) that will be printed when
1812: the word is executed. The string can contain any printable characters
1813: except @code{"}. A @code{"} has a special function; it is not a Forth
1814: word but it acts as a delimiter (the way that delimiters work is
1815: described in the next section). Finally, @code{dup} duplicates the value
1816: at the top of the stack. Try typing @code{5 dup .s} to see what it does.
1817:
1818: We already know that the text interpreter searches through the
1819: dictionary to locate names. If you've followed the examples earlier, you
1820: will already have a definition called @code{add-two}. Lets try modifying
1821: it by typing in a new definition:
1822:
1823: @example
1824: @kbd{: add-two dup . ." + 2 =" 2 + . ;@key{RET}} redefined add-two ok
1825: @end example
1826:
1827: Forth recognised that we were defining a word that already exists, and
1828: printed a message to warn us of that fact. Let's try out the new
1829: definition:
1830:
1831: @example
1832: @kbd{9 add-two@key{RET}} 9 + 2 =11 ok
1833: @end example
1834:
1835: @noindent
1836: All that we've actually done here, though, is to create a new
1837: definition, with a particular name. The fact that there was already a
1838: definition with the same name did not make any difference to the way
1839: that the new definition was created (except that Forth printed a warning
1840: message). The old definition of add-two still exists (try @code{demo}
1841: again to see that this is true). Any new definition will use the new
1842: definition of @code{add-two}, but old definitions continue to use the
1843: version that already existed at the time that they were @code{compiled}.
1844:
1845: Before you go on to the next section, try defining and redefining some
1846: words of your own.
1847:
1848: @comment ----------------------------------------------
1849: @node How does that work?, Forth is written in Forth, Your first definition, Introduction
1850: @section How does that work?
1851: @cindex parsing words
1852:
1853: @c That's pretty deep (IMO way too deep) for an introduction. - anton
1854:
1855: @c Is it a good idea to talk about the interpretation semantics of a
1856: @c number? We don't have an xt to go along with it. - anton
1857:
1858: @c Now that I have eliminated execution semantics, I wonder if it would not
1859: @c be better to keep them (or add run-time semantics), to make it easier to
1860: @c explain what compilation semantics usually does. - anton
1861:
1862: @c nac-> I removed the term ``default compilation sematics'' from the
1863: @c introductory chapter. Removing ``execution semantics'' was making
1864: @c everything simpler to explain, then I think the use of this term made
1865: @c everything more complex again. I replaced it with ``default
1866: @c semantics'' (which is used elsewhere in the manual) by which I mean
1867: @c ``a definition that has neither the immediate nor the compile-only
1868: @c flag set''. I reworded big chunks of the ``how does that work''
1869: @c section (and, unusually for me, I think I even made it shorter!). See
1870: @c what you think -- I know I have not addressed your primary concern
1871: @c that it is too heavy-going for an introduction. From what I understood
1872: @c of your course notes it looks as though they might be a good framework.
1873: @c Things that I've tried to capture here are some things that came as a
1874: @c great revelation here when I first understood them. Also, I like the
1875: @c fact that a very simple code example shows up almost all of the issues
1876: @c that you need to understand to see how Forth works. That's unique and
1877: @c worthwhile to emphasise.
1878:
1879: Now we're going to take another look at the definition of @code{add-two}
1880: from the previous section. From our knowledge of the way that the text
1881: interpreter works, we would have expected this result when we tried to
1882: define @code{add-two}:
1883:
1884: @example
1885: @kbd{: add-two 2 + . ;@key{RET}}
1886: ^^^^^^^
1887: Error: Undefined word
1888: @end example
1889:
1890: The reason that this didn't happen is bound up in the way that @code{:}
1891: works. The word @code{:} does two special things. The first special
1892: thing that it does prevents the text interpreter from ever seeing the
1893: characters @code{add-two}. The text interpreter uses a variable called
1894: @cindex modifying >IN
1895: @code{>IN} (pronounced ``to-in'') to keep track of where it is in the
1896: input line. When it encounters the word @code{:} it behaves in exactly
1897: the same way as it does for any other word; it looks it up in the name
1898: dictionary, finds its xt and executes it. When @code{:} executes, it
1899: looks at the input buffer, finds the word @code{add-two} and advances the
1900: value of @code{>IN} to point past it. It then does some other stuff
1901: associated with creating the new definition (including creating an entry
1902: for @code{add-two} in the name dictionary). When the execution of @code{:}
1903: completes, control returns to the text interpreter, which is oblivious
1904: to the fact that it has been tricked into ignoring part of the input
1905: line.
1906:
1907: @cindex parsing words
1908: Words like @code{:} -- words that advance the value of @code{>IN} and so
1909: prevent the text interpreter from acting on the whole of the input line
1910: -- are called @dfn{parsing words}.
1911:
1912: @cindex @code{state} - effect on the text interpreter
1913: @cindex text interpreter - effect of state
1914: The second special thing that @code{:} does is change the value of a
1915: variable called @code{state}, which affects the way that the text
1916: interpreter behaves. When Gforth starts up, @code{state} has the value
1917: 0, and the text interpreter is said to be @dfn{interpreting}. During a
1918: colon definition (started with @code{:}), @code{state} is set to -1 and
1919: the text interpreter is said to be @dfn{compiling}.
1920:
1921: In this example, the text interpreter is compiling when it processes the
1922: string ``@code{2 + . ;}''. It still breaks the string down into
1923: character sequences in the same way. However, instead of pushing the
1924: number @code{2} onto the stack, it lays down (@dfn{compiles}) some magic
1925: into the definition of @code{add-two} that will make the number @code{2} get
1926: pushed onto the stack when @code{add-two} is @dfn{executed}. Similarly,
1927: the behaviours of @code{+} and @code{.} are also compiled into the
1928: definition.
1929:
1930: One category of words don't get compiled. These so-called @dfn{immediate
1931: words} get executed (performed @i{now}) regardless of whether the text
1932: interpreter is interpreting or compiling. The word @code{;} is an
1933: immediate word. Rather than being compiled into the definition, it
1934: executes. Its effect is to terminate the current definition, which
1935: includes changing the value of @code{state} back to 0.
1936:
1937: When you execute @code{add-two}, it has a @dfn{run-time effect} that is
1938: exactly the same as if you had typed @code{2 + . @key{RET}} outside of a
1939: definition.
1940:
1941: In Forth, every word or number can be described in terms of two
1942: properties:
1943:
1944: @itemize @bullet
1945: @item
1946: @cindex interpretation semantics
1947: Its @dfn{interpretation semantics} describe how it will behave when the
1948: text interpreter encounters it in @dfn{interpret} state. The
1949: interpretation semantics of a word are represented by an @dfn{execution
1950: token}.
1951: @item
1952: @cindex compilation semantics
1953: Its @dfn{compilation semantics} describe how it will behave when the
1954: text interpreter encounters it in @dfn{compile} state. The compilation
1955: semantics of a word are represented in an implementation-dependent way;
1956: Gforth uses a @dfn{compilation token}.
1957: @end itemize
1958:
1959: @noindent
1960: Numbers are always treated in a fixed way:
1961:
1962: @itemize @bullet
1963: @item
1964: When the number is @dfn{interpreted}, its behaviour is to push the
1965: number onto the stack.
1966: @item
1967: When the number is @dfn{compiled}, a piece of code is appended to the
1968: current definition that pushes the number when it runs. (In other words,
1969: the compilation semantics of a number are to postpone its interpretation
1970: semantics until the run-time of the definition that it is being compiled
1971: into.)
1972: @end itemize
1973:
1974: Words don't behave in such a regular way, but most have @i{default
1975: semantics} which means that they behave like this:
1976:
1977: @itemize @bullet
1978: @item
1979: The @dfn{interpretation semantics} of the word are to do something useful.
1980: @item
1981: The @dfn{compilation semantics} of the word are to append its
1982: @dfn{interpretation semantics} to the current definition (so that its
1983: run-time behaviour is to do something useful).
1984: @end itemize
1985:
1986: @cindex immediate words
1987: The actual behaviour of any particular word can be controlled by using
1988: the words @code{immediate} and @code{compile-only} when the word is
1989: defined. These words set flags in the name dictionary entry of the most
1990: recently defined word, and these flags are retrieved by the text
1991: interpreter when it finds the word in the name dictionary.
1992:
1993: A word that is marked as @dfn{immediate} has compilation semantics that
1994: are identical to its interpretation semantics. In other words, it
1995: behaves like this:
1996:
1997: @itemize @bullet
1998: @item
1999: The @dfn{interpretation semantics} of the word are to do something useful.
2000: @item
2001: The @dfn{compilation semantics} of the word are to do something useful
2002: (and actually the same thing); i.e., it is executed during compilation.
2003: @end itemize
2004:
2005: Marking a word as @dfn{compile-only} prohibits the text interpreter from
2006: performing the interpretation semantics of the word directly; an attempt
2007: to do so will generate an error. It is never necessary to use
2008: @code{compile-only} (and it is not even part of ANS Forth, though it is
2009: provided by many implementations) but it is good etiquette to apply it
2010: to a word that will not behave correctly (and might have unexpected
2011: side-effects) in interpret state. For example, it is only legal to use
2012: the conditional word @code{IF} within a definition. If you forget this
2013: and try to use it elsewhere, the fact that (in Gforth) it is marked as
2014: @code{compile-only} allows the text interpreter to generate a helpful
2015: error message rather than subjecting you to the consequences of your
2016: folly.
2017:
2018: This example shows the difference between an immediate and a
2019: non-immediate word:
2020:
2021: @example
2022: : show-state state @@ . ;
2023: : show-state-now show-state ; immediate
2024: : word1 show-state ;
2025: : word2 show-state-now ;
2026: @end example
2027:
2028: The word @code{immediate} after the definition of @code{show-state-now}
2029: makes that word an immediate word. These definitions introduce a new
2030: word: @code{@@} (pronounced ``fetch''). This word fetches the value of a
2031: variable, and leaves it on the stack. Therefore, the behaviour of
2032: @code{show-state} is to print a number that represents the current value
2033: of @code{state}.
2034:
2035: When you execute @code{word1}, it prints the number 0, indicating that
2036: the system is interpreting. When the text interpreter compiled the
2037: definition of @code{word1}, it encountered @code{show-state} whose
2038: compilation semantics are to append its interpretation semantics to the
2039: current definition. When you execute @code{word1}, it performs the
2040: interpretation semantics of @code{show-state}. At the time that @code{word1}
2041: (and therefore @code{show-state}) are executed, the system is
2042: interpreting.
2043:
2044: When you pressed @key{RET} after entering the definition of @code{word2},
2045: you should have seen the number -1 printed, followed by ``@code{
2046: ok}''. When the text interpreter compiled the definition of
2047: @code{word2}, it encountered @code{show-state-now}, an immediate word,
2048: whose compilation semantics are therefore to perform its interpretation
2049: semantics. It is executed straight away (even before the text
2050: interpreter has moved on to process another group of characters; the
2051: @code{;} in this example). The effect of executing it are to display the
2052: value of @code{state} @i{at the time that the definition of}
2053: @code{word2} @i{is being defined}. Printing -1 demonstrates that the
2054: system is compiling at this time. If you execute @code{word2} it does
2055: nothing at all.
2056:
2057: @cindex @code{."}, how it works
2058: Before leaving the subject of immediate words, consider the behaviour of
2059: @code{."} in the definition of @code{greet}, in the previous
2060: section. This word is both a parsing word and an immediate word. Notice
2061: that there is a space between @code{."} and the start of the text
2062: @code{Hello and welcome}, but that there is no space between the last
2063: letter of @code{welcome} and the @code{"} character. The reason for this
2064: is that @code{."} is a Forth word; it must have a space after it so that
2065: the text interpreter can identify it. The @code{"} is not a Forth word;
2066: it is a @dfn{delimiter}. The examples earlier show that, when the string
2067: is displayed, there is neither a space before the @code{H} nor after the
2068: @code{e}. Since @code{."} is an immediate word, it executes at the time
2069: that @code{greet} is defined. When it executes, its behaviour is to
2070: search forward in the input line looking for the delimiter. When it
2071: finds the delimiter, it updates @code{>IN} to point past the
2072: delimiter. It also compiles some magic code into the definition of
2073: @code{greet}; the xt of a run-time routine that prints a text string. It
2074: compiles the string @code{Hello and welcome} into memory so that it is
2075: available to be printed later. When the text interpreter gains control,
2076: the next word it finds in the input stream is @code{;} and so it
2077: terminates the definition of @code{greet}.
2078:
2079:
2080: @comment ----------------------------------------------
2081: @node Forth is written in Forth, Review - elements of a Forth system, How does that work?, Introduction
2082: @section Forth is written in Forth
2083: @cindex structure of Forth programs
2084:
2085: When you start up a Forth compiler, a large number of definitions
2086: already exist. In Forth, you develop a new application using bottom-up
2087: programming techniques to create new definitions that are defined in
2088: terms of existing definitions. As you create each definition you can
2089: test and debug it interactively.
2090:
2091: If you have tried out the examples in this section, you will probably
2092: have typed them in by hand; when you leave Gforth, your definitions will
2093: be lost. You can avoid this by using a text editor to enter Forth source
2094: code into a file, and then loading code from the file using
2095: @code{include} (@xref{Forth source files}). A Forth source file is
2096: processed by the text interpreter, just as though you had typed it in by
2097: hand@footnote{Actually, there are some subtle differences -- see
2098: @ref{The Text Interpreter}.}.
2099:
2100: Gforth also supports the traditional Forth alternative to using text
2101: files for program entry (@xref{Blocks}).
2102:
2103: In common with many, if not most, Forth compilers, most of Gforth is
2104: actually written in Forth. All of the @file{.fs} files in the
2105: installation directory@footnote{For example,
2106: @file{/usr/local/share/gforth...}} are Forth source files, which you can
2107: study to see examples of Forth programming.
2108:
2109: Gforth maintains a history file that records every line that you type to
2110: the text interpreter. This file is preserved between sessions, and is
2111: used to provide a command-line recall facility. If you enter long
2112: definitions by hand, you can use a text editor to paste them out of the
2113: history file into a Forth source file for reuse at a later time
2114: (@pxref{Command-line editing} for more information).
2115:
2116:
2117: @comment ----------------------------------------------
2118: @node Review - elements of a Forth system, Where to go next, Forth is written in Forth, Introduction
2119: @section Review - elements of a Forth system
2120: @cindex elements of a Forth system
2121:
2122: To summarise this chapter:
2123:
2124: @itemize @bullet
2125: @item
2126: Forth programs use @dfn{factoring} to break a problem down into small
2127: fragments called @dfn{words} or @dfn{definitions}.
2128: @item
2129: Forth program development is an interactive process.
2130: @item
2131: The main command loop that accepts input, and controls both
2132: interpretation and compilation, is called the @dfn{text interpreter}
2133: (also known as the @dfn{outer interpreter}).
2134: @item
2135: Forth has a very simple syntax, consisting of words and numbers
2136: separated by spaces or carriage-return characters. Any additional syntax
2137: is imposed by @dfn{parsing words}.
2138: @item
2139: Forth uses a stack to pass parameters between words. As a result, it
2140: uses postfix notation.
2141: @item
2142: To use a word that has previously been defined, the text interpreter
2143: searches for the word in the @dfn{name dictionary}.
2144: @item
2145: Words have @dfn{interpretation semantics} and @dfn{compilation semantics}.
2146: @item
2147: The text interpreter uses the value of @code{state} to select between
2148: the use of the @dfn{interpretation semantics} and the @dfn{compilation
2149: semantics} of a word that it encounters.
2150: @item
2151: The relationship between the @dfn{interpretation semantics} and
2152: @dfn{compilation semantics} for a word
2153: depend upon the way in which the word was defined (for example, whether
2154: it is an @dfn{immediate} word).
2155: @item
2156: Forth definitions can be implemented in Forth (called @dfn{high-level
2157: definitions}) or in some other way (usually a lower-level language and
2158: as a result often called @dfn{low-level definitions}, @dfn{code
2159: definitions} or @dfn{primitives}).
2160: @item
2161: Many Forth systems are implemented mainly in Forth.
2162: @end itemize
2163:
2164:
2165: @comment ----------------------------------------------
2166: @node Where to go next,Exercises,Review - elements of a Forth system, Introduction
2167: @section Where To Go Next
2168: @cindex where to go next
2169:
2170: Amazing as it may seem, if you have read (and understood) this far, you
2171: know almost all the fundamentals about the inner workings of a Forth
2172: system. You certainly know enough to be able to read and understand the
2173: rest of this manual and the ANS Forth document, to learn more about the
2174: facilities that Forth in general and Gforth in particular provide. Even
2175: scarier, you know almost enough to implement your own Forth system.
2176: However, that's not a good idea just yet... better to try writing some
2177: programs in Gforth.
2178:
2179: Forth has such a rich vocabulary that it can be hard to know where to
2180: start in learning it. This section suggests a few sets of words that are
2181: enough to write small but useful programs. Use the word index in this
2182: document to learn more about each word, then try it out and try to write
2183: small definitions using it. Start by experimenting with these words:
2184:
2185: @itemize @bullet
2186: @item
2187: Arithmetic: @code{+ - * / /MOD */ ABS INVERT}
2188: @item
2189: Comparison: @code{MIN MAX =}
2190: @item
2191: Logic: @code{AND OR XOR NOT}
2192: @item
2193: Stack manipulation: @code{DUP DROP SWAP OVER}
2194: @item
2195: Loops and decisions: @code{IF ELSE ENDIF ?DO I LOOP}
2196: @item
2197: Input/Output: @code{. ." EMIT CR KEY}
2198: @item
2199: Defining words: @code{: ; CREATE}
2200: @item
2201: Memory allocation words: @code{ALLOT ,}
2202: @item
2203: Tools: @code{SEE WORDS .S MARKER}
2204: @end itemize
2205:
2206: When you have mastered those, go on to:
2207:
2208: @itemize @bullet
2209: @item
2210: More defining words: @code{VARIABLE CONSTANT VALUE TO CREATE DOES>}
2211: @item
2212: Memory access: @code{@@ !}
2213: @end itemize
2214:
2215: When you have mastered these, there's nothing for it but to read through
2216: the whole of this manual and find out what you've missed.
2217:
2218: @comment ----------------------------------------------
2219: @node Exercises, ,Where to go next, Introduction
2220: @section Exercises
2221: @cindex exercises
2222:
2223: TODO: provide a set of programming excercises linked into the stuff done
2224: already and into other sections of the manual. Provide solutions to all
2225: the exercises in a .fs file in the distribution.
2226:
2227: @c Get some inspiration from Starting Forth and Kelly&Spies.
2228:
2229: @c excercises:
2230: @c 1. take inches and convert to feet and inches.
2231: @c 2. take temperature and convert from fahrenheight to celcius;
2232: @c may need to care about symmetric vs floored??
2233: @c 3. take input line and do character substitution
2234: @c to encipher or decipher
2235: @c 4. as above but work on a file for in and out
2236: @c 5. take input line and convert to pig-latin
2237: @c
2238: @c thing of sets of things to exercise then come up with
2239: @c problems that need those things.
2240:
2241:
2242: @c ******************************************************************
2243: @node Words, Error messages, Introduction, Top
2244: @chapter Forth Words
2245: @cindex words
2246:
2247: @menu
2248: * Notation::
2249: * Comments::
2250: * Boolean Flags::
2251: * Arithmetic::
2252: * Stack Manipulation::
2253: * Memory::
2254: * Control Structures::
2255: * Defining Words::
2256: * The Text Interpreter::
2257: * Tokens for Words::
2258: * Word Lists::
2259: * Environmental Queries::
2260: * Files::
2261: * Blocks::
2262: * Other I/O::
2263: * Programming Tools::
2264: * Assembler and Code Words::
2265: * Threading Words::
2266: * Locals::
2267: * Structures::
2268: * Object-oriented Forth::
2269: * Passing Commands to the OS::
2270: * Miscellaneous Words::
2271: @end menu
2272:
2273: @node Notation, Comments, Words, Words
2274: @section Notation
2275: @cindex notation of glossary entries
2276: @cindex format of glossary entries
2277: @cindex glossary notation format
2278: @cindex word glossary entry format
2279:
2280: The Forth words are described in this section in the glossary notation
2281: that has become a de-facto standard for Forth texts, i.e.,
2282:
2283: @format
2284: @i{word} @i{Stack effect} @i{wordset} @i{pronunciation}
2285: @end format
2286: @i{Description}
2287:
2288: @table @var
2289: @item word
2290: The name of the word.
2291:
2292: @item Stack effect
2293: @cindex stack effect
2294: The stack effect is written in the notation @code{@i{before} --
2295: @i{after}}, where @i{before} and @i{after} describe the top of
2296: stack entries before and after the execution of the word. The rest of
2297: the stack is not touched by the word. The top of stack is rightmost,
2298: i.e., a stack sequence is written as it is typed in. Note that Gforth
2299: uses a separate floating point stack, but a unified stack
2300: notation. Also, return stack effects are not shown in @i{stack
2301: effect}, but in @i{Description}. The name of a stack item describes
2302: the type and/or the function of the item. See below for a discussion of
2303: the types.
2304:
2305: All words have two stack effects: A compile-time stack effect and a
2306: run-time stack effect. The compile-time stack-effect of most words is
2307: @i{ -- }. If the compile-time stack-effect of a word deviates from
2308: this standard behaviour, or the word does other unusual things at
2309: compile time, both stack effects are shown; otherwise only the run-time
2310: stack effect is shown.
2311:
2312: @cindex pronounciation of words
2313: @item pronunciation
2314: How the word is pronounced.
2315:
2316: @cindex wordset
2317: @item wordset
2318: The ANS Forth standard is divided into several word sets. A standard
2319: system need not support all of them. Therefore, in theory, the fewer
2320: word sets your program uses the more portable it will be. However, we
2321: suspect that most ANS Forth systems on personal machines will feature
2322: all word sets. Words that are not defined in ANS Forth have
2323: @code{gforth} or @code{gforth-internal} as word set. @code{gforth}
2324: describes words that will work in future releases of Gforth;
2325: @code{gforth-internal} words are more volatile. Environmental query
2326: strings are also displayed like words; you can recognize them by the
2327: @code{environment} in the word set field.
2328:
2329: @item Description
2330: A description of the behaviour of the word.
2331: @end table
2332:
2333: @cindex types of stack items
2334: @cindex stack item types
2335: The type of a stack item is specified by the character(s) the name
2336: starts with:
2337:
2338: @table @code
2339: @item f
2340: @cindex @code{f}, stack item type
2341: Boolean flags, i.e. @code{false} or @code{true}.
2342: @item c
2343: @cindex @code{c}, stack item type
2344: Char
2345: @item w
2346: @cindex @code{w}, stack item type
2347: Cell, can contain an integer or an address
2348: @item n
2349: @cindex @code{n}, stack item type
2350: signed integer
2351: @item u
2352: @cindex @code{u}, stack item type
2353: unsigned integer
2354: @item d
2355: @cindex @code{d}, stack item type
2356: double sized signed integer
2357: @item ud
2358: @cindex @code{ud}, stack item type
2359: double sized unsigned integer
2360: @item r
2361: @cindex @code{r}, stack item type
2362: Float (on the FP stack)
2363: @item a-
2364: @cindex @code{a_}, stack item type
2365: Cell-aligned address
2366: @item c-
2367: @cindex @code{c_}, stack item type
2368: Char-aligned address (note that a Char may have two bytes in Windows NT)
2369: @item f-
2370: @cindex @code{f_}, stack item type
2371: Float-aligned address
2372: @item df-
2373: @cindex @code{df_}, stack item type
2374: Address aligned for IEEE double precision float
2375: @item sf-
2376: @cindex @code{sf_}, stack item type
2377: Address aligned for IEEE single precision float
2378: @item xt
2379: @cindex @code{xt}, stack item type
2380: Execution token, same size as Cell
2381: @item wid
2382: @cindex @code{wid}, stack item type
2383: Word list ID, same size as Cell
2384: @item f83name
2385: @cindex @code{f83name}, stack item type
2386: Pointer to a name structure
2387: @item "
2388: @cindex @code{"}, stack item type
2389: string in the input stream (not on the stack). The terminating character
2390: is a blank by default. If it is not a blank, it is shown in @code{<>}
2391: quotes.
2392: @end table
2393:
2394: @node Comments, Boolean Flags, Notation, Words
2395: @section Comments
2396: @cindex comments
2397:
2398: Forth supports two styles of comment; the traditional @i{in-line} comment,
2399: @code{(} and its modern cousin, the @i{comment to end of line}; @code{\}.
2400:
2401:
2402: doc-(
2403: doc-\
2404: doc-\G
2405:
2406:
2407: @node Boolean Flags, Arithmetic, Comments, Words
2408: @section Boolean Flags
2409: @cindex Boolean flags
2410:
2411: A Boolean flag is cell-sized. A cell with all bits clear represents the
2412: flag @code{false} and a flag with all bits set represents the flag
2413: @code{true}. Words that check a flag (for example, @code{IF}) will treat
2414: a cell that has @i{any} bit set as @code{true}.
2415:
2416:
2417: doc-true
2418: doc-false
2419: doc-on
2420: doc-off
2421:
2422:
2423: @node Arithmetic, Stack Manipulation, Boolean Flags, Words
2424: @section Arithmetic
2425: @cindex arithmetic words
2426:
2427: @cindex division with potentially negative operands
2428: Forth arithmetic is not checked, i.e., you will not hear about integer
2429: overflow on addition or multiplication, you may hear about division by
2430: zero if you are lucky. The operator is written after the operands, but
2431: the operands are still in the original order. I.e., the infix @code{2-1}
2432: corresponds to @code{2 1 -}. Forth offers a variety of division
2433: operators. If you perform division with potentially negative operands,
2434: you do not want to use @code{/} or @code{/mod} with its undefined
2435: behaviour, but rather @code{fm/mod} or @code{sm/mod} (probably the
2436: former, @pxref{Mixed precision}).
2437: @comment TODO discuss the different division forms and the std approach
2438:
2439: @menu
2440: * Single precision::
2441: * Bitwise operations::
2442: * Double precision:: Double-cell integer arithmetic
2443: * Numeric comparison::
2444: * Mixed precision:: Operations with single and double-cell integers
2445: * Floating Point::
2446: @end menu
2447:
2448: @node Single precision, Bitwise operations, Arithmetic, Arithmetic
2449: @subsection Single precision
2450: @cindex single precision arithmetic words
2451:
2452: By default, numbers in Forth are single-precision integers that are 1
2453: cell in size. They can be signed or unsigned, depending upon how you
2454: treat them. @xref{Number Conversion} for the rules used by the text
2455: interpreter for recognising single-precision integers.
2456:
2457:
2458: doc-+
2459: doc-1+
2460: doc--
2461: doc-1-
2462: doc-*
2463: doc-/
2464: doc-mod
2465: doc-/mod
2466: doc-negate
2467: doc-abs
2468: doc-min
2469: doc-max
2470: doc-d>s
2471: doc-floored
2472:
2473:
2474: @node Bitwise operations, Double precision, Single precision, Arithmetic
2475: @subsection Bitwise operations
2476: @cindex bitwise operation words
2477:
2478:
2479: doc-and
2480: doc-or
2481: doc-xor
2482: doc-invert
2483: doc-lshift
2484: doc-rshift
2485: doc-2*
2486: doc-d2*
2487: doc-2/
2488: doc-d2/
2489:
2490:
2491: @node Double precision, Numeric comparison, Bitwise operations, Arithmetic
2492: @subsection Double precision
2493: @cindex double precision arithmetic words
2494:
2495: @xref{Number Conversion} for the rules used by the text interpreter for
2496: recognising double-precision integers.
2497:
2498: A double precision number is represented by a cell pair, with the most
2499: significant cell at the TOS. It is trivial to convert an unsigned
2500: single to an (unsigned) double; simply push a @code{0} onto the
2501: TOS. Since numbers are represented by Gforth using 2's complement
2502: arithmetic, converting a signed single to a (signed) double requires
2503: sign-extension across the most significant cell. This can be achieved
2504: using @code{s>d}. The moral of the story is that you cannot convert a
2505: number without knowing whether it represents an unsigned or a
2506: signed number.
2507:
2508:
2509: doc-s>d
2510: doc-d+
2511: doc-d-
2512: doc-dnegate
2513: doc-dabs
2514: doc-dmin
2515: doc-dmax
2516:
2517:
2518: @node Numeric comparison, Mixed precision, Double precision, Arithmetic
2519: @subsection Numeric comparison
2520: @cindex numeric comparison words
2521:
2522:
2523: doc-<
2524: doc-<=
2525: doc-<>
2526: doc-=
2527: doc->
2528: doc->=
2529:
2530: doc-0<
2531: doc-0<=
2532: doc-0<>
2533: doc-0=
2534: doc-0>
2535: doc-0>=
2536:
2537: doc-u<
2538: doc-u<=
2539: @c u<> and u= exist but are the same as <> and =
2540: @c doc-u<>
2541: @c doc-u=
2542: doc-u>
2543: doc-u>=
2544:
2545: doc-within
2546:
2547: doc-d<
2548: doc-d<=
2549: doc-d<>
2550: doc-d=
2551: doc-d>
2552: doc-d>=
2553:
2554: doc-d0<
2555: doc-d0<=
2556: doc-d0<>
2557: doc-d0=
2558: doc-d0>
2559: doc-d0>=
2560:
2561: doc-du<
2562: doc-du<=
2563: @c du<> and du= exist but are the same as d<> and d=
2564: @c doc-du<>
2565: @c doc-du=
2566: doc-du>
2567: doc-du>=
2568:
2569:
2570: @node Mixed precision, Floating Point, Numeric comparison, Arithmetic
2571: @subsection Mixed precision
2572: @cindex mixed precision arithmetic words
2573:
2574:
2575: doc-m+
2576: doc-*/
2577: doc-*/mod
2578: doc-m*
2579: doc-um*
2580: doc-m*/
2581: doc-um/mod
2582: doc-fm/mod
2583: doc-sm/rem
2584:
2585:
2586: @node Floating Point, , Mixed precision, Arithmetic
2587: @subsection Floating Point
2588: @cindex floating point arithmetic words
2589:
2590: @xref{Number Conversion} for the rules used by the text interpreter for
2591: recognising floating-point numbers.
2592:
2593: Gforth has a separate floating point
2594: stack, but the documentation uses the unified notation.
2595:
2596: @cindex floating-point arithmetic, pitfalls
2597: Floating point numbers have a number of unpleasant surprises for the
2598: unwary (e.g., floating point addition is not associative) and even a few
2599: for the wary. You should not use them unless you know what you are doing
2600: or you don't care that the results you get are totally bogus. If you
2601: want to learn about the problems of floating point numbers (and how to
2602: avoid them), you might start with @cite{David Goldberg, What Every
2603: Computer Scientist Should Know About Floating-Point Arithmetic, ACM
2604: Computing Surveys 23(1):5@minus{}48, March 1991}
2605: (@url{http://www.validgh.com/goldberg/paper.ps}).
2606:
2607:
2608: doc-d>f
2609: doc-f>d
2610: doc-f+
2611: doc-f-
2612: doc-f*
2613: doc-f/
2614: doc-fnegate
2615: doc-fabs
2616: doc-fmax
2617: doc-fmin
2618: doc-floor
2619: doc-fround
2620: doc-f**
2621: doc-fsqrt
2622: doc-fexp
2623: doc-fexpm1
2624: doc-fln
2625: doc-flnp1
2626: doc-flog
2627: doc-falog
2628: doc-f2*
2629: doc-f2/
2630: doc-1/f
2631: doc-precision
2632: doc-set-precision
2633:
2634: @cindex angles in trigonometric operations
2635: @cindex trigonometric operations
2636: Angles in floating point operations are given in radians (a full circle
2637: has 2 pi radians).
2638:
2639: doc-fsin
2640: doc-fcos
2641: doc-fsincos
2642: doc-ftan
2643: doc-fasin
2644: doc-facos
2645: doc-fatan
2646: doc-fatan2
2647: doc-fsinh
2648: doc-fcosh
2649: doc-ftanh
2650: doc-fasinh
2651: doc-facosh
2652: doc-fatanh
2653: doc-pi
2654:
2655: @cindex equality of floats
2656: @cindex floating-point comparisons
2657: One particular problem with floating-point arithmetic is that comparison
2658: for equality often fails when you would expect it to succeed. For this
2659: reason approximate equality is often preferred (but you still have to
2660: know what you are doing). The comparison words are:
2661:
2662: doc-f~rel
2663: doc-f~abs
2664: doc-f=
2665: doc-f~
2666: doc-f<>
2667:
2668: doc-f<
2669: doc-f<=
2670: doc-f>
2671: doc-f>=
2672:
2673: doc-f0<
2674: doc-f0<=
2675: doc-f0<>
2676: doc-f0=
2677: doc-f0>
2678: doc-f0>=
2679:
2680:
2681: @node Stack Manipulation, Memory, Arithmetic, Words
2682: @section Stack Manipulation
2683: @cindex stack manipulation words
2684:
2685: @cindex floating-point stack in the standard
2686: Gforth maintains a number of separate stacks:
2687:
2688: @cindex data stack
2689: @cindex parameter stack
2690: @itemize @bullet
2691: @item
2692: A data stack (also known as the @dfn{parameter stack}) -- for
2693: characters, cells, addresses, and double cells.
2694:
2695: @cindex floating-point stack
2696: @item
2697: A floating point stack -- for holding floating point (FP) numbers.
2698:
2699: @cindex return stack
2700: @item
2701: A return stack -- for holding the return addresses of colon
2702: definitions and other (non-FP) data.
2703:
2704: @cindex locals stack
2705: @item
2706: A locals stack -- for holding local variables.
2707: @end itemize
2708:
2709: @menu
2710: * Data stack::
2711: * Floating point stack::
2712: * Return stack::
2713: * Locals stack::
2714: * Stack pointer manipulation::
2715: @end menu
2716:
2717: @node Data stack, Floating point stack, Stack Manipulation, Stack Manipulation
2718: @subsection Data stack
2719: @cindex data stack manipulation words
2720: @cindex stack manipulations words, data stack
2721:
2722:
2723: doc-drop
2724: doc-nip
2725: doc-dup
2726: doc-over
2727: doc-tuck
2728: doc-swap
2729: doc-pick
2730: doc-rot
2731: doc--rot
2732: doc-?dup
2733: doc-roll
2734: doc-2drop
2735: doc-2nip
2736: doc-2dup
2737: doc-2over
2738: doc-2tuck
2739: doc-2swap
2740: doc-2rot
2741:
2742:
2743: @node Floating point stack, Return stack, Data stack, Stack Manipulation
2744: @subsection Floating point stack
2745: @cindex floating-point stack manipulation words
2746: @cindex stack manipulation words, floating-point stack
2747:
2748: Whilst every sane Forth has a separate floating-point stack, it is not
2749: strictly required; an ANS Forth system could theoretically keep
2750: floating-point numbers on the data stack. As an additional difficulty,
2751: you don't know how many cells a floating-point number takes. It is
2752: reportedly possible to write words in a way that they work also for a
2753: unified stack model, but we do not recommend trying it. Instead, just
2754: say that your program has an environmental dependency on a separate
2755: floating-point stack.
2756:
2757: doc-floating-stack
2758:
2759: doc-fdrop
2760: doc-fnip
2761: doc-fdup
2762: doc-fover
2763: doc-ftuck
2764: doc-fswap
2765: doc-fpick
2766: doc-frot
2767:
2768:
2769: @node Return stack, Locals stack, Floating point stack, Stack Manipulation
2770: @subsection Return stack
2771: @cindex return stack manipulation words
2772: @cindex stack manipulation words, return stack
2773:
2774: @cindex return stack and locals
2775: @cindex locals and return stack
2776: A Forth system is allowed to keep local variables on the
2777: return stack. This is reasonable, as local variables usually eliminate
2778: the need to use the return stack explicitly. So, if you want to produce
2779: a standard compliant program and you are using local variables in a
2780: word, forget about return stack manipulations in that word (refer to the
2781: standard document for the exact rules).
2782:
2783: doc->r
2784: doc-r>
2785: doc-r@
2786: doc-rdrop
2787: doc-2>r
2788: doc-2r>
2789: doc-2r@
2790: doc-2rdrop
2791:
2792:
2793: @node Locals stack, Stack pointer manipulation, Return stack, Stack Manipulation
2794: @subsection Locals stack
2795:
2796: @comment TODO
2797:
2798: @node Stack pointer manipulation, , Locals stack, Stack Manipulation
2799: @subsection Stack pointer manipulation
2800: @cindex stack pointer manipulation words
2801:
2802: @c removed s0 r0 l0 -- they are obsolete aliases for sp0 rp0 lp0
2803: doc-sp0
2804: doc-sp@
2805: doc-sp!
2806: doc-fp0
2807: doc-fp@
2808: doc-fp!
2809: doc-rp0
2810: doc-rp@
2811: doc-rp!
2812: doc-lp0
2813: doc-lp@
2814: doc-lp!
2815:
2816:
2817: @node Memory, Control Structures, Stack Manipulation, Words
2818: @section Memory
2819: @cindex memory words
2820:
2821: @menu
2822: * Memory model::
2823: * Dictionary allocation::
2824: * Heap Allocation::
2825: * Memory Access::
2826: * Address arithmetic::
2827: * Memory Blocks::
2828: @end menu
2829:
2830: @node Memory model, Dictionary allocation, Memory, Memory
2831: @subsection ANS Forth and Gforth memory models
2832:
2833: @c The ANS Forth description is a mess (e.g., is the heap part of
2834: @c the dictionary?), so let's not stick to closely with it.
2835:
2836: ANS Forth considers a Forth system as consisting of several memories, of
2837: which only @dfn{data space} is managed and accessible with the memory
2838: words. Memory not necessarily in data space includes the stacks, the
2839: code (called code space) and the headers (called name space). In Gforth
2840: everything is in data space, but the code for the primitives is usually
2841: read-only.
2842:
2843: Data space is divided into a number of areas: The (data space portion of
2844: the) dictionary@footnote{Sometimes, the term @dfn{dictionary} is used to
2845: refer to the search data structure embodied in word lists and headers,
2846: because it is used for looking up names, just as you would in a
2847: conventional dictionary.}, the heap, and a number of system-allocated
2848: buffers.
2849:
2850: In ANS Forth data space is also divided into contiguous regions. You
2851: can only use address arithmetic within a contiguous region, not between
2852: them. Usually each allocation gives you one contiguous region, but the
2853: dictionary allocation words have additional rules (@pxref{Dictionary
2854: allocation}).
2855:
2856: Gforth provides one big address space, and address arithmetic can be
2857: performed between any addresses. However, in the dictionary headers or
2858: code are interleaved with data, so almost the only contiguous data space
2859: regions there are those described by ANS Forth as contiguous; but you
2860: can be sure that the dictionary is allocated towards increasing
2861: addresses even between contiguous regions. The memory order of
2862: allocations in the heap is platform-dependent (and possibly different
2863: from one run to the next).
2864:
2865: @subsubsection ANS Forth dictionary details
2866:
2867: This section is just informative, you can skip it if you are in a hurry.
2868:
2869: When you create a colon definition, the text interpreter compiles the
2870: code for the definition into the code space and compiles the name
2871: of the definition into the header space, together with other
2872: information about the definition (such as its execution token).
2873:
2874: When you create a variable, the execution of @code{Variable} will
2875: compile some code, assign one cell in data space, and compile the name
2876: of the variable into the header space.
2877:
2878: @cindex memory regions - relationship between them
2879: ANS Forth does not specify the relationship between the three memory
2880: regions, and specifies that a Standard program must not access code or
2881: data space directly -- it may only access data space directly. In
2882: addition, the Standard defines what relationships you may and may not
2883: rely on when allocating regions in data space. These constraints are
2884: simply a reflection of the many diverse techniques that are used to
2885: implement Forth systems; understanding and following the requirements of
2886: the Standard allows you to write portable programs -- programs that run
2887: in the same way on any of these diverse systems. Another way of looking
2888: at this is to say that ANS Forth was designed to permit compliant Forth
2889: systems to be implemented in many diverse ways.
2890:
2891: @cindex memory regions - how they are assigned
2892: Here are some examples of ways in which name, code and data spaces
2893: might be assigned in different Forth implementations:
2894:
2895: @itemize @bullet
2896: @item
2897: For a Forth system that runs from RAM under a general-purpose operating
2898: system, it can be convenient to interleave name, code and data spaces in
2899: a single contiguous memory region. This organisation can be
2900: memory-efficient (for example, because the relationship between the name
2901: dictionary entry and the associated code space entry can be
2902: implicit, rather than requiring an explicit memory pointer to reference
2903: from the header space and the code space). This is the
2904: organisation used by Gforth, as this example@footnote{The addresses
2905: in the example have been truncated to fit it onto the page, and the
2906: addresses and data shown will not match the output from your system} shows:
2907: @example
2908: hex
2909: variable fred 123456 fred !
2910: variable jim abcd jim !
2911: : foo + / - ;
2912: ' fred 10 - 50 dump
2913: ..80: 5C 46 0E 40 84 66 72 65 - 64 20 20 20 20 20 20 20 \F.@.fred
2914: ..90: D0 9B 04 08 00 00 00 00 - 56 34 12 00 80 46 0E 40 ........V4...F.@
2915: ..A0: 83 6A 69 6D 20 20 20 20 - D0 9B 04 08 00 00 00 00 .jim ........
2916: ..B0: CD AB 00 00 9C 46 0E 40 - 83 66 6F 6F 20 20 20 20 .....F.@.foo
2917: ..C0: 80 9B 04 08 00 00 00 00 - E4 2E 05 08 0C 2F 05 08 ............./..
2918: @end example
2919:
2920: @item
2921: For a high-performance system running on a modern RISC processor with a
2922: modified Harvard architecture (one that has a unified main memory but
2923: separate instruction and data caches), it is desirable to separate
2924: processor instructions from processor data. This encourages a high cache
2925: density and therefore a high cache hit rate. The Forth code space
2926: is not necessarily made up entirely of processor instructions; its
2927: nature is dependent upon the Forth implementation.
2928:
2929: @item
2930: A Forth compiler that runs on a segmented 8086 processor could be
2931: designed to interleave the name, code and data spaces within a single
2932: 64Kbyte segment. A more common implementation choice is to use a
2933: separate 64Kbyte segment for each region, which provides more memory
2934: overall but provides an address map in which only the data space is
2935: accessible.
2936:
2937: @item
2938: Microprocessors exist that run Forth (or many of the primitives required
2939: to implement the Forth virtual machine efficiently) directly. On these
2940: processors, the relationship between name, code and data spaces may be
2941: imposed as a side-effect of the architecture of the processor.
2942:
2943: @item
2944: A Forth compiler that executes from ROM on an embedded system needs its
2945: data space separated from the name and code spaces so that the data
2946: space can be mapped to a RAM area.
2947:
2948: @item
2949: A Forth compiler that runs on an embedded system may have a requirement
2950: for a small memory footprint. On such a system it can be useful to
2951: separate the header space from the data and code spaces; once the
2952: application has been compiled, the header space is no longer
2953: required@footnote{more strictly speaking, most applications can be
2954: designed so that this is the case}. The header space can be deleted
2955: entirely, or could be stored in memory on a remote @i{host} system for
2956: debug and development purposes. In the latter case, the compiler running
2957: on the @i{target} system could implement a protocol across a
2958: communication link that would allow it to interrogate the header space.
2959: @end itemize
2960:
2961: @node Dictionary allocation, Heap Allocation, Memory model, Memory
2962: @subsection Dictionary allocation
2963: @cindex reserving data space
2964: @cindex data space - reserving some
2965:
2966: Dictionary allocation is a stack-oriented allocation scheme, i.e., if
2967: you want to deallocate X, you also deallocate everything
2968: allocated after X.
2969:
2970: The allocations using the words below are contiguous and grow the region
2971: towards increasing addresses. Other words that allocate dictionary
2972: memory of any kind (i.e., defining words including @code{:noname}) end
2973: the contiguous region and start a new one.
2974:
2975: In ANS Forth only @code{create}d words are guaranteed to produce an
2976: address that is the start of the following contiguous region. In
2977: particular, the cell allocated by @code{variable} is not guaranteed to
2978: be contiguous with following @code{allot}ed memory.
2979:
2980: You can deallocate memory by using @code{allot} with a negative argument
2981: (with some restrictions, see @code{allot}). For larger deallocations use
2982: @code{marker}.
2983:
2984:
2985: doc-here
2986: doc-unused
2987: doc-allot
2988: doc-c,
2989: doc-f,
2990: doc-,
2991: doc-2,
2992: @cindex user space
2993: doc-udp
2994: doc-uallot
2995:
2996: Memory accesses have to be aligned (@pxref{Address arithmetic}). So of
2997: course you should allocate memory in an aligned way, too. I.e., before
2998: allocating allocating a cell, @code{here} must be cell-aligned, etc.
2999: The words below align @code{here} if it is not already. Basically it is
3000: only already aligned for a type, if the last allocation was a multiple
3001: of the size of this type and if @code{here} was aligned for this type
3002: before.
3003:
3004: After freshly @code{create}ing a word, @code{here} is @code{align}ed in
3005: ANS Forth (@code{maxalign}ed in Gforth).
3006:
3007: doc-align
3008: doc-falign
3009: doc-sfalign
3010: doc-dfalign
3011: doc-maxalign
3012: doc-cfalign
3013:
3014:
3015: @node Heap Allocation, Memory Access, Dictionary allocation, Memory
3016: @subsection Heap allocation
3017: @cindex heap allocation
3018: @cindex dynamic allocation of memory
3019: @cindex memory-allocation word set
3020:
3021: Heap allocation supports deallocation of allocated memory in any
3022: order. Dictionary allocation is not affected by it (i.e., it does not
3023: end a contiguous region). In Gforth, these words are implemented using
3024: the standard C library calls malloc(), free() and resize().
3025:
3026: doc-allocate
3027: doc-free
3028: doc-resize
3029:
3030:
3031: @node Memory Access, Address arithmetic, Heap Allocation, Memory
3032: @subsection Memory Access
3033: @cindex memory access words
3034:
3035:
3036: doc-@
3037: doc-!
3038: doc-+!
3039: doc-c@
3040: doc-c!
3041: doc-2@
3042: doc-2!
3043: doc-f@
3044: doc-f!
3045: doc-sf@
3046: doc-sf!
3047: doc-df@
3048: doc-df!
3049:
3050: @node Address arithmetic, Memory Blocks, Memory Access, Memory
3051: @subsection Address arithmetic
3052: @cindex address arithmetic words
3053:
3054: Address arithmetic is the foundation on which data structures like
3055: arrays, records (@pxref{Structures}) and objects (@pxref{Object-oriented
3056: Forth}) are built.
3057:
3058: ANS Forth does not specify the sizes of the data types. Instead, it
3059: offers a number of words for computing sizes and doing address
3060: arithmetic. Address arithmetic is performed in terms of address units
3061: (aus); on most systems the address unit is one byte. Note that a
3062: character may have more than one au, so @code{chars} is no noop (on
3063: systems where it is a noop, it compiles to nothing).
3064:
3065: @cindex alignment of addresses for types
3066: ANS Forth also defines words for aligning addresses for specific
3067: types. Many computers require that accesses to specific data types
3068: must only occur at specific addresses; e.g., that cells may only be
3069: accessed at addresses divisible by 4. Even if a machine allows unaligned
3070: accesses, it can usually perform aligned accesses faster.
3071:
3072: For the performance-conscious: alignment operations are usually only
3073: necessary during the definition of a data structure, not during the
3074: (more frequent) accesses to it.
3075:
3076: ANS Forth defines no words for character-aligning addresses. This is not
3077: an oversight, but reflects the fact that addresses that are not
3078: char-aligned have no use in the standard and therefore will not be
3079: created.
3080:
3081: @cindex @code{CREATE} and alignment
3082: ANS Forth guarantees that addresses returned by @code{CREATE}d words
3083: are cell-aligned; in addition, Gforth guarantees that these addresses
3084: are aligned for all purposes.
3085:
3086: Note that the ANS Forth word @code{char} has nothing to do with address
3087: arithmetic.
3088:
3089:
3090: doc-chars
3091: doc-char+
3092: doc-cells
3093: doc-cell+
3094: doc-cell
3095: doc-aligned
3096: doc-floats
3097: doc-float+
3098: doc-float
3099: doc-faligned
3100: doc-sfloats
3101: doc-sfloat+
3102: doc-sfaligned
3103: doc-dfloats
3104: doc-dfloat+
3105: doc-dfaligned
3106: doc-maxaligned
3107: doc-cfaligned
3108: doc-address-unit-bits
3109:
3110:
3111: @node Memory Blocks, , Address arithmetic, Memory
3112: @subsection Memory Blocks
3113: @cindex memory block words
3114: @cindex character strings - moving and copying
3115:
3116: Memory blocks often represent character strings; @xref{String Formats}
3117: for ways of storing character strings in memory. @xref{Displaying
3118: characters and strings} for other string-processing words.
3119:
3120: Some of these words work on address units. Others work on character
3121: units (increments of @code{CHAR}), and expect a @code{CHAR}-aligned
3122: address. Choose the correct operation depending upon your data type.
3123:
3124: When copying characters between overlapping memory regions, choose
3125: carefully between @code{cmove} and @code{cmove>}.
3126:
3127: You can only use any of these words @i{portably} to access data space.
3128:
3129: @comment TODO - think the naming of the arguments is wrong for move
3130: @comment well, really it seems to be the Standard that's wrong; it
3131: @comment describes MOVE as a word that requires a CELL-aligned source
3132: @comment and destination address but a xtranfer count that need not
3133: @comment be a multiple of CELL.
3134:
3135: doc-move
3136: doc-erase
3137: doc-cmove
3138: doc-cmove>
3139: doc-fill
3140: doc-blank
3141: doc-compare
3142: doc-search
3143: doc--trailing
3144: doc-/string
3145:
3146:
3147: @comment TODO examples
3148:
3149:
3150: @node Control Structures, Defining Words, Memory, Words
3151: @section Control Structures
3152: @cindex control structures
3153:
3154: Control structures in Forth cannot be used interpretively, only in a
3155: colon definition@footnote{To be precise, they have no interpretation
3156: semantics (@pxref{Interpretation and Compilation Semantics}).}. We do
3157: not like this limitation, but have not seen a satisfying way around it
3158: yet, although many schemes have been proposed.
3159:
3160: @menu
3161: * Selection:: IF ... ELSE ... ENDIF
3162: * Simple Loops:: BEGIN ...
3163: * Counted Loops:: DO
3164: * Arbitrary control structures::
3165: * Calls and returns::
3166: * Exception Handling::
3167: @end menu
3168:
3169: @node Selection, Simple Loops, Control Structures, Control Structures
3170: @subsection Selection
3171: @cindex selection control structures
3172: @cindex control structures for selection
3173:
3174: @c what's the purpose of all these @i? Maybe we should define a macro
3175: @c so we can produce logical markup. - anton
3176:
3177: @c nac-> When I started working on the manual, a mixture of @i and @var
3178: @c were used inconsistently in code examples and \Glossary entries. These
3179: @c two behave differently in info format so I decided to standardize on @i.
3180: @c Logical markup would be better but texi isn't really upto it, and
3181: @c texi2html just ignores macros.
3182:
3183: @cindex @code{IF} control structure
3184: @example
3185: @i{flag}
3186: IF
3187: @i{code}
3188: ENDIF
3189: @end example
3190: @noindent
3191:
3192: If @i{flag} is non-zero (as far as @code{IF} etc. are concerned, a cell
3193: with any bit set represents truth) @i{code} is executed.
3194:
3195: @example
3196: @i{flag}
3197: IF
3198: @i{code1}
3199: ELSE
3200: @i{code2}
3201: ENDIF
3202: @end example
3203:
3204: If @var{flag} is true, @i{code1} is executed, otherwise @i{code2} is
3205: executed.
3206:
3207: You can use @code{THEN} instead of @code{ENDIF}. Indeed, @code{THEN} is
3208: standard, and @code{ENDIF} is not, although it is quite popular. We
3209: recommend using @code{ENDIF}, because it is less confusing for people
3210: who also know other languages (and is not prone to reinforcing negative
3211: prejudices against Forth in these people). Adding @code{ENDIF} to a
3212: system that only supplies @code{THEN} is simple:
3213: @example
3214: : ENDIF POSTPONE THEN ; immediate
3215: @end example
3216:
3217: [According to @cite{Webster's New Encyclopedic Dictionary}, @dfn{then
3218: (adv.)} has the following meanings:
3219: @quotation
3220: ... 2b: following next after in order ... 3d: as a necessary consequence
3221: (if you were there, then you saw them).
3222: @end quotation
3223: Forth's @code{THEN} has the meaning 2b, whereas @code{THEN} in Pascal
3224: and many other programming languages has the meaning 3d.]
3225:
3226: Gforth also provides the words @code{?DUP-IF} and @code{?DUP-0=-IF}, so
3227: you can avoid using @code{?dup}. Using these alternatives is also more
3228: efficient than using @code{?dup}. Definitions in ANS Forth
3229: for @code{ENDIF}, @code{?DUP-IF} and @code{?DUP-0=-IF} are provided in
3230: @file{compat/control.fs}.
3231:
3232: @cindex @code{CASE} control structure
3233: @example
3234: @i{n}
3235: CASE
3236: @i{n1} OF @i{code1} ENDOF
3237: @i{n2} OF @i{code2} ENDOF
3238: @dots{}
3239: ENDCASE
3240: @end example
3241:
3242: Executes the first @i{codei}, where the @i{ni} is equal to
3243: @i{n}. A default case can be added by simply writing the code after
3244: the last @code{ENDOF}. It may use @i{n}, which is on top of the stack,
3245: but must not consume it.
3246:
3247: @node Simple Loops, Counted Loops, Selection, Control Structures
3248: @subsection Simple Loops
3249: @cindex simple loops
3250: @cindex loops without count
3251:
3252: @cindex @code{WHILE} loop
3253: @example
3254: BEGIN
3255: @i{code1}
3256: @i{flag}
3257: WHILE
3258: @i{code2}
3259: REPEAT
3260: @end example
3261:
3262: @i{code1} is executed and @i{flag} is computed. If it is true,
3263: @i{code2} is executed and the loop is restarted; If @i{flag} is
3264: false, execution continues after the @code{REPEAT}.
3265:
3266: @cindex @code{UNTIL} loop
3267: @example
3268: BEGIN
3269: @i{code}
3270: @i{flag}
3271: UNTIL
3272: @end example
3273:
3274: @i{code} is executed. The loop is restarted if @code{flag} is false.
3275:
3276: @cindex endless loop
3277: @cindex loops, endless
3278: @example
3279: BEGIN
3280: @i{code}
3281: AGAIN
3282: @end example
3283:
3284: This is an endless loop.
3285:
3286: @node Counted Loops, Arbitrary control structures, Simple Loops, Control Structures
3287: @subsection Counted Loops
3288: @cindex counted loops
3289: @cindex loops, counted
3290: @cindex @code{DO} loops
3291:
3292: The basic counted loop is:
3293: @example
3294: @i{limit} @i{start}
3295: ?DO
3296: @i{body}
3297: LOOP
3298: @end example
3299:
3300: This performs one iteration for every integer, starting from @i{start}
3301: and up to, but excluding @i{limit}. The counter, or @i{index}, can be
3302: accessed with @code{i}. For example, the loop:
3303: @example
3304: 10 0 ?DO
3305: i .
3306: LOOP
3307: @end example
3308: @noindent
3309: prints @code{0 1 2 3 4 5 6 7 8 9}
3310:
3311: The index of the innermost loop can be accessed with @code{i}, the index
3312: of the next loop with @code{j}, and the index of the third loop with
3313: @code{k}.
3314:
3315:
3316: doc-i
3317: doc-j
3318: doc-k
3319:
3320:
3321: The loop control data are kept on the return stack, so there are some
3322: restrictions on mixing return stack accesses and counted loop words. In
3323: particuler, if you put values on the return stack outside the loop, you
3324: cannot read them inside the loop@footnote{well, not in a way that is
3325: portable.}. If you put values on the return stack within a loop, you
3326: have to remove them before the end of the loop and before accessing the
3327: index of the loop.
3328:
3329: There are several variations on the counted loop:
3330:
3331: @itemize @bullet
3332: @item
3333: @code{LEAVE} leaves the innermost counted loop immediately; execution
3334: continues after the associated @code{LOOP} or @code{NEXT}. For example:
3335:
3336: @example
3337: 10 0 ?DO i DUP . 3 = IF LEAVE THEN LOOP
3338: @end example
3339: prints @code{0 1 2 3}
3340:
3341:
3342: @item
3343: @code{UNLOOP} prepares for an abnormal loop exit, e.g., via
3344: @code{EXIT}. @code{UNLOOP} removes the loop control parameters from the
3345: return stack so @code{EXIT} can get to its return address. For example:
3346:
3347: @example
3348: : demo 10 0 ?DO i DUP . 3 = IF UNLOOP EXIT THEN LOOP ." Done" ;
3349: @end example
3350: prints @code{0 1 2 3}
3351:
3352:
3353: @item
3354: If @i{start} is greater than @i{limit}, a @code{?DO} loop is entered
3355: (and @code{LOOP} iterates until they become equal by wrap-around
3356: arithmetic). This behaviour is usually not what you want. Therefore,
3357: Gforth offers @code{+DO} and @code{U+DO} (as replacements for
3358: @code{?DO}), which do not enter the loop if @i{start} is greater than
3359: @i{limit}; @code{+DO} is for signed loop parameters, @code{U+DO} for
3360: unsigned loop parameters.
3361:
3362: @item
3363: @code{?DO} can be replaced by @code{DO}. @code{DO} always enters
3364: the loop, independent of the loop parameters. Do not use @code{DO}, even
3365: if you know that the loop is entered in any case. Such knowledge tends
3366: to become invalid during maintenance of a program, and then the
3367: @code{DO} will make trouble.
3368:
3369: @item
3370: @code{LOOP} can be replaced with @code{@i{n} +LOOP}; this updates the
3371: index by @i{n} instead of by 1. The loop is terminated when the border
3372: between @i{limit-1} and @i{limit} is crossed. E.g.:
3373:
3374: @example
3375: 4 0 +DO i . 2 +LOOP
3376: @end example
3377: @noindent
3378: prints @code{0 2}
3379:
3380: @example
3381: 4 1 +DO i . 2 +LOOP
3382: @end example
3383: @noindent
3384: prints @code{1 3}
3385:
3386:
3387: @cindex negative increment for counted loops
3388: @cindex counted loops with negative increment
3389: The behaviour of @code{@i{n} +LOOP} is peculiar when @i{n} is negative:
3390:
3391: @example
3392: -1 0 ?DO i . -1 +LOOP
3393: @end example
3394: @noindent
3395: prints @code{0 -1}
3396:
3397: @example
3398: 0 0 ?DO i . -1 +LOOP
3399: @end example
3400: prints nothing.
3401:
3402: Therefore we recommend avoiding @code{@i{n} +LOOP} with negative
3403: @i{n}. One alternative is @code{@i{u} -LOOP}, which reduces the
3404: index by @i{u} each iteration. The loop is terminated when the border
3405: between @i{limit+1} and @i{limit} is crossed. Gforth also provides
3406: @code{-DO} and @code{U-DO} for down-counting loops. E.g.:
3407:
3408: @example
3409: -2 0 -DO i . 1 -LOOP
3410: @end example
3411: @noindent
3412: prints @code{0 -1}
3413:
3414: @example
3415: -1 0 -DO i . 1 -LOOP
3416: @end example
3417: @noindent
3418: prints @code{0}
3419:
3420: @example
3421: 0 0 -DO i . 1 -LOOP
3422: @end example
3423: @noindent
3424: prints nothing.
3425:
3426: @end itemize
3427:
3428: Unfortunately, @code{+DO}, @code{U+DO}, @code{-DO}, @code{U-DO} and
3429: @code{-LOOP} are not defined in ANS Forth. However, an implementation
3430: for these words that uses only standard words is provided in
3431: @file{compat/loops.fs}.
3432:
3433:
3434: @cindex @code{FOR} loops
3435: Another counted loop is:
3436: @example
3437: @i{n}
3438: FOR
3439: @i{body}
3440: NEXT
3441: @end example
3442: This is the preferred loop of native code compiler writers who are too
3443: lazy to optimize @code{?DO} loops properly. This loop structure is not
3444: defined in ANS Forth. In Gforth, this loop iterates @i{n+1} times;
3445: @code{i} produces values starting with @i{n} and ending with 0. Other
3446: Forth systems may behave differently, even if they support @code{FOR}
3447: loops. To avoid problems, don't use @code{FOR} loops.
3448:
3449: @node Arbitrary control structures, Calls and returns, Counted Loops, Control Structures
3450: @subsection Arbitrary control structures
3451: @cindex control structures, user-defined
3452:
3453: @cindex control-flow stack
3454: ANS Forth permits and supports using control structures in a non-nested
3455: way. Information about incomplete control structures is stored on the
3456: control-flow stack. This stack may be implemented on the Forth data
3457: stack, and this is what we have done in Gforth.
3458:
3459: @cindex @code{orig}, control-flow stack item
3460: @cindex @code{dest}, control-flow stack item
3461: An @i{orig} entry represents an unresolved forward branch, a @i{dest}
3462: entry represents a backward branch target. A few words are the basis for
3463: building any control structure possible (except control structures that
3464: need storage, like calls, coroutines, and backtracking).
3465:
3466:
3467: doc-if
3468: doc-ahead
3469: doc-then
3470: doc-begin
3471: doc-until
3472: doc-again
3473: doc-cs-pick
3474: doc-cs-roll
3475:
3476:
3477: The Standard words @code{CS-PICK} and @code{CS-ROLL} allow you to
3478: manipulate the control-flow stack in a portable way. Without them, you
3479: would need to know how many stack items are occupied by a control-flow
3480: entry (many systems use one cell. In Gforth they currently take three,
3481: but this may change in the future).
3482:
3483: Some standard control structure words are built from these words:
3484:
3485:
3486: doc-else
3487: doc-while
3488: doc-repeat
3489:
3490:
3491: @noindent
3492: Gforth adds some more control-structure words:
3493:
3494:
3495: doc-endif
3496: doc-?dup-if
3497: doc-?dup-0=-if
3498:
3499:
3500: @noindent
3501: Counted loop words constitute a separate group of words:
3502:
3503:
3504: doc-?do
3505: doc-+do
3506: doc-u+do
3507: doc--do
3508: doc-u-do
3509: doc-do
3510: doc-for
3511: doc-loop
3512: doc-+loop
3513: doc--loop
3514: doc-next
3515: doc-leave
3516: doc-?leave
3517: doc-unloop
3518: doc-done
3519:
3520:
3521: The standard does not allow using @code{CS-PICK} and @code{CS-ROLL} on
3522: @i{do-sys}. Gforth allows it, but it's your job to ensure that for
3523: every @code{?DO} etc. there is exactly one @code{UNLOOP} on any path
3524: through the definition (@code{LOOP} etc. compile an @code{UNLOOP} on the
3525: fall-through path). Also, you have to ensure that all @code{LEAVE}s are
3526: resolved (by using one of the loop-ending words or @code{DONE}).
3527:
3528: @noindent
3529: Another group of control structure words are:
3530:
3531:
3532: doc-case
3533: doc-endcase
3534: doc-of
3535: doc-endof
3536:
3537:
3538: @i{case-sys} and @i{of-sys} cannot be processed using @code{CS-PICK} and
3539: @code{CS-ROLL}.
3540:
3541: @subsubsection Programming Style
3542:
3543: In order to ensure readability we recommend that you do not create
3544: arbitrary control structures directly, but define new control structure
3545: words for the control structure you want and use these words in your
3546: program. For example, instead of writing:
3547:
3548: @example
3549: BEGIN
3550: ...
3551: IF [ 1 CS-ROLL ]
3552: ...
3553: AGAIN THEN
3554: @end example
3555:
3556: @noindent
3557: we recommend defining control structure words, e.g.,
3558:
3559: @example
3560: : WHILE ( DEST -- ORIG DEST )
3561: POSTPONE IF
3562: 1 CS-ROLL ; immediate
3563:
3564: : REPEAT ( orig dest -- )
3565: POSTPONE AGAIN
3566: POSTPONE THEN ; immediate
3567: @end example
3568:
3569: @noindent
3570: and then using these to create the control structure:
3571:
3572: @example
3573: BEGIN
3574: ...
3575: WHILE
3576: ...
3577: REPEAT
3578: @end example
3579:
3580: That's much easier to read, isn't it? Of course, @code{REPEAT} and
3581: @code{WHILE} are predefined, so in this example it would not be
3582: necessary to define them.
3583:
3584: @node Calls and returns, Exception Handling, Arbitrary control structures, Control Structures
3585: @subsection Calls and returns
3586: @cindex calling a definition
3587: @cindex returning from a definition
3588:
3589: @cindex recursive definitions
3590: A definition can be called simply be writing the name of the definition
3591: to be called. Normally a definition is invisible during its own
3592: definition. If you want to write a directly recursive definition, you
3593: can use @code{recursive} to make the current definition visible, or
3594: @code{recurse} to call the current definition directly.
3595:
3596:
3597: doc-recursive
3598: doc-recurse
3599:
3600:
3601: @comment TODO add example of the two recursion methods
3602: @quotation
3603: @progstyle
3604: I prefer using @code{recursive} to @code{recurse}, because calling the
3605: definition by name is more descriptive (if the name is well-chosen) than
3606: the somewhat cryptic @code{recurse}. E.g., in a quicksort
3607: implementation, it is much better to read (and think) ``now sort the
3608: partitions'' than to read ``now do a recursive call''.
3609: @end quotation
3610:
3611: For mutual recursion, use @code{Defer}red words, like this:
3612:
3613: @example
3614: Defer foo
3615:
3616: : bar ( ... -- ... )
3617: ... foo ... ;
3618:
3619: :noname ( ... -- ... )
3620: ... bar ... ;
3621: IS foo
3622: @end example
3623:
3624: Deferred words are discussed in more detail in @ref{Deferred words}.
3625:
3626: The current definition returns control to the calling definition when
3627: the end of the definition is reached or @code{EXIT} is encountered.
3628:
3629: doc-exit
3630: doc-;s
3631:
3632:
3633: @node Exception Handling, , Calls and returns, Control Structures
3634: @subsection Exception Handling
3635: @cindex exceptions
3636:
3637: If your program detects a fatal error condition, the simplest action
3638: that it can take is to @code{quit}. This resets the return stack and
3639: restarts the text interpreter, but does not print any error message.
3640:
3641: The next stage in severity is to execute @code{abort}, which has the
3642: same effect as @code{quit}, with the addition that it resets the data
3643: stack.
3644:
3645: A slightly more sophisticated approach is use use @code{abort"}, which
3646: compiles a string to be used as an error message and does a conditional
3647: @code{abort} at run-time. For example:
3648:
3649: @example
3650: @kbd{: checker abort" That flag was true" ." A false flag" ;@key{RET}} ok
3651: @kbd{0 checker@key{RET}} A false flag ok
3652: @kbd{1 checker@key{RET}}
3653: :1: That flag was true
3654: 1 checker
3655: ^^^^^^^
3656: $400D1648 throw
3657: $400E4660
3658: @end example
3659:
3660: These simple techniques allow a program to react to a fatal error
3661: condition, but they are not exactly user-friendly. The ANS Forth
3662: Exception word set provides the pair of words @code{throw} and
3663: @code{catch}, which can be used to provide sophisticated error-handling.
3664:
3665: @code{catch} has a similar behaviour to @code{execute}, in that it takes
3666: an @i{xt} as a parameter and starts execution of the xt. However,
3667: before passing control to the xt, @code{catch} pushes an
3668: @dfn{exception frame} onto the @dfn{exception stack}. This exception
3669: frame is used to restore the system to a known state if a detected error
3670: occurs during the execution of the xt. A typical way to use @code{catch}
3671: would be:
3672:
3673: @example
3674: ... ['] foo catch IF ...
3675: @end example
3676:
3677: @c TOS is undefined. - anton
3678:
3679: @c nac-> TODO -- I need to look at this example again.
3680:
3681: Whilst @code{foo} executes, it can call other words to any level of
3682: nesting, as usual. If @code{foo} (and all the words that it calls)
3683: execute successfully, control will ultimately pass to the word following
3684: the @code{catch}, and there will be a 0 at TOS. However, if any word
3685: detects an error, it can terminate the execution of @code{foo} by
3686: pushing a non-zero error code onto the stack and then performing a
3687: @code{throw}. The execution of @code{throw} will pass control to the
3688: word following the @code{catch}, but this time the TOS will hold the
3689: error code. Therefore, the @code{IF} in the example can be used to
3690: determine whether @code{foo} executed successfully.
3691:
3692: This simple example shows how you can use @code{throw} and @code{catch}
3693: to ``take over'' exception handling from the system:
3694: @example
3695: : my-div ['] / catch if ." DIVIDE ERROR" else ." OK.. " . then ;
3696: @end example
3697:
3698: The next example is more sophisticated and shows a multi-level
3699: @code{throw} and @code{catch}. To understand this example, start at the
3700: definition of @code{top-level} and work backwards:
3701:
3702: @example
3703: : lowest-level ( -- c )
3704: key dup 27 = if
3705: 1 throw \ ESCAPE key pressed
3706: else
3707: ." lowest-level successful" CR
3708: then
3709: ;
3710:
3711: : lower-level ( -- c )
3712: lowest-level
3713: \ at this level consider a CTRL-U to be a fatal error
3714: dup 21 = if \ CTRL-U
3715: 2 throw
3716: else
3717: ." lower-level successful" CR
3718: then
3719: ;
3720:
3721: : low-level ( -- c )
3722: ['] lower-level catch
3723: ?dup if
3724: \ error occurred - do we recognise it?
3725: dup 1 = if
3726: \ ESCAPE key pressed.. pretend it was an E
3727: [char] E
3728: else throw \ propogate the error upwards
3729: then
3730: then
3731: ." low-level successfull" CR
3732: ;
3733:
3734: : top-level ( -- )
3735: CR ['] low-level catch \ CATCH is used like EXECUTE
3736: ?dup if \ error occurred..
3737: ." Error " . ." occurred - contact your supplier"
3738: else
3739: ." The '" emit ." ' key was pressed" CR
3740: then
3741: ;
3742: @end example
3743:
3744: The ANS Forth document assigns @code{throw} codes thus:
3745:
3746: @itemize @bullet
3747: @item
3748: codes in the range -1 -- -255 are reserved to be assigned by the
3749: Standard. Assignments for codes in the range -1 -- -58 are currently
3750: documented in the Standard. In particular, @code{-1 throw} is equivalent
3751: to @code{abort} and @code{-2 throw} is equivalent to @code{abort"}.
3752: @item
3753: codes in the range -256 -- -4095 are reserved to be assigned by the system.
3754: @item
3755: all other codes may be assigned by programs.
3756: @end itemize
3757:
3758: Gforth provides the word @code{exception} as a mechanism for assigning
3759: system throw codes to applications. This allows multiple applications to
3760: co-exist in memory without any clash of @code{throw} codes. A definition
3761: of @code{exception} in ANS Forth is provided in
3762: @file{compat/exception.fs}.
3763:
3764:
3765: doc-quit
3766: doc-abort
3767: doc-abort"
3768:
3769: doc-catch
3770: doc-throw
3771: doc---exception-exception
3772:
3773:
3774:
3775: @c -------------------------------------------------------------
3776: @node Defining Words, The Text Interpreter, Control Structures, Words
3777: @section Defining Words
3778: @cindex defining words
3779:
3780: @menu
3781: * CREATE::
3782: * Variables:: Variables and user variables
3783: * Constants::
3784: * Values:: Initialised variables
3785: * Colon Definitions::
3786: * Anonymous Definitions:: Definitions without names
3787: * User-defined Defining Words::
3788: * Deferred words:: Allow forward references
3789: * Aliases::
3790: * Supplying names::
3791: * Interpretation and Compilation Semantics::
3792: * Combined words::
3793: @end menu
3794:
3795: @node CREATE, Variables, Defining Words, Defining Words
3796: @subsection @code{CREATE}
3797: @cindex simple defining words
3798: @cindex defining words, simple
3799:
3800: Defining words are used to create new entries in the dictionary. The
3801: simplest defining word is @code{CREATE}. @code{CREATE} is used like
3802: this:
3803:
3804: @example
3805: CREATE new-word1
3806: @end example
3807:
3808: @code{CREATE} is a parsing word that generates a dictionary entry for
3809: @code{new-word1}. When @code{new-word1} is executed, all that it does is
3810: leave an address on the stack. The address represents the value of
3811: the data space pointer (@code{HERE}) at the time that @code{new-word1}
3812: was defined. Therefore, @code{CREATE} is a way of associating a name
3813: with the address of a region of memory.
3814:
3815: doc-create
3816:
3817: By extending this example to reserve some memory in data space, we end
3818: up with a @i{variable}. Here are two different ways to do it:
3819:
3820: @example
3821: CREATE new-word2 1 cells allot \ reserve 1 cell - initial value undefined
3822: CREATE new-word3 4 , \ reserve 1 cell and initialise it (to 4)
3823: @end example
3824:
3825: The variable can be examined and modified using @code{@@} (``fetch'') and
3826: @code{!} (``store'') like this:
3827:
3828: @example
3829: new-word2 @@ . \ get address, fetch from it and display
3830: 1234 new-word2 ! \ new value, get address, store to it
3831: @end example
3832:
3833: @cindex arrays
3834: A similar mechanism can be used to create arrays. For example, an
3835: 80-character text input buffer:
3836:
3837: @example
3838: CREATE text-buf 80 chars allot
3839:
3840: text-buf 0 chars c@@ \ the 1st character (offset 0)
3841: text-buf 3 chars c@@ \ the 4th character (offset 3)
3842: @end example
3843:
3844: You can build arbitrarily complex data structures by allocating
3845: appropriate areas of memory. @xref{Structures} for further discussions
3846: of this, and to learn about some Gforth tools that make it easier.
3847:
3848:
3849: @node Variables, Constants, CREATE, Defining Words
3850: @subsection Variables
3851: @cindex variables
3852:
3853: The previous section showed how a sequence of commands could be used to
3854: generate a variable. As a final refinement, the whole code sequence can
3855: be wrapped up in a defining word (pre-empting the subject of the next
3856: section), making it easier to create new variables:
3857:
3858: @example
3859: : myvariableX ( "name" -- a-addr ) CREATE 1 cells allot ;
3860: : myvariable0 ( "name" -- a-addr ) CREATE 0 , ;
3861:
3862: myvariableX foo \ variable foo starts off with an unknown value
3863: myvariable0 joe \ whilst joe is initialised to 0
3864:
3865: 45 3 * foo ! \ set foo to 135
3866: 1234 joe ! \ set joe to 1234
3867: 3 joe +! \ increment joe by 3.. to 1237
3868: @end example
3869:
3870: Not surprisingly, there is no need to define @code{myvariable}, since
3871: Forth already has a definition @code{Variable}. ANS Forth does not
3872: require a @code{Variable} to be initialised when it is created (i.e., it
3873: behaves like @code{myvariableX}). In contrast, Gforth's @code{Variable}
3874: initialises the variable to 0 (i.e., it behaves exactly like
3875: @code{myvariable0}). Forth also provides @code{2Variable} and
3876: @code{fvariable} for double and floating-point variables,
3877: respectively -- both are initialised to 0 in Gforth.
3878:
3879: doc-variable
3880: doc-2variable
3881: doc-fvariable
3882:
3883:
3884: @cindex user variables
3885: @cindex user space
3886: The defining word @code{User} behaves in the same way as @code{Variable}.
3887: The difference is that it reserves space in @i{user (data) space} rather
3888: than normal data space. In a Forth system that has a multi-tasker, each
3889: task has its own set of user variables.
3890:
3891: doc-user
3892:
3893: @comment TODO is that stuff about user variables strictly correct? Is it
3894: @comment just terminal tasks that have user variables?
3895: @comment should document tasker.fs (with some examples) elsewhere
3896: @comment in this manual, then expand on user space and user variables.
3897:
3898:
3899: @node Constants, Values, Variables, Defining Words
3900: @subsection Constants
3901: @cindex constants
3902:
3903: @code{Constant} allows you to declare a fixed value and refer to it by
3904: name. For example:
3905:
3906: @example
3907: 12 Constant INCHES-PER-FOOT
3908: 3E+08 fconstant SPEED-O-LIGHT
3909: @end example
3910:
3911: A @code{Variable} can be both read and written, so its run-time
3912: behaviour is to supply an address through which its current value can be
3913: manipulated. In contrast, the value of a @code{Constant} cannot be
3914: changed once it has been declared@footnote{Well, often it can be -- but
3915: not in a Standard, portable way. It's safer to use a @code{Value} (read
3916: on).} so it's not necessary to supply the address -- it is more
3917: efficient to return the value of the constant directly. That's exactly
3918: what happens; the run-time effect of a constant is to put its value on
3919: the top of the stack (@ref{User-defined Defining Words} describes one
3920: way of implementing @code{Constant}).
3921:
3922: Gforth also provides @code{2Constant} and @code{fconstant} for defining
3923: double and floating-point constants, respectively.
3924:
3925: doc-constant
3926: doc-2constant
3927: doc-fconstant
3928:
3929: @c that's too deep, and it's not necessarily true for all ANS Forths. - anton
3930: @c nac-> How could that not be true in an ANS Forth? You can't define a
3931: @c constant, use it and then delete the definition of the constant..
3932: @c I agree that it's rather deep, but IMO it is an important difference
3933: @c relative to other programming languages.. often it's annoying: it
3934: @c certainly changes my programming style relative to C.
3935:
3936: Constants in Forth behave differently from their equivalents in other
3937: programming languages. In other languages, a constant (such as an EQU in
3938: assembler or a #define in C) only exists at compile-time; in the
3939: executable program the constant has been translated into an absolute
3940: number and, unless you are using a symbolic debugger, it's impossible to
3941: know what abstract thing that number represents. In Forth a constant has
3942: an entry in the header space and remains there after the code that uses
3943: it has been defined. In fact, it must remain in the dictionary since it
3944: has run-time duties to perform. For example:
3945:
3946: @example
3947: 12 Constant INCHES-PER-FOOT
3948: : FEET-TO-INCHES ( n1 -- n2 ) INCHES-PER-FOOT * ;
3949: @end example
3950:
3951: @cindex in-lining of constants
3952: When @code{FEET-TO-INCHES} is executed, it will in turn execute the xt
3953: associated with the constant @code{INCHES-PER-FOOT}. If you use
3954: @code{see} to decompile the definition of @code{FEET-TO-INCHES}, you can
3955: see that it makes a call to @code{INCHES-PER-FOOT}. Some Forth compilers
3956: attempt to optimise constants by in-lining them where they are used. You
3957: can force Gforth to in-line a constant like this:
3958:
3959: @example
3960: : FEET-TO-INCHES ( n1 -- n2 ) [ INCHES-PER-FOOT ] LITERAL * ;
3961: @end example
3962:
3963: If you use @code{see} to decompile @i{this} version of
3964: @code{FEET-TO-INCHES}, you can see that @code{INCHES-PER-FOOT} is no
3965: longer present. @xref{Interpret/Compile states} and @ref{Literals} on
3966: how this works.
3967:
3968: In-lining constants in this way might improve execution time
3969: fractionally, and can ensure that a constant is now only referenced at
3970: compile-time. However, the definition of the constant still remains in
3971: the dictionary. Some Forth compilers provide a mechanism for controlling
3972: a second dictionary for holding transient words such that this second
3973: dictionary can be deleted later in order to recover memory
3974: space. However, there is no standard way of doing this.
3975:
3976:
3977: @node Values, Colon Definitions, Constants, Defining Words
3978: @subsection Values
3979: @cindex values
3980:
3981: A @code{Value} is like a @code{Variable} but with two important
3982: differences:
3983:
3984: @itemize @bullet
3985: @item
3986: A @code{Value} is initialised when it is declared; like a
3987: @code{Constant} but unlike a @code{Variable}.
3988: @item
3989: A @code{Value} returns its value rather than its address when it is
3990: executed; i.e., it has the same run-time behaviour as @code{Constant}.
3991: @end itemize
3992:
3993: A @code{Value} needs an additional word, @code{TO} to allow its value to
3994: be changed. Here are some examples:
3995:
3996: @example
3997: 12 Value APPLES \ Define APPLES with an initial value of 12
3998: 34 TO APPLES \ Change the value of APPLES. TO is a parsing word
3999: APPLES \ puts 34 on the top of the stack.
4000: @end example
4001:
4002: doc-value
4003: doc-to
4004:
4005:
4006: @node Colon Definitions, Anonymous Definitions, Values, Defining Words
4007: @subsection Colon Definitions
4008: @cindex colon definitions
4009:
4010: @example
4011: : name ( ... -- ... )
4012: word1 word2 word3 ;
4013: @end example
4014:
4015: @noindent
4016: Creates a word called @code{name} that, upon execution, executes
4017: @code{word1 word2 word3}. @code{name} is a @dfn{(colon) definition}.
4018:
4019: The explanation above is somewhat superficial. @xref{Your first
4020: definition} for simple examples of colon definitions, then
4021: @xref{Interpretation and Compilation Semantics} for an in-depth
4022: discussion of some of the issues involved.
4023:
4024: doc-:
4025: doc-;
4026:
4027:
4028: @node Anonymous Definitions, User-defined Defining Words, Colon Definitions, Defining Words
4029: @subsection Anonymous Definitions
4030: @cindex colon definitions
4031: @cindex defining words without name
4032:
4033: Sometimes you want to define an @dfn{anonymous word}; a word without a
4034: name. You can do this with:
4035:
4036: doc-:noname
4037:
4038: This leaves the execution token for the word on the stack after the
4039: closing @code{;}. Here's an example in which a deferred word is
4040: initialised with an @code{xt} from an anonymous colon definition:
4041:
4042: @example
4043: Defer deferred
4044: :noname ( ... -- ... )
4045: ... ;
4046: IS deferred
4047: @end example
4048:
4049: @noindent
4050: Gforth provides an alternative way of doing this, using two separate
4051: words:
4052:
4053: doc-noname
4054: @cindex execution token of last defined word
4055: doc-lastxt
4056:
4057: @noindent
4058: The previous example can be rewritten using @code{noname} and
4059: @code{lastxt}:
4060:
4061: @example
4062: Defer deferred
4063: noname : ( ... -- ... )
4064: ... ;
4065: lastxt IS deferred
4066: @end example
4067:
4068: @noindent
4069: @code{noname} works with any defining word, not just @code{:}.
4070:
4071: @code{lastxt} also works when the last word was not defined as
4072: @code{noname}. It also has the useful property that is is valid as soon
4073: as the header for a definition has been built. Thus:
4074:
4075: @example
4076: lastxt . : foo [ lastxt . ] ; ' foo .
4077: @end example
4078:
4079: @noindent
4080: prints 3 numbers; the last two are the same.
4081:
4082:
4083: @node User-defined Defining Words, Deferred words, Anonymous Definitions, Defining Words
4084: @subsection User-defined Defining Words
4085: @cindex user-defined defining words
4086: @cindex defining words, user-defined
4087:
4088: You can create a new defining word by wrapping defining-time code around
4089: an existing defining word and putting the sequence in a colon
4090: definition. For example, suppose that you have a word @code{stats} that
4091: gathers statistics about colon definitions given the @i{xt} of the
4092: definition, and you want every colon definition in your application to
4093: make a call to @code{stats}. You can define and use a new version of
4094: @code{:} like this:
4095:
4096: @example
4097: : stats ( xt -- ) DUP ." (Gathering statistics for " . ." )"
4098: ... ; \ other code
4099:
4100: : my: : lastxt postpone literal ['] stats compile, ;
4101:
4102: my: foo + - ;
4103: @end example
4104:
4105: When @code{foo} is defined using @code{my:} these steps occur:
4106:
4107: @itemize @bullet
4108: @item
4109: @code{my:} is executed.
4110: @item
4111: The @code{:} within the definition (the one between @code{my:} and
4112: @code{lastxt}) is executed, and does just what it always does; it parses
4113: the input stream for a name, builds a dictionary header for the name
4114: @code{foo} and switches @code{state} from interpret to compile.
4115: @item
4116: The word @code{lastxt} is executed. It puts the @i{xt} for the word that is
4117: being defined -- @code{foo} -- onto the stack.
4118: @item
4119: The code that was produced by @code{postpone literal} is executed; this
4120: causes the value on the stack to be compiled as a literal in the code
4121: area of @code{foo}.
4122: @item
4123: The code @code{['] stats} compiles a literal into the definition of
4124: @code{my:}. When @code{compile,} is executed, that literal -- the
4125: execution token for @code{stats} -- is layed down in the code area of
4126: @code{foo} , following the literal@footnote{Strictly speaking, the
4127: mechanism that @code{compile,} uses to convert an @i{xt} into something
4128: in the code area is implementation-dependent. A threaded implementation
4129: might spit out the execution token directly whilst another
4130: implementation might spit out a native code sequence.}.
4131: @item
4132: At this point, the execution of @code{my:} is complete, and control
4133: returns to the text interpreter. The text interpreter is in compile
4134: state, so subsequent text @code{+ -} is compiled into the definition of
4135: @code{foo} and the @code{;} terminates the definition as always.
4136: @end itemize
4137:
4138: You can use @code{see} to decompile a word that was defined using
4139: @code{my:} and see how it is different from a normal @code{:}
4140: definition. For example:
4141:
4142: @example
4143: : bar + - ; \ like foo but using : rather than my:
4144: see bar
4145: : bar
4146: + - ;
4147: see foo
4148: : foo
4149: 107645672 stats + - ;
4150:
4151: \ use ' stats . to show that 107645672 is the xt for stats
4152: @end example
4153:
4154: You can use techniques like this to make new defining words in terms of
4155: @i{any} existing defining word.
4156:
4157:
4158: @cindex defining defining words
4159: @cindex @code{CREATE} ... @code{DOES>}
4160: If you want the words defined with your defining words to behave
4161: differently from words defined with standard defining words, you can
4162: write your defining word like this:
4163:
4164: @example
4165: : def-word ( "name" -- )
4166: CREATE @i{code1}
4167: DOES> ( ... -- ... )
4168: @i{code2} ;
4169:
4170: def-word name
4171: @end example
4172:
4173: @cindex child words
4174: This fragment defines a @dfn{defining word} @code{def-word} and then
4175: executes it. When @code{def-word} executes, it @code{CREATE}s a new
4176: word, @code{name}, and executes the code @i{code1}. The code @i{code2}
4177: is not executed at this time. The word @code{name} is sometimes called a
4178: @dfn{child} of @code{def-word}.
4179:
4180: When you execute @code{name}, the address of the body of @code{name} is
4181: put on the data stack and @i{code2} is executed (the address of the body
4182: of @code{name} is the address @code{HERE} returns immediately after the
4183: @code{CREATE}).
4184:
4185: @cindex atavism in child words
4186: You can use @code{def-word} to define a set of child words that behave
4187: differently, though atavistically; they all have a common run-time
4188: behaviour determined by @i{code2}. Typically, the @i{code1} sequence
4189: builds a data area in the body of the child word. The structure of the
4190: data is common to all children of @code{def-word}, but the data values
4191: are specific -- and private -- to each child word. When a child word is
4192: executed, the address of its private data area is passed as a parameter
4193: on TOS to be used and manipulated@footnote{It is legitimate both to read
4194: and write to this data area.} by @i{code2}.
4195:
4196: The two fragments of code that make up the defining words act (are
4197: executed) at two completely separate times:
4198:
4199: @itemize @bullet
4200: @item
4201: At @i{define time}, the defining word executes @i{code1} to generate a
4202: child word
4203: @item
4204: At @i{child execution time}, when a child word is invoked, @i{code2}
4205: is executed, using parameters (data) that are private and specific to
4206: the child word.
4207: @end itemize
4208:
4209: Another way of understanding the behaviour of @code{def-word} and
4210: @code{name} is to say that, if you make the following definitions:
4211: @example
4212: : def-word1 ( "name" -- )
4213: CREATE @i{code1} ;
4214:
4215: : action1 ( ... -- ... )
4216: @i{code2} ;
4217:
4218: def-word1 name1
4219: @end example
4220:
4221: @noindent
4222: Then using @code{name1 action1} is equivalent to using @code{name}.
4223:
4224: The classic example is that you can define @code{CONSTANT} in this way:
4225:
4226: @example
4227: : CONSTANT ( w "name" -- )
4228: CREATE ,
4229: DOES> ( -- w )
4230: @@ ;
4231: @end example
4232:
4233: @comment There is a beautiful description of how this works and what
4234: @comment it does in the Forthwrite 100th edition.. as well as an elegant
4235: @comment commentary on the Counting Fruits problem.
4236:
4237: When you create a constant with @code{5 CONSTANT five}, a set of
4238: define-time actions take place; first a new word @code{five} is created,
4239: then the value 5 is laid down in the body of @code{five} with
4240: @code{,}. When @code{five} is executed, the address of the body is put on
4241: the stack, and @code{@@} retrieves the value 5. The word @code{five} has
4242: no code of its own; it simply contains a data field and a pointer to the
4243: code that follows @code{DOES>} in its defining word. That makes words
4244: created in this way very compact.
4245:
4246: The final example in this section is intended to remind you that space
4247: reserved in @code{CREATE}d words is @i{data} space and therefore can be
4248: both read and written by a Standard program@footnote{Exercise: use this
4249: example as a starting point for your own implementation of @code{Value}
4250: and @code{TO} -- if you get stuck, investigate the behaviour of @code{'} and
4251: @code{[']}.}:
4252:
4253: @example
4254: : foo ( "name" -- )
4255: CREATE -1 ,
4256: DOES> ( -- )
4257: @@ . ;
4258:
4259: foo first-word
4260: foo second-word
4261:
4262: 123 ' first-word >BODY !
4263: @end example
4264:
4265: If @code{first-word} had been a @code{CREATE}d word, we could simply
4266: have executed it to get the address of its data field. However, since it
4267: was defined to have @code{DOES>} actions, its execution semantics are to
4268: perform those @code{DOES>} actions. To get the address of its data field
4269: it's necessary to use @code{'} to get its xt, then @code{>BODY} to
4270: translate the xt into the address of the data field. When you execute
4271: @code{first-word}, it will display @code{123}. When you execute
4272: @code{second-word} it will display @code{-1}.
4273:
4274: @cindex stack effect of @code{DOES>}-parts
4275: @cindex @code{DOES>}-parts, stack effect
4276: In the examples above the stack comment after the @code{DOES>} specifies
4277: the stack effect of the defined words, not the stack effect of the
4278: following code (the following code expects the address of the body on
4279: the top of stack, which is not reflected in the stack comment). This is
4280: the convention that I use and recommend (it clashes a bit with using
4281: locals declarations for stack effect specification, though).
4282:
4283: @subsubsection Applications of @code{CREATE..DOES>}
4284: @cindex @code{CREATE} ... @code{DOES>}, applications
4285:
4286: You may wonder how to use this feature. Here are some usage patterns:
4287:
4288: @cindex factoring similar colon definitions
4289: When you see a sequence of code occurring several times, and you can
4290: identify a meaning, you will factor it out as a colon definition. When
4291: you see similar colon definitions, you can factor them using
4292: @code{CREATE..DOES>}. E.g., an assembler usually defines several words
4293: that look very similar:
4294: @example
4295: : ori, ( reg-target reg-source n -- )
4296: 0 asm-reg-reg-imm ;
4297: : andi, ( reg-target reg-source n -- )
4298: 1 asm-reg-reg-imm ;
4299: @end example
4300:
4301: @noindent
4302: This could be factored with:
4303: @example
4304: : reg-reg-imm ( op-code -- )
4305: CREATE ,
4306: DOES> ( reg-target reg-source n -- )
4307: @@ asm-reg-reg-imm ;
4308:
4309: 0 reg-reg-imm ori,
4310: 1 reg-reg-imm andi,
4311: @end example
4312:
4313: @cindex currying
4314: Another view of @code{CREATE..DOES>} is to consider it as a crude way to
4315: supply a part of the parameters for a word (known as @dfn{currying} in
4316: the functional language community). E.g., @code{+} needs two
4317: parameters. Creating versions of @code{+} with one parameter fixed can
4318: be done like this:
4319: @example
4320: : curry+ ( n1 -- )
4321: CREATE ,
4322: DOES> ( n2 -- n1+n2 )
4323: @@ + ;
4324:
4325: 3 curry+ 3+
4326: -2 curry+ 2-
4327: @end example
4328:
4329: @subsubsection The gory details of @code{CREATE..DOES>}
4330: @cindex @code{CREATE} ... @code{DOES>}, details
4331:
4332: doc-does>
4333:
4334: @cindex @code{DOES>} in a separate definition
4335: This means that you need not use @code{CREATE} and @code{DOES>} in the
4336: same definition; you can put the @code{DOES>}-part in a separate
4337: definition. This allows us to, e.g., select among different @code{DOES>}-parts:
4338: @example
4339: : does1
4340: DOES> ( ... -- ... )
4341: ... ;
4342:
4343: : does2
4344: DOES> ( ... -- ... )
4345: ... ;
4346:
4347: : def-word ( ... -- ... )
4348: create ...
4349: IF
4350: does1
4351: ELSE
4352: does2
4353: ENDIF ;
4354: @end example
4355:
4356: In this example, the selection of whether to use @code{does1} or
4357: @code{does2} is made at compile-time; at the time that the child word is
4358: @code{CREATE}d.
4359:
4360: @cindex @code{DOES>} in interpretation state
4361: In a standard program you can apply a @code{DOES>}-part only if the last
4362: word was defined with @code{CREATE}. In Gforth, the @code{DOES>}-part
4363: will override the behaviour of the last word defined in any case. In a
4364: standard program, you can use @code{DOES>} only in a colon
4365: definition. In Gforth, you can also use it in interpretation state, in a
4366: kind of one-shot mode; for example:
4367: @example
4368: CREATE name ( ... -- ... )
4369: @i{initialization}
4370: DOES>
4371: @i{code} ;
4372: @end example
4373:
4374: @noindent
4375: is equivalent to the standard:
4376: @example
4377: :noname
4378: DOES>
4379: @i{code} ;
4380: CREATE name EXECUTE ( ... -- ... )
4381: @i{initialization}
4382: @end example
4383:
4384:
4385: doc->body
4386:
4387:
4388: @node Deferred words, Aliases, User-defined Defining Words, Defining Words
4389: @subsection Deferred words
4390: @cindex deferred words
4391:
4392: The defining word @code{Defer} allows you to define a word by name
4393: without defining its behaviour; the definition of its behaviour is
4394: deferred. Here are two situation where this can be useful:
4395:
4396: @itemize @bullet
4397: @item
4398: Where you want to allow the behaviour of a word to be altered later, and
4399: for all precompiled references to the word to change when its behaviour
4400: is changed.
4401: @item
4402: For mutual recursion; @xref{Calls and returns}.
4403: @end itemize
4404:
4405: In the following example, @code{foo} always invokes the version of
4406: @code{greet} that prints ``@code{Good morning}'' whilst @code{bar}
4407: always invokes the version that prints ``@code{Hello}''. There is no way
4408: of getting @code{foo} to use the later version without re-ordering the
4409: source code and recompiling it.
4410:
4411: @example
4412: : greet ." Good morning" ;
4413: : foo ... greet ... ;
4414: : greet ." Hello" ;
4415: : bar ... greet ... ;
4416: @end example
4417:
4418: This problem can be solved by defining @code{greet} as a @code{Defer}red
4419: word. The behaviour of a @code{Defer}red word can be defined and
4420: redefined at any time by using @code{IS} to associate the xt of a
4421: previously-defined word with it. The previous example becomes:
4422:
4423: @example
4424: Defer greet
4425: : foo ... greet ... ;
4426: : bar ... greet ... ;
4427: : greet1 ." Good morning" ;
4428: : greet2 ." Hello" ;
4429: ' greet2 <IS> greet \ make greet behave like greet2
4430: @end example
4431:
4432: A deferred word can be used to improve the statistics-gathering example
4433: from @ref{User-defined Defining Words}; rather than edit the
4434: application's source code to change every @code{:} to a @code{my:}, do
4435: this:
4436:
4437: @example
4438: : real: : ; \ retain access to the original
4439: defer : \ redefine as a deferred word
4440: ' my: IS : \ use special version of :
4441: \
4442: \ load application here
4443: \
4444: ' real: IS : \ go back to the original
4445: @end example
4446:
4447:
4448: One thing to note is that @code{<IS>} consumes its name when it is
4449: executed. If you want to specify the name at compile time, use
4450: @code{[IS]}:
4451:
4452: @example
4453: : set-greet ( xt -- )
4454: [IS] greet ;
4455:
4456: ' greet1 set-greet
4457: @end example
4458:
4459: A deferred word can only inherit default semantics from the xt (because
4460: that is all that an xt can represent -- @pxref{Tokens for Words} for
4461: more discussion of this). However, the semantics of the deferred word
4462: itself can be modified at the time that it is defined. For example:
4463:
4464: @example
4465: : bar .... ; compile-only
4466: Defer fred immediate
4467: Defer jim
4468:
4469: ' bar <IS> jim \ jim has default semantics
4470: ' bar <IS> fred \ fred is immediate
4471: @end example
4472:
4473: doc-defer
4474: doc-<is>
4475: doc-[is]
4476: doc-is
4477: @comment TODO document these: what's defers [is]
4478: doc-what's
4479: doc-defers
4480:
4481: @c Use @code{words-deferred} to see a list of deferred words.
4482:
4483: Definitions in ANS Forth for @code{defer}, @code{<is>} and @code{[is]}
4484: are provided in @file{compat/defer.fs}.
4485:
4486:
4487: @node Aliases, Supplying names, Deferred words, Defining Words
4488: @subsection Aliases
4489: @cindex aliases
4490:
4491: The defining word @code{Alias} allows you to define a word by name that
4492: has the same behaviour as some other word. Here are two situation where
4493: this can be useful:
4494:
4495: @itemize @bullet
4496: @item
4497: When you want access to a word's definition from a different word list
4498: (for an example of this, see the definition of the @code{Root} word list
4499: in the Gforth source).
4500: @item
4501: When you want to create a synonym; a definition that can be known by
4502: either of two names (for example, @code{THEN} and @code{ENDIF} are
4503: aliases).
4504: @end itemize
4505:
4506: The word whose behaviour the alias is to inherit is represented by an
4507: xt. Therefore, the alias only inherits default semantics from its
4508: ancestor. The semantics of the alias itself can be modified at the time
4509: that it is defined. For example:
4510:
4511: @example
4512: : foo ... ; immediate
4513:
4514: ' foo Alias bar \ bar is not an immediate word
4515: ' foo Alias fooby immediate \ fooby is an immediate word
4516: @end example
4517:
4518: Words that are aliases have the same xt, different headers in the
4519: dictionary, and consequently different name tokens (@pxref{Tokens for
4520: Words}) and possibly different immediate flags. An alias can only have
4521: default or immediate compilation semantics; you can define aliases for
4522: combined words with @code{interpret/compile:} -- see @ref{Combined words}.
4523:
4524: doc-alias
4525:
4526:
4527: @node Supplying names, Interpretation and Compilation Semantics, Aliases, Defining Words
4528: @subsection Supplying the name of a defined word
4529: @cindex names for defined words
4530: @cindex defining words, name given in a string
4531:
4532: By default, a defining word takes the name for the defined word from the
4533: input stream. Sometimes you want to supply the name from a string. You
4534: can do this with:
4535:
4536: doc-nextname
4537:
4538: For example:
4539:
4540: @example
4541: s" foo" nextname create
4542: @end example
4543:
4544: @noindent
4545: is equivalent to:
4546:
4547: @example
4548: create foo
4549: @end example
4550:
4551: @noindent
4552: @code{nextname} works with any defining word, not just @code{:}.
4553:
4554:
4555: @node Interpretation and Compilation Semantics, Combined words, Supplying names, Defining Words
4556: @subsection Interpretation and Compilation Semantics
4557: @cindex semantics, interpretation and compilation
4558:
4559: @cindex interpretation semantics
4560: The @dfn{interpretation semantics} of a word are what the text
4561: interpreter does when it encounters the word in interpret state. It also
4562: appears in some other contexts, e.g., the execution token returned by
4563: @code{' @i{word}} identifies the interpretation semantics of
4564: @i{word} (in other words, @code{' @i{word} execute} is equivalent to
4565: interpret-state text interpretation of @code{@i{word}}).
4566:
4567: @cindex compilation semantics
4568: The @dfn{compilation semantics} of a word are what the text interpreter
4569: does when it encounters the word in compile state. It also appears in
4570: other contexts, e.g, @code{POSTPONE @i{word}} compiles@footnote{In
4571: standard terminology, ``appends to the current definition''.} the
4572: compilation semantics of @i{word}.
4573:
4574: @cindex execution semantics
4575: The standard also talks about @dfn{execution semantics}. They are used
4576: only for defining the interpretation and compilation semantics of many
4577: words. By default, the interpretation semantics of a word are to
4578: @code{execute} its execution semantics, and the compilation semantics of
4579: a word are to @code{compile,} its execution semantics.@footnote{In
4580: standard terminology: The default interpretation semantics are its
4581: execution semantics; the default compilation semantics are to append its
4582: execution semantics to the execution semantics of the current
4583: definition.}
4584:
4585: @comment TODO expand, make it co-operate with new sections on text interpreter.
4586:
4587: @cindex immediate words
4588: @cindex compile-only words
4589: You can change the semantics of the most-recently defined word:
4590:
4591:
4592: doc-immediate
4593: doc-compile-only
4594: doc-restrict
4595:
4596:
4597: Note that ticking (@code{'}) a compile-only word gives an error
4598: (``Interpreting a compile-only word'').
4599:
4600:
4601: @node Combined words, ,Interpretation and Compilation Semantics, Defining Words
4602: @subsection Combined Words
4603: @cindex combined words
4604:
4605: Gforth allows you to define @dfn{combined words} -- words that have an
4606: arbitrary combination of interpretation and compilation semantics.
4607:
4608:
4609: doc-interpret/compile:
4610:
4611:
4612: This feature was introduced for implementing @code{TO} and @code{S"}. I
4613: recommend that you do not define such words, as cute as they may be:
4614: they make it hard to get at both parts of the word in some contexts.
4615: E.g., assume you want to get an execution token for the compilation
4616: part. Instead, define two words, one that embodies the interpretation
4617: part, and one that embodies the compilation part. Once you have done
4618: that, you can define a combined word with @code{interpret/compile:} for
4619: the convenience of your users.
4620:
4621: You might try to use this feature to provide an optimizing
4622: implementation of the default compilation semantics of a word. For
4623: example, by defining:
4624: @example
4625: :noname
4626: foo bar ;
4627: :noname
4628: POSTPONE foo POSTPONE bar ;
4629: interpret/compile: opti-foobar
4630: @end example
4631:
4632: @noindent
4633: as an optimizing version of:
4634:
4635: @example
4636: : foobar
4637: foo bar ;
4638: @end example
4639:
4640: Unfortunately, this does not work correctly with @code{[compile]},
4641: because @code{[compile]} assumes that the compilation semantics of all
4642: @code{interpret/compile:} words are non-default. I.e., @code{[compile]
4643: opti-foobar} would compile compilation semantics, whereas
4644: @code{[compile] foobar} would compile interpretation semantics.
4645:
4646: @cindex state-smart words (are a bad idea)
4647: Some people try to use @dfn{state-smart} words to emulate the feature provided
4648: by @code{interpret/compile:} (words are state-smart if they check
4649: @code{STATE} during execution). E.g., they would try to code
4650: @code{foobar} like this:
4651:
4652: @example
4653: : foobar
4654: STATE @@
4655: IF ( compilation state )
4656: POSTPONE foo POSTPONE bar
4657: ELSE
4658: foo bar
4659: ENDIF ; immediate
4660: @end example
4661:
4662: Although this works if @code{foobar} is only processed by the text
4663: interpreter, it does not work in other contexts (like @code{'} or
4664: @code{POSTPONE}). E.g., @code{' foobar} will produce an execution token
4665: for a state-smart word, not for the interpretation semantics of the
4666: original @code{foobar}; when you execute this execution token (directly
4667: with @code{EXECUTE} or indirectly through @code{COMPILE,}) in compile
4668: state, the result will not be what you expected (i.e., it will not
4669: perform @code{foo bar}). State-smart words are a bad idea. Simply don't
4670: write them@footnote{For a more detailed discussion of this topic, see
4671: @cite{@code{State}-smartness -- Why it is Evil and How to Exorcise it} by Anton
4672: Ertl; presented at EuroForth '98 and available from
4673: @url{http://www.complang.tuwien.ac.at/papers/ertl98.ps.gz}}!
4674:
4675: @cindex defining words with arbitrary semantics combinations
4676: It is also possible to write defining words that define words with
4677: arbitrary combinations of interpretation and compilation semantics. In
4678: general, they look like this:
4679:
4680: @example
4681: : def-word
4682: create-interpret/compile
4683: @i{code1}
4684: interpretation>
4685: @i{code2}
4686: <interpretation
4687: compilation>
4688: @i{code3}
4689: <compilation ;
4690: @end example
4691:
4692: For a @i{word} defined with @code{def-word}, the interpretation
4693: semantics are to push the address of the body of @i{word} and perform
4694: @i{code2}, and the compilation semantics are to push the address of
4695: the body of @i{word} and perform @i{code3}. E.g., @code{constant}
4696: can also be defined like this (except that the defined constants don't
4697: behave correctly when @code{[compile]}d):
4698:
4699: @example
4700: : constant ( n "name" -- )
4701: create-interpret/compile
4702: ,
4703: interpretation> ( -- n )
4704: @@
4705: <interpretation
4706: compilation> ( compilation. -- ; run-time. -- n )
4707: @@ postpone literal
4708: <compilation ;
4709: @end example
4710:
4711:
4712: doc-create-interpret/compile
4713: doc-interpretation>
4714: doc-<interpretation
4715: doc-compilation>
4716: doc-<compilation
4717:
4718:
4719: Words defined with @code{interpret/compile:} and
4720: @code{create-interpret/compile} have an extended header structure that
4721: differs from other words; however, unless you try to access them with
4722: plain address arithmetic, you should not notice this. Words for
4723: accessing the header structure usually know how to deal with this; e.g.,
4724: @code{'} @i{word} @code{>body} also gives you the body of a word created
4725: with @code{create-interpret/compile}.
4726:
4727:
4728: doc-postpone
4729:
4730: @comment TODO -- expand glossary text for POSTPONE
4731:
4732: @c ----------------------------------------------------------
4733: @node The Text Interpreter, Tokens for Words, Defining Words, Words
4734: @section The Text Interpreter
4735: @cindex interpreter - outer
4736: @cindex text interpreter
4737: @cindex outer interpreter
4738:
4739: @c Should we really describe all these ugly details? IMO the text
4740: @c interpreter should be much cleaner, but that may not be possible within
4741: @c ANS Forth. - anton
4742: @c nac-> I wanted to explain how it works to show how you can exploit
4743: @c it in your own programs. When I was writing a cross-compiler, figuring out
4744: @c some of these gory details was very helpful to me. None of the textbooks
4745: @c I've seen cover it, and the most modern Forth textbook -- Forth Inc's,
4746: @c seems to positively avoid going into too much detail for some of
4747: @c the internals.
4748:
4749: The text interpreter@footnote{This is an expanded version of the
4750: material in @ref{Introducing the Text Interpreter}.} is an endless loop
4751: that processes input from the current input device. It is also called
4752: the outer interpreter, in contrast to the inner interpreter
4753: (@pxref{Engine}) which executes the compiled Forth code on interpretive
4754: implementations.
4755:
4756: @cindex interpret state
4757: @cindex compile state
4758: The text interpreter operates in one of two states: @dfn{interpret
4759: state} and @dfn{compile state}. The current state is defined by the
4760: aptly-named variable, @code{state}.
4761:
4762: This section starts by describing how the text interpreter behaves when
4763: it is in interpret state, processing input from the user input device --
4764: the keyboard. This is the mode that a Forth system is in after it starts
4765: up.
4766:
4767: @cindex input buffer
4768: @cindex terminal input buffer
4769: The text interpreter works from an area of memory called the @dfn{input
4770: buffer}@footnote{When the text interpreter is processing input from the
4771: keyboard, this area of memory is called the @dfn{terminal input buffer}
4772: (TIB) and is addressed by the (obsolescent) words @code{TIB} and
4773: @code{#TIB}.}, which stores your keyboard input when you press the
4774: @key{RET} key. Starting at the beginning of the input buffer, it skips
4775: leading spaces (called @dfn{delimiters}) then parses a string (a
4776: sequence of non-space characters) until it reaches either a space
4777: character or the end of the buffer. Having parsed a string, it makes two
4778: attempts to process it:
4779:
4780: @cindex dictionary
4781: @itemize @bullet
4782: @item
4783: It looks for the string in a @dfn{dictionary} of definitions. If the
4784: string is found, the string names a @dfn{definition} (also known as a
4785: @dfn{word}) and the dictionary search returns information that allows
4786: the text interpreter to perform the word's @dfn{interpretation
4787: semantics}. In most cases, this simply means that the word will be
4788: executed.
4789: @item
4790: If the string is not found in the dictionary, the text interpreter
4791: attempts to treat it as a number, using the rules described in
4792: @ref{Number Conversion}. If the string represents a legal number in the
4793: current radix, the number is pushed onto a parameter stack (the data
4794: stack for integers, the floating-point stack for floating-point
4795: numbers).
4796: @end itemize
4797:
4798: If both attempts fail, or if the word is found in the dictionary but has
4799: no interpretation semantics@footnote{This happens if the word was
4800: defined as @code{COMPILE-ONLY}.} the text interpreter discards the
4801: remainder of the input buffer, issues an error message and waits for
4802: more input. If one of the attempts succeeds, the text interpreter
4803: repeats the parsing process until the whole of the input buffer has been
4804: processed, at which point it prints the status message ``@code{ ok}''
4805: and waits for more input.
4806:
4807: @cindex parse area
4808: The text interpreter keeps track of its position in the input buffer by
4809: updating a variable called @code{>IN} (pronounced ``to-in''). The value
4810: of @code{>IN} starts out as 0, indicating an offset of 0 from the start
4811: of the input buffer. The region from offset @code{>IN @@} to the end of
4812: the input buffer is called the @dfn{parse area}@footnote{In other words,
4813: the text interpreter processes the contents of the input buffer by
4814: parsing strings from the parse area until the parse area is empty.}.
4815: This example shows how @code{>IN} changes as the text interpreter parses
4816: the input buffer:
4817:
4818: @example
4819: : remaining >IN @@ SOURCE 2 PICK - -ROT + SWAP
4820: CR ." ->" TYPE ." <-" ; IMMEDIATE
4821:
4822: 1 2 3 remaining + remaining .
4823:
4824: : foo 1 2 3 remaining SWAP remaining ;
4825: @end example
4826:
4827: @noindent
4828: The result is:
4829:
4830: @example
4831: ->+ remaining .<-
4832: ->.<-5 ok
4833:
4834: ->SWAP remaining ;-<
4835: ->;<- ok
4836: @end example
4837:
4838: @cindex parsing words
4839: The value of @code{>IN} can also be modified by a word in the input
4840: buffer that is executed by the text interpreter. This means that a word
4841: can ``trick'' the text interpreter into either skipping a section of the
4842: input buffer@footnote{This is how parsing words work.} or into parsing a
4843: section twice. For example:
4844:
4845: @example
4846: : lat ." <<lat>>" ;
4847: : flat ." <<flat>>" >IN DUP @@ 3 - SWAP ! ;
4848: @end example
4849:
4850: @noindent
4851: When @code{flat} is executed, this output is produced@footnote{Exercise
4852: for the reader: what would happen if the @code{3} were replaced with
4853: @code{4}?}:
4854:
4855: @example
4856: <<flat>><<lat>>
4857: @end example
4858:
4859: @noindent
4860: Two important notes about the behaviour of the text interpreter:
4861:
4862: @itemize @bullet
4863: @item
4864: It processes each input string to completion before parsing additional
4865: characters from the input buffer.
4866: @item
4867: It treats the input buffer as a read-only region (and so must your code).
4868: @end itemize
4869:
4870: @noindent
4871: When the text interpreter is in compile state, its behaviour changes in
4872: these ways:
4873:
4874: @itemize @bullet
4875: @item
4876: If a parsed string is found in the dictionary, the text interpreter will
4877: perform the word's @dfn{compilation semantics}. In most cases, this
4878: simply means that the execution semantics of the word will be appended
4879: to the current definition.
4880: @item
4881: When a number is encountered, it is compiled into the current definition
4882: (as a literal) rather than being pushed onto a parameter stack.
4883: @item
4884: If an error occurs, @code{state} is modified to put the text interpreter
4885: back into interpret state.
4886: @item
4887: Each time a line is entered from the keyboard, Gforth prints
4888: ``@code{ compiled}'' rather than `` @code{ok}''.
4889: @end itemize
4890:
4891: @cindex text interpreter - input sources
4892: When the text interpreter is using an input device other than the
4893: keyboard, its behaviour changes in these ways:
4894:
4895: @itemize @bullet
4896: @item
4897: When the parse area is empty, the text interpreter attempts to refill
4898: the input buffer from the input source. When the input source is
4899: exhausted, the input source is set back to the user input device.
4900: @item
4901: It doesn't print out ``@code{ ok}'' or ``@code{ compiled}'' messages each
4902: time the parse area is emptied.
4903: @item
4904: If an error occurs, the input source is set back to the user input
4905: device.
4906: @end itemize
4907:
4908: @ref{Input Sources} describes this in more detail.
4909:
4910:
4911: doc->in
4912: doc-source
4913:
4914: doc-tib
4915: doc-#tib
4916:
4917:
4918: @menu
4919: * Input Sources::
4920: * Number Conversion::
4921: * Interpret/Compile states::
4922: * Literals::
4923: * Interpreter Directives::
4924: @end menu
4925:
4926: @node Input Sources, Number Conversion, The Text Interpreter, The Text Interpreter
4927: @subsection Input Sources
4928: @cindex input sources
4929: @cindex text interpreter - input sources
4930:
4931: By default, the text interpreter processes input from the user input
4932: device (the keyboard) when Forth starts up. The text interpreter can
4933: process input from any of these sources:
4934:
4935: @itemize @bullet
4936: @item
4937: The user input device -- the keyboard.
4938: @item
4939: A file, using the words described in @ref{Forth source files}.
4940: @item
4941: A block, using the words described in @ref{Blocks}.
4942: @item
4943: A text string, using @code{evaluate}.
4944: @end itemize
4945:
4946: A program can identify the current input device from the values of
4947: @code{source-id} and @code{blk}.
4948:
4949:
4950: doc-source-id
4951: doc-blk
4952:
4953: doc-save-input
4954: doc-restore-input
4955:
4956: doc-evaluate
4957:
4958:
4959:
4960: @node Number Conversion, Interpret/Compile states, Input Sources, The Text Interpreter
4961: @subsection Number Conversion
4962: @cindex number conversion
4963: @cindex double-cell numbers, input format
4964: @cindex input format for double-cell numbers
4965: @cindex single-cell numbers, input format
4966: @cindex input format for single-cell numbers
4967: @cindex floating-point numbers, input format
4968: @cindex input format for floating-point numbers
4969:
4970: This section describes the rules that the text interpreter uses when it
4971: tries to convert a string into a number.
4972:
4973: Let <digit> represent any character that is a legal digit in the current
4974: number base@footnote{For example, 0-9 when the number base is decimal or
4975: 0-9, A-F when the number base is hexadecimal.}.
4976:
4977: Let <decimal digit> represent any character in the range 0-9.
4978:
4979: Let @{@i{a b}@} represent the @i{optional} presence of any of the characters
4980: in the braces (@i{a} or @i{b} or neither).
4981:
4982: Let * represent any number of instances of the previous character
4983: (including none).
4984:
4985: Let any other character represent itself.
4986:
4987: @noindent
4988: Now, the conversion rules are:
4989:
4990: @itemize @bullet
4991: @item
4992: A string of the form <digit><digit>* is treated as a single-precision
4993: (cell-sized) positive integer. Examples are 0 123 6784532 32343212343456 42
4994: @item
4995: A string of the form -<digit><digit>* is treated as a single-precision
4996: (cell-sized) negative integer, and is represented using 2's-complement
4997: arithmetic. Examples are -45 -5681 -0
4998: @item
4999: A string of the form <digit><digit>*.<digit>* is treated as a double-precision
5000: (double-cell-sized) positive integer. Examples are 3465. 3.465 34.65
5001: (all three of these represent the same number).
5002: @item
5003: A string of the form -<digit><digit>*.<digit>* is treated as a
5004: double-precision (double-cell-sized) negative integer, and is
5005: represented using 2's-complement arithmetic. Examples are -3465. -3.465
5006: -34.65 (all three of these represent the same number).
5007: @item
5008: A string of the form @{+ -@}<decimal digit>@{.@}<decimal digit>*@{e
5009: E@}@{+ -@}<decimal digit><decimal digit>* is treated as a floating-point
5010: number. Examples are 1e 1e0 1.e 1.e0 +1e+0 (which all represent the same
5011: number) +12.E-4
5012: @end itemize
5013:
5014: By default, the number base used for integer number conversion is given
5015: by the contents of the variable @code{base}. Note that a lot of
5016: confusion can result from unexpected values of @code{base}. If you
5017: change @code{base} anywhere, make sure to save the old value and restore
5018: it afterwards. In general I recommend keeping @code{base} decimal, and
5019: using the prefixes described below for the popular non-decimal bases.
5020:
5021: doc-dpl
5022: doc-base
5023: doc-hex
5024: doc-decimal
5025:
5026:
5027: @cindex '-prefix for character strings
5028: @cindex &-prefix for decimal numbers
5029: @cindex %-prefix for binary numbers
5030: @cindex $-prefix for hexadecimal numbers
5031: Gforth allows you to override the value of @code{base} by using a
5032: prefix@footnote{Some Forth implementations provide a similar scheme by
5033: implementing @code{$} etc. as parsing words that process the subsequent
5034: number in the input stream and push it onto the stack. For example, see
5035: @cite{Number Conversion and Literals}, by Wil Baden; Forth Dimensions
5036: 20(3) pages 26--27. In such implementations, unlike in Gforth, a space
5037: is required between the prefix and the number.} before the first digit
5038: of an (integer) number. Four prefixes are supported:
5039:
5040: @itemize @bullet
5041: @item
5042: @code{&} -- decimal
5043: @item
5044: @code{%} -- binary
5045: @item
5046: @code{$} -- hexadecimal
5047: @item
5048: @code{'} -- base @code{max-char+1}
5049: @end itemize
5050:
5051: Here are some examples, with the equivalent decimal number shown after
5052: in braces:
5053:
5054: -$41 (-65), %1001101 (205), %1001.0001 (145 - a double-precision number),
5055: 'AB (16706; ascii A is 65, ascii B is 66, number is 65*256 + 66),
5056: 'ab (24930; ascii a is 97, ascii B is 98, number is 97*256 + 98),
5057: &905 (905), $abc (2478), $ABC (2478).
5058:
5059: @cindex number conversion - traps for the unwary
5060: @noindent
5061: Number conversion has a number of traps for the unwary:
5062:
5063: @itemize @bullet
5064: @item
5065: You cannot determine the current number base using the code sequence
5066: @code{base @@ .} -- the number base is always 10 in the current number
5067: base. Instead, use something like @code{base @@ dec.}
5068: @item
5069: If the number base is set to a value greater than 14 (for example,
5070: hexadecimal), the number 123E4 is ambiguous; the conversion rules allow
5071: it to be intepreted as either a single-precision integer or a
5072: floating-point number (Gforth treats it as an integer). The ambiguity
5073: can be resolved by explicitly stating the sign of the mantissa and/or
5074: exponent: 123E+4 or +123E4 -- if the number base is decimal, no
5075: ambiguity arises; either representation will be treated as a
5076: floating-point number.
5077: @item
5078: There is a word @code{bin} but it does @i{not} set the number base!
5079: It is used to specify file types.
5080: @item
5081: ANS Forth requires the @code{.} of a double-precision number to
5082: be the final character in the string. Allowing the @code{.} to be
5083: anywhere after the first digit is a Gforth extension.
5084: @item
5085: The number conversion process does not check for overflow.
5086: @item
5087: In Gforth, number conversion to floating-point numbers always use base
5088: 10, irrespective of the value of @code{base}. In ANS Forth,
5089: conversion to floating-point numbers whilst the value of
5090: @code{base} is not 10 is an ambiguous condition.
5091: @end itemize
5092:
5093: @ref{Input} describes words that you can use to read numbers into your
5094: programs.
5095:
5096: @node Interpret/Compile states, Literals, Number Conversion, The Text Interpreter
5097: @subsection Interpret/Compile states
5098: @cindex Interpret/Compile states
5099:
5100: A standard program is not permitted to change @code{state}
5101: explicitly. However, it can change @code{state} implicitly, using the
5102: words @code{[} and @code{]}. When @code{[} is executed it switches
5103: @code{state} to interpret state, and therefore the text interpreter
5104: starts interpreting. When @code{]} is executed it switches @code{state}
5105: to compile state and therefore the text interpreter starts
5106: compiling. The most common usage for these words is for switching into
5107: interpret state and back from within a colon definition; this technique
5108: can be used to compile a literal (@pxref{Literals} for an example) or
5109: for conditional compilation (@pxref{Interpreter Directives} for an
5110: example).
5111:
5112:
5113: @c This is a bad example: It's non-standard, and it's not necessary.
5114: @c However, I can't think of a good example for switching into compile
5115: @c state when there is no current word (@code{state}-smart words are not a
5116: @c good reason). So maybe we should use an example for switching into
5117: @c interpret @code{state} in a colon def. - anton
5118: @c nac-> I agree. I started out by putting in the example, then realised
5119: @c that it was non-ANS, so wrote more words around it. I hope this
5120: @c re-written version is acceptable to you. I do want to keep the example
5121: @c as it is helpful for showing what is and what is not portable, particularly
5122: @c where it outlaws a style in common use.
5123:
5124:
5125: @code{[} and @code{]} also give you the ability to switch into compile
5126: state and back, but we cannot think of any useful Standard application
5127: for this ability. Pre-ANS Forth textbooks have examples like this:
5128:
5129: @example
5130: : AA ." this is A" ;
5131: : BB ." this is B" ;
5132: : CC ." this is C" ;
5133:
5134: create table ] aa bb cc [
5135:
5136: : go ( n -- ) \ n is offset into table.. 0 for 1st entry
5137: cells table + @ execute ;
5138: @end example
5139:
5140: This example builds a jump table; @code{0 go} will display ``@code{this
5141: is A}''. Using @code{[} and @code{]} in this example is equivalent to
5142: defining @code{table} like this:
5143:
5144: @example
5145: create table ' aa COMPILE, ' bb COMPILE, ' cc COMPILE,
5146: @end example
5147:
5148: The problem with this code is that the definition of @code{table} is not
5149: portable -- it @i{compile}s execution tokens into code space. Whilst it
5150: @i{may} work on systems where code space and data space co-incide, the
5151: Standard only allows data space to be assigned for a @code{CREATE}d
5152: word. In addition, the Standard only allows @code{@@} to access data
5153: space, whilst this example is using it to access code space. The only
5154: portable, Standard way to build this table is to build it in data space,
5155: like this:
5156:
5157: @example
5158: create table ' aa , ' bb , ' cc ,
5159: @end example
5160:
5161: doc-state
5162: doc-[
5163: doc-]
5164:
5165:
5166: @node Literals, Interpreter Directives, Interpret/Compile states, The Text Interpreter
5167: @subsection Literals
5168: @cindex Literals
5169:
5170: Often, you want to use a number within a colon definition. When you do
5171: this, the text interpreter automatically compiles the number as a
5172: @i{literal}. A literal is a number whose run-time effect is to be pushed
5173: onto the stack. If you had to do some maths to generate the number, you
5174: might write it like this:
5175:
5176: @example
5177: : HOUR-TO-SEC ( n1 -- n2 )
5178: 60 * \ to minutes
5179: 60 * ; \ to seconds
5180: @end example
5181:
5182: It is very clear what this definition is doing, but it's inefficient
5183: since it is performing 2 multiples at run-time. An alternative would be
5184: to write:
5185:
5186: @example
5187: : HOUR-TO-SEC ( n1 -- n2 )
5188: 3600 * ; \ to seconds
5189: @end example
5190:
5191: Which does the same thing, and has the advantage of using a single
5192: multiply. Ideally, we'd like the efficiency of the second with the
5193: readability of the first.
5194:
5195: @code{Literal} allows us to achieve that. It takes a number from the
5196: stack and lays it down in the current definition just as though the
5197: number had been typed directly into the definition. Our first attempt
5198: might look like this:
5199:
5200: @example
5201: 60 \ mins per hour
5202: 60 * \ seconds per minute
5203: : HOUR-TO-SEC ( n1 -- n2 )
5204: Literal * ; \ to seconds
5205: @end example
5206:
5207: But this produces the error message @code{unstructured}. What happened?
5208: The stack notation for @code{:} is (@i{ -- colon-sys}) and the size of
5209: @i{colon-sys} is implementation-defined. In other words, once we start a
5210: colon definition we can't portably access anything that was on the stack
5211: before the definition began@footnote{@cite{Two Problems in ANS Forth},
5212: by Thomas Worthington; Forth Dimensions 20(2) pages 32--34 describes
5213: some situations where you might want to access stack items above
5214: colon-sys, and provides a solution to the problem.}. The correct way of
5215: solving this problem in this instance is to use @code{[ ]} like this:
5216:
5217: @example
5218: : HOUR-TO-SEC ( n1 -- n2 )
5219: [ 60 \ minutes per hour
5220: 60 * ] \ seconds per minute
5221: LITERAL * ; \ to seconds
5222: @end example
5223:
5224:
5225: doc-literal
5226: doc-]L
5227: doc-2literal
5228: doc-fliteral
5229:
5230:
5231: @node Interpreter Directives, , Literals, The Text Interpreter
5232: @subsection Interpreter Directives
5233: @cindex interpreter directives
5234:
5235: These words are usually used in interpret state; typically to control
5236: which parts of a source file are processed by the text
5237: interpreter. There are only a few ANS Forth Standard words, but Gforth
5238: supplements these with a rich set of immediate control structure words
5239: to compensate for the fact that the non-immediate versions can only be
5240: used in compile state (@pxref{Control Structures}). Typical usages:
5241:
5242: @example
5243: FALSE Constant ASSEMBLER
5244: .
5245: .
5246: ASSEMBLER [IF]
5247: : ASSEMBLER-FEATURE
5248: ...
5249: ;
5250: [ENDIF]
5251: .
5252: .
5253: : SEE
5254: ... \ general-purpose SEE code
5255: [ ASSEMBLER [IF] ]
5256: ... \ assembler-specific SEE code
5257: [ [ENDIF] ]
5258: ;
5259: @end example
5260:
5261:
5262: doc-[IF]
5263: doc-[ELSE]
5264: doc-[THEN]
5265: doc-[ENDIF]
5266:
5267: doc-[IFDEF]
5268: doc-[IFUNDEF]
5269:
5270: doc-[?DO]
5271: doc-[DO]
5272: doc-[FOR]
5273: doc-[LOOP]
5274: doc-[+LOOP]
5275: doc-[NEXT]
5276:
5277: doc-[BEGIN]
5278: doc-[UNTIL]
5279: doc-[AGAIN]
5280: doc-[WHILE]
5281: doc-[REPEAT]
5282:
5283:
5284:
5285:
5286: @c -------------------------------------------------------------
5287: @node Tokens for Words, Word Lists, The Text Interpreter, Words
5288: @section Tokens for Words
5289: @cindex tokens for words
5290:
5291: This section describes the creation and use of tokens that represent
5292: words.
5293:
5294: Named words have information stored in their header space entries to
5295: indicate any non-default semantics (@pxref{Interpretation and
5296: Compilation Semantics}). The semantics can be modified, using
5297: @code{immediate} and/or @code{compile-only}, at the time that the words
5298: are defined. Unnamed words have (by definition) no header space
5299: entry, and therefore must have default semantics.
5300:
5301: Named words have interpretation and compilation semantics. Unnamed words
5302: just have execution semantics.
5303:
5304: @cindex xt
5305: @cindex execution token
5306: The execution semantics of an unnamed word are represented by an
5307: @dfn{execution token} (@i{xt}). As explained in @ref{Supplying names},
5308: the execution token of the last word defined can be produced with
5309: @code{lastxt}.
5310:
5311: The interpretation semantics of a named word are also represented by an
5312: execution token. You can produce the execution token using @code{'} or
5313: @code{[']}. A simple example shows the difference between the two:
5314:
5315: @example
5316: : greet ( -- ) ." Hello" ;
5317: : foo ( -- xt ) ['] greet execute ; \ ['] parses greet at compile-time
5318: : bar ( -- ) ' execute ; \ ' parses at run-time
5319:
5320: \ the next four lines all do the same thing
5321: foo
5322: bar greet
5323: greet
5324: ' greet EXECUTE
5325: @end example
5326:
5327: An execution token occupies one cell.
5328: @cindex code field address
5329: @cindex CFA
5330: In Gforth, the abstract data type @i{execution token} is implemented
5331: as a code field address (CFA).
5332: @comment TODO note that the standard does not say what it represents..
5333: @comment and you cannot necessarily compile it in all Forths (eg native
5334: @comment compilers?).
5335:
5336: For literals, use @code{'} in interpreted code and @code{[']} in
5337: compiled code. Gforth's @code{'} and @code{[']} behave somewhat
5338: unusually by complaining about compile-only words. To get the execution
5339: token for a compile-only word @i{name}, use @code{COMP' @i{name} DROP}
5340: or @code{[COMP'] @i{name} DROP}.
5341:
5342: @cindex compilation token
5343: The compilation semantics of a named word are represented by a
5344: @dfn{compilation token} consisting of two cells: @i{w xt}. The top cell
5345: @i{xt} is an execution token. The compilation semantics represented by
5346: the compilation token can be performed with @code{execute}, which
5347: consumes the whole compilation token, with an additional stack effect
5348: determined by the represented compilation semantics.
5349:
5350: At present, the @i{w} part of a compilation token is an execution token,
5351: and the @i{xt} part represents either @code{execute} or
5352: @code{compile,}@footnote{Depending upon the compilation semantics of the
5353: word. If the word has default compilation semantics, the @i{xt} will
5354: represent @code{compile,}. Otherwise (e.g., for immediate words), the
5355: @i{xt} will represent @code{execute}.}. However, don't rely on that
5356: knowledge, unless necessary; future versions of Gforth may introduce
5357: unusual compilation tokens (e.g., a compilation token that represents
5358: the compilation semantics of a literal).
5359:
5360: You can compile the compilation semantics with @code{postpone,}. I.e.,
5361: @code{COMP' @i{word} postpone,} is equivalent to @code{postpone
5362: @i{word}}.
5363:
5364: @cindex name token
5365: @cindex name field address
5366: @cindex NFA
5367: Named words are also represented by the @dfn{name token}, (@i{nt}). In
5368: Gforth, the abstract data type @emph{name token} is implemented as a
5369: name field address (NFA).
5370:
5371:
5372: doc-execute
5373: doc-compile,
5374: doc-[']
5375: doc-'
5376: doc-[comp']
5377: doc-comp'
5378: doc-postpone,
5379:
5380: doc-find-name
5381: doc-name>int
5382: doc-name?int
5383: doc-name>comp
5384: doc-name>string
5385:
5386:
5387: @c -------------------------------------------------------------
5388: @node Word Lists, Environmental Queries, Tokens for Words, Words
5389: @section Word Lists
5390: @cindex word lists
5391: @cindex header space
5392:
5393: A wordlist is a list of named words; you can add new words and look up
5394: words by name (and you can remove words in a restricted way with
5395: markers). Every named (and @code{reveal}ed) word is in one wordlist.
5396:
5397: @cindex search order stack
5398: The text interpreter searches the wordlists present in the search order
5399: (a stack of wordlists), from the top to the bottom. Within each
5400: wordlist, the search starts conceptually at the newest word; i.e., if
5401: two words in a wordlist have the same name, the newer word is found.
5402:
5403: @cindex compilation word list
5404: New words are added to the @dfn{compilation wordlist} (aka current
5405: wordlist).
5406:
5407: @cindex wid
5408: A word list is identified by a cell-sized word list identifier (@i{wid})
5409: in much the same way as a file is identified by a file handle. The
5410: numerical value of the wid has no (portable) meaning, and might change
5411: from session to session.
5412:
5413: The ANS Forth ``Search order'' word set is intended to provide a set of
5414: low-level tools that allow various different schemes to be
5415: implemented. Gforth provides @code{vocabulary}, a traditional Forth
5416: word. @file{compat/vocabulary.fs} provides an implementation in ANS
5417: Forth.
5418:
5419: @comment TODO: locals section refers to here, saying that every word list (aka
5420: @comment vocabulary) has its own methods for searching etc. Need to document that.
5421:
5422: @comment TODO: document markers, reveal, tables, mappedwordlist
5423:
5424: @comment the gforthman- prefix is used to pick out the true definition of a
5425: @comment word from the source files, rather than some alias.
5426:
5427: doc-forth-wordlist
5428: doc-definitions
5429: doc-get-current
5430: doc-set-current
5431: doc-get-order
5432: doc---gforthman-set-order
5433: doc-wordlist
5434: doc-table
5435: doc-push-order
5436: doc-previous
5437: doc-also
5438: doc---gforthman-forth
5439: doc-only
5440: doc---gforthman-order
5441:
5442: doc-find
5443: doc-search-wordlist
5444:
5445: doc-words
5446: doc-vlist
5447: @c doc-words-deferred
5448:
5449: doc-mappedwordlist
5450: doc-root
5451: doc-vocabulary
5452: doc-seal
5453: doc-vocs
5454: doc-current
5455: doc-context
5456:
5457:
5458: @menu
5459: * Why use word lists?::
5460: * Word list examples::
5461: @end menu
5462:
5463: @node Why use word lists?, Word list examples, Word Lists, Word Lists
5464: @subsection Why use word lists?
5465: @cindex word lists - why use them?
5466:
5467: Here are some reasons for using multiple word lists:
5468:
5469: @itemize @bullet
5470: @item
5471: To improve compilation speed by reducing the number of header space
5472: entries that must be searched. This is achieved by creating a new
5473: word list that contains all of the definitions that are used in the
5474: definition of a Forth system but which would not usually be used by
5475: programs running on that system. That word list would be on the search
5476: list when the Forth system was compiled but would be removed from the
5477: search list for normal operation. This can be a useful technique for
5478: low-performance systems (for example, 8-bit processors in embedded
5479: systems) but is unlikely to be necessary in high-performance desktop
5480: systems.
5481: @item
5482: To prevent a set of words from being used outside the context in which
5483: they are valid. Two classic examples of this are an integrated editor
5484: (all of the edit commands are defined in a separate word list; the
5485: search order is set to the editor word list when the editor is invoked;
5486: the old search order is restored when the editor is terminated) and an
5487: integrated assembler (the op-codes for the machine are defined in a
5488: separate word list which is used when a @code{CODE} word is defined).
5489: @item
5490: To prevent a name-space clash between multiple definitions with the same
5491: name. For example, when building a cross-compiler you might have a word
5492: @code{IF} that generates conditional code for your target system. By
5493: placing this definition in a different word list you can control whether
5494: the host system's @code{IF} or the target system's @code{IF} get used in
5495: any particular context by controlling the order of the word lists on the
5496: search order stack.
5497: @end itemize
5498:
5499: @node Word list examples, ,Why use word lists?, Word Lists
5500: @subsection Word list examples
5501: @cindex word lists - examples
5502:
5503: Here is an example of creating and using a new wordlist using ANS
5504: Forth Standard words:
5505:
5506: @example
5507: wordlist constant my-new-words-wordlist
5508: : my-new-words get-order nip my-new-words-wordlist swap set-order ;
5509:
5510: \ add it to the search order
5511: also my-new-words
5512:
5513: \ alternatively, add it to the search order and make it
5514: \ the compilation word list
5515: also my-new-words definitions
5516: \ type "order" to see the problem
5517: @end example
5518:
5519: The problem with this example is that @code{order} has no way to
5520: associate the name @code{my-new-words} with the wid of the word list (in
5521: Gforth, @code{order} and @code{vocs} will display @code{???} for a wid
5522: that has no associated name). There is no Standard way of associating a
5523: name with a wid.
5524:
5525: In Gforth, this example can be re-coded using @code{vocabulary}, which
5526: associates a name with a wid:
5527:
5528: @example
5529: vocabulary my-new-words
5530:
5531: \ add it to the search order
5532: also my-new-words
5533:
5534: \ alternatively, add it to the search order and make it
5535: \ the compilation word list
5536: my-new-words definitions
5537: \ type "order" to see that the problem is solved
5538: @end example
5539:
5540: @c -------------------------------------------------------------
5541: @node Environmental Queries, Files, Word Lists, Words
5542: @section Environmental Queries
5543: @cindex environmental queries
5544:
5545: ANS Forth introduced the idea of ``environmental queries'' as a way
5546: for a program running on a system to determine certain characteristics of the system.
5547: The Standard specifies a number of strings that might be recognised by a system.
5548:
5549: The Standard requires that the header space used for environmental queries
5550: be distinct from the header space used for definitions.
5551:
5552: Typically, environmental queries are supported by creating a set of
5553: definitions in a word list that is @i{only} used during environmental
5554: queries; that is what Gforth does. There is no Standard way of adding
5555: definitions to the set of recognised environmental queries, but any
5556: implementation that supports the loading of optional word sets must have
5557: some mechanism for doing this (after loading the word set, the
5558: associated environmental query string must return @code{true}). In
5559: Gforth, the word list used to honour environmental queries can be
5560: manipulated just like any other word list.
5561:
5562:
5563: doc-environment?
5564: doc-environment-wordlist
5565:
5566: doc-gforth
5567: doc-os-class
5568:
5569:
5570: Note that, whilst the documentation for (e.g.) @code{gforth} shows it
5571: returning two items on the stack, querying it using @code{environment?}
5572: will return an additional item; the @code{true} flag that shows that the
5573: string was recognised.
5574:
5575: @comment TODO Document the standard strings or note where they are documented herein
5576:
5577: Here are some examples of using environmental queries:
5578:
5579: @example
5580: s" address-unit-bits" environment? 0=
5581: [IF]
5582: cr .( environmental attribute address-units-bits unknown... ) cr
5583: [THEN]
5584:
5585: s" block" environment? [IF] DROP include block.fs [THEN]
5586:
5587: s" gforth" environment? [IF] 2DROP include compat/vocabulary.fs [THEN]
5588:
5589: s" gforth" environment? [IF] .( Gforth version ) TYPE
5590: [ELSE] .( Not Gforth..) [THEN]
5591: @end example
5592:
5593:
5594: Here is an example of adding a definition to the environment word list:
5595:
5596: @example
5597: get-current environment-wordlist set-current
5598: true constant block
5599: true constant block-ext
5600: set-current
5601: @end example
5602:
5603: You can see what definitions are in the environment word list like this:
5604:
5605: @example
5606: get-order 1+ environment-wordlist swap set-order words previous
5607: @end example
5608:
5609:
5610: @c -------------------------------------------------------------
5611: @node Files, Blocks, Environmental Queries, Words
5612: @section Files
5613: @cindex files
5614: @cindex I/O - file-handling
5615:
5616: Gforth provides facilities for accessing files that are stored in the
5617: host operating system's file-system. Files that are processed by Gforth
5618: can be divided into two categories:
5619:
5620: @itemize @bullet
5621: @item
5622: Files that are processed by the Text Interpreter (@dfn{Forth source files}).
5623: @item
5624: Files that are processed by some other program (@dfn{general files}).
5625: @end itemize
5626:
5627: doc-loadfilename
5628: doc-sourcefilename
5629: doc-sourceline#
5630:
5631: @menu
5632: * Forth source files::
5633: * General files::
5634: * Search Paths::
5635: * Forth Search Paths::
5636: * General Search Paths::
5637: @end menu
5638:
5639:
5640: @c -------------------------------------------------------------
5641: @node Forth source files, General files, Files, Files
5642: @subsection Forth source files
5643: @cindex including files
5644: @cindex Forth source files
5645:
5646: The simplest way to interpret the contents of a file is to use one of
5647: these two formats:
5648:
5649: @example
5650: include mysource.fs
5651: s" mysource.fs" included
5652: @end example
5653:
5654: Sometimes you want to include a file only if it is not included already
5655: (by, say, another source file). In that case, you can use one of these
5656: three formats:
5657:
5658: @example
5659: require mysource.fs
5660: needs mysource.fs
5661: s" mysource.fs" required
5662: @end example
5663:
5664: @cindex stack effect of included files
5665: @cindex including files, stack effect
5666: It is good practice to write your source files such that interpreting them
5667: does not change the stack. Source files designed in this way can be used with
5668: @code{required} and friends without complications. For example:
5669:
5670: @example
5671: 1 require foo.fs drop
5672: @end example
5673:
5674:
5675: doc-include-file
5676: doc-included
5677: doc-included?
5678: doc-include
5679: doc-required
5680: doc-require
5681: doc-needs
5682: doc-init-included-files
5683:
5684:
5685: A definition in ANS Forth for @code{required} is provided in
5686: @file{compat/required.fs}.
5687:
5688: @c -------------------------------------------------------------
5689: @node General files, Search Paths, Forth source files, Files
5690: @subsection General files
5691: @cindex general files
5692: @cindex file-handling
5693:
5694: Files are opened/created by name and type. The following types are
5695: recognised:
5696:
5697:
5698: doc-r/o
5699: doc-r/w
5700: doc-w/o
5701: doc-bin
5702:
5703:
5704: When a file is opened/created, it returns a file identifier,
5705: @i{wfileid} that is used for all other file commands. All file
5706: commands also return a status value, @i{wior}, that is 0 for a
5707: successful operation and an implementation-defined non-zero value in the
5708: case of an error.
5709:
5710:
5711: doc-open-file
5712: doc-create-file
5713:
5714: doc-close-file
5715: doc-delete-file
5716: doc-rename-file
5717: doc-read-file
5718: doc-read-line
5719: doc-write-file
5720: doc-write-line
5721: doc-emit-file
5722: doc-flush-file
5723:
5724: doc-file-status
5725: doc-file-position
5726: doc-reposition-file
5727: doc-file-size
5728: doc-resize-file
5729:
5730:
5731: @c ---------------------------------------------------------
5732: @node Search Paths, Forth Search Paths, General files, Files
5733: @subsection Search Paths
5734: @cindex path for @code{included}
5735: @cindex file search path
5736: @cindex @code{include} search path
5737: @cindex search path for files
5738:
5739: If you specify an absolute filename (i.e., a filename starting with
5740: @file{/} or @file{~}, or with @file{:} in the second position (as in
5741: @samp{C:...})) for @code{included} and friends, that file is included
5742: just as you would expect.
5743:
5744: For relative filenames, Gforth uses a search path similar to Forth's
5745: search order (@pxref{Word Lists}). It tries to find the given filename
5746: in the directories present in the path, and includes the first one it
5747: finds. There are separate search paths for Forth source files and
5748: general files.
5749:
5750: If the search path contains the directory @file{.} (as it should), this
5751: refers to the directory that the present file was @code{included}
5752: from. This allows files to include other files relative to their own
5753: position (irrespective of the current working directory or the absolute
5754: position). This feature is essential for libraries consisting of
5755: several files, where a file may include other files from the library.
5756: It corresponds to @code{#include "..."} in C. If the current input
5757: source is not a file, @file{.} refers to the directory of the innermost
5758: file being included, or, if there is no file being included, to the
5759: current working directory.
5760:
5761: Use @file{~+} to refer to the current working directory (as in the
5762: @code{bash}).
5763:
5764: If the filename starts with @file{./}, the search path is not searched
5765: (just as with absolute filenames), and the @file{.} has the same meaning
5766: as described above.
5767:
5768: @c ---------------------------------------------------------
5769: @node Forth Search Paths, General Search Paths, Search Paths, Files
5770: @subsubsection Forth Search Paths
5771: @cindex search path control - Forth
5772:
5773: The search path is initialized when you start Gforth (@pxref{Invoking
5774: Gforth}). You can display it and change it using these words:
5775:
5776:
5777: doc-.fpath
5778: doc-fpath+
5779: doc-fpath=
5780: doc-open-fpath-file
5781:
5782:
5783: @noindent
5784: Here is an example of using @code{fpath} and @code{require}:
5785:
5786: @example
5787: fpath= /usr/lib/forth/|./
5788: require timer.fs
5789: @end example
5790:
5791: @c ---------------------------------------------------------
5792: @node General Search Paths, , Forth Search Paths, Files
5793: @subsubsection General Search Paths
5794: @cindex search path control - for user applications
5795:
5796: Your application may need to search files in several directories, like
5797: @code{included} does. To facilitate this, Gforth allows you to define
5798: and use your own search paths, by providing generic equivalents of the
5799: Forth search path words:
5800:
5801:
5802: doc-.path
5803: doc-path+
5804: doc-path=
5805: doc-open-path-file
5806:
5807:
5808: Here's an example of creating a search path:
5809:
5810: @example
5811: \ Make a buffer for the path:
5812: create mypath 100 chars , \ maximum length (is checked)
5813: 0 , \ real len
5814: 100 chars allot \ space for path
5815: @end example
5816:
5817: @c -------------------------------------------------------------
5818: @node Blocks, Other I/O, Files, Words
5819: @section Blocks
5820: @cindex I/O - blocks
5821: @cindex blocks
5822:
5823: When you run Gforth on a modern desk-top computer, it runs under the
5824: control of an operating system which provides certain services. One of
5825: these services is @var{file services}, which allows Forth source code
5826: and data to be stored in files and read into Gforth (@pxref{Files}).
5827:
5828: Traditionally, Forth has been an important programming language on
5829: systems where it has interfaced directly to the underlying hardware with
5830: no intervening operating system. Forth provides a mechanism, called
5831: @dfn{blocks}, for accessing mass storage on such systems.
5832:
5833: A block is a 1024-byte data area, which can be used to hold data or
5834: Forth source code. No structure is imposed on the contents of the
5835: block. A block is identified by its number; blocks are numbered
5836: contiguously from 1 to an implementation-defined maximum.
5837:
5838: A typical system that used blocks but no operating system might use a
5839: single floppy-disk drive for mass storage, with the disks formatted to
5840: provide 256-byte sectors. Blocks would be implemented by assigning the
5841: first four sectors of the disk to block 1, the second four sectors to
5842: block 2 and so on, up to the limit of the capacity of the disk. The disk
5843: would not contain any file system information, just the set of blocks.
5844:
5845: @cindex blocks file
5846: On systems that do provide file services, blocks are typically
5847: implemented by storing a sequence of blocks within a single @dfn{blocks
5848: file}. The size of the blocks file will be an exact multiple of 1024
5849: bytes, corresponding to the number of blocks it contains. This is the
5850: mechanism that Gforth uses.
5851:
5852: @cindex @file{blocks.fb}
5853: Only 1 blocks file can be open at a time. If you use block words without
5854: having specified a blocks file, Gforth defaults to the blocks file
5855: @file{blocks.fb}. Gforth uses the Forth search path when attempting to
5856: locate a blocks file (@pxref{Forth Search Paths}).
5857:
5858: @cindex block buffers
5859: When you read and write blocks under program control, Gforth uses a
5860: number of @dfn{block buffers} as intermediate storage. These buffers are
5861: not used when you use @code{load} to interpret the contents of a block.
5862:
5863: The behaviour of the block buffers is directly analagous to that of a
5864: cache. Each block buffer has three states:
5865:
5866: @itemize @bullet
5867: @item
5868: Unassigned
5869: @item
5870: Assigned-clean
5871: @item
5872: Assigned-dirty
5873: @end itemize
5874:
5875: Initially, all block buffers are @i{unassigned}. In order to access a
5876: block, the block (specified by its block number) must be assigned to a
5877: block buffer.
5878:
5879: The assignment of a block to a block buffer is performed by @code{block}
5880: or @code{buffer}. Use @code{block} when you wish to modify the existing
5881: contents of a block. Use @code{buffer} when you don't care about the
5882: existing contents of the block@footnote{The ANS Forth definition of
5883: @code{buffer} is intended not to cause disk I/O; if the data associated
5884: with the particular block is already stored in a block buffer due to an
5885: earlier @code{block} command, @code{buffer} will return that block
5886: buffer and the existing contents of the block will be
5887: available. Otherwise, @code{buffer} will simply assign a new, empty
5888: block buffer for the block.}.
5889:
5890: Once a block has been assigned to a block buffer, the block buffer state
5891: becomes @i{assigned-clean}. Data can now be manipulated within the
5892: block buffer.
5893:
5894: When the contents of a block buffer is changed it is necessary,
5895: @i{before calling} @code{block} @i{or} @code{buffer} @i{again}, to
5896: either abandon the changes (by doing nothing) or commit the changes,
5897: using @code{update}. Using @code{update} does not change the blocks
5898: file; it simply changes a block buffer's state to @i{assigned-dirty}.
5899:
5900: The word @code{flush} causes all @i{assigned-dirty} blocks to be
5901: written back to the blocks file on disk. Leaving Gforth using @code{bye}
5902: also causes a @code{flush} to be performed.
5903:
5904: In Gforth, @code{block} and @code{buffer} use a @i{direct-mapped}
5905: algorithm to assign a block buffer to a block. That means that any
5906: particular block can only be assigned to one specific block buffer,
5907: called (for the particular operation) the @i{victim buffer}. If the
5908: victim buffer is @i{unassigned} or @i{assigned-clean} it can be
5909: allocated to the new block immediately. If it is @i{assigned-dirty}
5910: its current contents must be written out to disk before it can be
5911: allocated to the new block.
5912:
5913: Although no structure is imposed on the contents of a block, it is
5914: traditional to display the contents as 16 lines each of 64 characters. A
5915: block provides a single, continuous stream of input (for example, it
5916: acts as a single parse area) -- there are no end-of-line characters
5917: within a block, and no end-of-file character at the end of a
5918: block. There are two consequences of this:
5919:
5920: @itemize @bullet
5921: @item
5922: The last character of one line wraps straight into the first character
5923: of the following line
5924: @item
5925: The word @code{\} -- comment to end of line -- requires special
5926: treatment; in the context of a block it causes all characters until the
5927: end of the current 64-character ``line'' to be ignored.
5928: @end itemize
5929:
5930: In Gforth, when you use @code{block} with a non-existent block number,
5931: the current blocks file will be extended to the appropriate size and the
5932: block buffer will be initialised with spaces.
5933:
5934: Gforth includes a simple block editor (type @code{use blocked.fb} then
5935: @i{0 list} for details) but doesn't encourage the use of blocks; the
5936: mechanism is only provided for backward compatibility -- ANS Forth
5937: requires blocks to be available when files are.
5938:
5939: Common techniques that are used when working with blocks include:
5940:
5941: @itemize @bullet
5942: @item
5943: A screen editor that allows you to edit blocks without leaving the Forth
5944: environment.
5945: @item
5946: Shadow screens; where every code block has an associated block
5947: containing comments (for example: code in odd block numbers, comments in
5948: even block numbers). Typically, the block editor provides a convenient
5949: mechanism to toggle between code and comments.
5950: @item
5951: Load blocks; a single block (typically block 1) contains a number of
5952: @code{thru} commands which @code{load} the whole of the application.
5953: @end itemize
5954:
5955: See Frank Sergeant's Pygmy Forth to see just how well blocks can be
5956: integrated into a Forth programming environment.
5957:
5958: @comment TODO what about errors on open-blocks?
5959:
5960: doc-open-blocks
5961: doc-use
5962: doc-get-block-fid
5963: doc-block-position
5964:
5965: doc-scr
5966: doc-list
5967:
5968: doc---gforthman-block
5969: doc-buffer
5970:
5971: doc-update
5972: doc-updated?
5973: doc-save-buffers
5974: doc-empty-buffers
5975: doc-empty-buffer
5976: doc-flush
5977:
5978: doc-load
5979: doc-thru
5980: doc-+load
5981: doc-+thru
5982: doc---gforthman--->
5983: doc-block-included
5984:
5985:
5986: @c -------------------------------------------------------------
5987: @node Other I/O, Programming Tools, Blocks, Words
5988: @section Other I/O
5989: @cindex I/O - keyboard and display
5990:
5991: @menu
5992: * Simple numeric output:: Predefined formats
5993: * Formatted numeric output:: Formatted (pictured) output
5994: * String Formats:: How Forth stores strings in memory
5995: * Displaying characters and strings:: Other stuff
5996: * Input:: Input
5997: @end menu
5998:
5999: @node Simple numeric output, Formatted numeric output, Other I/O, Other I/O
6000: @subsection Simple numeric output
6001: @cindex numeric output - simple/free-format
6002:
6003: The simplest output functions are those that display numbers from the
6004: data or floating-point stacks. Floating-point output is always displayed
6005: using base 10. Numbers displayed from the data stack use the value stored
6006: in @code{base}.
6007:
6008:
6009: doc-.
6010: doc-dec.
6011: doc-hex.
6012: doc-u.
6013: doc-.r
6014: doc-u.r
6015: doc-d.
6016: doc-ud.
6017: doc-d.r
6018: doc-ud.r
6019: doc-f.
6020: doc-fe.
6021: doc-fs.
6022:
6023:
6024: Examples of printing the number 1234.5678E23 in the different floating-point output
6025: formats are shown below:
6026:
6027: @example
6028: f. 123456779999999000000000000.
6029: fe. 123.456779999999E24
6030: fs. 1.23456779999999E26
6031: @end example
6032:
6033:
6034: @node Formatted numeric output, String Formats, Simple numeric output, Other I/O
6035: @subsection Formatted numeric output
6036: @cindex formatted numeric output
6037: @cindex pictured numeric output
6038: @cindex numeric output - formatted
6039:
6040: Forth traditionally uses a technique called @dfn{pictured numeric
6041: output} for formatted printing of integers. In this technique, digits
6042: are extracted from the number (using the current output radix defined by
6043: @code{base}), converted to ASCII codes and appended to a string that is
6044: built in a scratch-pad area of memory (@pxref{core-idef,
6045: Implementation-defined options, Implementation-defined
6046: options}). Arbitrary characters can be appended to the string during the
6047: extraction process. The completed string is specified by an address
6048: and length and can be manipulated (@code{TYPE}ed, copied, modified)
6049: under program control.
6050:
6051: All of the words described in the previous section for simple numeric
6052: output are implemented in Gforth using pictured numeric output.
6053:
6054: Three important things to remember about Pictured Numeric Output:
6055:
6056: @itemize @bullet
6057: @item
6058: It always operates on double-precision numbers; to display a
6059: single-precision number, convert it first (@pxref{Double precision} for
6060: ways of doing this).
6061: @item
6062: It always treats the double-precision number as though it were
6063: unsigned. The examples below show ways of printing signed numbers.
6064: @item
6065: The string is built up from right to left; least significant digit first.
6066: @end itemize
6067:
6068:
6069: doc-<#
6070: doc-#
6071: doc-#s
6072: doc-hold
6073: doc-sign
6074: doc-#>
6075:
6076: doc-represent
6077:
6078:
6079: @noindent
6080: Here are some examples of using pictured numeric output:
6081:
6082: @example
6083: : my-u. ( u -- )
6084: \ Simplest use of pns.. behaves like Standard u.
6085: 0 \ convert to unsigned double
6086: <# \ start conversion
6087: #s \ convert all digits
6088: #> \ complete conversion
6089: TYPE SPACE ; \ display, with trailing space
6090:
6091: : cents-only ( u -- )
6092: 0 \ convert to unsigned double
6093: <# \ start conversion
6094: # # \ convert two least-significant digits
6095: #> \ complete conversion, discard other digits
6096: TYPE SPACE ; \ display, with trailing space
6097:
6098: : dollars-and-cents ( u -- )
6099: 0 \ convert to unsigned double
6100: <# \ start conversion
6101: # # \ convert two least-significant digits
6102: [char] . hold \ insert decimal point
6103: #s \ convert remaining digits
6104: [char] $ hold \ append currency symbol
6105: #> \ complete conversion
6106: TYPE SPACE ; \ display, with trailing space
6107:
6108: : my-. ( n -- )
6109: \ handling negatives.. behaves like Standard .
6110: s>d \ convert to signed double
6111: swap over dabs \ leave sign byte followed by unsigned double
6112: <# \ start conversion
6113: #s \ convert all digits
6114: rot sign \ get at sign byte, append "-" if needed
6115: #> \ complete conversion
6116: TYPE SPACE ; \ display, with trailing space
6117:
6118: : account. ( n -- )
6119: \ accountants don't like minus signs, they use braces
6120: \ for negative numbers
6121: s>d \ convert to signed double
6122: swap over dabs \ leave sign byte followed by unsigned double
6123: <# \ start conversion
6124: 2 pick \ get copy of sign byte
6125: 0< IF [char] ) hold THEN \ right-most character of output
6126: #s \ convert all digits
6127: rot \ get at sign byte
6128: 0< IF [char] ( hold THEN
6129: #> \ complete conversion
6130: TYPE SPACE ; \ display, with trailing space
6131: @end example
6132:
6133: Here are some examples of using these words:
6134:
6135: @example
6136: 1 my-u. 1
6137: hex -1 my-u. decimal FFFFFFFF
6138: 1 cents-only 01
6139: 1234 cents-only 34
6140: 2 dollars-and-cents $0.02
6141: 1234 dollars-and-cents $12.34
6142: 123 my-. 123
6143: -123 my. -123
6144: 123 account. 123
6145: -456 account. (456)
6146: @end example
6147:
6148:
6149: @node String Formats, Displaying characters and strings, Formatted numeric output, Other I/O
6150: @subsection String Formats
6151: @cindex strings - see character strings
6152: @cindex character strings - formats
6153: @cindex I/O - see character strings
6154:
6155: Forth commonly uses two different methods for representing character
6156: strings:
6157:
6158: @itemize @bullet
6159: @item
6160: @cindex address of counted string
6161: @cindex counted string
6162: As a @dfn{counted string}, represented by a @i{c-addr}. The char
6163: addressed by @i{c-addr} contains a character-count, @i{n}, of the
6164: string and the string occupies the subsequent @i{n} char addresses in
6165: memory.
6166: @item
6167: As cell pair on the stack; @i{c-addr u}, where @i{u} is the length
6168: of the string in characters, and @i{c-addr} is the address of the
6169: first byte of the string.
6170: @end itemize
6171:
6172: ANS Forth encourages the use of the second format when representing
6173: strings on the stack, whilst conceeding that the counted string format
6174: remains useful as a way of storing strings in memory.
6175:
6176:
6177: doc-count
6178:
6179:
6180: @xref{Memory Blocks} for words that move, copy and search
6181: for strings. @xref{Displaying characters and strings,} for words that
6182: display characters and strings.
6183:
6184:
6185: @node Displaying characters and strings, Input, String Formats, Other I/O
6186: @subsection Displaying characters and strings
6187: @cindex characters - compiling and displaying
6188: @cindex character strings - compiling and displaying
6189:
6190: This section starts with a glossary of Forth words and ends with a set
6191: of examples.
6192:
6193:
6194: doc-bl
6195: doc-space
6196: doc-spaces
6197: doc-emit
6198: doc-toupper
6199: doc-."
6200: doc-.(
6201: doc-type
6202: doc-typewhite
6203: doc-cr
6204: @cindex cursor control
6205: doc-at-xy
6206: doc-page
6207: doc-s"
6208: doc-c"
6209: doc-char
6210: doc-[char]
6211: doc-sliteral
6212:
6213:
6214: @noindent
6215: As an example, consider the following text, stored in a file @file{test.fs}:
6216:
6217: @example
6218: .( text-1)
6219: : my-word
6220: ." text-2" cr
6221: .( text-3)
6222: ;
6223:
6224: ." text-4"
6225:
6226: : my-char
6227: [char] ALPHABET emit
6228: char emit
6229: ;
6230: @end example
6231:
6232: When you load this code into Gforth, the following output is generated:
6233:
6234: @example
6235: @kbd{include test.fs @key{RET}} text-1text-3text-4 ok
6236: @end example
6237:
6238: @itemize @bullet
6239: @item
6240: Messages @code{text-1} and @code{text-3} are displayed because @code{.(}
6241: is an immediate word; it behaves in the same way whether it is used inside
6242: or outside a colon definition.
6243: @item
6244: Message @code{text-4} is displayed because of Gforth's added interpretation
6245: semantics for @code{."}.
6246: @item
6247: Message @code{text-2} is @i{not} displayed, because the text interpreter
6248: performs the compilation semantics for @code{."} within the definition of
6249: @code{my-word}.
6250: @end itemize
6251:
6252: Here are some examples of executing @code{my-word} and @code{my-char}:
6253:
6254: @example
6255: @kbd{my-word @key{RET}} text-2
6256: ok
6257: @kbd{my-char fred @key{RET}} Af ok
6258: @kbd{my-char jim @key{RET}} Aj ok
6259: @end example
6260:
6261: @itemize @bullet
6262: @item
6263: Message @code{text-2} is displayed because of the run-time behaviour of
6264: @code{."}.
6265: @item
6266: @code{[char]} compiles the ``A'' from ``ALPHABET'' and puts its display code
6267: on the stack at run-time. @code{emit} always displays the character
6268: when @code{my-char} is executed.
6269: @item
6270: @code{char} parses a string at run-time and the second @code{emit} displays
6271: the first character of the string.
6272: @item
6273: If you type @code{see my-char} you can see that @code{[char]} discarded
6274: the text ``LPHABET'' and only compiled the display code for ``A'' into the
6275: definition of @code{my-char}.
6276: @end itemize
6277:
6278:
6279:
6280: @node Input, , Displaying characters and strings, Other I/O
6281: @subsection Input
6282: @cindex input
6283: @cindex I/O - see input
6284: @cindex parsing a string
6285:
6286: @xref{String Formats} for ways of storing character strings in memory.
6287:
6288: @comment TODO examples for >number >float accept key key? pad parse word refill
6289: @comment then index them
6290:
6291:
6292: doc-key
6293: doc-key?
6294: doc-ekey
6295: doc-ekey?
6296: doc-ekey>char
6297: doc->number
6298: doc->float
6299: doc-accept
6300: doc-pad
6301: doc-parse
6302: doc-word
6303: doc-sword
6304: doc-(name)
6305: doc-refill
6306: @comment obsolescent words..
6307: doc-convert
6308: doc-query
6309: doc-expect
6310: doc-span
6311:
6312:
6313:
6314: @c -------------------------------------------------------------
6315: @node Programming Tools, Assembler and Code Words, Other I/O, Words
6316: @section Programming Tools
6317: @cindex programming tools
6318:
6319: @menu
6320: * Debugging:: Simple and quick.
6321: * Assertions:: Making your programs self-checking.
6322: * Singlestep Debugger:: Executing your program word by word.
6323: @end menu
6324:
6325: @node Debugging, Assertions, Programming Tools, Programming Tools
6326: @subsection Debugging
6327: @cindex debugging
6328:
6329: Languages with a slow edit/compile/link/test development loop tend to
6330: require sophisticated tracing/stepping debuggers to facilate
6331: productive debugging.
6332:
6333: A much better (faster) way in fast-compiling languages is to add
6334: printing code at well-selected places, let the program run, look at
6335: the output, see where things went wrong, add more printing code, etc.,
6336: until the bug is found.
6337:
6338: The simple debugging aids provided in @file{debugs.fs}
6339: are meant to support this style of debugging. In addition, there are
6340: words for non-destructively inspecting the stack and memory:
6341:
6342:
6343: doc-.s
6344: doc-f.s
6345:
6346:
6347: There is a word @code{.r} but it does @i{not} display the return
6348: stack! It is used for formatted numeric output.
6349:
6350:
6351: doc-depth
6352: doc-fdepth
6353: doc-clearstack
6354: doc-?
6355: doc-dump
6356:
6357:
6358: The word @code{~~} prints debugging information (by default the source
6359: location and the stack contents). It is easy to insert. If you use Emacs
6360: it is also easy to remove (@kbd{C-x ~} in the Emacs Forth mode to
6361: query-replace them with nothing). The deferred words
6362: @code{printdebugdata} and @code{printdebugline} control the output of
6363: @code{~~}. The default source location output format works well with
6364: Emacs' compilation mode, so you can step through the program at the
6365: source level using @kbd{C-x `} (the advantage over a stepping debugger
6366: is that you can step in any direction and you know where the crash has
6367: happened or where the strange data has occurred).
6368:
6369: The default actions of @code{~~} clobber the contents of the pictured
6370: numeric output string, so you should not use @code{~~}, e.g., between
6371: @code{<#} and @code{#>}.
6372:
6373:
6374: doc-~~
6375: doc-printdebugdata
6376: doc-printdebugline
6377:
6378: doc-see
6379: doc-marker
6380:
6381:
6382: Here's an example of using @code{marker} at the start of a source file
6383: that you are debugging; it ensures that you only ever have one copy of
6384: the file's definitions compiled at any time:
6385:
6386: @example
6387: [IFDEF] my-code
6388: my-code
6389: [ENDIF]
6390:
6391: marker my-code
6392: init-included-files
6393:
6394: \ .. definitions start here
6395: \ .
6396: \ .
6397: \ end
6398: @end example
6399:
6400:
6401:
6402: @node Assertions, Singlestep Debugger, Debugging, Programming Tools
6403: @subsection Assertions
6404: @cindex assertions
6405:
6406: It is a good idea to make your programs self-checking, especially if you
6407: make an assumption that may become invalid during maintenance (for
6408: example, that a certain field of a data structure is never zero). Gforth
6409: supports @dfn{assertions} for this purpose. They are used like this:
6410:
6411: @example
6412: assert( @i{flag} )
6413: @end example
6414:
6415: The code between @code{assert(} and @code{)} should compute a flag, that
6416: should be true if everything is alright and false otherwise. It should
6417: not change anything else on the stack. The overall stack effect of the
6418: assertion is @code{( -- )}. E.g.
6419:
6420: @example
6421: assert( 1 1 + 2 = ) \ what we learn in school
6422: assert( dup 0<> ) \ assert that the top of stack is not zero
6423: assert( false ) \ this code should not be reached
6424: @end example
6425:
6426: The need for assertions is different at different times. During
6427: debugging, we want more checking, in production we sometimes care more
6428: for speed. Therefore, assertions can be turned off, i.e., the assertion
6429: becomes a comment. Depending on the importance of an assertion and the
6430: time it takes to check it, you may want to turn off some assertions and
6431: keep others turned on. Gforth provides several levels of assertions for
6432: this purpose:
6433:
6434:
6435: doc-assert0(
6436: doc-assert1(
6437: doc-assert2(
6438: doc-assert3(
6439: doc-assert(
6440: doc-)
6441:
6442:
6443: The variable @code{assert-level} specifies the highest assertions that
6444: are turned on. I.e., at the default @code{assert-level} of one,
6445: @code{assert0(} and @code{assert1(} assertions perform checking, while
6446: @code{assert2(} and @code{assert3(} assertions are treated as comments.
6447:
6448: The value of @code{assert-level} is evaluated at compile-time, not at
6449: run-time. Therefore you cannot turn assertions on or off at run-time;
6450: you have to set the @code{assert-level} appropriately before compiling a
6451: piece of code. You can compile different pieces of code at different
6452: @code{assert-level}s (e.g., a trusted library at level 1 and
6453: newly-written code at level 3).
6454:
6455:
6456: doc-assert-level
6457:
6458:
6459: If an assertion fails, a message compatible with Emacs' compilation mode
6460: is produced and the execution is aborted (currently with @code{ABORT"}.
6461: If there is interest, we will introduce a special throw code. But if you
6462: intend to @code{catch} a specific condition, using @code{throw} is
6463: probably more appropriate than an assertion).
6464:
6465: Definitions in ANS Forth for these assertion words are provided
6466: in @file{compat/assert.fs}.
6467:
6468:
6469: @node Singlestep Debugger, , Assertions, Programming Tools
6470: @subsection Singlestep Debugger
6471: @cindex singlestep Debugger
6472: @cindex debugging Singlestep
6473:
6474: When you create a new word there's often the need to check whether it
6475: behaves correctly or not. You can do this by typing @code{dbg
6476: badword}. A debug session might look like this:
6477:
6478: @example
6479: : badword 0 DO i . LOOP ; ok
6480: 2 dbg badword
6481: : badword
6482: Scanning code...
6483:
6484: Nesting debugger ready!
6485:
6486: 400D4738 8049BC4 0 -> [ 2 ] 00002 00000
6487: 400D4740 8049F68 DO -> [ 0 ]
6488: 400D4744 804A0C8 i -> [ 1 ] 00000
6489: 400D4748 400C5E60 . -> 0 [ 0 ]
6490: 400D474C 8049D0C LOOP -> [ 0 ]
6491: 400D4744 804A0C8 i -> [ 1 ] 00001
6492: 400D4748 400C5E60 . -> 1 [ 0 ]
6493: 400D474C 8049D0C LOOP -> [ 0 ]
6494: 400D4758 804B384 ; -> ok
6495: @end example
6496:
6497: Each line displayed is one step. You always have to hit return to
6498: execute the next word that is displayed. If you don't want to execute
6499: the next word in a whole, you have to type @kbd{n} for @code{nest}. Here is
6500: an overview what keys are available:
6501:
6502: @table @i
6503:
6504: @item @key{RET}
6505: Next; Execute the next word.
6506:
6507: @item n
6508: Nest; Single step through next word.
6509:
6510: @item u
6511: Unnest; Stop debugging and execute rest of word. If we got to this word
6512: with nest, continue debugging with the calling word.
6513:
6514: @item d
6515: Done; Stop debugging and execute rest.
6516:
6517: @item s
6518: Stop; Abort immediately.
6519:
6520: @end table
6521:
6522: Debugging large application with this mechanism is very difficult, because
6523: you have to nest very deeply into the program before the interesting part
6524: begins. This takes a lot of time.
6525:
6526: To do it more directly put a @code{BREAK:} command into your source code.
6527: When program execution reaches @code{BREAK:} the single step debugger is
6528: invoked and you have all the features described above.
6529:
6530: If you have more than one part to debug it is useful to know where the
6531: program has stopped at the moment. You can do this by the
6532: @code{BREAK" string"} command. This behaves like @code{BREAK:} except that
6533: string is typed out when the ``breakpoint'' is reached.
6534:
6535:
6536: doc-dbg
6537: doc-break:
6538: doc-break"
6539:
6540:
6541:
6542: @c -------------------------------------------------------------
6543: @node Assembler and Code Words, Threading Words, Programming Tools, Words
6544: @section Assembler and Code Words
6545: @cindex assembler
6546: @cindex code words
6547:
6548: Gforth provides some words for defining primitives (words written in
6549: machine code), and for defining the machine-code equivalent of
6550: @code{DOES>}-based defining words. However, the machine-independent
6551: nature of Gforth poses a few problems: First of all, Gforth runs on
6552: several architectures, so it can provide no standard assembler. What's
6553: worse is that the register allocation not only depends on the processor,
6554: but also on the @code{gcc} version and options used.
6555:
6556: The words that Gforth offers encapsulate some system dependences (e.g.,
6557: the header structure), so a system-independent assembler may be used in
6558: Gforth. If you do not have an assembler, you can compile machine code
6559: directly with @code{,} and @code{c,}@footnote{This isn't portable,
6560: because these words emit stuff in @i{data} space; it works because
6561: Gforth has unified code/data spaces. Assembler isn't likely to be
6562: portable anyway.}.
6563:
6564:
6565: doc-assembler
6566: doc-init-asm
6567: doc-code
6568: doc-end-code
6569: doc-;code
6570: doc-flush-icache
6571:
6572:
6573: If @code{flush-icache} does not work correctly, @code{code} words
6574: etc. will not work (reliably), either.
6575:
6576: The typical usage of these @code{code} words can be shown most easily by
6577: analogy to the equivalent high-level defining words:
6578:
6579: @example
6580: : foo code foo
6581: <high-level Forth words> <assembler>
6582: ; end-code
6583:
6584: : bar : bar
6585: <high-level Forth words> <high-level Forth words>
6586: CREATE CREATE
6587: <high-level Forth words> <high-level Forth words>
6588: DOES> ;code
6589: <high-level Forth words> <assembler>
6590: ; end-code
6591: @end example
6592:
6593: @code{flush-icache} is always present. The other words are rarely used
6594: and reside in @code{code.fs}, which is usually not loaded. You can load
6595: it with @code{require code.fs}.
6596:
6597: @cindex registers of the inner interpreter
6598: In the assembly code you will want to refer to the inner interpreter's
6599: registers (e.g., the data stack pointer) and you may want to use other
6600: registers for temporary storage. Unfortunately, the register allocation
6601: is installation-dependent.
6602:
6603: The easiest solution is to use explicit register declarations
6604: (@pxref{Explicit Reg Vars, , Variables in Specified Registers, gcc.info,
6605: GNU C Manual}) for all of the inner interpreter's registers: You have to
6606: compile Gforth with @code{-DFORCE_REG} (configure option
6607: @code{--enable-force-reg}) and the appropriate declarations must be
6608: present in the @code{machine.h} file (see @code{mips.h} for an example;
6609: you can find a full list of all declarable register symbols with
6610: @code{grep register engine.c}). If you give explicit registers to all
6611: variables that are declared at the beginning of @code{engine()}, you
6612: should be able to use the other caller-saved registers for temporary
6613: storage. Alternatively, you can use the @code{gcc} option
6614: @code{-ffixed-REG} (@pxref{Code Gen Options, , Options for Code
6615: Generation Conventions, gcc.info, GNU C Manual}) to reserve a register
6616: (however, this restriction on register allocation may slow Gforth
6617: significantly).
6618:
6619: If this solution is not viable (e.g., because @code{gcc} does not allow
6620: you to explicitly declare all the registers you need), you have to find
6621: out by looking at the code where the inner interpreter's registers
6622: reside and which registers can be used for temporary storage. You can
6623: get an assembly listing of the engine's code with @code{make engine.s}.
6624:
6625: In any case, it is good practice to abstract your assembly code from the
6626: actual register allocation. E.g., if the data stack pointer resides in
6627: register @code{$17}, create an alias for this register called @code{sp},
6628: and use that in your assembly code.
6629:
6630: @cindex code words, portable
6631: Another option for implementing normal and defining words efficiently
6632: is to add the desired functionality to the source of Gforth. For normal
6633: words you just have to edit @file{primitives} (@pxref{Automatic
6634: Generation}). Defining words (equivalent to @code{;CODE} words, for fast
6635: defined words) may require changes in @file{engine.c}, @file{kernel.fs},
6636: @file{prims2x.fs}, and possibly @file{cross.fs}.
6637:
6638:
6639: @c -------------------------------------------------------------
6640: @node Threading Words, Locals, Assembler and Code Words, Words
6641: @section Threading Words
6642: @cindex threading words
6643:
6644: @cindex code address
6645: These words provide access to code addresses and other threading stuff
6646: in Gforth (and, possibly, other interpretive Forths). It more or less
6647: abstracts away the differences between direct and indirect threading
6648: (and, for direct threading, the machine dependences). However, at
6649: present this wordset is still incomplete. It is also pretty low-level;
6650: some day it will hopefully be made unnecessary by an internals wordset
6651: that abstracts implementation details away completely.
6652:
6653:
6654: doc-threading-method
6655: doc->code-address
6656: doc->does-code
6657: doc-code-address!
6658: doc-does-code!
6659: doc-does-handler!
6660: doc-/does-handler
6661:
6662:
6663: The code addresses produced by various defining words are produced by
6664: the following words:
6665:
6666:
6667: doc-docol:
6668: doc-docon:
6669: doc-dovar:
6670: doc-douser:
6671: doc-dodefer:
6672: doc-dofield:
6673:
6674:
6675: You can recognize words defined by a @code{CREATE}...@code{DOES>} word
6676: with @code{>does-code}. If the word was defined in that way, the value
6677: returned is non-zero and identifies the @code{DOES>} used by the
6678: defining word.
6679: @comment TODO should that be ``identifies the xt of the DOES> ??''
6680:
6681: @c -------------------------------------------------------------
6682: @node Locals, Structures, Threading Words, Words
6683: @section Locals
6684: @cindex locals
6685:
6686: Local variables can make Forth programming more enjoyable and Forth
6687: programs easier to read. Unfortunately, the locals of ANS Forth are
6688: laden with restrictions. Therefore, we provide not only the ANS Forth
6689: locals wordset, but also our own, more powerful locals wordset (we
6690: implemented the ANS Forth locals wordset through our locals wordset).
6691:
6692: The ideas in this section have also been published in the paper
6693: @cite{Automatic Scoping of Local Variables} by M. Anton Ertl, presented
6694: at EuroForth '94; it is available at
6695: @*@url{http://www.complang.tuwien.ac.at/papers/ertl94l.ps.gz}.
6696:
6697: @menu
6698: * Gforth locals::
6699: * ANS Forth locals::
6700: @end menu
6701:
6702: @node Gforth locals, ANS Forth locals, Locals, Locals
6703: @subsection Gforth locals
6704: @cindex Gforth locals
6705: @cindex locals, Gforth style
6706:
6707: Locals can be defined with
6708:
6709: @example
6710: @{ local1 local2 ... -- comment @}
6711: @end example
6712: or
6713: @example
6714: @{ local1 local2 ... @}
6715: @end example
6716:
6717: E.g.,
6718: @example
6719: : max @{ n1 n2 -- n3 @}
6720: n1 n2 > if
6721: n1
6722: else
6723: n2
6724: endif ;
6725: @end example
6726:
6727: The similarity of locals definitions with stack comments is intended. A
6728: locals definition often replaces the stack comment of a word. The order
6729: of the locals corresponds to the order in a stack comment and everything
6730: after the @code{--} is really a comment.
6731:
6732: This similarity has one disadvantage: It is too easy to confuse locals
6733: declarations with stack comments, causing bugs and making them hard to
6734: find. However, this problem can be avoided by appropriate coding
6735: conventions: Do not use both notations in the same program. If you do,
6736: they should be distinguished using additional means, e.g. by position.
6737:
6738: @cindex types of locals
6739: @cindex locals types
6740: The name of the local may be preceded by a type specifier, e.g.,
6741: @code{F:} for a floating point value:
6742:
6743: @example
6744: : CX* @{ F: Ar F: Ai F: Br F: Bi -- Cr Ci @}
6745: \ complex multiplication
6746: Ar Br f* Ai Bi f* f-
6747: Ar Bi f* Ai Br f* f+ ;
6748: @end example
6749:
6750: @cindex flavours of locals
6751: @cindex locals flavours
6752: @cindex value-flavoured locals
6753: @cindex variable-flavoured locals
6754: Gforth currently supports cells (@code{W:}, @code{W^}), doubles
6755: (@code{D:}, @code{D^}), floats (@code{F:}, @code{F^}) and characters
6756: (@code{C:}, @code{C^}) in two flavours: a value-flavoured local (defined
6757: with @code{W:}, @code{D:} etc.) produces its value and can be changed
6758: with @code{TO}. A variable-flavoured local (defined with @code{W^} etc.)
6759: produces its address (which becomes invalid when the variable's scope is
6760: left). E.g., the standard word @code{emit} can be defined in terms of
6761: @code{type} like this:
6762:
6763: @example
6764: : emit @{ C^ char* -- @}
6765: char* 1 type ;
6766: @end example
6767:
6768: @cindex default type of locals
6769: @cindex locals, default type
6770: A local without type specifier is a @code{W:} local. Both flavours of
6771: locals are initialized with values from the data or FP stack.
6772:
6773: Currently there is no way to define locals with user-defined data
6774: structures, but we are working on it.
6775:
6776: Gforth allows defining locals everywhere in a colon definition. This
6777: poses the following questions:
6778:
6779: @menu
6780: * Where are locals visible by name?::
6781: * How long do locals live?::
6782: * Programming Style::
6783: * Implementation::
6784: @end menu
6785:
6786: @node Where are locals visible by name?, How long do locals live?, Gforth locals, Gforth locals
6787: @subsubsection Where are locals visible by name?
6788: @cindex locals visibility
6789: @cindex visibility of locals
6790: @cindex scope of locals
6791:
6792: Basically, the answer is that locals are visible where you would expect
6793: it in block-structured languages, and sometimes a little longer. If you
6794: want to restrict the scope of a local, enclose its definition in
6795: @code{SCOPE}...@code{ENDSCOPE}.
6796:
6797:
6798: doc-scope
6799: doc-endscope
6800:
6801:
6802: These words behave like control structure words, so you can use them
6803: with @code{CS-PICK} and @code{CS-ROLL} to restrict the scope in
6804: arbitrary ways.
6805:
6806: If you want a more exact answer to the visibility question, here's the
6807: basic principle: A local is visible in all places that can only be
6808: reached through the definition of the local@footnote{In compiler
6809: construction terminology, all places dominated by the definition of the
6810: local.}. In other words, it is not visible in places that can be reached
6811: without going through the definition of the local. E.g., locals defined
6812: in @code{IF}...@code{ENDIF} are visible until the @code{ENDIF}, locals
6813: defined in @code{BEGIN}...@code{UNTIL} are visible after the
6814: @code{UNTIL} (until, e.g., a subsequent @code{ENDSCOPE}).
6815:
6816: The reasoning behind this solution is: We want to have the locals
6817: visible as long as it is meaningful. The user can always make the
6818: visibility shorter by using explicit scoping. In a place that can
6819: only be reached through the definition of a local, the meaning of a
6820: local name is clear. In other places it is not: How is the local
6821: initialized at the control flow path that does not contain the
6822: definition? Which local is meant, if the same name is defined twice in
6823: two independent control flow paths?
6824:
6825: This should be enough detail for nearly all users, so you can skip the
6826: rest of this section. If you really must know all the gory details and
6827: options, read on.
6828:
6829: In order to implement this rule, the compiler has to know which places
6830: are unreachable. It knows this automatically after @code{AHEAD},
6831: @code{AGAIN}, @code{EXIT} and @code{LEAVE}; in other cases (e.g., after
6832: most @code{THROW}s), you can use the word @code{UNREACHABLE} to tell the
6833: compiler that the control flow never reaches that place. If
6834: @code{UNREACHABLE} is not used where it could, the only consequence is
6835: that the visibility of some locals is more limited than the rule above
6836: says. If @code{UNREACHABLE} is used where it should not (i.e., if you
6837: lie to the compiler), buggy code will be produced.
6838:
6839:
6840: doc-unreachable
6841:
6842:
6843: Another problem with this rule is that at @code{BEGIN}, the compiler
6844: does not know which locals will be visible on the incoming
6845: back-edge. All problems discussed in the following are due to this
6846: ignorance of the compiler (we discuss the problems using @code{BEGIN}
6847: loops as examples; the discussion also applies to @code{?DO} and other
6848: loops). Perhaps the most insidious example is:
6849: @example
6850: AHEAD
6851: BEGIN
6852: x
6853: [ 1 CS-ROLL ] THEN
6854: @{ x @}
6855: ...
6856: UNTIL
6857: @end example
6858:
6859: This should be legal according to the visibility rule. The use of
6860: @code{x} can only be reached through the definition; but that appears
6861: textually below the use.
6862:
6863: From this example it is clear that the visibility rules cannot be fully
6864: implemented without major headaches. Our implementation treats common
6865: cases as advertised and the exceptions are treated in a safe way: The
6866: compiler makes a reasonable guess about the locals visible after a
6867: @code{BEGIN}; if it is too pessimistic, the
6868: user will get a spurious error about the local not being defined; if the
6869: compiler is too optimistic, it will notice this later and issue a
6870: warning. In the case above the compiler would complain about @code{x}
6871: being undefined at its use. You can see from the obscure examples in
6872: this section that it takes quite unusual control structures to get the
6873: compiler into trouble, and even then it will often do fine.
6874:
6875: If the @code{BEGIN} is reachable from above, the most optimistic guess
6876: is that all locals visible before the @code{BEGIN} will also be
6877: visible after the @code{BEGIN}. This guess is valid for all loops that
6878: are entered only through the @code{BEGIN}, in particular, for normal
6879: @code{BEGIN}...@code{WHILE}...@code{REPEAT} and
6880: @code{BEGIN}...@code{UNTIL} loops and it is implemented in our
6881: compiler. When the branch to the @code{BEGIN} is finally generated by
6882: @code{AGAIN} or @code{UNTIL}, the compiler checks the guess and
6883: warns the user if it was too optimistic:
6884: @example
6885: IF
6886: @{ x @}
6887: BEGIN
6888: \ x ?
6889: [ 1 cs-roll ] THEN
6890: ...
6891: UNTIL
6892: @end example
6893:
6894: Here, @code{x} lives only until the @code{BEGIN}, but the compiler
6895: optimistically assumes that it lives until the @code{THEN}. It notices
6896: this difference when it compiles the @code{UNTIL} and issues a
6897: warning. The user can avoid the warning, and make sure that @code{x}
6898: is not used in the wrong area by using explicit scoping:
6899: @example
6900: IF
6901: SCOPE
6902: @{ x @}
6903: ENDSCOPE
6904: BEGIN
6905: [ 1 cs-roll ] THEN
6906: ...
6907: UNTIL
6908: @end example
6909:
6910: Since the guess is optimistic, there will be no spurious error messages
6911: about undefined locals.
6912:
6913: If the @code{BEGIN} is not reachable from above (e.g., after
6914: @code{AHEAD} or @code{EXIT}), the compiler cannot even make an
6915: optimistic guess, as the locals visible after the @code{BEGIN} may be
6916: defined later. Therefore, the compiler assumes that no locals are
6917: visible after the @code{BEGIN}. However, the user can use
6918: @code{ASSUME-LIVE} to make the compiler assume that the same locals are
6919: visible at the BEGIN as at the point where the top control-flow stack
6920: item was created.
6921:
6922:
6923: doc-assume-live
6924:
6925:
6926: @noindent
6927: E.g.,
6928: @example
6929: @{ x @}
6930: AHEAD
6931: ASSUME-LIVE
6932: BEGIN
6933: x
6934: [ 1 CS-ROLL ] THEN
6935: ...
6936: UNTIL
6937: @end example
6938:
6939: Other cases where the locals are defined before the @code{BEGIN} can be
6940: handled by inserting an appropriate @code{CS-ROLL} before the
6941: @code{ASSUME-LIVE} (and changing the control-flow stack manipulation
6942: behind the @code{ASSUME-LIVE}).
6943:
6944: Cases where locals are defined after the @code{BEGIN} (but should be
6945: visible immediately after the @code{BEGIN}) can only be handled by
6946: rearranging the loop. E.g., the ``most insidious'' example above can be
6947: arranged into:
6948: @example
6949: BEGIN
6950: @{ x @}
6951: ... 0=
6952: WHILE
6953: x
6954: REPEAT
6955: @end example
6956:
6957: @node How long do locals live?, Programming Style, Where are locals visible by name?, Gforth locals
6958: @subsubsection How long do locals live?
6959: @cindex locals lifetime
6960: @cindex lifetime of locals
6961:
6962: The right answer for the lifetime question would be: A local lives at
6963: least as long as it can be accessed. For a value-flavoured local this
6964: means: until the end of its visibility. However, a variable-flavoured
6965: local could be accessed through its address far beyond its visibility
6966: scope. Ultimately, this would mean that such locals would have to be
6967: garbage collected. Since this entails un-Forth-like implementation
6968: complexities, I adopted the same cowardly solution as some other
6969: languages (e.g., C): The local lives only as long as it is visible;
6970: afterwards its address is invalid (and programs that access it
6971: afterwards are erroneous).
6972:
6973: @node Programming Style, Implementation, How long do locals live?, Gforth locals
6974: @subsubsection Programming Style
6975: @cindex locals programming style
6976: @cindex programming style, locals
6977:
6978: The freedom to define locals anywhere has the potential to change
6979: programming styles dramatically. In particular, the need to use the
6980: return stack for intermediate storage vanishes. Moreover, all stack
6981: manipulations (except @code{PICK}s and @code{ROLL}s with run-time
6982: determined arguments) can be eliminated: If the stack items are in the
6983: wrong order, just write a locals definition for all of them; then
6984: write the items in the order you want.
6985:
6986: This seems a little far-fetched and eliminating stack manipulations is
6987: unlikely to become a conscious programming objective. Still, the number
6988: of stack manipulations will be reduced dramatically if local variables
6989: are used liberally (e.g., compare @code{max} in @ref{Gforth locals} with
6990: a traditional implementation of @code{max}).
6991:
6992: This shows one potential benefit of locals: making Forth programs more
6993: readable. Of course, this benefit will only be realized if the
6994: programmers continue to honour the principle of factoring instead of
6995: using the added latitude to make the words longer.
6996:
6997: @cindex single-assignment style for locals
6998: Using @code{TO} can and should be avoided. Without @code{TO},
6999: every value-flavoured local has only a single assignment and many
7000: advantages of functional languages apply to Forth. I.e., programs are
7001: easier to analyse, to optimize and to read: It is clear from the
7002: definition what the local stands for, it does not turn into something
7003: different later.
7004:
7005: E.g., a definition using @code{TO} might look like this:
7006: @example
7007: : strcmp @{ addr1 u1 addr2 u2 -- n @}
7008: u1 u2 min 0
7009: ?do
7010: addr1 c@@ addr2 c@@ -
7011: ?dup-if
7012: unloop exit
7013: then
7014: addr1 char+ TO addr1
7015: addr2 char+ TO addr2
7016: loop
7017: u1 u2 - ;
7018: @end example
7019: Here, @code{TO} is used to update @code{addr1} and @code{addr2} at
7020: every loop iteration. @code{strcmp} is a typical example of the
7021: readability problems of using @code{TO}. When you start reading
7022: @code{strcmp}, you think that @code{addr1} refers to the start of the
7023: string. Only near the end of the loop you realize that it is something
7024: else.
7025:
7026: This can be avoided by defining two locals at the start of the loop that
7027: are initialized with the right value for the current iteration.
7028: @example
7029: : strcmp @{ addr1 u1 addr2 u2 -- n @}
7030: addr1 addr2
7031: u1 u2 min 0
7032: ?do @{ s1 s2 @}
7033: s1 c@@ s2 c@@ -
7034: ?dup-if
7035: unloop exit
7036: then
7037: s1 char+ s2 char+
7038: loop
7039: 2drop
7040: u1 u2 - ;
7041: @end example
7042: Here it is clear from the start that @code{s1} has a different value
7043: in every loop iteration.
7044:
7045: @node Implementation, , Programming Style, Gforth locals
7046: @subsubsection Implementation
7047: @cindex locals implementation
7048: @cindex implementation of locals
7049:
7050: @cindex locals stack
7051: Gforth uses an extra locals stack. The most compelling reason for
7052: this is that the return stack is not float-aligned; using an extra stack
7053: also eliminates the problems and restrictions of using the return stack
7054: as locals stack. Like the other stacks, the locals stack grows toward
7055: lower addresses. A few primitives allow an efficient implementation:
7056:
7057:
7058: doc-@local#
7059: doc-f@local#
7060: doc-laddr#
7061: doc-lp+!#
7062: doc-lp!
7063: doc->l
7064: doc-f>l
7065:
7066:
7067: In addition to these primitives, some specializations of these
7068: primitives for commonly occurring inline arguments are provided for
7069: efficiency reasons, e.g., @code{@@local0} as specialization of
7070: @code{@@local#} for the inline argument 0. The following compiling words
7071: compile the right specialized version, or the general version, as
7072: appropriate:
7073:
7074:
7075: doc-compile-@local
7076: doc-compile-f@local
7077: doc-compile-lp+!
7078:
7079:
7080: Combinations of conditional branches and @code{lp+!#} like
7081: @code{?branch-lp+!#} (the locals pointer is only changed if the branch
7082: is taken) are provided for efficiency and correctness in loops.
7083:
7084: A special area in the dictionary space is reserved for keeping the
7085: local variable names. @code{@{} switches the dictionary pointer to this
7086: area and @code{@}} switches it back and generates the locals
7087: initializing code. @code{W:} etc.@ are normal defining words. This
7088: special area is cleared at the start of every colon definition.
7089:
7090: @cindex word list for defining locals
7091: A special feature of Gforth's dictionary is used to implement the
7092: definition of locals without type specifiers: every word list (aka
7093: vocabulary) has its own methods for searching
7094: etc. (@pxref{Word Lists}). For the present purpose we defined a word list
7095: with a special search method: When it is searched for a word, it
7096: actually creates that word using @code{W:}. @code{@{} changes the search
7097: order to first search the word list containing @code{@}}, @code{W:} etc.,
7098: and then the word list for defining locals without type specifiers.
7099:
7100: The lifetime rules support a stack discipline within a colon
7101: definition: The lifetime of a local is either nested with other locals
7102: lifetimes or it does not overlap them.
7103:
7104: At @code{BEGIN}, @code{IF}, and @code{AHEAD} no code for locals stack
7105: pointer manipulation is generated. Between control structure words
7106: locals definitions can push locals onto the locals stack. @code{AGAIN}
7107: is the simplest of the other three control flow words. It has to
7108: restore the locals stack depth of the corresponding @code{BEGIN}
7109: before branching. The code looks like this:
7110: @format
7111: @code{lp+!#} current-locals-size @minus{} dest-locals-size
7112: @code{branch} <begin>
7113: @end format
7114:
7115: @code{UNTIL} is a little more complicated: If it branches back, it
7116: must adjust the stack just like @code{AGAIN}. But if it falls through,
7117: the locals stack must not be changed. The compiler generates the
7118: following code:
7119: @format
7120: @code{?branch-lp+!#} <begin> current-locals-size @minus{} dest-locals-size
7121: @end format
7122: The locals stack pointer is only adjusted if the branch is taken.
7123:
7124: @code{THEN} can produce somewhat inefficient code:
7125: @format
7126: @code{lp+!#} current-locals-size @minus{} orig-locals-size
7127: <orig target>:
7128: @code{lp+!#} orig-locals-size @minus{} new-locals-size
7129: @end format
7130: The second @code{lp+!#} adjusts the locals stack pointer from the
7131: level at the @i{orig} point to the level after the @code{THEN}. The
7132: first @code{lp+!#} adjusts the locals stack pointer from the current
7133: level to the level at the orig point, so the complete effect is an
7134: adjustment from the current level to the right level after the
7135: @code{THEN}.
7136:
7137: @cindex locals information on the control-flow stack
7138: @cindex control-flow stack items, locals information
7139: In a conventional Forth implementation a dest control-flow stack entry
7140: is just the target address and an orig entry is just the address to be
7141: patched. Our locals implementation adds a word list to every orig or dest
7142: item. It is the list of locals visible (or assumed visible) at the point
7143: described by the entry. Our implementation also adds a tag to identify
7144: the kind of entry, in particular to differentiate between live and dead
7145: (reachable and unreachable) orig entries.
7146:
7147: A few unusual operations have to be performed on locals word lists:
7148:
7149:
7150: doc-common-list
7151: doc-sub-list?
7152: doc-list-size
7153:
7154:
7155: Several features of our locals word list implementation make these
7156: operations easy to implement: The locals word lists are organised as
7157: linked lists; the tails of these lists are shared, if the lists
7158: contain some of the same locals; and the address of a name is greater
7159: than the address of the names behind it in the list.
7160:
7161: Another important implementation detail is the variable
7162: @code{dead-code}. It is used by @code{BEGIN} and @code{THEN} to
7163: determine if they can be reached directly or only through the branch
7164: that they resolve. @code{dead-code} is set by @code{UNREACHABLE},
7165: @code{AHEAD}, @code{EXIT} etc., and cleared at the start of a colon
7166: definition, by @code{BEGIN} and usually by @code{THEN}.
7167:
7168: Counted loops are similar to other loops in most respects, but
7169: @code{LEAVE} requires special attention: It performs basically the same
7170: service as @code{AHEAD}, but it does not create a control-flow stack
7171: entry. Therefore the information has to be stored elsewhere;
7172: traditionally, the information was stored in the target fields of the
7173: branches created by the @code{LEAVE}s, by organizing these fields into a
7174: linked list. Unfortunately, this clever trick does not provide enough
7175: space for storing our extended control flow information. Therefore, we
7176: introduce another stack, the leave stack. It contains the control-flow
7177: stack entries for all unresolved @code{LEAVE}s.
7178:
7179: Local names are kept until the end of the colon definition, even if
7180: they are no longer visible in any control-flow path. In a few cases
7181: this may lead to increased space needs for the locals name area, but
7182: usually less than reclaiming this space would cost in code size.
7183:
7184:
7185: @node ANS Forth locals, , Gforth locals, Locals
7186: @subsection ANS Forth locals
7187: @cindex locals, ANS Forth style
7188:
7189: The ANS Forth locals wordset does not define a syntax for locals, but
7190: words that make it possible to define various syntaxes. One of the
7191: possible syntaxes is a subset of the syntax we used in the Gforth locals
7192: wordset, i.e.:
7193:
7194: @example
7195: @{ local1 local2 ... -- comment @}
7196: @end example
7197: @noindent
7198: or
7199: @example
7200: @{ local1 local2 ... @}
7201: @end example
7202:
7203: The order of the locals corresponds to the order in a stack comment. The
7204: restrictions are:
7205:
7206: @itemize @bullet
7207: @item
7208: Locals can only be cell-sized values (no type specifiers are allowed).
7209: @item
7210: Locals can be defined only outside control structures.
7211: @item
7212: Locals can interfere with explicit usage of the return stack. For the
7213: exact (and long) rules, see the standard. If you don't use return stack
7214: accessing words in a definition using locals, you will be all right. The
7215: purpose of this rule is to make locals implementation on the return
7216: stack easier.
7217: @item
7218: The whole definition must be in one line.
7219: @end itemize
7220:
7221: Locals defined in this way behave like @code{VALUE}s
7222: (@xref{Values}). I.e., they are initialized from the stack. Using their
7223: name produces their value. Their value can be changed using @code{TO}.
7224:
7225: Since this syntax is supported by Gforth directly, you need not do
7226: anything to use it. If you want to port a program using this syntax to
7227: another ANS Forth system, use @file{compat/anslocal.fs} to implement the
7228: syntax on the other system.
7229:
7230: Note that a syntax shown in the standard, section A.13 looks
7231: similar, but is quite different in having the order of locals
7232: reversed. Beware!
7233:
7234: The ANS Forth locals wordset itself consists of a word:
7235:
7236:
7237: doc-(local)
7238:
7239:
7240: The ANS Forth locals extension wordset defines a syntax using @code{locals|}, but it is so
7241: awful that we strongly recommend not to use it. We have implemented this
7242: syntax to make porting to Gforth easy, but do not document it here. The
7243: problem with this syntax is that the locals are defined in an order
7244: reversed with respect to the standard stack comment notation, making
7245: programs harder to read, and easier to misread and miswrite. The only
7246: merit of this syntax is that it is easy to implement using the ANS Forth
7247: locals wordset.
7248:
7249:
7250: @c ----------------------------------------------------------
7251: @node Structures, Object-oriented Forth, Locals, Words
7252: @section Structures
7253: @cindex structures
7254: @cindex records
7255:
7256: This section presents the structure package that comes with Gforth. A
7257: version of the package implemented in ANS Forth is available in
7258: @file{compat/struct.fs}. This package was inspired by a posting on
7259: comp.lang.forth in 1989 (unfortunately I don't remember, by whom;
7260: possibly John Hayes). A version of this section has been published in
7261: ???. Marcel Hendrix provided helpful comments.
7262:
7263: @menu
7264: * Why explicit structure support?::
7265: * Structure Usage::
7266: * Structure Naming Convention::
7267: * Structure Implementation::
7268: * Structure Glossary::
7269: @end menu
7270:
7271: @node Why explicit structure support?, Structure Usage, Structures, Structures
7272: @subsection Why explicit structure support?
7273:
7274: @cindex address arithmetic for structures
7275: @cindex structures using address arithmetic
7276: If we want to use a structure containing several fields, we could simply
7277: reserve memory for it, and access the fields using address arithmetic
7278: (@pxref{Address arithmetic}). As an example, consider a structure with
7279: the following fields
7280:
7281: @table @code
7282: @item a
7283: is a float
7284: @item b
7285: is a cell
7286: @item c
7287: is a float
7288: @end table
7289:
7290: Given the (float-aligned) base address of the structure we get the
7291: address of the field
7292:
7293: @table @code
7294: @item a
7295: without doing anything further.
7296: @item b
7297: with @code{float+}
7298: @item c
7299: with @code{float+ cell+ faligned}
7300: @end table
7301:
7302: It is easy to see that this can become quite tiring.
7303:
7304: Moreover, it is not very readable, because seeing a
7305: @code{cell+} tells us neither which kind of structure is
7306: accessed nor what field is accessed; we have to somehow infer the kind
7307: of structure, and then look up in the documentation, which field of
7308: that structure corresponds to that offset.
7309:
7310: Finally, this kind of address arithmetic also causes maintenance
7311: troubles: If you add or delete a field somewhere in the middle of the
7312: structure, you have to find and change all computations for the fields
7313: afterwards.
7314:
7315: So, instead of using @code{cell+} and friends directly, how
7316: about storing the offsets in constants:
7317:
7318: @example
7319: 0 constant a-offset
7320: 0 float+ constant b-offset
7321: 0 float+ cell+ faligned c-offset
7322: @end example
7323:
7324: Now we can get the address of field @code{x} with @code{x-offset
7325: +}. This is much better in all respects. Of course, you still
7326: have to change all later offset definitions if you add a field. You can
7327: fix this by declaring the offsets in the following way:
7328:
7329: @example
7330: 0 constant a-offset
7331: a-offset float+ constant b-offset
7332: b-offset cell+ faligned constant c-offset
7333: @end example
7334:
7335: Since we always use the offsets with @code{+}, we could use a defining
7336: word @code{cfield} that includes the @code{+} in the action of the
7337: defined word:
7338:
7339: @example
7340: : cfield ( n "name" -- )
7341: create ,
7342: does> ( name execution: addr1 -- addr2 )
7343: @@ + ;
7344:
7345: 0 cfield a
7346: 0 a float+ cfield b
7347: 0 b cell+ faligned cfield c
7348: @end example
7349:
7350: Instead of @code{x-offset +}, we now simply write @code{x}.
7351:
7352: The structure field words now can be used quite nicely. However,
7353: their definition is still a bit cumbersome: We have to repeat the
7354: name, the information about size and alignment is distributed before
7355: and after the field definitions etc. The structure package presented
7356: here addresses these problems.
7357:
7358: @node Structure Usage, Structure Naming Convention, Why explicit structure support?, Structures
7359: @subsection Structure Usage
7360: @cindex structure usage
7361:
7362: @cindex @code{field} usage
7363: @cindex @code{struct} usage
7364: @cindex @code{end-struct} usage
7365: You can define a structure for a (data-less) linked list with:
7366: @example
7367: struct
7368: cell% field list-next
7369: end-struct list%
7370: @end example
7371:
7372: With the address of the list node on the stack, you can compute the
7373: address of the field that contains the address of the next node with
7374: @code{list-next}. E.g., you can determine the length of a list
7375: with:
7376:
7377: @example
7378: : list-length ( list -- n )
7379: \ "list" is a pointer to the first element of a linked list
7380: \ "n" is the length of the list
7381: 0 BEGIN ( list1 n1 )
7382: over
7383: WHILE ( list1 n1 )
7384: 1+ swap list-next @@ swap
7385: REPEAT
7386: nip ;
7387: @end example
7388:
7389: You can reserve memory for a list node in the dictionary with
7390: @code{list% %allot}, which leaves the address of the list node on the
7391: stack. For the equivalent allocation on the heap you can use @code{list%
7392: %alloc} (or, for an @code{allocate}-like stack effect (i.e., with ior),
7393: use @code{list% %allocate}). You can get the the size of a list
7394: node with @code{list% %size} and its alignment with @code{list%
7395: %alignment}.
7396:
7397: Note that in ANS Forth the body of a @code{create}d word is
7398: @code{aligned} but not necessarily @code{faligned};
7399: therefore, if you do a:
7400: @example
7401: create @emph{name} foo% %allot
7402: @end example
7403:
7404: @noindent
7405: then the memory alloted for @code{foo%} is
7406: guaranteed to start at the body of @code{@emph{name}} only if
7407: @code{foo%} contains only character, cell and double fields.
7408:
7409: @cindex structures containing structures
7410: You can include a structure @code{foo%} as a field of
7411: another structure, like this:
7412: @example
7413: struct
7414: ...
7415: foo% field ...
7416: ...
7417: end-struct ...
7418: @end example
7419:
7420: @cindex structure extension
7421: @cindex extended records
7422: Instead of starting with an empty structure, you can extend an
7423: existing structure. E.g., a plain linked list without data, as defined
7424: above, is hardly useful; You can extend it to a linked list of integers,
7425: like this:@footnote{This feature is also known as @emph{extended
7426: records}. It is the main innovation in the Oberon language; in other
7427: words, adding this feature to Modula-2 led Wirth to create a new
7428: language, write a new compiler etc. Adding this feature to Forth just
7429: required a few lines of code.}
7430:
7431: @example
7432: list%
7433: cell% field intlist-int
7434: end-struct intlist%
7435: @end example
7436:
7437: @code{intlist%} is a structure with two fields:
7438: @code{list-next} and @code{intlist-int}.
7439:
7440: @cindex structures containing arrays
7441: You can specify an array type containing @emph{n} elements of
7442: type @code{foo%} like this:
7443:
7444: @example
7445: foo% @emph{n} *
7446: @end example
7447:
7448: You can use this array type in any place where you can use a normal
7449: type, e.g., when defining a @code{field}, or with
7450: @code{%allot}.
7451:
7452: @cindex first field optimization
7453: The first field is at the base address of a structure and the word
7454: for this field (e.g., @code{list-next}) actually does not change
7455: the address on the stack. You may be tempted to leave it away in the
7456: interest of run-time and space efficiency. This is not necessary,
7457: because the structure package optimizes this case and compiling such
7458: words does not generate any code. So, in the interest of readability
7459: and maintainability you should include the word for the field when
7460: accessing the field.
7461:
7462: @node Structure Naming Convention, Structure Implementation, Structure Usage, Structures
7463: @subsection Structure Naming Convention
7464: @cindex structure naming convention
7465:
7466: The field names that come to (my) mind are often quite generic, and,
7467: if used, would cause frequent name clashes. E.g., many structures
7468: probably contain a @code{counter} field. The structure names
7469: that come to (my) mind are often also the logical choice for the names
7470: of words that create such a structure.
7471:
7472: Therefore, I have adopted the following naming conventions:
7473:
7474: @itemize @bullet
7475: @cindex field naming convention
7476: @item
7477: The names of fields are of the form
7478: @code{@emph{struct}-@emph{field}}, where
7479: @code{@emph{struct}} is the basic name of the structure, and
7480: @code{@emph{field}} is the basic name of the field. You can
7481: think of field words as converting the (address of the)
7482: structure into the (address of the) field.
7483:
7484: @cindex structure naming convention
7485: @item
7486: The names of structures are of the form
7487: @code{@emph{struct}%}, where
7488: @code{@emph{struct}} is the basic name of the structure.
7489: @end itemize
7490:
7491: This naming convention does not work that well for fields of extended
7492: structures; e.g., the integer list structure has a field
7493: @code{intlist-int}, but has @code{list-next}, not
7494: @code{intlist-next}.
7495:
7496: @node Structure Implementation, Structure Glossary, Structure Naming Convention, Structures
7497: @subsection Structure Implementation
7498: @cindex structure implementation
7499: @cindex implementation of structures
7500:
7501: The central idea in the implementation is to pass the data about the
7502: structure being built on the stack, not in some global
7503: variable. Everything else falls into place naturally once this design
7504: decision is made.
7505:
7506: The type description on the stack is of the form @emph{align
7507: size}. Keeping the size on the top-of-stack makes dealing with arrays
7508: very simple.
7509:
7510: @code{field} is a defining word that uses @code{Create}
7511: and @code{DOES>}. The body of the field contains the offset
7512: of the field, and the normal @code{DOES>} action is simply:
7513:
7514: @example
7515: @ +
7516: @end example
7517:
7518: @noindent
7519: i.e., add the offset to the address, giving the stack effect
7520: @i{addr1 -- addr2} for a field.
7521:
7522: @cindex first field optimization, implementation
7523: This simple structure is slightly complicated by the optimization
7524: for fields with offset 0, which requires a different
7525: @code{DOES>}-part (because we cannot rely on there being
7526: something on the stack if such a field is invoked during
7527: compilation). Therefore, we put the different @code{DOES>}-parts
7528: in separate words, and decide which one to invoke based on the
7529: offset. For a zero offset, the field is basically a noop; it is
7530: immediate, and therefore no code is generated when it is compiled.
7531:
7532: @node Structure Glossary, , Structure Implementation, Structures
7533: @subsection Structure Glossary
7534: @cindex structure glossary
7535:
7536:
7537: doc-%align
7538: doc-%alignment
7539: doc-%alloc
7540: doc-%allocate
7541: doc-%allot
7542: doc-cell%
7543: doc-char%
7544: doc-dfloat%
7545: doc-double%
7546: doc-end-struct
7547: doc-field
7548: doc-float%
7549: doc-naligned
7550: doc-sfloat%
7551: doc-%size
7552: doc-struct
7553:
7554:
7555: @c -------------------------------------------------------------
7556: @node Object-oriented Forth, Passing Commands to the OS, Structures, Words
7557: @section Object-oriented Forth
7558:
7559: Gforth comes with three packages for object-oriented programming:
7560: @file{objects.fs}, @file{oof.fs}, and @file{mini-oof.fs}; none of them
7561: is preloaded, so you have to @code{include} them before use. The most
7562: important differences between these packages (and others) are discussed
7563: in @ref{Comparison with other object models}. All packages are written
7564: in ANS Forth and can be used with any other ANS Forth.
7565:
7566: @menu
7567: * Why object-oriented programming?::
7568: * Object-Oriented Terminology::
7569: * Objects::
7570: * OOF::
7571: * Mini-OOF::
7572: * Comparison with other object models::
7573: @end menu
7574:
7575:
7576: @node Why object-oriented programming?, Object-Oriented Terminology, , Object-oriented Forth
7577: @subsubsection Why object-oriented programming?
7578: @cindex object-oriented programming motivation
7579: @cindex motivation for object-oriented programming
7580:
7581: Often we have to deal with several data structures (@emph{objects}),
7582: that have to be treated similarly in some respects, but differently in
7583: others. Graphical objects are the textbook example: circles, triangles,
7584: dinosaurs, icons, and others, and we may want to add more during program
7585: development. We want to apply some operations to any graphical object,
7586: e.g., @code{draw} for displaying it on the screen. However, @code{draw}
7587: has to do something different for every kind of object.
7588: @comment TODO add some other operations eg perimeter, area
7589: @comment and tie in to concrete examples later..
7590:
7591: We could implement @code{draw} as a big @code{CASE}
7592: control structure that executes the appropriate code depending on the
7593: kind of object to be drawn. This would be not be very elegant, and,
7594: moreover, we would have to change @code{draw} every time we add
7595: a new kind of graphical object (say, a spaceship).
7596:
7597: What we would rather do is: When defining spaceships, we would tell
7598: the system: ``Here's how you @code{draw} a spaceship; you figure
7599: out the rest''.
7600:
7601: This is the problem that all systems solve that (rightfully) call
7602: themselves object-oriented; the object-oriented packages presented here
7603: solve this problem (and not much else).
7604: @comment TODO ?list properties of oo systems.. oo vs o-based?
7605:
7606: @node Object-Oriented Terminology, Objects, Why object-oriented programming?, Object-oriented Forth
7607: @subsubsection Object-Oriented Terminology
7608: @cindex object-oriented terminology
7609: @cindex terminology for object-oriented programming
7610:
7611: This section is mainly for reference, so you don't have to understand
7612: all of it right away. The terminology is mainly Smalltalk-inspired. In
7613: short:
7614:
7615: @table @emph
7616: @cindex class
7617: @item class
7618: a data structure definition with some extras.
7619:
7620: @cindex object
7621: @item object
7622: an instance of the data structure described by the class definition.
7623:
7624: @cindex instance variables
7625: @item instance variables
7626: fields of the data structure.
7627:
7628: @cindex selector
7629: @cindex method selector
7630: @cindex virtual function
7631: @item selector
7632: (or @emph{method selector}) a word (e.g.,
7633: @code{draw}) that performs an operation on a variety of data
7634: structures (classes). A selector describes @emph{what} operation to
7635: perform. In C++ terminology: a (pure) virtual function.
7636:
7637: @cindex method
7638: @item method
7639: the concrete definition that performs the operation
7640: described by the selector for a specific class. A method specifies
7641: @emph{how} the operation is performed for a specific class.
7642:
7643: @cindex selector invocation
7644: @cindex message send
7645: @cindex invoking a selector
7646: @item selector invocation
7647: a call of a selector. One argument of the call (the TOS (top-of-stack))
7648: is used for determining which method is used. In Smalltalk terminology:
7649: a message (consisting of the selector and the other arguments) is sent
7650: to the object.
7651:
7652: @cindex receiving object
7653: @item receiving object
7654: the object used for determining the method executed by a selector
7655: invocation. In the @file{objects.fs} model, it is the object that is on
7656: the TOS when the selector is invoked. (@emph{Receiving} comes from
7657: the Smalltalk @emph{message} terminology.)
7658:
7659: @cindex child class
7660: @cindex parent class
7661: @cindex inheritance
7662: @item child class
7663: a class that has (@emph{inherits}) all properties (instance variables,
7664: selectors, methods) from a @emph{parent class}. In Smalltalk
7665: terminology: The subclass inherits from the superclass. In C++
7666: terminology: The derived class inherits from the base class.
7667:
7668: @end table
7669:
7670: @c If you wonder about the message sending terminology, it comes from
7671: @c a time when each object had it's own task and objects communicated via
7672: @c message passing; eventually the Smalltalk developers realized that
7673: @c they can do most things through simple (indirect) calls. They kept the
7674: @c terminology.
7675:
7676:
7677: @node Objects, OOF, Object-Oriented Terminology, Object-oriented Forth
7678: @subsection The @file{objects.fs} model
7679: @cindex objects
7680: @cindex object-oriented programming
7681:
7682: @cindex @file{objects.fs}
7683: @cindex @file{oof.fs}
7684:
7685: This section describes the @file{objects.fs} package. This material also
7686: has been published in @cite{Yet Another Forth Objects Package} by Anton
7687: Ertl and appeared in Forth Dimensions 19(2), pages 37--43
7688: (@url{http://www.complang.tuwien.ac.at/forth/objects/objects.html}).
7689: @c McKewan's and Zsoter's packages
7690:
7691: This section assumes that you have read @ref{Structures}.
7692:
7693: The techniques on which this model is based have been used to implement
7694: the parser generator, Gray, and have also been used in Gforth for
7695: implementing the various flavours of word lists (hashed or not,
7696: case-sensitive or not, special-purpose word lists for locals etc.).
7697:
7698:
7699: @menu
7700: * Properties of the Objects model::
7701: * Basic Objects Usage::
7702: * The Objects base class::
7703: * Creating objects::
7704: * Object-Oriented Programming Style::
7705: * Class Binding::
7706: * Method conveniences::
7707: * Classes and Scoping::
7708: * Dividing classes::
7709: * Object Interfaces::
7710: * Objects Implementation::
7711: * Objects Glossary::
7712: @end menu
7713:
7714: Marcel Hendrix provided helpful comments on this section. Andras Zsoter
7715: and Bernd Paysan helped me with the related works section.
7716:
7717: @node Properties of the Objects model, Basic Objects Usage, Objects, Objects
7718: @subsubsection Properties of the @file{objects.fs} model
7719: @cindex @file{objects.fs} properties
7720:
7721: @itemize @bullet
7722: @item
7723: It is straightforward to pass objects on the stack. Passing
7724: selectors on the stack is a little less convenient, but possible.
7725:
7726: @item
7727: Objects are just data structures in memory, and are referenced by their
7728: address. You can create words for objects with normal defining words
7729: like @code{constant}. Likewise, there is no difference between instance
7730: variables that contain objects and those that contain other data.
7731:
7732: @item
7733: Late binding is efficient and easy to use.
7734:
7735: @item
7736: It avoids parsing, and thus avoids problems with state-smartness
7737: and reduced extensibility; for convenience there are a few parsing
7738: words, but they have non-parsing counterparts. There are also a few
7739: defining words that parse. This is hard to avoid, because all standard
7740: defining words parse (except @code{:noname}); however, such
7741: words are not as bad as many other parsing words, because they are not
7742: state-smart.
7743:
7744: @item
7745: It does not try to incorporate everything. It does a few things and does
7746: them well (IMO). In particular, this model was not designed to support
7747: information hiding (although it has features that may help); you can use
7748: a separate package for achieving this.
7749:
7750: @item
7751: It is layered; you don't have to learn and use all features to use this
7752: model. Only a few features are necessary (@xref{Basic Objects Usage},
7753: @xref{The Objects base class}, @xref{Creating objects}.), the others
7754: are optional and independent of each other.
7755:
7756: @item
7757: An implementation in ANS Forth is available.
7758:
7759: @end itemize
7760:
7761:
7762: @node Basic Objects Usage, The Objects base class, Properties of the Objects model, Objects
7763: @subsubsection Basic @file{objects.fs} Usage
7764: @cindex basic objects usage
7765: @cindex objects, basic usage
7766:
7767: You can define a class for graphical objects like this:
7768:
7769: @cindex @code{class} usage
7770: @cindex @code{end-class} usage
7771: @cindex @code{selector} usage
7772: @example
7773: object class \ "object" is the parent class
7774: selector draw ( x y graphical -- )
7775: end-class graphical
7776: @end example
7777:
7778: This code defines a class @code{graphical} with an
7779: operation @code{draw}. We can perform the operation
7780: @code{draw} on any @code{graphical} object, e.g.:
7781:
7782: @example
7783: 100 100 t-rex draw
7784: @end example
7785:
7786: @noindent
7787: where @code{t-rex} is a word (say, a constant) that produces a
7788: graphical object.
7789:
7790: @comment TODO add a 2nd operation eg perimeter.. and use for
7791: @comment a concrete example
7792:
7793: @cindex abstract class
7794: How do we create a graphical object? With the present definitions,
7795: we cannot create a useful graphical object. The class
7796: @code{graphical} describes graphical objects in general, but not
7797: any concrete graphical object type (C++ users would call it an
7798: @emph{abstract class}); e.g., there is no method for the selector
7799: @code{draw} in the class @code{graphical}.
7800:
7801: For concrete graphical objects, we define child classes of the
7802: class @code{graphical}, e.g.:
7803:
7804: @cindex @code{overrides} usage
7805: @cindex @code{field} usage in class definition
7806: @example
7807: graphical class \ "graphical" is the parent class
7808: cell% field circle-radius
7809:
7810: :noname ( x y circle -- )
7811: circle-radius @@ draw-circle ;
7812: overrides draw
7813:
7814: :noname ( n-radius circle -- )
7815: circle-radius ! ;
7816: overrides construct
7817:
7818: end-class circle
7819: @end example
7820:
7821: Here we define a class @code{circle} as a child of @code{graphical},
7822: with field @code{circle-radius} (which behaves just like a field
7823: (@pxref{Structures}); it defines (using @code{overrides}) new methods
7824: for the selectors @code{draw} and @code{construct} (@code{construct} is
7825: defined in @code{object}, the parent class of @code{graphical}).
7826:
7827: Now we can create a circle on the heap (i.e.,
7828: @code{allocate}d memory) with:
7829:
7830: @cindex @code{heap-new} usage
7831: @example
7832: 50 circle heap-new constant my-circle
7833: @end example
7834:
7835: @noindent
7836: @code{heap-new} invokes @code{construct}, thus
7837: initializing the field @code{circle-radius} with 50. We can draw
7838: this new circle at (100,100) with:
7839:
7840: @example
7841: 100 100 my-circle draw
7842: @end example
7843:
7844: @cindex selector invocation, restrictions
7845: @cindex class definition, restrictions
7846: Note: You can only invoke a selector if the object on the TOS
7847: (the receiving object) belongs to the class where the selector was
7848: defined or one of its descendents; e.g., you can invoke
7849: @code{draw} only for objects belonging to @code{graphical}
7850: or its descendents (e.g., @code{circle}). Immediately before
7851: @code{end-class}, the search order has to be the same as
7852: immediately after @code{class}.
7853:
7854: @node The Objects base class, Creating objects, Basic Objects Usage, Objects
7855: @subsubsection The @file{object.fs} base class
7856: @cindex @code{object} class
7857:
7858: When you define a class, you have to specify a parent class. So how do
7859: you start defining classes? There is one class available from the start:
7860: @code{object}. It is ancestor for all classes and so is the
7861: only class that has no parent. It has two selectors: @code{construct}
7862: and @code{print}.
7863:
7864: @node Creating objects, Object-Oriented Programming Style, The Objects base class, Objects
7865: @subsubsection Creating objects
7866: @cindex creating objects
7867: @cindex object creation
7868: @cindex object allocation options
7869:
7870: @cindex @code{heap-new} discussion
7871: @cindex @code{dict-new} discussion
7872: @cindex @code{construct} discussion
7873: You can create and initialize an object of a class on the heap with
7874: @code{heap-new} ( ... class -- object ) and in the dictionary
7875: (allocation with @code{allot}) with @code{dict-new} (
7876: ... class -- object ). Both words invoke @code{construct}, which
7877: consumes the stack items indicated by "..." above.
7878:
7879: @cindex @code{init-object} discussion
7880: @cindex @code{class-inst-size} discussion
7881: If you want to allocate memory for an object yourself, you can get its
7882: alignment and size with @code{class-inst-size 2@@} ( class --
7883: align size ). Once you have memory for an object, you can initialize
7884: it with @code{init-object} ( ... class object -- );
7885: @code{construct} does only a part of the necessary work.
7886:
7887: @node Object-Oriented Programming Style, Class Binding, Creating objects, Objects
7888: @subsubsection Object-Oriented Programming Style
7889: @cindex object-oriented programming style
7890:
7891: This section is not exhaustive.
7892:
7893: @cindex stack effects of selectors
7894: @cindex selectors and stack effects
7895: In general, it is a good idea to ensure that all methods for the
7896: same selector have the same stack effect: when you invoke a selector,
7897: you often have no idea which method will be invoked, so, unless all
7898: methods have the same stack effect, you will not know the stack effect
7899: of the selector invocation.
7900:
7901: One exception to this rule is methods for the selector
7902: @code{construct}. We know which method is invoked, because we
7903: specify the class to be constructed at the same place. Actually, I
7904: defined @code{construct} as a selector only to give the users a
7905: convenient way to specify initialization. The way it is used, a
7906: mechanism different from selector invocation would be more natural
7907: (but probably would take more code and more space to explain).
7908:
7909: @node Class Binding, Method conveniences, Object-Oriented Programming Style, Objects
7910: @subsubsection Class Binding
7911: @cindex class binding
7912: @cindex early binding
7913:
7914: @cindex late binding
7915: Normal selector invocations determine the method at run-time depending
7916: on the class of the receiving object. This run-time selection is called
7917: @i{late binding}.
7918:
7919: Sometimes it's preferable to invoke a different method. For example,
7920: you might want to use the simple method for @code{print}ing
7921: @code{object}s instead of the possibly long-winded @code{print} method
7922: of the receiver class. You can achieve this by replacing the invocation
7923: of @code{print} with:
7924:
7925: @cindex @code{[bind]} usage
7926: @example
7927: [bind] object print
7928: @end example
7929:
7930: @noindent
7931: in compiled code or:
7932:
7933: @cindex @code{bind} usage
7934: @example
7935: bind object print
7936: @end example
7937:
7938: @cindex class binding, alternative to
7939: @noindent
7940: in interpreted code. Alternatively, you can define the method with a
7941: name (e.g., @code{print-object}), and then invoke it through the
7942: name. Class binding is just a (often more convenient) way to achieve
7943: the same effect; it avoids name clutter and allows you to invoke
7944: methods directly without naming them first.
7945:
7946: @cindex superclass binding
7947: @cindex parent class binding
7948: A frequent use of class binding is this: When we define a method
7949: for a selector, we often want the method to do what the selector does
7950: in the parent class, and a little more. There is a special word for
7951: this purpose: @code{[parent]}; @code{[parent]
7952: @emph{selector}} is equivalent to @code{[bind] @emph{parent
7953: selector}}, where @code{@emph{parent}} is the parent
7954: class of the current class. E.g., a method definition might look like:
7955:
7956: @cindex @code{[parent]} usage
7957: @example
7958: :noname
7959: dup [parent] foo \ do parent's foo on the receiving object
7960: ... \ do some more
7961: ; overrides foo
7962: @end example
7963:
7964: @cindex class binding as optimization
7965: In @cite{Object-oriented programming in ANS Forth} (Forth Dimensions,
7966: March 1997), Andrew McKewan presents class binding as an optimization
7967: technique. I recommend not using it for this purpose unless you are in
7968: an emergency. Late binding is pretty fast with this model anyway, so the
7969: benefit of using class binding is small; the cost of using class binding
7970: where it is not appropriate is reduced maintainability.
7971:
7972: While we are at programming style questions: You should bind
7973: selectors only to ancestor classes of the receiving object. E.g., say,
7974: you know that the receiving object is of class @code{foo} or its
7975: descendents; then you should bind only to @code{foo} and its
7976: ancestors.
7977:
7978: @node Method conveniences, Classes and Scoping, Class Binding, Objects
7979: @subsubsection Method conveniences
7980: @cindex method conveniences
7981:
7982: In a method you usually access the receiving object pretty often. If
7983: you define the method as a plain colon definition (e.g., with
7984: @code{:noname}), you may have to do a lot of stack
7985: gymnastics. To avoid this, you can define the method with @code{m:
7986: ... ;m}. E.g., you could define the method for
7987: @code{draw}ing a @code{circle} with
7988:
7989: @cindex @code{this} usage
7990: @cindex @code{m:} usage
7991: @cindex @code{;m} usage
7992: @example
7993: m: ( x y circle -- )
7994: ( x y ) this circle-radius @@ draw-circle ;m
7995: @end example
7996:
7997: @cindex @code{exit} in @code{m: ... ;m}
7998: @cindex @code{exitm} discussion
7999: @cindex @code{catch} in @code{m: ... ;m}
8000: When this method is executed, the receiver object is removed from the
8001: stack; you can access it with @code{this} (admittedly, in this
8002: example the use of @code{m: ... ;m} offers no advantage). Note
8003: that I specify the stack effect for the whole method (i.e. including
8004: the receiver object), not just for the code between @code{m:}
8005: and @code{;m}. You cannot use @code{exit} in
8006: @code{m:...;m}; instead, use
8007: @code{exitm}.@footnote{Moreover, for any word that calls
8008: @code{catch} and was defined before loading
8009: @code{objects.fs}, you have to redefine it like I redefined
8010: @code{catch}: @code{: catch this >r catch r> to-this ;}}
8011:
8012: @cindex @code{inst-var} usage
8013: You will frequently use sequences of the form @code{this
8014: @emph{field}} (in the example above: @code{this
8015: circle-radius}). If you use the field only in this way, you can
8016: define it with @code{inst-var} and eliminate the
8017: @code{this} before the field name. E.g., the @code{circle}
8018: class above could also be defined with:
8019:
8020: @example
8021: graphical class
8022: cell% inst-var radius
8023:
8024: m: ( x y circle -- )
8025: radius @@ draw-circle ;m
8026: overrides draw
8027:
8028: m: ( n-radius circle -- )
8029: radius ! ;m
8030: overrides construct
8031:
8032: end-class circle
8033: @end example
8034:
8035: @code{radius} can only be used in @code{circle} and its
8036: descendent classes and inside @code{m:...;m}.
8037:
8038: @cindex @code{inst-value} usage
8039: You can also define fields with @code{inst-value}, which is
8040: to @code{inst-var} what @code{value} is to
8041: @code{variable}. You can change the value of such a field with
8042: @code{[to-inst]}. E.g., we could also define the class
8043: @code{circle} like this:
8044:
8045: @example
8046: graphical class
8047: inst-value radius
8048:
8049: m: ( x y circle -- )
8050: radius draw-circle ;m
8051: overrides draw
8052:
8053: m: ( n-radius circle -- )
8054: [to-inst] radius ;m
8055: overrides construct
8056:
8057: end-class circle
8058: @end example
8059:
8060: Finally, you can define named methods with @code{:m}. One use of this
8061: feature is the definition of words that occur only in one class and are
8062: not intended to be overridden, but which still need method context
8063: (e.g., for accessing @code{inst-var}s). Another use is for methods that
8064: would be bound frequently, if defined anonymously.
8065:
8066:
8067: @node Classes and Scoping, Dividing classes, Method conveniences, Objects
8068: @subsubsection Classes and Scoping
8069: @cindex classes and scoping
8070: @cindex scoping and classes
8071:
8072: Inheritance is frequent, unlike structure extension. This exacerbates
8073: the problem with the field name convention (@pxref{Structure Naming
8074: Convention}): One always has to remember in which class the field was
8075: originally defined; changing a part of the class structure would require
8076: changes for renaming in otherwise unaffected code.
8077:
8078: @cindex @code{inst-var} visibility
8079: @cindex @code{inst-value} visibility
8080: To solve this problem, I added a scoping mechanism (which was not in my
8081: original charter): A field defined with @code{inst-var} (or
8082: @code{inst-value}) is visible only in the class where it is defined and in
8083: the descendent classes of this class. Using such fields only makes
8084: sense in @code{m:}-defined methods in these classes anyway.
8085:
8086: This scoping mechanism allows us to use the unadorned field name,
8087: because name clashes with unrelated words become much less likely.
8088:
8089: @cindex @code{protected} discussion
8090: @cindex @code{private} discussion
8091: Once we have this mechanism, we can also use it for controlling the
8092: visibility of other words: All words defined after
8093: @code{protected} are visible only in the current class and its
8094: descendents. @code{public} restores the compilation
8095: (i.e. @code{current}) word list that was in effect before. If you
8096: have several @code{protected}s without an intervening
8097: @code{public} or @code{set-current}, @code{public}
8098: will restore the compilation word list in effect before the first of
8099: these @code{protected}s.
8100:
8101: @node Dividing classes, Object Interfaces, Classes and Scoping, Objects
8102: @subsubsection Dividing classes
8103: @cindex Dividing classes
8104: @cindex @code{methods}...@code{end-methods}
8105:
8106: You may want to do the definition of methods separate from the
8107: definition of the class, its selectors, fields, and instance variables,
8108: i.e., separate the implementation from the definition. You can do this
8109: in the following way:
8110:
8111: @example
8112: graphical class
8113: inst-value radius
8114: end-class circle
8115:
8116: ... \ do some other stuff
8117:
8118: circle methods \ now we are ready
8119:
8120: m: ( x y circle -- )
8121: radius draw-circle ;m
8122: overrides draw
8123:
8124: m: ( n-radius circle -- )
8125: [to-inst] radius ;m
8126: overrides construct
8127:
8128: end-methods
8129: @end example
8130:
8131: You can use several @code{methods}...@code{end-methods} sections. The
8132: only things you can do to the class in these sections are: defining
8133: methods, and overriding the class's selectors. You must not define new
8134: selectors or fields.
8135:
8136: Note that you often have to override a selector before using it. In
8137: particular, you usually have to override @code{construct} with a new
8138: method before you can invoke @code{heap-new} and friends. E.g., you
8139: must not create a circle before the @code{overrides construct} sequence
8140: in the example above.
8141:
8142: @node Object Interfaces, Objects Implementation, Dividing classes, Objects
8143: @subsubsection Object Interfaces
8144: @cindex object interfaces
8145: @cindex interfaces for objects
8146:
8147: In this model you can only call selectors defined in the class of the
8148: receiving objects or in one of its ancestors. If you call a selector
8149: with a receiving object that is not in one of these classes, the
8150: result is undefined; if you are lucky, the program crashes
8151: immediately.
8152:
8153: @cindex selectors common to hardly-related classes
8154: Now consider the case when you want to have a selector (or several)
8155: available in two classes: You would have to add the selector to a
8156: common ancestor class, in the worst case to @code{object}. You
8157: may not want to do this, e.g., because someone else is responsible for
8158: this ancestor class.
8159:
8160: The solution for this problem is interfaces. An interface is a
8161: collection of selectors. If a class implements an interface, the
8162: selectors become available to the class and its descendents. A class
8163: can implement an unlimited number of interfaces. For the problem
8164: discussed above, we would define an interface for the selector(s), and
8165: both classes would implement the interface.
8166:
8167: As an example, consider an interface @code{storage} for
8168: writing objects to disk and getting them back, and a class
8169: @code{foo} that implements it. The code would look like this:
8170:
8171: @cindex @code{interface} usage
8172: @cindex @code{end-interface} usage
8173: @cindex @code{implementation} usage
8174: @example
8175: interface
8176: selector write ( file object -- )
8177: selector read1 ( file object -- )
8178: end-interface storage
8179:
8180: bar class
8181: storage implementation
8182:
8183: ... overrides write
8184: ... overrides read1
8185: ...
8186: end-class foo
8187: @end example
8188:
8189: @noindent
8190: (I would add a word @code{read} @i{( file -- object )} that uses
8191: @code{read1} internally, but that's beyond the point illustrated
8192: here.)
8193:
8194: Note that you cannot use @code{protected} in an interface; and
8195: of course you cannot define fields.
8196:
8197: In the Neon model, all selectors are available for all classes;
8198: therefore it does not need interfaces. The price you pay in this model
8199: is slower late binding, and therefore, added complexity to avoid late
8200: binding.
8201:
8202: @node Objects Implementation, Objects Glossary, Object Interfaces, Objects
8203: @subsubsection @file{objects.fs} Implementation
8204: @cindex @file{objects.fs} implementation
8205:
8206: @cindex @code{object-map} discussion
8207: An object is a piece of memory, like one of the data structures
8208: described with @code{struct...end-struct}. It has a field
8209: @code{object-map} that points to the method map for the object's
8210: class.
8211:
8212: @cindex method map
8213: @cindex virtual function table
8214: The @emph{method map}@footnote{This is Self terminology; in C++
8215: terminology: virtual function table.} is an array that contains the
8216: execution tokens (@i{xt}s) of the methods for the object's class. Each
8217: selector contains an offset into a method map.
8218:
8219: @cindex @code{selector} implementation, class
8220: @code{selector} is a defining word that uses
8221: @code{CREATE} and @code{DOES>}. The body of the
8222: selector contains the offset; the @code{DOES>} action for a
8223: class selector is, basically:
8224:
8225: @example
8226: ( object addr ) @@ over object-map @@ + @@ execute
8227: @end example
8228:
8229: Since @code{object-map} is the first field of the object, it
8230: does not generate any code. As you can see, calling a selector has a
8231: small, constant cost.
8232:
8233: @cindex @code{current-interface} discussion
8234: @cindex class implementation and representation
8235: A class is basically a @code{struct} combined with a method
8236: map. During the class definition the alignment and size of the class
8237: are passed on the stack, just as with @code{struct}s, so
8238: @code{field} can also be used for defining class
8239: fields. However, passing more items on the stack would be
8240: inconvenient, so @code{class} builds a data structure in memory,
8241: which is accessed through the variable
8242: @code{current-interface}. After its definition is complete, the
8243: class is represented on the stack by a pointer (e.g., as parameter for
8244: a child class definition).
8245:
8246: A new class starts off with the alignment and size of its parent,
8247: and a copy of the parent's method map. Defining new fields extends the
8248: size and alignment; likewise, defining new selectors extends the
8249: method map. @code{overrides} just stores a new @i{xt} in the method
8250: map at the offset given by the selector.
8251:
8252: @cindex class binding, implementation
8253: Class binding just gets the @i{xt} at the offset given by the selector
8254: from the class's method map and @code{compile,}s (in the case of
8255: @code{[bind]}) it.
8256:
8257: @cindex @code{this} implementation
8258: @cindex @code{catch} and @code{this}
8259: @cindex @code{this} and @code{catch}
8260: I implemented @code{this} as a @code{value}. At the
8261: start of an @code{m:...;m} method the old @code{this} is
8262: stored to the return stack and restored at the end; and the object on
8263: the TOS is stored @code{TO this}. This technique has one
8264: disadvantage: If the user does not leave the method via
8265: @code{;m}, but via @code{throw} or @code{exit},
8266: @code{this} is not restored (and @code{exit} may
8267: crash). To deal with the @code{throw} problem, I have redefined
8268: @code{catch} to save and restore @code{this}; the same
8269: should be done with any word that can catch an exception. As for
8270: @code{exit}, I simply forbid it (as a replacement, there is
8271: @code{exitm}).
8272:
8273: @cindex @code{inst-var} implementation
8274: @code{inst-var} is just the same as @code{field}, with
8275: a different @code{DOES>} action:
8276: @example
8277: @@ this +
8278: @end example
8279: Similar for @code{inst-value}.
8280:
8281: @cindex class scoping implementation
8282: Each class also has a word list that contains the words defined with
8283: @code{inst-var} and @code{inst-value}, and its protected
8284: words. It also has a pointer to its parent. @code{class} pushes
8285: the word lists of the class and all its ancestors onto the search order stack,
8286: and @code{end-class} drops them.
8287:
8288: @cindex interface implementation
8289: An interface is like a class without fields, parent and protected
8290: words; i.e., it just has a method map. If a class implements an
8291: interface, its method map contains a pointer to the method map of the
8292: interface. The positive offsets in the map are reserved for class
8293: methods, therefore interface map pointers have negative
8294: offsets. Interfaces have offsets that are unique throughout the
8295: system, unlike class selectors, whose offsets are only unique for the
8296: classes where the selector is available (invokable).
8297:
8298: This structure means that interface selectors have to perform one
8299: indirection more than class selectors to find their method. Their body
8300: contains the interface map pointer offset in the class method map, and
8301: the method offset in the interface method map. The
8302: @code{does>} action for an interface selector is, basically:
8303:
8304: @example
8305: ( object selector-body )
8306: 2dup selector-interface @@ ( object selector-body object interface-offset )
8307: swap object-map @@ + @@ ( object selector-body map )
8308: swap selector-offset @@ + @@ execute
8309: @end example
8310:
8311: where @code{object-map} and @code{selector-offset} are
8312: first fields and generate no code.
8313:
8314: As a concrete example, consider the following code:
8315:
8316: @example
8317: interface
8318: selector if1sel1
8319: selector if1sel2
8320: end-interface if1
8321:
8322: object class
8323: if1 implementation
8324: selector cl1sel1
8325: cell% inst-var cl1iv1
8326:
8327: ' m1 overrides construct
8328: ' m2 overrides if1sel1
8329: ' m3 overrides if1sel2
8330: ' m4 overrides cl1sel2
8331: end-class cl1
8332:
8333: create obj1 object dict-new drop
8334: create obj2 cl1 dict-new drop
8335: @end example
8336:
8337: The data structure created by this code (including the data structure
8338: for @code{object}) is shown in the <a
8339: href="objects-implementation.eps">figure</a>, assuming a cell size of 4.
8340: @comment TODO add this diagram..
8341:
8342: @node Objects Glossary, , Objects Implementation, Objects
8343: @subsubsection @file{objects.fs} Glossary
8344: @cindex @file{objects.fs} Glossary
8345:
8346:
8347: doc---objects-bind
8348: doc---objects-<bind>
8349: doc---objects-bind'
8350: doc---objects-[bind]
8351: doc---objects-class
8352: doc---objects-class->map
8353: doc---objects-class-inst-size
8354: doc---objects-class-override!
8355: doc---objects-construct
8356: doc---objects-current'
8357: doc---objects-[current]
8358: doc---objects-current-interface
8359: doc---objects-dict-new
8360: doc---objects-drop-order
8361: doc---objects-end-class
8362: doc---objects-end-class-noname
8363: doc---objects-end-interface
8364: doc---objects-end-interface-noname
8365: doc---objects-end-methods
8366: doc---objects-exitm
8367: doc---objects-heap-new
8368: doc---objects-implementation
8369: doc---objects-init-object
8370: doc---objects-inst-value
8371: doc---objects-inst-var
8372: doc---objects-interface
8373: doc---objects-m:
8374: doc---objects-:m
8375: doc---objects-;m
8376: doc---objects-method
8377: doc---objects-methods
8378: doc---objects-object
8379: doc---objects-overrides
8380: doc---objects-[parent]
8381: doc---objects-print
8382: doc---objects-protected
8383: doc---objects-public
8384: doc---objects-push-order
8385: doc---objects-selector
8386: doc---objects-this
8387: doc---objects-<to-inst>
8388: doc---objects-[to-inst]
8389: doc---objects-to-this
8390: doc---objects-xt-new
8391:
8392:
8393: @c -------------------------------------------------------------
8394: @node OOF, Mini-OOF, Objects, Object-oriented Forth
8395: @subsection The @file{oof.fs} model
8396: @cindex oof
8397: @cindex object-oriented programming
8398:
8399: @cindex @file{objects.fs}
8400: @cindex @file{oof.fs}
8401:
8402: This section describes the @file{oof.fs} package.
8403:
8404: The package described in this section has been used in bigFORTH since 1991, and
8405: used for two large applications: a chromatographic system used to
8406: create new medicaments, and a graphic user interface library (MINOS).
8407:
8408: You can find a description (in German) of @file{oof.fs} in @cite{Object
8409: oriented bigFORTH} by Bernd Paysan, published in @cite{Vierte Dimension}
8410: 10(2), 1994.
8411:
8412: @menu
8413: * Properties of the OOF model::
8414: * Basic OOF Usage::
8415: * The OOF base class::
8416: * Class Declaration::
8417: * Class Implementation::
8418: @end menu
8419:
8420: @node Properties of the OOF model, Basic OOF Usage, OOF, OOF
8421: @subsubsection Properties of the @file{oof.fs} model
8422: @cindex @file{oof.fs} properties
8423:
8424: @itemize @bullet
8425: @item
8426: This model combines object oriented programming with information
8427: hiding. It helps you writing large application, where scoping is
8428: necessary, because it provides class-oriented scoping.
8429:
8430: @item
8431: Named objects, object pointers, and object arrays can be created,
8432: selector invocation uses the ``object selector'' syntax. Selector invocation
8433: to objects and/or selectors on the stack is a bit less convenient, but
8434: possible.
8435:
8436: @item
8437: Selector invocation and instance variable usage of the active object is
8438: straightforward, since both make use of the active object.
8439:
8440: @item
8441: Late binding is efficient and easy to use.
8442:
8443: @item
8444: State-smart objects parse selectors. However, extensibility is provided
8445: using a (parsing) selector @code{postpone} and a selector @code{'}.
8446:
8447: @item
8448: An implementation in ANS Forth is available.
8449:
8450: @end itemize
8451:
8452:
8453: @node Basic OOF Usage, The OOF base class, Properties of the OOF model, OOF
8454: @subsubsection Basic @file{oof.fs} Usage
8455: @cindex @file{oof.fs} usage
8456:
8457: This section uses the same example as for @code{objects} (@pxref{Basic Objects Usage}).
8458:
8459: You can define a class for graphical objects like this:
8460:
8461: @cindex @code{class} usage
8462: @cindex @code{class;} usage
8463: @cindex @code{method} usage
8464: @example
8465: object class graphical \ "object" is the parent class
8466: method draw ( x y graphical -- )
8467: class;
8468: @end example
8469:
8470: This code defines a class @code{graphical} with an
8471: operation @code{draw}. We can perform the operation
8472: @code{draw} on any @code{graphical} object, e.g.:
8473:
8474: @example
8475: 100 100 t-rex draw
8476: @end example
8477:
8478: @noindent
8479: where @code{t-rex} is an object or object pointer, created with e.g.
8480: @code{graphical : t-rex}.
8481:
8482: @cindex abstract class
8483: How do we create a graphical object? With the present definitions,
8484: we cannot create a useful graphical object. The class
8485: @code{graphical} describes graphical objects in general, but not
8486: any concrete graphical object type (C++ users would call it an
8487: @emph{abstract class}); e.g., there is no method for the selector
8488: @code{draw} in the class @code{graphical}.
8489:
8490: For concrete graphical objects, we define child classes of the
8491: class @code{graphical}, e.g.:
8492:
8493: @example
8494: graphical class circle \ "graphical" is the parent class
8495: cell var circle-radius
8496: how:
8497: : draw ( x y -- )
8498: circle-radius @@ draw-circle ;
8499:
8500: : init ( n-radius -- (
8501: circle-radius ! ;
8502: class;
8503: @end example
8504:
8505: Here we define a class @code{circle} as a child of @code{graphical},
8506: with a field @code{circle-radius}; it defines new methods for the
8507: selectors @code{draw} and @code{init} (@code{init} is defined in
8508: @code{object}, the parent class of @code{graphical}).
8509:
8510: Now we can create a circle in the dictionary with:
8511:
8512: @example
8513: 50 circle : my-circle
8514: @end example
8515:
8516: @noindent
8517: @code{:} invokes @code{init}, thus initializing the field
8518: @code{circle-radius} with 50. We can draw this new circle at (100,100)
8519: with:
8520:
8521: @example
8522: 100 100 my-circle draw
8523: @end example
8524:
8525: @cindex selector invocation, restrictions
8526: @cindex class definition, restrictions
8527: Note: You can only invoke a selector if the receiving object belongs to
8528: the class where the selector was defined or one of its descendents;
8529: e.g., you can invoke @code{draw} only for objects belonging to
8530: @code{graphical} or its descendents (e.g., @code{circle}). The scoping
8531: mechanism will check if you try to invoke a selector that is not
8532: defined in this class hierarchy, so you'll get an error at compilation
8533: time.
8534:
8535:
8536: @node The OOF base class, Class Declaration, Basic OOF Usage, OOF
8537: @subsubsection The @file{oof.fs} base class
8538: @cindex @file{oof.fs} base class
8539:
8540: When you define a class, you have to specify a parent class. So how do
8541: you start defining classes? There is one class available from the start:
8542: @code{object}. You have to use it as ancestor for all classes. It is the
8543: only class that has no parent. Classes are also objects, except that
8544: they don't have instance variables; class manipulation such as
8545: inheritance or changing definitions of a class is handled through
8546: selectors of the class @code{object}.
8547:
8548: @code{object} provides a number of selectors:
8549:
8550: @itemize @bullet
8551: @item
8552: @code{class} for subclassing, @code{definitions} to add definitions
8553: later on, and @code{class?} to get type informations (is the class a
8554: subclass of the class passed on the stack?).
8555:
8556: doc---object-class
8557: doc---object-definitions
8558: doc---object-class?
8559:
8560:
8561: @item
8562: @code{init} and @code{dispose} as constructor and destructor of the
8563: object. @code{init} is invocated after the object's memory is allocated,
8564: while @code{dispose} also handles deallocation. Thus if you redefine
8565: @code{dispose}, you have to call the parent's dispose with @code{super
8566: dispose}, too.
8567:
8568: doc---object-init
8569: doc---object-dispose
8570:
8571:
8572: @item
8573: @code{new}, @code{new[]}, @code{:}, @code{ptr}, @code{asptr}, and
8574: @code{[]} to create named and unnamed objects and object arrays or
8575: object pointers.
8576:
8577: doc---object-new
8578: doc---object-new[]
8579: doc---object-:
8580: doc---object-ptr
8581: doc---object-asptr
8582: doc---object-[]
8583:
8584:
8585: @item
8586: @code{::} and @code{super} for explicit scoping. You should use explicit
8587: scoping only for super classes or classes with the same set of instance
8588: variables. Explicitly-scoped selectors use early binding.
8589:
8590: doc---object-::
8591: doc---object-super
8592:
8593:
8594: @item
8595: @code{self} to get the address of the object
8596:
8597: doc---object-self
8598:
8599:
8600: @item
8601: @code{bind}, @code{bound}, @code{link}, and @code{is} to assign object
8602: pointers and instance defers.
8603:
8604: doc---object-bind
8605: doc---object-bound
8606: doc---object-link
8607: doc---object-is
8608:
8609:
8610: @item
8611: @code{'} to obtain selector tokens, @code{send} to invocate selectors
8612: form the stack, and @code{postpone} to generate selector invocation code.
8613:
8614: doc---object-'
8615: doc---object-postpone
8616:
8617:
8618: @item
8619: @code{with} and @code{endwith} to select the active object from the
8620: stack, and enable its scope. Using @code{with} and @code{endwith}
8621: also allows you to create code using selector @code{postpone} without being
8622: trapped by the state-smart objects.
8623:
8624: doc---object-with
8625: doc---object-endwith
8626:
8627:
8628: @end itemize
8629:
8630: @node Class Declaration, Class Implementation, The OOF base class, OOF
8631: @subsubsection Class Declaration
8632: @cindex class declaration
8633:
8634: @itemize @bullet
8635: @item
8636: Instance variables
8637:
8638: doc---oof-var
8639:
8640:
8641: @item
8642: Object pointers
8643:
8644: doc---oof-ptr
8645: doc---oof-asptr
8646:
8647:
8648: @item
8649: Instance defers
8650:
8651: doc---oof-defer
8652:
8653:
8654: @item
8655: Method selectors
8656:
8657: doc---oof-early
8658: doc---oof-method
8659:
8660:
8661: @item
8662: Class-wide variables
8663:
8664: doc---oof-static
8665:
8666:
8667: @item
8668: End declaration
8669:
8670: doc---oof-how:
8671: doc---oof-class;
8672:
8673:
8674: @end itemize
8675:
8676: @c -------------------------------------------------------------
8677: @node Class Implementation, , Class Declaration, OOF
8678: @subsubsection Class Implementation
8679: @cindex class implementation
8680:
8681: @c -------------------------------------------------------------
8682: @node Mini-OOF, Comparison with other object models, OOF, Object-oriented Forth
8683: @subsection The @file{mini-oof.fs} model
8684: @cindex mini-oof
8685:
8686: Gforth's third object oriented Forth package is a 12-liner. It uses a
8687: mixture of the @file{object.fs} and the @file{oof.fs} syntax,
8688: and reduces to the bare minimum of features. This is based on a posting
8689: of Bernd Paysan in comp.arch.
8690:
8691: @menu
8692: * Basic Mini-OOF Usage::
8693: * Mini-OOF Example::
8694: * Mini-OOF Implementation::
8695: @end menu
8696:
8697: @c -------------------------------------------------------------
8698: @node Basic Mini-OOF Usage, Mini-OOF Example, , Mini-OOF
8699: @subsubsection Basic @file{mini-oof.fs} Usage
8700: @cindex mini-oof usage
8701:
8702: There is a base class (@code{class}, which allocates one cell for the
8703: object pointer) plus seven other words: to define a method, a variable,
8704: a class; to end a class, to resolve binding, to allocate an object and
8705: to compile a class method.
8706: @comment TODO better description of the last one
8707:
8708:
8709: doc-object
8710: doc-method
8711: doc-var
8712: doc-class
8713: doc-end-class
8714: doc-defines
8715: doc-new
8716: doc-::
8717:
8718:
8719:
8720: @c -------------------------------------------------------------
8721: @node Mini-OOF Example, Mini-OOF Implementation, Basic Mini-OOF Usage, Mini-OOF
8722: @subsubsection Mini-OOF Example
8723: @cindex mini-oof example
8724:
8725: A short example shows how to use this package. This example, in slightly
8726: extended form, is supplied as @file{moof-exm.fs}
8727: @comment TODO could flesh this out with some comments from the Forthwrite article
8728:
8729: @example
8730: object class
8731: method init
8732: method draw
8733: end-class graphical
8734: @end example
8735:
8736: This code defines a class @code{graphical} with an
8737: operation @code{draw}. We can perform the operation
8738: @code{draw} on any @code{graphical} object, e.g.:
8739:
8740: @example
8741: 100 100 t-rex draw
8742: @end example
8743:
8744: where @code{t-rex} is an object or object pointer, created with e.g.
8745: @code{graphical new Constant t-rex}.
8746:
8747: For concrete graphical objects, we define child classes of the
8748: class @code{graphical}, e.g.:
8749:
8750: @example
8751: graphical class
8752: cell var circle-radius
8753: end-class circle \ "graphical" is the parent class
8754:
8755: :noname ( x y -- )
8756: circle-radius @@ draw-circle ; circle defines draw
8757: :noname ( r -- )
8758: circle-radius ! ; circle defines init
8759: @end example
8760:
8761: There is no implicit init method, so we have to define one. The creation
8762: code of the object now has to call init explicitely.
8763:
8764: @example
8765: circle new Constant my-circle
8766: 50 my-circle init
8767: @end example
8768:
8769: It is also possible to add a function to create named objects with
8770: automatic call of @code{init}, given that all objects have @code{init}
8771: on the same place:
8772:
8773: @example
8774: : new: ( .. o "name" -- )
8775: new dup Constant init ;
8776: 80 circle new: large-circle
8777: @end example
8778:
8779: We can draw this new circle at (100,100) with:
8780:
8781: @example
8782: 100 100 my-circle draw
8783: @end example
8784:
8785: @node Mini-OOF Implementation, , Mini-OOF Example, Mini-OOF
8786: @subsubsection @file{mini-oof.fs} Implementation
8787:
8788: Object-oriented systems with late binding typically use a
8789: ``vtable''-approach: the first variable in each object is a pointer to a
8790: table, which contains the methods as function pointers. The vtable
8791: may also contain other information.
8792:
8793: So first, let's declare methods:
8794:
8795: @example
8796: : method ( m v -- m' v ) Create over , swap cell+ swap
8797: DOES> ( ... o -- ... ) @ over @ + @ execute ;
8798: @end example
8799:
8800: During method declaration, the number of methods and instance
8801: variables is on the stack (in address units). @code{method} creates
8802: one method and increments the method number. To execute a method, it
8803: takes the object, fetches the vtable pointer, adds the offset, and
8804: executes the @i{xt} stored there. Each method takes the object it is
8805: invoked from as top of stack parameter. The method itself should
8806: consume that object.
8807:
8808: Now, we also have to declare instance variables
8809:
8810: @example
8811: : var ( m v size -- m v' ) Create over , +
8812: DOES> ( o -- addr ) @ + ;
8813: @end example
8814:
8815: As before, a word is created with the current offset. Instance
8816: variables can have different sizes (cells, floats, doubles, chars), so
8817: all we do is take the size and add it to the offset. If your machine
8818: has alignment restrictions, put the proper @code{aligned} or
8819: @code{faligned} before the variable, to adjust the variable
8820: offset. That's why it is on the top of stack.
8821:
8822: We need a starting point (the base object) and some syntactic sugar:
8823:
8824: @example
8825: Create object 1 cells , 2 cells ,
8826: : class ( class -- class methods vars ) dup 2@ ;
8827: @end example
8828:
8829: For inheritance, the vtable of the parent object has to be
8830: copied when a new, derived class is declared. This gives all the
8831: methods of the parent class, which can be overridden, though.
8832:
8833: @example
8834: : end-class ( class methods vars -- )
8835: Create here >r , dup , 2 cells ?DO ['] noop , 1 cells +LOOP
8836: cell+ dup cell+ r> rot @ 2 cells /string move ;
8837: @end example
8838:
8839: The first line creates the vtable, initialized with
8840: @code{noop}s. The second line is the inheritance mechanism, it
8841: copies the xts from the parent vtable.
8842:
8843: We still have no way to define new methods, let's do that now:
8844:
8845: @example
8846: : defines ( xt class -- ) ' >body @ + ! ;
8847: @end example
8848:
8849: To allocate a new object, we need a word, too:
8850:
8851: @example
8852: : new ( class -- o ) here over @ allot swap over ! ;
8853: @end example
8854:
8855: Sometimes derived classes want to access the method of the
8856: parent object. There are two ways to achieve this with Mini-OOF:
8857: first, you could use named words, and second, you could look up the
8858: vtable of the parent object.
8859:
8860: @example
8861: : :: ( class "name" -- ) ' >body @ + @ compile, ;
8862: @end example
8863:
8864:
8865: Nothing can be more confusing than a good example, so here is
8866: one. First let's declare a text object (called
8867: @code{button}), that stores text and position:
8868:
8869: @example
8870: object class
8871: cell var text
8872: cell var len
8873: cell var x
8874: cell var y
8875: method init
8876: method draw
8877: end-class button
8878: @end example
8879:
8880: @noindent
8881: Now, implement the two methods, @code{draw} and @code{init}:
8882:
8883: @example
8884: :noname ( o -- )
8885: >r r@ x @ r@ y @ at-xy r@ text @ r> len @ type ;
8886: button defines draw
8887: :noname ( addr u o -- )
8888: >r 0 r@ x ! 0 r@ y ! r@ len ! r> text ! ;
8889: button defines init
8890: @end example
8891:
8892: @noindent
8893: To demonstrate inheritance, we define a class @code{bold-button}, with no
8894: new data and no new methods:
8895:
8896: @example
8897: button class
8898: end-class bold-button
8899:
8900: : bold 27 emit ." [1m" ;
8901: : normal 27 emit ." [0m" ;
8902: @end example
8903:
8904: @noindent
8905: The class @code{bold-button} has a different draw method to
8906: @code{button}, but the new method is defined in terms of the draw method
8907: for @code{button}:
8908:
8909: @example
8910: :noname bold [ button :: draw ] normal ; bold-button defines draw
8911: @end example
8912:
8913: @noindent
8914: Finally, create two objects and apply methods:
8915:
8916: @example
8917: button new Constant foo
8918: s" thin foo" foo init
8919: page
8920: foo draw
8921: bold-button new Constant bar
8922: s" fat bar" bar init
8923: 1 bar y !
8924: bar draw
8925: @end example
8926:
8927:
8928: @node Comparison with other object models, , Mini-OOF, Object-oriented Forth
8929: @subsubsection Comparison with other object models
8930: @cindex comparison of object models
8931: @cindex object models, comparison
8932:
8933: Many object-oriented Forth extensions have been proposed (@cite{A survey
8934: of object-oriented Forths} (SIGPLAN Notices, April 1996) by Bradford
8935: J. Rodriguez and W. F. S. Poehlman lists 17). This section discusses the
8936: relation of the object models described here to two well-known and two
8937: closely-related (by the use of method maps) models.
8938:
8939: @cindex Neon model
8940: The most popular model currently seems to be the Neon model (see
8941: @cite{Object-oriented programming in ANS Forth} (Forth Dimensions, March
8942: 1997) by Andrew McKewan) but this model has a number of limitations
8943: @footnote{A longer version of this critique can be
8944: found in @cite{On Standardizing Object-Oriented Forth Extensions} (Forth
8945: Dimensions, May 1997) by Anton Ertl.}:
8946:
8947: @itemize @bullet
8948: @item
8949: It uses a @code{@emph{selector
8950: object}} syntax, which makes it unnatural to pass objects on the
8951: stack.
8952:
8953: @item
8954: It requires that the selector parses the input stream (at
8955: compile time); this leads to reduced extensibility and to bugs that are+
8956: hard to find.
8957:
8958: @item
8959: It allows using every selector to every object;
8960: this eliminates the need for classes, but makes it harder to create
8961: efficient implementations.
8962: @end itemize
8963:
8964: @cindex Pountain's object-oriented model
8965: Another well-known publication is @cite{Object-Oriented Forth} (Academic
8966: Press, London, 1987) by Dick Pountain. However, it is not really about
8967: object-oriented programming, because it hardly deals with late
8968: binding. Instead, it focuses on features like information hiding and
8969: overloading that are characteristic of modular languages like Ada (83).
8970:
8971: @cindex Zsoter's object-oriented model
8972: In @cite{Does late binding have to be slow?} (Forth Dimensions 18(1) 1996, pages 31-35)
8973: Andras Zsoter describes a model that makes heavy use of an active object
8974: (like @code{this} in @file{objects.fs}): The active object is not only
8975: used for accessing all fields, but also specifies the receiving object
8976: of every selector invocation; you have to change the active object
8977: explicitly with @code{@{ ... @}}, whereas in @file{objects.fs} it
8978: changes more or less implicitly at @code{m: ... ;m}. Such a change at
8979: the method entry point is unnecessary with the Zsoter's model, because
8980: the receiving object is the active object already. On the other hand, the explicit
8981: change is absolutely necessary in that model, because otherwise no one
8982: could ever change the active object. An ANS Forth implementation of this
8983: model is available at @url{http://www.forth.org/fig/oopf.html}.
8984:
8985: @cindex @file{oof.fs}, differences to other models
8986: The @file{oof.fs} model combines information hiding and overloading
8987: resolution (by keeping names in various word lists) with object-oriented
8988: programming. It sets the active object implicitly on method entry, but
8989: also allows explicit changing (with @code{>o...o>} or with
8990: @code{with...endwith}). It uses parsing and state-smart objects and
8991: classes for resolving overloading and for early binding: the object or
8992: class parses the selector and determines the method from this. If the
8993: selector is not parsed by an object or class, it performs a call to the
8994: selector for the active object (late binding), like Zsoter's model.
8995: Fields are always accessed through the active object. The big
8996: disadvantage of this model is the parsing and the state-smartness, which
8997: reduces extensibility and increases the opportunities for subtle bugs;
8998: essentially, you are only safe if you never tick or @code{postpone} an
8999: object or class (Bernd disagrees, but I (Anton) am not convinced).
9000:
9001: @cindex @file{mini-oof.fs}, differences to other models
9002: The @file{mini-oof.fs} model is quite similar to a very stripped-down version of
9003: the @file{objects.fs} model, but syntactically it is a mixture of the @file{objects.fs} and
9004: @file{oof.fs} models.
9005:
9006: @c -------------------------------------------------------------
9007: @node Passing Commands to the OS, Miscellaneous Words, Object-oriented Forth, Words
9008: @section Passing Commands to the Operating System
9009: @cindex operating system - passing commands
9010: @cindex shell commands
9011:
9012: Gforth allows you to pass an arbitrary string to the host operating
9013: system shell (if such a thing exists) for execution.
9014:
9015:
9016: doc-sh
9017: doc-system
9018: doc-$?
9019: doc-getenv
9020:
9021:
9022: @c -------------------------------------------------------------
9023: @node Miscellaneous Words, , Passing Commands to the OS, Words
9024: @section Miscellaneous Words
9025: @cindex miscellaneous words
9026:
9027: @comment TODO find homes for these
9028:
9029: These section lists the ANS Forth words that are not documented
9030: elsewhere in this manual. Ultimately, they all need proper homes.
9031:
9032:
9033: doc-ms
9034: doc-time&date
9035:
9036:
9037:
9038: doc-[compile]
9039:
9040:
9041: The following ANS Forth words are not currently supported by Gforth
9042: (@pxref{ANS conformance}):
9043:
9044: @code{EDITOR}
9045: @code{EMIT?}
9046: @code{FORGET}
9047:
9048: @c ******************************************************************
9049: @node Error messages, Tools, Words, Top
9050: @chapter Error messages
9051: @cindex error messages
9052: @cindex backtrace
9053:
9054: A typical Gforth error message looks like this:
9055:
9056: @example
9057: in file included from :-1
9058: in file included from ./yyy.fs:1
9059: ./xxx.fs:4: Invalid memory address
9060: bar
9061: ^^^
9062: $400E664C @@
9063: $400E6664 foo
9064: @end example
9065:
9066: The message identifying the error is @code{Invalid memory address}. The
9067: error happened when text-interpreting line 4 of the file
9068: @file{./xxx.fs}. This line is given (it contains @code{bar}), and the
9069: word on the line where the error happened, is pointed out (with
9070: @code{^^^}).
9071:
9072: The file containing the error was included in line 1 of @file{./yyy.fs},
9073: and @file{yyy.fs} was included from a non-file (in this case, by giving
9074: @file{yyy.fs} as command-line parameter to Gforth).
9075:
9076: At the end of the error message you find a return stack dump that can be
9077: interpreted as a backtrace (possibly empty). On top you find the top of
9078: the return stack when the @code{throw} happened, and at the bottom you
9079: find the return stack entry just above the return stack of the topmost
9080: text interpreter.
9081:
9082: To the right of most return stack entries you see a guess for the word
9083: that pushed that return stack entry as its return address. This gives a
9084: backtrace. In our case we see that @code{bar} called @code{foo}, and
9085: @code{foo} called @code{@@} (and @code{@@} had an @emph{Invalid memory
9086: address} exception).
9087:
9088: Note that the backtrace is not perfect: We don't know which return stack
9089: entries are return addresses (so we may get false positives); and in
9090: some cases (e.g., for @code{abort"}) we cannot determine from the return
9091: address the word that pushed the return address, so for some return
9092: addresses you see no names in the return stack dump.
9093:
9094: @cindex @code{catch} and backtraces
9095: The return stack dump represents the return stack at the time when a
9096: specific @code{throw} was executed. In programs that make use of
9097: @code{catch}, it is not necessarily clear which @code{throw} should be
9098: used for the return stack dump (e.g., consider one @code{throw} that
9099: indicates an error, which is caught, and during recovery another error
9100: happens; which @code{throw} should be used for the stack dump?). Gforth
9101: presents the return stack dump for the first @code{throw} after the last
9102: executed (not returned-to) @code{catch}; this works well in the usual
9103: case.
9104:
9105: @cindex @code{gforth-fast} and backtraces
9106: @cindex @code{gforth-fast}, difference from @code{gforth}
9107: @cindex backtraces with @code{gforth-fast}
9108: @cindex return stack dump with @code{gforth-fast}
9109: @code{gforth} is able to do a return stack dump for throws generated
9110: from primitives (e.g., invalid memory address, stack empty etc.);
9111: @code{gforth-fast} is only able to do a return stack dump from a
9112: directly called @code{throw} (including @code{abort} etc.). This is the
9113: only difference (apart from a speed factor of between 1.15 (K6-2) and
9114: 1.6 (21164A)) between @code{gforth} and @code{gforth-fast}. Given an
9115: exception caused by a primitive in @code{gforth-fast}, you will
9116: typically see no return stack dump at all; however, if the exception is
9117: caught by @code{catch} (e.g., for restoring some state), and then
9118: @code{throw}n again, the return stack dump will be for the first such
9119: @code{throw}.
9120:
9121: @c ******************************************************************
9122: @node Tools, ANS conformance, Error messages, Top
9123: @chapter Tools
9124:
9125: @menu
9126: * ANS Report:: Report the words used, sorted by wordset.
9127: @end menu
9128:
9129: See also @ref{Emacs and Gforth}.
9130:
9131: @node ANS Report, , Tools, Tools
9132: @section @file{ans-report.fs}: Report the words used, sorted by wordset
9133: @cindex @file{ans-report.fs}
9134: @cindex report the words used in your program
9135: @cindex words used in your program
9136:
9137: If you want to label a Forth program as ANS Forth Program, you must
9138: document which wordsets the program uses; for extension wordsets, it is
9139: helpful to list the words the program requires from these wordsets
9140: (because Forth systems are allowed to provide only some words of them).
9141:
9142: The @file{ans-report.fs} tool makes it easy for you to determine which
9143: words from which wordset and which non-ANS words your application
9144: uses. You simply have to include @file{ans-report.fs} before loading the
9145: program you want to check. After loading your program, you can get the
9146: report with @code{print-ans-report}. A typical use is to run this as
9147: batch job like this:
9148: @example
9149: gforth ans-report.fs myprog.fs -e "print-ans-report bye"
9150: @end example
9151:
9152: The output looks like this (for @file{compat/control.fs}):
9153: @example
9154: The program uses the following words
9155: from CORE :
9156: : POSTPONE THEN ; immediate ?dup IF 0=
9157: from BLOCK-EXT :
9158: \
9159: from FILE :
9160: (
9161: @end example
9162:
9163: @subsection Caveats
9164:
9165: Note that @file{ans-report.fs} just checks which words are used, not whether
9166: they are used in an ANS Forth conforming way!
9167:
9168: Some words are defined in several wordsets in the
9169: standard. @file{ans-report.fs} reports them for only one of the
9170: wordsets, and not necessarily the one you expect. It depends on usage
9171: which wordset is the right one to specify. E.g., if you only use the
9172: compilation semantics of @code{S"}, it is a Core word; if you also use
9173: its interpretation semantics, it is a File word.
9174:
9175: @c ******************************************************************
9176: @node ANS conformance, Model, Tools, Top
9177: @chapter ANS conformance
9178: @cindex ANS conformance of Gforth
9179:
9180: To the best of our knowledge, Gforth is an
9181:
9182: ANS Forth System
9183: @itemize @bullet
9184: @item providing the Core Extensions word set
9185: @item providing the Block word set
9186: @item providing the Block Extensions word set
9187: @item providing the Double-Number word set
9188: @item providing the Double-Number Extensions word set
9189: @item providing the Exception word set
9190: @item providing the Exception Extensions word set
9191: @item providing the Facility word set
9192: @item providing @code{EKEY}, @code{EKEY>CHAR}, @code{EKEY?}, @code{MS} and @code{TIME&DATE} from the Facility Extensions word set
9193: @item providing the File Access word set
9194: @item providing the File Access Extensions word set
9195: @item providing the Floating-Point word set
9196: @item providing the Floating-Point Extensions word set
9197: @item providing the Locals word set
9198: @item providing the Locals Extensions word set
9199: @item providing the Memory-Allocation word set
9200: @item providing the Memory-Allocation Extensions word set (that one's easy)
9201: @item providing the Programming-Tools word set
9202: @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
9203: @item providing the Search-Order word set
9204: @item providing the Search-Order Extensions word set
9205: @item providing the String word set
9206: @item providing the String Extensions word set (another easy one)
9207: @end itemize
9208:
9209: @cindex system documentation
9210: In addition, ANS Forth systems are required to document certain
9211: implementation choices. This chapter tries to meet these
9212: requirements. In many cases it gives a way to ask the system for the
9213: information instead of providing the information directly, in
9214: particular, if the information depends on the processor, the operating
9215: system or the installation options chosen, or if they are likely to
9216: change during the maintenance of Gforth.
9217:
9218: @comment The framework for the rest has been taken from pfe.
9219:
9220: @menu
9221: * The Core Words::
9222: * The optional Block word set::
9223: * The optional Double Number word set::
9224: * The optional Exception word set::
9225: * The optional Facility word set::
9226: * The optional File-Access word set::
9227: * The optional Floating-Point word set::
9228: * The optional Locals word set::
9229: * The optional Memory-Allocation word set::
9230: * The optional Programming-Tools word set::
9231: * The optional Search-Order word set::
9232: @end menu
9233:
9234:
9235: @c =====================================================================
9236: @node The Core Words, The optional Block word set, ANS conformance, ANS conformance
9237: @comment node-name, next, previous, up
9238: @section The Core Words
9239: @c =====================================================================
9240: @cindex core words, system documentation
9241: @cindex system documentation, core words
9242:
9243: @menu
9244: * core-idef:: Implementation Defined Options
9245: * core-ambcond:: Ambiguous Conditions
9246: * core-other:: Other System Documentation
9247: @end menu
9248:
9249: @c ---------------------------------------------------------------------
9250: @node core-idef, core-ambcond, The Core Words, The Core Words
9251: @subsection Implementation Defined Options
9252: @c ---------------------------------------------------------------------
9253: @cindex core words, implementation-defined options
9254: @cindex implementation-defined options, core words
9255:
9256:
9257: @table @i
9258: @item (Cell) aligned addresses:
9259: @cindex cell-aligned addresses
9260: @cindex aligned addresses
9261: processor-dependent. Gforth's alignment words perform natural alignment
9262: (e.g., an address aligned for a datum of size 8 is divisible by
9263: 8). Unaligned accesses usually result in a @code{-23 THROW}.
9264:
9265: @item @code{EMIT} and non-graphic characters:
9266: @cindex @code{EMIT} and non-graphic characters
9267: @cindex non-graphic characters and @code{EMIT}
9268: The character is output using the C library function (actually, macro)
9269: @code{putc}.
9270:
9271: @item character editing of @code{ACCEPT} and @code{EXPECT}:
9272: @cindex character editing of @code{ACCEPT} and @code{EXPECT}
9273: @cindex editing in @code{ACCEPT} and @code{EXPECT}
9274: @cindex @code{ACCEPT}, editing
9275: @cindex @code{EXPECT}, editing
9276: This is modeled on the GNU readline library (@pxref{Readline
9277: Interaction, , Command Line Editing, readline, The GNU Readline
9278: Library}) with Emacs-like key bindings. @kbd{Tab} deviates a little by
9279: producing a full word completion every time you type it (instead of
9280: producing the common prefix of all completions). @xref{Command-line editing}.
9281:
9282: @item character set:
9283: @cindex character set
9284: The character set of your computer and display device. Gforth is
9285: 8-bit-clean (but some other component in your system may make trouble).
9286:
9287: @item Character-aligned address requirements:
9288: @cindex character-aligned address requirements
9289: installation-dependent. Currently a character is represented by a C
9290: @code{unsigned char}; in the future we might switch to @code{wchar_t}
9291: (Comments on that requested).
9292:
9293: @item character-set extensions and matching of names:
9294: @cindex character-set extensions and matching of names
9295: @cindex case-sensitivity for name lookup
9296: @cindex name lookup, case-sensitivity
9297: @cindex locale and case-sensitivity
9298: Any character except the ASCII NUL character can be used in a
9299: name. Matching is case-insensitive (except in @code{TABLE}s). The
9300: matching is performed using the C function @code{strncasecmp}, whose
9301: function is probably influenced by the locale. E.g., the @code{C} locale
9302: does not know about accents and umlauts, so they are matched
9303: case-sensitively in that locale. For portability reasons it is best to
9304: write programs such that they work in the @code{C} locale. Then one can
9305: use libraries written by a Polish programmer (who might use words
9306: containing ISO Latin-2 encoded characters) and by a French programmer
9307: (ISO Latin-1) in the same program (of course, @code{WORDS} will produce
9308: funny results for some of the words (which ones, depends on the font you
9309: are using)). Also, the locale you prefer may not be available in other
9310: operating systems. Hopefully, Unicode will solve these problems one day.
9311:
9312: @item conditions under which control characters match a space delimiter:
9313: @cindex space delimiters
9314: @cindex control characters as delimiters
9315: If @code{WORD} is called with the space character as a delimiter, all
9316: white-space characters (as identified by the C macro @code{isspace()})
9317: are delimiters. @code{PARSE}, on the other hand, treats space like other
9318: delimiters. @code{SWORD} treats space like @code{WORD}, but behaves
9319: like @code{PARSE} otherwise. @code{(NAME)}, which is used by the outer
9320: interpreter (aka text interpreter) by default, treats all white-space
9321: characters as delimiters.
9322:
9323: @item format of the control-flow stack:
9324: @cindex control-flow stack, format
9325: The data stack is used as control-flow stack. The size of a control-flow
9326: stack item in cells is given by the constant @code{cs-item-size}. At the
9327: time of this writing, an item consists of a (pointer to a) locals list
9328: (third), an address in the code (second), and a tag for identifying the
9329: item (TOS). The following tags are used: @code{defstart},
9330: @code{live-orig}, @code{dead-orig}, @code{dest}, @code{do-dest},
9331: @code{scopestart}.
9332:
9333: @item conversion of digits > 35
9334: @cindex digits > 35
9335: The characters @code{[\]^_'} are the digits with the decimal value
9336: 36@minus{}41. There is no way to input many of the larger digits.
9337:
9338: @item display after input terminates in @code{ACCEPT} and @code{EXPECT}:
9339: @cindex @code{EXPECT}, display after end of input
9340: @cindex @code{ACCEPT}, display after end of input
9341: The cursor is moved to the end of the entered string. If the input is
9342: terminated using the @kbd{Return} key, a space is typed.
9343:
9344: @item exception abort sequence of @code{ABORT"}:
9345: @cindex exception abort sequence of @code{ABORT"}
9346: @cindex @code{ABORT"}, exception abort sequence
9347: The error string is stored into the variable @code{"error} and a
9348: @code{-2 throw} is performed.
9349:
9350: @item input line terminator:
9351: @cindex input line terminator
9352: @cindex line terminator on input
9353: @cindex newline character on input
9354: For interactive input, @kbd{C-m} (CR) and @kbd{C-j} (LF) terminate
9355: lines. One of these characters is typically produced when you type the
9356: @kbd{Enter} or @kbd{Return} key.
9357:
9358: @item maximum size of a counted string:
9359: @cindex maximum size of a counted string
9360: @cindex counted string, maximum size
9361: @code{s" /counted-string" environment? drop .}. Currently 255 characters
9362: on all ports, but this may change.
9363:
9364: @item maximum size of a parsed string:
9365: @cindex maximum size of a parsed string
9366: @cindex parsed string, maximum size
9367: Given by the constant @code{/line}. Currently 255 characters.
9368:
9369: @item maximum size of a definition name, in characters:
9370: @cindex maximum size of a definition name, in characters
9371: @cindex name, maximum length
9372: 31
9373:
9374: @item maximum string length for @code{ENVIRONMENT?}, in characters:
9375: @cindex maximum string length for @code{ENVIRONMENT?}, in characters
9376: @cindex @code{ENVIRONMENT?} string length, maximum
9377: 31
9378:
9379: @item method of selecting the user input device:
9380: @cindex user input device, method of selecting
9381: The user input device is the standard input. There is currently no way to
9382: change it from within Gforth. However, the input can typically be
9383: redirected in the command line that starts Gforth.
9384:
9385: @item method of selecting the user output device:
9386: @cindex user output device, method of selecting
9387: @code{EMIT} and @code{TYPE} output to the file-id stored in the value
9388: @code{outfile-id} (@code{stdout} by default). Gforth uses unbuffered
9389: output when the user output device is a terminal, otherwise the output
9390: is buffered.
9391:
9392: @item methods of dictionary compilation:
9393: What are we expected to document here?
9394:
9395: @item number of bits in one address unit:
9396: @cindex number of bits in one address unit
9397: @cindex address unit, size in bits
9398: @code{s" address-units-bits" environment? drop .}. 8 in all current
9399: ports.
9400:
9401: @item number representation and arithmetic:
9402: @cindex number representation and arithmetic
9403: Processor-dependent. Binary two's complement on all current ports.
9404:
9405: @item ranges for integer types:
9406: @cindex ranges for integer types
9407: @cindex integer types, ranges
9408: Installation-dependent. Make environmental queries for @code{MAX-N},
9409: @code{MAX-U}, @code{MAX-D} and @code{MAX-UD}. The lower bounds for
9410: unsigned (and positive) types is 0. The lower bound for signed types on
9411: two's complement and one's complement machines machines can be computed
9412: by adding 1 to the upper bound.
9413:
9414: @item read-only data space regions:
9415: @cindex read-only data space regions
9416: @cindex data-space, read-only regions
9417: The whole Forth data space is writable.
9418:
9419: @item size of buffer at @code{WORD}:
9420: @cindex size of buffer at @code{WORD}
9421: @cindex @code{WORD} buffer size
9422: @code{PAD HERE - .}. 104 characters on 32-bit machines. The buffer is
9423: shared with the pictured numeric output string. If overwriting
9424: @code{PAD} is acceptable, it is as large as the remaining dictionary
9425: space, although only as much can be sensibly used as fits in a counted
9426: string.
9427:
9428: @item size of one cell in address units:
9429: @cindex cell size
9430: @code{1 cells .}.
9431:
9432: @item size of one character in address units:
9433: @cindex char size
9434: @code{1 chars .}. 1 on all current ports.
9435:
9436: @item size of the keyboard terminal buffer:
9437: @cindex size of the keyboard terminal buffer
9438: @cindex terminal buffer, size
9439: Varies. You can determine the size at a specific time using @code{lp@@
9440: tib - .}. It is shared with the locals stack and TIBs of files that
9441: include the current file. You can change the amount of space for TIBs
9442: and locals stack at Gforth startup with the command line option
9443: @code{-l}.
9444:
9445: @item size of the pictured numeric output buffer:
9446: @cindex size of the pictured numeric output buffer
9447: @cindex pictured numeric output buffer, size
9448: @code{PAD HERE - .}. 104 characters on 32-bit machines. The buffer is
9449: shared with @code{WORD}.
9450:
9451: @item size of the scratch area returned by @code{PAD}:
9452: @cindex size of the scratch area returned by @code{PAD}
9453: @cindex @code{PAD} size
9454: The remainder of dictionary space. @code{unused pad here - - .}.
9455:
9456: @item system case-sensitivity characteristics:
9457: @cindex case-sensitivity characteristics
9458: Dictionary searches are case-insensitive (except in
9459: @code{TABLE}s). However, as explained above under @i{character-set
9460: extensions}, the matching for non-ASCII characters is determined by the
9461: locale you are using. In the default @code{C} locale all non-ASCII
9462: characters are matched case-sensitively.
9463:
9464: @item system prompt:
9465: @cindex system prompt
9466: @cindex prompt
9467: @code{ ok} in interpret state, @code{ compiled} in compile state.
9468:
9469: @item division rounding:
9470: @cindex division rounding
9471: installation dependent. @code{s" floored" environment? drop .}. We leave
9472: the choice to @code{gcc} (what to use for @code{/}) and to you (whether
9473: to use @code{fm/mod}, @code{sm/rem} or simply @code{/}).
9474:
9475: @item values of @code{STATE} when true:
9476: @cindex @code{STATE} values
9477: -1.
9478:
9479: @item values returned after arithmetic overflow:
9480: On two's complement machines, arithmetic is performed modulo
9481: 2**bits-per-cell for single arithmetic and 4**bits-per-cell for double
9482: arithmetic (with appropriate mapping for signed types). Division by zero
9483: typically results in a @code{-55 throw} (Floating-point unidentified
9484: fault), although a @code{-10 throw} (divide by zero) would be more
9485: appropriate.
9486:
9487: @item whether the current definition can be found after @t{DOES>}:
9488: @cindex @t{DOES>}, visibility of current definition
9489: No.
9490:
9491: @end table
9492:
9493: @c ---------------------------------------------------------------------
9494: @node core-ambcond, core-other, core-idef, The Core Words
9495: @subsection Ambiguous conditions
9496: @c ---------------------------------------------------------------------
9497: @cindex core words, ambiguous conditions
9498: @cindex ambiguous conditions, core words
9499:
9500: @table @i
9501:
9502: @item a name is neither a word nor a number:
9503: @cindex name not found
9504: @cindex undefined word
9505: @code{-13 throw} (Undefined word). Actually, @code{-13 bounce}, which
9506: preserves the data and FP stack, so you don't lose more work than
9507: necessary.
9508:
9509: @item a definition name exceeds the maximum length allowed:
9510: @cindex word name too long
9511: @code{-19 throw} (Word name too long)
9512:
9513: @item addressing a region not inside the various data spaces of the forth system:
9514: @cindex Invalid memory address
9515: The stacks, code space and header space are accessible. Machine code space is
9516: typically readable. Accessing other addresses gives results dependent on
9517: the operating system. On decent systems: @code{-9 throw} (Invalid memory
9518: address).
9519:
9520: @item argument type incompatible with parameter:
9521: @cindex argument type mismatch
9522: This is usually not caught. Some words perform checks, e.g., the control
9523: flow words, and issue a @code{ABORT"} or @code{-12 THROW} (Argument type
9524: mismatch).
9525:
9526: @item attempting to obtain the execution token of a word with undefined execution semantics:
9527: @cindex Interpreting a compile-only word, for @code{'} etc.
9528: @cindex execution token of words with undefined execution semantics
9529: @code{-14 throw} (Interpreting a compile-only word). In some cases, you
9530: get an execution token for @code{compile-only-error} (which performs a
9531: @code{-14 throw} when executed).
9532:
9533: @item dividing by zero:
9534: @cindex dividing by zero
9535: @cindex floating point unidentified fault, integer division
9536: On better platforms, this produces a @code{-10 throw} (Division by
9537: zero); on other systems, this typically results in a @code{-55 throw}
9538: (Floating-point unidentified fault).
9539:
9540: @item insufficient data stack or return stack space:
9541: @cindex insufficient data stack or return stack space
9542: @cindex stack overflow
9543: @cindex address alignment exception, stack overflow
9544: @cindex Invalid memory address, stack overflow
9545: Depending on the operating system, the installation, and the invocation
9546: of Gforth, this is either checked by the memory management hardware, or
9547: it is not checked. If it is checked, you typically get a @code{-3 throw}
9548: (Stack overflow), @code{-5 throw} (Return stack overflow), or @code{-9
9549: throw} (Invalid memory address) (depending on the platform and how you
9550: achieved the overflow) as soon as the overflow happens. If it is not
9551: checked, overflows typically result in mysterious illegal memory
9552: accesses, producing @code{-9 throw} (Invalid memory address) or
9553: @code{-23 throw} (Address alignment exception); they might also destroy
9554: the internal data structure of @code{ALLOCATE} and friends, resulting in
9555: various errors in these words.
9556:
9557: @item insufficient space for loop control parameters:
9558: @cindex insufficient space for loop control parameters
9559: like other return stack overflows.
9560:
9561: @item insufficient space in the dictionary:
9562: @cindex insufficient space in the dictionary
9563: @cindex dictionary overflow
9564: If you try to allot (either directly with @code{allot}, or indirectly
9565: with @code{,}, @code{create} etc.) more memory than available in the
9566: dictionary, you get a @code{-8 throw} (Dictionary overflow). If you try
9567: to access memory beyond the end of the dictionary, the results are
9568: similar to stack overflows.
9569:
9570: @item interpreting a word with undefined interpretation semantics:
9571: @cindex interpreting a word with undefined interpretation semantics
9572: @cindex Interpreting a compile-only word
9573: For some words, we have defined interpretation semantics. For the
9574: others: @code{-14 throw} (Interpreting a compile-only word).
9575:
9576: @item modifying the contents of the input buffer or a string literal:
9577: @cindex modifying the contents of the input buffer or a string literal
9578: These are located in writable memory and can be modified.
9579:
9580: @item overflow of the pictured numeric output string:
9581: @cindex overflow of the pictured numeric output string
9582: @cindex pictured numeric output string, overflow
9583: @code{-17 throw} (Pictured numeric ouput string overflow).
9584:
9585: @item parsed string overflow:
9586: @cindex parsed string overflow
9587: @code{PARSE} cannot overflow. @code{WORD} does not check for overflow.
9588:
9589: @item producing a result out of range:
9590: @cindex result out of range
9591: On two's complement machines, arithmetic is performed modulo
9592: 2**bits-per-cell for single arithmetic and 4**bits-per-cell for double
9593: arithmetic (with appropriate mapping for signed types). Division by zero
9594: typically results in a @code{-10 throw} (divide by zero) or @code{-55
9595: throw} (floating point unidentified fault). @code{convert} and
9596: @code{>number} currently overflow silently.
9597:
9598: @item reading from an empty data or return stack:
9599: @cindex stack empty
9600: @cindex stack underflow
9601: @cindex return stack underflow
9602: The data stack is checked by the outer (aka text) interpreter after
9603: every word executed. If it has underflowed, a @code{-4 throw} (Stack
9604: underflow) is performed. Apart from that, stacks may be checked or not,
9605: depending on operating system, installation, and invocation. If they are
9606: caught by a check, they typically result in @code{-4 throw} (Stack
9607: underflow), @code{-6 throw} (Return stack underflow) or @code{-9 throw}
9608: (Invalid memory address), depending on the platform and which stack
9609: underflows and by how much. Note that even if the system uses checking
9610: (through the MMU), your program may have to underflow by a significant
9611: number of stack items to trigger the reaction (the reason for this is
9612: that the MMU, and therefore the checking, works with a page-size
9613: granularity). If there is no checking, the symptoms resulting from an
9614: underflow are similar to those from an overflow. Unbalanced return
9615: stack errors result in a variaty of symptoms, including @code{-9 throw}
9616: (Invalid memory address) and Illegal Instruction (typically @code{-260
9617: throw}).
9618:
9619: @item unexpected end of the input buffer, resulting in an attempt to use a zero-length string as a name:
9620: @cindex unexpected end of the input buffer
9621: @cindex zero-length string as a name
9622: @cindex Attempt to use zero-length string as a name
9623: @code{Create} and its descendants perform a @code{-16 throw} (Attempt to
9624: use zero-length string as a name). Words like @code{'} probably will not
9625: find what they search. Note that it is possible to create zero-length
9626: names with @code{nextname} (should it not?).
9627:
9628: @item @code{>IN} greater than input buffer:
9629: @cindex @code{>IN} greater than input buffer
9630: The next invocation of a parsing word returns a string with length 0.
9631:
9632: @item @code{RECURSE} appears after @code{DOES>}:
9633: @cindex @code{RECURSE} appears after @code{DOES>}
9634: Compiles a recursive call to the defining word, not to the defined word.
9635:
9636: @item argument input source different than current input source for @code{RESTORE-INPUT}:
9637: @cindex argument input source different than current input source for @code{RESTORE-INPUT}
9638: @cindex argument type mismatch, @code{RESTORE-INPUT}
9639: @cindex @code{RESTORE-INPUT}, Argument type mismatch
9640: @code{-12 THROW}. Note that, once an input file is closed (e.g., because
9641: the end of the file was reached), its source-id may be
9642: reused. Therefore, restoring an input source specification referencing a
9643: closed file may lead to unpredictable results instead of a @code{-12
9644: THROW}.
9645:
9646: In the future, Gforth may be able to restore input source specifications
9647: from other than the current input source.
9648:
9649: @item data space containing definitions gets de-allocated:
9650: @cindex data space containing definitions gets de-allocated
9651: Deallocation with @code{allot} is not checked. This typically results in
9652: memory access faults or execution of illegal instructions.
9653:
9654: @item data space read/write with incorrect alignment:
9655: @cindex data space read/write with incorrect alignment
9656: @cindex alignment faults
9657: @cindex address alignment exception
9658: Processor-dependent. Typically results in a @code{-23 throw} (Address
9659: alignment exception). Under Linux-Intel on a 486 or later processor with
9660: alignment turned on, incorrect alignment results in a @code{-9 throw}
9661: (Invalid memory address). There are reportedly some processors with
9662: alignment restrictions that do not report violations.
9663:
9664: @item data space pointer not properly aligned, @code{,}, @code{C,}:
9665: @cindex data space pointer not properly aligned, @code{,}, @code{C,}
9666: Like other alignment errors.
9667:
9668: @item less than u+2 stack items (@code{PICK} and @code{ROLL}):
9669: Like other stack underflows.
9670:
9671: @item loop control parameters not available:
9672: @cindex loop control parameters not available
9673: Not checked. The counted loop words simply assume that the top of return
9674: stack items are loop control parameters and behave accordingly.
9675:
9676: @item most recent definition does not have a name (@code{IMMEDIATE}):
9677: @cindex most recent definition does not have a name (@code{IMMEDIATE})
9678: @cindex last word was headerless
9679: @code{abort" last word was headerless"}.
9680:
9681: @item name not defined by @code{VALUE} used by @code{TO}:
9682: @cindex name not defined by @code{VALUE} used by @code{TO}
9683: @cindex @code{TO} on non-@code{VALUE}s
9684: @cindex Invalid name argument, @code{TO}
9685: @code{-32 throw} (Invalid name argument) (unless name is a local or was
9686: defined by @code{CONSTANT}; in the latter case it just changes the constant).
9687:
9688: @item name not found (@code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]}):
9689: @cindex name not found (@code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]})
9690: @cindex undefined word, @code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]}
9691: @code{-13 throw} (Undefined word)
9692:
9693: @item parameters are not of the same type (@code{DO}, @code{?DO}, @code{WITHIN}):
9694: @cindex parameters are not of the same type (@code{DO}, @code{?DO}, @code{WITHIN})
9695: Gforth behaves as if they were of the same type. I.e., you can predict
9696: the behaviour by interpreting all parameters as, e.g., signed.
9697:
9698: @item @code{POSTPONE} or @code{[COMPILE]} applied to @code{TO}:
9699: @cindex @code{POSTPONE} or @code{[COMPILE]} applied to @code{TO}
9700: Assume @code{: X POSTPONE TO ; IMMEDIATE}. @code{X} performs the
9701: compilation semantics of @code{TO}.
9702:
9703: @item String longer than a counted string returned by @code{WORD}:
9704: @cindex string longer than a counted string returned by @code{WORD}
9705: @cindex @code{WORD}, string overflow
9706: Not checked. The string will be ok, but the count will, of course,
9707: contain only the least significant bits of the length.
9708:
9709: @item u greater than or equal to the number of bits in a cell (@code{LSHIFT}, @code{RSHIFT}):
9710: @cindex @code{LSHIFT}, large shift counts
9711: @cindex @code{RSHIFT}, large shift counts
9712: Processor-dependent. Typical behaviours are returning 0 and using only
9713: the low bits of the shift count.
9714:
9715: @item word not defined via @code{CREATE}:
9716: @cindex @code{>BODY} of non-@code{CREATE}d words
9717: @code{>BODY} produces the PFA of the word no matter how it was defined.
9718:
9719: @cindex @code{DOES>} of non-@code{CREATE}d words
9720: @code{DOES>} changes the execution semantics of the last defined word no
9721: matter how it was defined. E.g., @code{CONSTANT DOES>} is equivalent to
9722: @code{CREATE , DOES>}.
9723:
9724: @item words improperly used outside @code{<#} and @code{#>}:
9725: Not checked. As usual, you can expect memory faults.
9726:
9727: @end table
9728:
9729:
9730: @c ---------------------------------------------------------------------
9731: @node core-other, , core-ambcond, The Core Words
9732: @subsection Other system documentation
9733: @c ---------------------------------------------------------------------
9734: @cindex other system documentation, core words
9735: @cindex core words, other system documentation
9736:
9737: @table @i
9738: @item nonstandard words using @code{PAD}:
9739: @cindex @code{PAD} use by nonstandard words
9740: None.
9741:
9742: @item operator's terminal facilities available:
9743: @cindex operator's terminal facilities available
9744: After processing the command line, Gforth goes into interactive mode,
9745: and you can give commands to Gforth interactively. The actual facilities
9746: available depend on how you invoke Gforth.
9747:
9748: @item program data space available:
9749: @cindex program data space available
9750: @cindex data space available
9751: @code{UNUSED .} gives the remaining dictionary space. The total
9752: dictionary space can be specified with the @code{-m} switch
9753: (@pxref{Invoking Gforth}) when Gforth starts up.
9754:
9755: @item return stack space available:
9756: @cindex return stack space available
9757: You can compute the total return stack space in cells with
9758: @code{s" RETURN-STACK-CELLS" environment? drop .}. You can specify it at
9759: startup time with the @code{-r} switch (@pxref{Invoking Gforth}).
9760:
9761: @item stack space available:
9762: @cindex stack space available
9763: You can compute the total data stack space in cells with
9764: @code{s" STACK-CELLS" environment? drop .}. You can specify it at
9765: startup time with the @code{-d} switch (@pxref{Invoking Gforth}).
9766:
9767: @item system dictionary space required, in address units:
9768: @cindex system dictionary space required, in address units
9769: Type @code{here forthstart - .} after startup. At the time of this
9770: writing, this gives 80080 (bytes) on a 32-bit system.
9771: @end table
9772:
9773:
9774: @c =====================================================================
9775: @node The optional Block word set, The optional Double Number word set, The Core Words, ANS conformance
9776: @section The optional Block word set
9777: @c =====================================================================
9778: @cindex system documentation, block words
9779: @cindex block words, system documentation
9780:
9781: @menu
9782: * block-idef:: Implementation Defined Options
9783: * block-ambcond:: Ambiguous Conditions
9784: * block-other:: Other System Documentation
9785: @end menu
9786:
9787:
9788: @c ---------------------------------------------------------------------
9789: @node block-idef, block-ambcond, The optional Block word set, The optional Block word set
9790: @subsection Implementation Defined Options
9791: @c ---------------------------------------------------------------------
9792: @cindex implementation-defined options, block words
9793: @cindex block words, implementation-defined options
9794:
9795: @table @i
9796: @item the format for display by @code{LIST}:
9797: @cindex @code{LIST} display format
9798: First the screen number is displayed, then 16 lines of 64 characters,
9799: each line preceded by the line number.
9800:
9801: @item the length of a line affected by @code{\}:
9802: @cindex length of a line affected by @code{\}
9803: @cindex @code{\}, line length in blocks
9804: 64 characters.
9805: @end table
9806:
9807:
9808: @c ---------------------------------------------------------------------
9809: @node block-ambcond, block-other, block-idef, The optional Block word set
9810: @subsection Ambiguous conditions
9811: @c ---------------------------------------------------------------------
9812: @cindex block words, ambiguous conditions
9813: @cindex ambiguous conditions, block words
9814:
9815: @table @i
9816: @item correct block read was not possible:
9817: @cindex block read not possible
9818: Typically results in a @code{throw} of some OS-derived value (between
9819: -512 and -2048). If the blocks file was just not long enough, blanks are
9820: supplied for the missing portion.
9821:
9822: @item I/O exception in block transfer:
9823: @cindex I/O exception in block transfer
9824: @cindex block transfer, I/O exception
9825: Typically results in a @code{throw} of some OS-derived value (between
9826: -512 and -2048).
9827:
9828: @item invalid block number:
9829: @cindex invalid block number
9830: @cindex block number invalid
9831: @code{-35 throw} (Invalid block number)
9832:
9833: @item a program directly alters the contents of @code{BLK}:
9834: @cindex @code{BLK}, altering @code{BLK}
9835: The input stream is switched to that other block, at the same
9836: position. If the storing to @code{BLK} happens when interpreting
9837: non-block input, the system will get quite confused when the block ends.
9838:
9839: @item no current block buffer for @code{UPDATE}:
9840: @cindex @code{UPDATE}, no current block buffer
9841: @code{UPDATE} has no effect.
9842:
9843: @end table
9844:
9845: @c ---------------------------------------------------------------------
9846: @node block-other, , block-ambcond, The optional Block word set
9847: @subsection Other system documentation
9848: @c ---------------------------------------------------------------------
9849: @cindex other system documentation, block words
9850: @cindex block words, other system documentation
9851:
9852: @table @i
9853: @item any restrictions a multiprogramming system places on the use of buffer addresses:
9854: No restrictions (yet).
9855:
9856: @item the number of blocks available for source and data:
9857: depends on your disk space.
9858:
9859: @end table
9860:
9861:
9862: @c =====================================================================
9863: @node The optional Double Number word set, The optional Exception word set, The optional Block word set, ANS conformance
9864: @section The optional Double Number word set
9865: @c =====================================================================
9866: @cindex system documentation, double words
9867: @cindex double words, system documentation
9868:
9869: @menu
9870: * double-ambcond:: Ambiguous Conditions
9871: @end menu
9872:
9873:
9874: @c ---------------------------------------------------------------------
9875: @node double-ambcond, , The optional Double Number word set, The optional Double Number word set
9876: @subsection Ambiguous conditions
9877: @c ---------------------------------------------------------------------
9878: @cindex double words, ambiguous conditions
9879: @cindex ambiguous conditions, double words
9880:
9881: @table @i
9882: @item @i{d} outside of range of @i{n} in @code{D>S}:
9883: @cindex @code{D>S}, @i{d} out of range of @i{n}
9884: The least significant cell of @i{d} is produced.
9885:
9886: @end table
9887:
9888:
9889: @c =====================================================================
9890: @node The optional Exception word set, The optional Facility word set, The optional Double Number word set, ANS conformance
9891: @section The optional Exception word set
9892: @c =====================================================================
9893: @cindex system documentation, exception words
9894: @cindex exception words, system documentation
9895:
9896: @menu
9897: * exception-idef:: Implementation Defined Options
9898: @end menu
9899:
9900:
9901: @c ---------------------------------------------------------------------
9902: @node exception-idef, , The optional Exception word set, The optional Exception word set
9903: @subsection Implementation Defined Options
9904: @c ---------------------------------------------------------------------
9905: @cindex implementation-defined options, exception words
9906: @cindex exception words, implementation-defined options
9907:
9908: @table @i
9909: @item @code{THROW}-codes used in the system:
9910: @cindex @code{THROW}-codes used in the system
9911: The codes -256@minus{}-511 are used for reporting signals. The mapping
9912: from OS signal numbers to throw codes is -256@minus{}@i{signal}. The
9913: codes -512@minus{}-2047 are used for OS errors (for file and memory
9914: allocation operations). The mapping from OS error numbers to throw codes
9915: is -512@minus{}@code{errno}. One side effect of this mapping is that
9916: undefined OS errors produce a message with a strange number; e.g.,
9917: @code{-1000 THROW} results in @code{Unknown error 488} on my system.
9918: @end table
9919:
9920: @c =====================================================================
9921: @node The optional Facility word set, The optional File-Access word set, The optional Exception word set, ANS conformance
9922: @section The optional Facility word set
9923: @c =====================================================================
9924: @cindex system documentation, facility words
9925: @cindex facility words, system documentation
9926:
9927: @menu
9928: * facility-idef:: Implementation Defined Options
9929: * facility-ambcond:: Ambiguous Conditions
9930: @end menu
9931:
9932:
9933: @c ---------------------------------------------------------------------
9934: @node facility-idef, facility-ambcond, The optional Facility word set, The optional Facility word set
9935: @subsection Implementation Defined Options
9936: @c ---------------------------------------------------------------------
9937: @cindex implementation-defined options, facility words
9938: @cindex facility words, implementation-defined options
9939:
9940: @table @i
9941: @item encoding of keyboard events (@code{EKEY}):
9942: @cindex keyboard events, encoding in @code{EKEY}
9943: @cindex @code{EKEY}, encoding of keyboard events
9944: Keys corresponding to ASCII characters are encoded as ASCII characters.
9945: Other keys are encoded with the constants @code{k-left}, @code{k-right},
9946: @code{k-up}, @code{k-down}, @code{k-home}, @code{k-end}, @code{k1},
9947: @code{k2}, @code{k3}, @code{k4}, @code{k5}, @code{k6}, @code{k7},
9948: @code{k8}, @code{k9}, @code{k10}, @code{k11}, @code{k12}.
9949:
9950:
9951: @item duration of a system clock tick:
9952: @cindex duration of a system clock tick
9953: @cindex clock tick duration
9954: System dependent. With respect to @code{MS}, the time is specified in
9955: microseconds. How well the OS and the hardware implement this, is
9956: another question.
9957:
9958: @item repeatability to be expected from the execution of @code{MS}:
9959: @cindex repeatability to be expected from the execution of @code{MS}
9960: @cindex @code{MS}, repeatability to be expected
9961: System dependent. On Unix, a lot depends on load. If the system is
9962: lightly loaded, and the delay is short enough that Gforth does not get
9963: swapped out, the performance should be acceptable. Under MS-DOS and
9964: other single-tasking systems, it should be good.
9965:
9966: @end table
9967:
9968:
9969: @c ---------------------------------------------------------------------
9970: @node facility-ambcond, , facility-idef, The optional Facility word set
9971: @subsection Ambiguous conditions
9972: @c ---------------------------------------------------------------------
9973: @cindex facility words, ambiguous conditions
9974: @cindex ambiguous conditions, facility words
9975:
9976: @table @i
9977: @item @code{AT-XY} can't be performed on user output device:
9978: @cindex @code{AT-XY} can't be performed on user output device
9979: Largely terminal dependent. No range checks are done on the arguments.
9980: No errors are reported. You may see some garbage appearing, you may see
9981: simply nothing happen.
9982:
9983: @end table
9984:
9985:
9986: @c =====================================================================
9987: @node The optional File-Access word set, The optional Floating-Point word set, The optional Facility word set, ANS conformance
9988: @section The optional File-Access word set
9989: @c =====================================================================
9990: @cindex system documentation, file words
9991: @cindex file words, system documentation
9992:
9993: @menu
9994: * file-idef:: Implementation Defined Options
9995: * file-ambcond:: Ambiguous Conditions
9996: @end menu
9997:
9998: @c ---------------------------------------------------------------------
9999: @node file-idef, file-ambcond, The optional File-Access word set, The optional File-Access word set
10000: @subsection Implementation Defined Options
10001: @c ---------------------------------------------------------------------
10002: @cindex implementation-defined options, file words
10003: @cindex file words, implementation-defined options
10004:
10005: @table @i
10006: @item file access methods used:
10007: @cindex file access methods used
10008: @code{R/O}, @code{R/W} and @code{BIN} work as you would
10009: expect. @code{W/O} translates into the C file opening mode @code{w} (or
10010: @code{wb}): The file is cleared, if it exists, and created, if it does
10011: not (with both @code{open-file} and @code{create-file}). Under Unix
10012: @code{create-file} creates a file with 666 permissions modified by your
10013: umask.
10014:
10015: @item file exceptions:
10016: @cindex file exceptions
10017: The file words do not raise exceptions (except, perhaps, memory access
10018: faults when you pass illegal addresses or file-ids).
10019:
10020: @item file line terminator:
10021: @cindex file line terminator
10022: System-dependent. Gforth uses C's newline character as line
10023: terminator. What the actual character code(s) of this are is
10024: system-dependent.
10025:
10026: @item file name format:
10027: @cindex file name format
10028: System dependent. Gforth just uses the file name format of your OS.
10029:
10030: @item information returned by @code{FILE-STATUS}:
10031: @cindex @code{FILE-STATUS}, returned information
10032: @code{FILE-STATUS} returns the most powerful file access mode allowed
10033: for the file: Either @code{R/O}, @code{W/O} or @code{R/W}. If the file
10034: cannot be accessed, @code{R/O BIN} is returned. @code{BIN} is applicable
10035: along with the returned mode.
10036:
10037: @item input file state after an exception when including source:
10038: @cindex exception when including source
10039: All files that are left via the exception are closed.
10040:
10041: @item @i{ior} values and meaning:
10042: @cindex @i{ior} values and meaning
10043: The @i{ior}s returned by the file and memory allocation words are
10044: intended as throw codes. They typically are in the range
10045: -512@minus{}-2047 of OS errors. The mapping from OS error numbers to
10046: @i{ior}s is -512@minus{}@i{errno}.
10047:
10048: @item maximum depth of file input nesting:
10049: @cindex maximum depth of file input nesting
10050: @cindex file input nesting, maximum depth
10051: limited by the amount of return stack, locals/TIB stack, and the number
10052: of open files available. This should not give you troubles.
10053:
10054: @item maximum size of input line:
10055: @cindex maximum size of input line
10056: @cindex input line size, maximum
10057: @code{/line}. Currently 255.
10058:
10059: @item methods of mapping block ranges to files:
10060: @cindex mapping block ranges to files
10061: @cindex files containing blocks
10062: @cindex blocks in files
10063: By default, blocks are accessed in the file @file{blocks.fb} in the
10064: current working directory. The file can be switched with @code{USE}.
10065:
10066: @item number of string buffers provided by @code{S"}:
10067: @cindex @code{S"}, number of string buffers
10068: 1
10069:
10070: @item size of string buffer used by @code{S"}:
10071: @cindex @code{S"}, size of string buffer
10072: @code{/line}. currently 255.
10073:
10074: @end table
10075:
10076: @c ---------------------------------------------------------------------
10077: @node file-ambcond, , file-idef, The optional File-Access word set
10078: @subsection Ambiguous conditions
10079: @c ---------------------------------------------------------------------
10080: @cindex file words, ambiguous conditions
10081: @cindex ambiguous conditions, file words
10082:
10083: @table @i
10084: @item attempting to position a file outside its boundaries:
10085: @cindex @code{REPOSITION-FILE}, outside the file's boundaries
10086: @code{REPOSITION-FILE} is performed as usual: Afterwards,
10087: @code{FILE-POSITION} returns the value given to @code{REPOSITION-FILE}.
10088:
10089: @item attempting to read from file positions not yet written:
10090: @cindex reading from file positions not yet written
10091: End-of-file, i.e., zero characters are read and no error is reported.
10092:
10093: @item @i{file-id} is invalid (@code{INCLUDE-FILE}):
10094: @cindex @code{INCLUDE-FILE}, @i{file-id} is invalid
10095: An appropriate exception may be thrown, but a memory fault or other
10096: problem is more probable.
10097:
10098: @item I/O exception reading or closing @i{file-id} (@code{INCLUDE-FILE}, @code{INCLUDED}):
10099: @cindex @code{INCLUDE-FILE}, I/O exception reading or closing @i{file-id}
10100: @cindex @code{INCLUDED}, I/O exception reading or closing @i{file-id}
10101: The @i{ior} produced by the operation, that discovered the problem, is
10102: thrown.
10103:
10104: @item named file cannot be opened (@code{INCLUDED}):
10105: @cindex @code{INCLUDED}, named file cannot be opened
10106: The @i{ior} produced by @code{open-file} is thrown.
10107:
10108: @item requesting an unmapped block number:
10109: @cindex unmapped block numbers
10110: There are no unmapped legal block numbers. On some operating systems,
10111: writing a block with a large number may overflow the file system and
10112: have an error message as consequence.
10113:
10114: @item using @code{source-id} when @code{blk} is non-zero:
10115: @cindex @code{SOURCE-ID}, behaviour when @code{BLK} is non-zero
10116: @code{source-id} performs its function. Typically it will give the id of
10117: the source which loaded the block. (Better ideas?)
10118:
10119: @end table
10120:
10121:
10122: @c =====================================================================
10123: @node The optional Floating-Point word set, The optional Locals word set, The optional File-Access word set, ANS conformance
10124: @section The optional Floating-Point word set
10125: @c =====================================================================
10126: @cindex system documentation, floating-point words
10127: @cindex floating-point words, system documentation
10128:
10129: @menu
10130: * floating-idef:: Implementation Defined Options
10131: * floating-ambcond:: Ambiguous Conditions
10132: @end menu
10133:
10134:
10135: @c ---------------------------------------------------------------------
10136: @node floating-idef, floating-ambcond, The optional Floating-Point word set, The optional Floating-Point word set
10137: @subsection Implementation Defined Options
10138: @c ---------------------------------------------------------------------
10139: @cindex implementation-defined options, floating-point words
10140: @cindex floating-point words, implementation-defined options
10141:
10142: @table @i
10143: @item format and range of floating point numbers:
10144: @cindex format and range of floating point numbers
10145: @cindex floating point numbers, format and range
10146: System-dependent; the @code{double} type of C.
10147:
10148: @item results of @code{REPRESENT} when @i{float} is out of range:
10149: @cindex @code{REPRESENT}, results when @i{float} is out of range
10150: System dependent; @code{REPRESENT} is implemented using the C library
10151: function @code{ecvt()} and inherits its behaviour in this respect.
10152:
10153: @item rounding or truncation of floating-point numbers:
10154: @cindex rounding of floating-point numbers
10155: @cindex truncation of floating-point numbers
10156: @cindex floating-point numbers, rounding or truncation
10157: System dependent; the rounding behaviour is inherited from the hosting C
10158: compiler. IEEE-FP-based (i.e., most) systems by default round to
10159: nearest, and break ties by rounding to even (i.e., such that the last
10160: bit of the mantissa is 0).
10161:
10162: @item size of floating-point stack:
10163: @cindex floating-point stack size
10164: @code{s" FLOATING-STACK" environment? drop .} gives the total size of
10165: the floating-point stack (in floats). You can specify this on startup
10166: with the command-line option @code{-f} (@pxref{Invoking Gforth}).
10167:
10168: @item width of floating-point stack:
10169: @cindex floating-point stack width
10170: @code{1 floats}.
10171:
10172: @end table
10173:
10174:
10175: @c ---------------------------------------------------------------------
10176: @node floating-ambcond, , floating-idef, The optional Floating-Point word set
10177: @subsection Ambiguous conditions
10178: @c ---------------------------------------------------------------------
10179: @cindex floating-point words, ambiguous conditions
10180: @cindex ambiguous conditions, floating-point words
10181:
10182: @table @i
10183: @item @code{df@@} or @code{df!} used with an address that is not double-float aligned:
10184: @cindex @code{df@@} or @code{df!} used with an address that is not double-float aligned
10185: System-dependent. Typically results in a @code{-23 THROW} like other
10186: alignment violations.
10187:
10188: @item @code{f@@} or @code{f!} used with an address that is not float aligned:
10189: @cindex @code{f@@} used with an address that is not float aligned
10190: @cindex @code{f!} used with an address that is not float aligned
10191: System-dependent. Typically results in a @code{-23 THROW} like other
10192: alignment violations.
10193:
10194: @item floating-point result out of range:
10195: @cindex floating-point result out of range
10196: System-dependent. Can result in a @code{-55 THROW} (Floating-point
10197: unidentified fault), or can produce a special value representing, e.g.,
10198: Infinity.
10199:
10200: @item @code{sf@@} or @code{sf!} used with an address that is not single-float aligned:
10201: @cindex @code{sf@@} or @code{sf!} used with an address that is not single-float aligned
10202: System-dependent. Typically results in an alignment fault like other
10203: alignment violations.
10204:
10205: @item @code{base} is not decimal (@code{REPRESENT}, @code{F.}, @code{FE.}, @code{FS.}):
10206: @cindex @code{base} is not decimal (@code{REPRESENT}, @code{F.}, @code{FE.}, @code{FS.})
10207: The floating-point number is converted into decimal nonetheless.
10208:
10209: @item Both arguments are equal to zero (@code{FATAN2}):
10210: @cindex @code{FATAN2}, both arguments are equal to zero
10211: System-dependent. @code{FATAN2} is implemented using the C library
10212: function @code{atan2()}.
10213:
10214: @item Using @code{FTAN} on an argument @i{r1} where cos(@i{r1}) is zero:
10215: @cindex @code{FTAN} on an argument @i{r1} where cos(@i{r1}) is zero
10216: System-dependent. Anyway, typically the cos of @i{r1} will not be zero
10217: because of small errors and the tan will be a very large (or very small)
10218: but finite number.
10219:
10220: @item @i{d} cannot be presented precisely as a float in @code{D>F}:
10221: @cindex @code{D>F}, @i{d} cannot be presented precisely as a float
10222: The result is rounded to the nearest float.
10223:
10224: @item dividing by zero:
10225: @cindex dividing by zero, floating-point
10226: @cindex floating-point dividing by zero
10227: @cindex floating-point unidentified fault, FP divide-by-zero
10228: @code{-55 throw} (Floating-point unidentified fault)
10229:
10230: @item exponent too big for conversion (@code{DF!}, @code{DF@@}, @code{SF!}, @code{SF@@}):
10231: @cindex exponent too big for conversion (@code{DF!}, @code{DF@@}, @code{SF!}, @code{SF@@})
10232: System dependent. On IEEE-FP based systems the number is converted into
10233: an infinity.
10234:
10235: @item @i{float}<1 (@code{FACOSH}):
10236: @cindex @code{FACOSH}, @i{float}<1
10237: @cindex floating-point unidentified fault, @code{FACOSH}
10238: @code{-55 throw} (Floating-point unidentified fault)
10239:
10240: @item @i{float}=<-1 (@code{FLNP1}):
10241: @cindex @code{FLNP1}, @i{float}=<-1
10242: @cindex floating-point unidentified fault, @code{FLNP1}
10243: @code{-55 throw} (Floating-point unidentified fault). On IEEE-FP systems
10244: negative infinity is typically produced for @i{float}=-1.
10245:
10246: @item @i{float}=<0 (@code{FLN}, @code{FLOG}):
10247: @cindex @code{FLN}, @i{float}=<0
10248: @cindex @code{FLOG}, @i{float}=<0
10249: @cindex floating-point unidentified fault, @code{FLN} or @code{FLOG}
10250: @code{-55 throw} (Floating-point unidentified fault). On IEEE-FP systems
10251: negative infinity is typically produced for @i{float}=0.
10252:
10253: @item @i{float}<0 (@code{FASINH}, @code{FSQRT}):
10254: @cindex @code{FASINH}, @i{float}<0
10255: @cindex @code{FSQRT}, @i{float}<0
10256: @cindex floating-point unidentified fault, @code{FASINH} or @code{FSQRT}
10257: @code{-55 throw} (Floating-point unidentified fault). @code{fasinh}
10258: produces values for these inputs on my Linux box (Bug in the C library?)
10259:
10260: @item |@i{float}|>1 (@code{FACOS}, @code{FASIN}, @code{FATANH}):
10261: @cindex @code{FACOS}, |@i{float}|>1
10262: @cindex @code{FASIN}, |@i{float}|>1
10263: @cindex @code{FATANH}, |@i{float}|>1
10264: @cindex floating-point unidentified fault, @code{FACOS}, @code{FASIN} or @code{FATANH}
10265: @code{-55 throw} (Floating-point unidentified fault).
10266:
10267: @item integer part of float cannot be represented by @i{d} in @code{F>D}:
10268: @cindex @code{F>D}, integer part of float cannot be represented by @i{d}
10269: @cindex floating-point unidentified fault, @code{F>D}
10270: @code{-55 throw} (Floating-point unidentified fault).
10271:
10272: @item string larger than pictured numeric output area (@code{f.}, @code{fe.}, @code{fs.}):
10273: @cindex string larger than pictured numeric output area (@code{f.}, @code{fe.}, @code{fs.})
10274: This does not happen.
10275: @end table
10276:
10277: @c =====================================================================
10278: @node The optional Locals word set, The optional Memory-Allocation word set, The optional Floating-Point word set, ANS conformance
10279: @section The optional Locals word set
10280: @c =====================================================================
10281: @cindex system documentation, locals words
10282: @cindex locals words, system documentation
10283:
10284: @menu
10285: * locals-idef:: Implementation Defined Options
10286: * locals-ambcond:: Ambiguous Conditions
10287: @end menu
10288:
10289:
10290: @c ---------------------------------------------------------------------
10291: @node locals-idef, locals-ambcond, The optional Locals word set, The optional Locals word set
10292: @subsection Implementation Defined Options
10293: @c ---------------------------------------------------------------------
10294: @cindex implementation-defined options, locals words
10295: @cindex locals words, implementation-defined options
10296:
10297: @table @i
10298: @item maximum number of locals in a definition:
10299: @cindex maximum number of locals in a definition
10300: @cindex locals, maximum number in a definition
10301: @code{s" #locals" environment? drop .}. Currently 15. This is a lower
10302: bound, e.g., on a 32-bit machine there can be 41 locals of up to 8
10303: characters. The number of locals in a definition is bounded by the size
10304: of locals-buffer, which contains the names of the locals.
10305:
10306: @end table
10307:
10308:
10309: @c ---------------------------------------------------------------------
10310: @node locals-ambcond, , locals-idef, The optional Locals word set
10311: @subsection Ambiguous conditions
10312: @c ---------------------------------------------------------------------
10313: @cindex locals words, ambiguous conditions
10314: @cindex ambiguous conditions, locals words
10315:
10316: @table @i
10317: @item executing a named local in interpretation state:
10318: @cindex local in interpretation state
10319: @cindex Interpreting a compile-only word, for a local
10320: Locals have no interpretation semantics. If you try to perform the
10321: interpretation semantics, you will get a @code{-14 throw} somewhere
10322: (Interpreting a compile-only word). If you perform the compilation
10323: semantics, the locals access will be compiled (irrespective of state).
10324:
10325: @item @i{name} not defined by @code{VALUE} or @code{(LOCAL)} (@code{TO}):
10326: @cindex name not defined by @code{VALUE} or @code{(LOCAL)} used by @code{TO}
10327: @cindex @code{TO} on non-@code{VALUE}s and non-locals
10328: @cindex Invalid name argument, @code{TO}
10329: @code{-32 throw} (Invalid name argument)
10330:
10331: @end table
10332:
10333:
10334: @c =====================================================================
10335: @node The optional Memory-Allocation word set, The optional Programming-Tools word set, The optional Locals word set, ANS conformance
10336: @section The optional Memory-Allocation word set
10337: @c =====================================================================
10338: @cindex system documentation, memory-allocation words
10339: @cindex memory-allocation words, system documentation
10340:
10341: @menu
10342: * memory-idef:: Implementation Defined Options
10343: @end menu
10344:
10345:
10346: @c ---------------------------------------------------------------------
10347: @node memory-idef, , The optional Memory-Allocation word set, The optional Memory-Allocation word set
10348: @subsection Implementation Defined Options
10349: @c ---------------------------------------------------------------------
10350: @cindex implementation-defined options, memory-allocation words
10351: @cindex memory-allocation words, implementation-defined options
10352:
10353: @table @i
10354: @item values and meaning of @i{ior}:
10355: @cindex @i{ior} values and meaning
10356: The @i{ior}s returned by the file and memory allocation words are
10357: intended as throw codes. They typically are in the range
10358: -512@minus{}-2047 of OS errors. The mapping from OS error numbers to
10359: @i{ior}s is -512@minus{}@i{errno}.
10360:
10361: @end table
10362:
10363: @c =====================================================================
10364: @node The optional Programming-Tools word set, The optional Search-Order word set, The optional Memory-Allocation word set, ANS conformance
10365: @section The optional Programming-Tools word set
10366: @c =====================================================================
10367: @cindex system documentation, programming-tools words
10368: @cindex programming-tools words, system documentation
10369:
10370: @menu
10371: * programming-idef:: Implementation Defined Options
10372: * programming-ambcond:: Ambiguous Conditions
10373: @end menu
10374:
10375:
10376: @c ---------------------------------------------------------------------
10377: @node programming-idef, programming-ambcond, The optional Programming-Tools word set, The optional Programming-Tools word set
10378: @subsection Implementation Defined Options
10379: @c ---------------------------------------------------------------------
10380: @cindex implementation-defined options, programming-tools words
10381: @cindex programming-tools words, implementation-defined options
10382:
10383: @table @i
10384: @item ending sequence for input following @code{;CODE} and @code{CODE}:
10385: @cindex @code{;CODE} ending sequence
10386: @cindex @code{CODE} ending sequence
10387: @code{END-CODE}
10388:
10389: @item manner of processing input following @code{;CODE} and @code{CODE}:
10390: @cindex @code{;CODE}, processing input
10391: @cindex @code{CODE}, processing input
10392: The @code{ASSEMBLER} vocabulary is pushed on the search order stack, and
10393: the input is processed by the text interpreter, (starting) in interpret
10394: state.
10395:
10396: @item search order capability for @code{EDITOR} and @code{ASSEMBLER}:
10397: @cindex @code{ASSEMBLER}, search order capability
10398: The ANS Forth search order word set.
10399:
10400: @item source and format of display by @code{SEE}:
10401: @cindex @code{SEE}, source and format of output
10402: The source for @code{see} is the intermediate code used by the inner
10403: interpreter. The current @code{see} tries to output Forth source code
10404: as well as possible.
10405:
10406: @end table
10407:
10408: @c ---------------------------------------------------------------------
10409: @node programming-ambcond, , programming-idef, The optional Programming-Tools word set
10410: @subsection Ambiguous conditions
10411: @c ---------------------------------------------------------------------
10412: @cindex programming-tools words, ambiguous conditions
10413: @cindex ambiguous conditions, programming-tools words
10414:
10415: @table @i
10416:
10417: @item deleting the compilation word list (@code{FORGET}):
10418: @cindex @code{FORGET}, deleting the compilation word list
10419: Not implemented (yet).
10420:
10421: @item fewer than @i{u}+1 items on the control-flow stack (@code{CS-PICK}, @code{CS-ROLL}):
10422: @cindex @code{CS-PICK}, fewer than @i{u}+1 items on the control flow-stack
10423: @cindex @code{CS-ROLL}, fewer than @i{u}+1 items on the control flow-stack
10424: @cindex control-flow stack underflow
10425: This typically results in an @code{abort"} with a descriptive error
10426: message (may change into a @code{-22 throw} (Control structure mismatch)
10427: in the future). You may also get a memory access error. If you are
10428: unlucky, this ambiguous condition is not caught.
10429:
10430: @item @i{name} can't be found (@code{FORGET}):
10431: @cindex @code{FORGET}, @i{name} can't be found
10432: Not implemented (yet).
10433:
10434: @item @i{name} not defined via @code{CREATE}:
10435: @cindex @code{;CODE}, @i{name} not defined via @code{CREATE}
10436: @code{;CODE} behaves like @code{DOES>} in this respect, i.e., it changes
10437: the execution semantics of the last defined word no matter how it was
10438: defined.
10439:
10440: @item @code{POSTPONE} applied to @code{[IF]}:
10441: @cindex @code{POSTPONE} applied to @code{[IF]}
10442: @cindex @code{[IF]} and @code{POSTPONE}
10443: After defining @code{: X POSTPONE [IF] ; IMMEDIATE}. @code{X} is
10444: equivalent to @code{[IF]}.
10445:
10446: @item reaching the end of the input source before matching @code{[ELSE]} or @code{[THEN]}:
10447: @cindex @code{[IF]}, end of the input source before matching @code{[ELSE]} or @code{[THEN]}
10448: Continue in the same state of conditional compilation in the next outer
10449: input source. Currently there is no warning to the user about this.
10450:
10451: @item removing a needed definition (@code{FORGET}):
10452: @cindex @code{FORGET}, removing a needed definition
10453: Not implemented (yet).
10454:
10455: @end table
10456:
10457:
10458: @c =====================================================================
10459: @node The optional Search-Order word set, , The optional Programming-Tools word set, ANS conformance
10460: @section The optional Search-Order word set
10461: @c =====================================================================
10462: @cindex system documentation, search-order words
10463: @cindex search-order words, system documentation
10464:
10465: @menu
10466: * search-idef:: Implementation Defined Options
10467: * search-ambcond:: Ambiguous Conditions
10468: @end menu
10469:
10470:
10471: @c ---------------------------------------------------------------------
10472: @node search-idef, search-ambcond, The optional Search-Order word set, The optional Search-Order word set
10473: @subsection Implementation Defined Options
10474: @c ---------------------------------------------------------------------
10475: @cindex implementation-defined options, search-order words
10476: @cindex search-order words, implementation-defined options
10477:
10478: @table @i
10479: @item maximum number of word lists in search order:
10480: @cindex maximum number of word lists in search order
10481: @cindex search order, maximum depth
10482: @code{s" wordlists" environment? drop .}. Currently 16.
10483:
10484: @item minimum search order:
10485: @cindex minimum search order
10486: @cindex search order, minimum
10487: @code{root root}.
10488:
10489: @end table
10490:
10491: @c ---------------------------------------------------------------------
10492: @node search-ambcond, , search-idef, The optional Search-Order word set
10493: @subsection Ambiguous conditions
10494: @c ---------------------------------------------------------------------
10495: @cindex search-order words, ambiguous conditions
10496: @cindex ambiguous conditions, search-order words
10497:
10498: @table @i
10499: @item changing the compilation word list (during compilation):
10500: @cindex changing the compilation word list (during compilation)
10501: @cindex compilation word list, change before definition ends
10502: The word is entered into the word list that was the compilation word list
10503: at the start of the definition. Any changes to the name field (e.g.,
10504: @code{immediate}) or the code field (e.g., when executing @code{DOES>})
10505: are applied to the latest defined word (as reported by @code{last} or
10506: @code{lastxt}), if possible, irrespective of the compilation word list.
10507:
10508: @item search order empty (@code{previous}):
10509: @cindex @code{previous}, search order empty
10510: @cindex vocstack empty, @code{previous}
10511: @code{abort" Vocstack empty"}.
10512:
10513: @item too many word lists in search order (@code{also}):
10514: @cindex @code{also}, too many word lists in search order
10515: @cindex vocstack full, @code{also}
10516: @code{abort" Vocstack full"}.
10517:
10518: @end table
10519:
10520: @c ***************************************************************
10521: @node Model, Integrating Gforth, ANS conformance, Top
10522: @chapter Model
10523:
10524: This chapter has yet to be written. It will contain information, on
10525: which internal structures you can rely.
10526:
10527: @c ***************************************************************
10528: @node Integrating Gforth, Emacs and Gforth, Model, Top
10529: @chapter Integrating Gforth into C programs
10530:
10531: This is not yet implemented.
10532:
10533: Several people like to use Forth as scripting language for applications
10534: that are otherwise written in C, C++, or some other language.
10535:
10536: The Forth system ATLAST provides facilities for embedding it into
10537: applications; unfortunately it has several disadvantages: most
10538: importantly, it is not based on ANS Forth, and it is apparently dead
10539: (i.e., not developed further and not supported). The facilities
10540: provided by Gforth in this area are inspired by ATLAST's facilities, so
10541: making the switch should not be hard.
10542:
10543: We also tried to design the interface such that it can easily be
10544: implemented by other Forth systems, so that we may one day arrive at a
10545: standardized interface. Such a standard interface would allow you to
10546: replace the Forth system without having to rewrite C code.
10547:
10548: You embed the Gforth interpreter by linking with the library
10549: @code{libgforth.a} (give the compiler the option @code{-lgforth}). All
10550: global symbols in this library that belong to the interface, have the
10551: prefix @code{forth_}. (Global symbols that are used internally have the
10552: prefix @code{gforth_}).
10553:
10554: You can include the declarations of Forth types and the functions and
10555: variables of the interface with @code{#include <forth.h>}.
10556:
10557: Types.
10558:
10559: Variables.
10560:
10561: Data and FP Stack pointer. Area sizes.
10562:
10563: functions.
10564:
10565: forth_init(imagefile)
10566: forth_evaluate(string) exceptions?
10567: forth_goto(address) (or forth_execute(xt)?)
10568: forth_continue() (a corountining mechanism)
10569:
10570: Adding primitives.
10571:
10572: No checking.
10573:
10574: Signals?
10575:
10576: Accessing the Stacks
10577:
10578: @c ******************************************************************
10579: @node Emacs and Gforth, Image Files, Integrating Gforth, Top
10580: @chapter Emacs and Gforth
10581: @cindex Emacs and Gforth
10582:
10583: @cindex @file{gforth.el}
10584: @cindex @file{forth.el}
10585: @cindex Rydqvist, Goran
10586: @cindex comment editing commands
10587: @cindex @code{\}, editing with Emacs
10588: @cindex debug tracer editing commands
10589: @cindex @code{~~}, removal with Emacs
10590: @cindex Forth mode in Emacs
10591: Gforth comes with @file{gforth.el}, an improved version of
10592: @file{forth.el} by Goran Rydqvist (included in the TILE package). The
10593: improvements are:
10594:
10595: @itemize @bullet
10596: @item
10597: A better (but still not perfect) handling of indentation.
10598: @item
10599: Comment paragraph filling (@kbd{M-q})
10600: @item
10601: Commenting (@kbd{C-x \}) and uncommenting (@kbd{C-u C-x \}) of regions
10602: @item
10603: Removal of debugging tracers (@kbd{C-x ~}, @pxref{Debugging}).
10604: @item
10605: Support of the @code{info-lookup} feature for looking up the
10606: documentation of a word.
10607: @end itemize
10608:
10609: I left the stuff I do not use alone, even though some of it only makes
10610: sense for TILE. To get a description of these features, enter Forth mode
10611: and type @kbd{C-h m}.
10612:
10613: @cindex source location of error or debugging output in Emacs
10614: @cindex error output, finding the source location in Emacs
10615: @cindex debugging output, finding the source location in Emacs
10616: In addition, Gforth supports Emacs quite well: The source code locations
10617: given in error messages, debugging output (from @code{~~}) and failed
10618: assertion messages are in the right format for Emacs' compilation mode
10619: (@pxref{Compilation, , Running Compilations under Emacs, emacs, Emacs
10620: Manual}) so the source location corresponding to an error or other
10621: message is only a few keystrokes away (@kbd{C-x `} for the next error,
10622: @kbd{C-c C-c} for the error under the cursor).
10623:
10624: @cindex @file{TAGS} file
10625: @cindex @file{etags.fs}
10626: @cindex viewing the source of a word in Emacs
10627: @cindex @code{require}, placement in files
10628: @cindex @code{include}, placement in files
10629: Also, if you @code{require} @file{etags.fs}, a new @file{TAGS} file will
10630: be produced (@pxref{Tags, , Tags Tables, emacs, Emacs Manual}) that
10631: contains the definitions of all words defined afterwards. You can then
10632: find the source for a word using @kbd{M-.}. Note that emacs can use
10633: several tags files at the same time (e.g., one for the Gforth sources
10634: and one for your program, @pxref{Select Tags Table,,Selecting a Tags
10635: Table,emacs, Emacs Manual}). The TAGS file for the preloaded words is
10636: @file{$(datadir)/gforth/$(VERSION)/TAGS} (e.g.,
10637: @file{/usr/local/share/gforth/0.2.0/TAGS}). To get the best behaviour
10638: with @file{etags.fs}, you should avoid putting definitions both before
10639: and after @code{require} etc., otherwise you will see the same file
10640: visited several times by commands like @code{tags-search}.
10641:
10642: @cindex viewing the documentation of a word in Emacs
10643: @cindex context-sensitive help
10644: Moreover, for words documented in this manual, you can look up the
10645: glossary entry quickly by using @kbd{C-h TAB}
10646: (@code{info-lookup-symbol}, see @pxref{Documentation, ,Documentation
10647: Commands, emacs, Emacs Manual}). This feature requires Emacs 20.3 or
10648: later and does not work for words containing @code{:}.
10649:
10650:
10651: @cindex @file{.emacs}
10652: To get all these benefits, add the following lines to your @file{.emacs}
10653: file:
10654:
10655: @example
10656: (autoload 'forth-mode "gforth.el")
10657: (setq auto-mode-alist (cons '("\\.fs\\'" . forth-mode) auto-mode-alist))
10658: @end example
10659:
10660: @c ******************************************************************
10661: @node Image Files, Engine, Emacs and Gforth, Top
10662: @chapter Image Files
10663: @cindex image file
10664: @cindex @file{.fi} files
10665: @cindex precompiled Forth code
10666: @cindex dictionary in persistent form
10667: @cindex persistent form of dictionary
10668:
10669: An image file is a file containing an image of the Forth dictionary,
10670: i.e., compiled Forth code and data residing in the dictionary. By
10671: convention, we use the extension @code{.fi} for image files.
10672:
10673: @menu
10674: * Image Licensing Issues:: Distribution terms for images.
10675: * Image File Background:: Why have image files?
10676: * Non-Relocatable Image Files:: don't always work.
10677: * Data-Relocatable Image Files:: are better.
10678: * Fully Relocatable Image Files:: better yet.
10679: * Stack and Dictionary Sizes:: Setting the default sizes for an image.
10680: * Running Image Files:: @code{gforth -i @i{file}} or @i{file}.
10681: * Modifying the Startup Sequence:: and turnkey applications.
10682: @end menu
10683:
10684: @node Image Licensing Issues, Image File Background, Image Files, Image Files
10685: @section Image Licensing Issues
10686: @cindex license for images
10687: @cindex image license
10688:
10689: An image created with @code{gforthmi} (@pxref{gforthmi}) or
10690: @code{savesystem} (@pxref{Non-Relocatable Image Files}) includes the
10691: original image; i.e., according to copyright law it is a derived work of
10692: the original image.
10693:
10694: Since Gforth is distributed under the GNU GPL, the newly created image
10695: falls under the GNU GPL, too. In particular, this means that if you
10696: distribute the image, you have to make all of the sources for the image
10697: available, including those you wrote. For details see @ref{License, ,
10698: GNU General Public License (Section 3)}.
10699:
10700: If you create an image with @code{cross} (@pxref{cross.fs}), the image
10701: contains only code compiled from the sources you gave it; if none of
10702: these sources is under the GPL, the terms discussed above do not apply
10703: to the image. However, if your image needs an engine (a gforth binary)
10704: that is under the GPL, you should make sure that you distribute both in
10705: a way that is at most a @emph{mere aggregation}, if you don't want the
10706: terms of the GPL to apply to the image.
10707:
10708: @node Image File Background, Non-Relocatable Image Files, Image Licensing Issues, Image Files
10709: @section Image File Background
10710: @cindex image file background
10711:
10712: Our Forth system consists not only of primitives, but also of
10713: definitions written in Forth. Since the Forth compiler itself belongs to
10714: those definitions, it is not possible to start the system with the
10715: primitives and the Forth source alone. Therefore we provide the Forth
10716: code as an image file in nearly executable form. When Gforth starts up,
10717: a C routine loads the image file into memory, optionally relocates the
10718: addresses, then sets up the memory (stacks etc.) according to
10719: information in the image file, and (finally) starts executing Forth
10720: code.
10721:
10722: The image file variants represent different compromises between the
10723: goals of making it easy to generate image files and making them
10724: portable.
10725:
10726: @cindex relocation at run-time
10727: Win32Forth 3.4 and Mitch Bradley's @code{cforth} use relocation at
10728: run-time. This avoids many of the complications discussed below (image
10729: files are data relocatable without further ado), but costs performance
10730: (one addition per memory access).
10731:
10732: @cindex relocation at load-time
10733: By contrast, the Gforth loader performs relocation at image load time. The
10734: loader also has to replace tokens that represent primitive calls with the
10735: appropriate code-field addresses (or code addresses in the case of
10736: direct threading).
10737:
10738: There are three kinds of image files, with different degrees of
10739: relocatability: non-relocatable, data-relocatable, and fully relocatable
10740: image files.
10741:
10742: @cindex image file loader
10743: @cindex relocating loader
10744: @cindex loader for image files
10745: These image file variants have several restrictions in common; they are
10746: caused by the design of the image file loader:
10747:
10748: @itemize @bullet
10749: @item
10750: There is only one segment; in particular, this means, that an image file
10751: cannot represent @code{ALLOCATE}d memory chunks (and pointers to
10752: them). The contents of the stacks are not represented, either.
10753:
10754: @item
10755: The only kinds of relocation supported are: adding the same offset to
10756: all cells that represent data addresses; and replacing special tokens
10757: with code addresses or with pieces of machine code.
10758:
10759: If any complex computations involving addresses are performed, the
10760: results cannot be represented in the image file. Several applications that
10761: use such computations come to mind:
10762: @itemize @minus
10763: @item
10764: Hashing addresses (or data structures which contain addresses) for table
10765: lookup. If you use Gforth's @code{table}s or @code{wordlist}s for this
10766: purpose, you will have no problem, because the hash tables are
10767: recomputed automatically when the system is started. If you use your own
10768: hash tables, you will have to do something similar.
10769:
10770: @item
10771: There's a cute implementation of doubly-linked lists that uses
10772: @code{XOR}ed addresses. You could represent such lists as singly-linked
10773: in the image file, and restore the doubly-linked representation on
10774: startup.@footnote{In my opinion, though, you should think thrice before
10775: using a doubly-linked list (whatever implementation).}
10776:
10777: @item
10778: The code addresses of run-time routines like @code{docol:} cannot be
10779: represented in the image file (because their tokens would be replaced by
10780: machine code in direct threaded implementations). As a workaround,
10781: compute these addresses at run-time with @code{>code-address} from the
10782: executions tokens of appropriate words (see the definitions of
10783: @code{docol:} and friends in @file{kernel.fs}).
10784:
10785: @item
10786: On many architectures addresses are represented in machine code in some
10787: shifted or mangled form. You cannot put @code{CODE} words that contain
10788: absolute addresses in this form in a relocatable image file. Workarounds
10789: are representing the address in some relative form (e.g., relative to
10790: the CFA, which is present in some register), or loading the address from
10791: a place where it is stored in a non-mangled form.
10792: @end itemize
10793: @end itemize
10794:
10795: @node Non-Relocatable Image Files, Data-Relocatable Image Files, Image File Background, Image Files
10796: @section Non-Relocatable Image Files
10797: @cindex non-relocatable image files
10798: @cindex image file, non-relocatable
10799:
10800: These files are simple memory dumps of the dictionary. They are specific
10801: to the executable (i.e., @file{gforth} file) they were created
10802: with. What's worse, they are specific to the place on which the
10803: dictionary resided when the image was created. Now, there is no
10804: guarantee that the dictionary will reside at the same place the next
10805: time you start Gforth, so there's no guarantee that a non-relocatable
10806: image will work the next time (Gforth will complain instead of crashing,
10807: though).
10808:
10809: You can create a non-relocatable image file with
10810:
10811:
10812: doc-savesystem
10813:
10814:
10815: @node Data-Relocatable Image Files, Fully Relocatable Image Files, Non-Relocatable Image Files, Image Files
10816: @section Data-Relocatable Image Files
10817: @cindex data-relocatable image files
10818: @cindex image file, data-relocatable
10819:
10820: These files contain relocatable data addresses, but fixed code addresses
10821: (instead of tokens). They are specific to the executable (i.e.,
10822: @file{gforth} file) they were created with. For direct threading on some
10823: architectures (e.g., the i386), data-relocatable images do not work. You
10824: get a data-relocatable image, if you use @file{gforthmi} with a
10825: Gforth binary that is not doubly indirect threaded (@pxref{Fully
10826: Relocatable Image Files}).
10827:
10828: @node Fully Relocatable Image Files, Stack and Dictionary Sizes, Data-Relocatable Image Files, Image Files
10829: @section Fully Relocatable Image Files
10830: @cindex fully relocatable image files
10831: @cindex image file, fully relocatable
10832:
10833: @cindex @file{kern*.fi}, relocatability
10834: @cindex @file{gforth.fi}, relocatability
10835: These image files have relocatable data addresses, and tokens for code
10836: addresses. They can be used with different binaries (e.g., with and
10837: without debugging) on the same machine, and even across machines with
10838: the same data formats (byte order, cell size, floating point
10839: format). However, they are usually specific to the version of Gforth
10840: they were created with. The files @file{gforth.fi} and @file{kernl*.fi}
10841: are fully relocatable.
10842:
10843: There are two ways to create a fully relocatable image file:
10844:
10845: @menu
10846: * gforthmi:: The normal way
10847: * cross.fs:: The hard way
10848: @end menu
10849:
10850: @node gforthmi, cross.fs, Fully Relocatable Image Files, Fully Relocatable Image Files
10851: @subsection @file{gforthmi}
10852: @cindex @file{comp-i.fs}
10853: @cindex @file{gforthmi}
10854:
10855: You will usually use @file{gforthmi}. If you want to create an
10856: image @i{file} that contains everything you would load by invoking
10857: Gforth with @code{gforth @i{options}}, you simply say:
10858: @example
10859: gforthmi @i{file} @i{options}
10860: @end example
10861:
10862: E.g., if you want to create an image @file{asm.fi} that has the file
10863: @file{asm.fs} loaded in addition to the usual stuff, you could do it
10864: like this:
10865:
10866: @example
10867: gforthmi asm.fi asm.fs
10868: @end example
10869:
10870: @file{gforthmi} is implemented as a sh script and works like this: It
10871: produces two non-relocatable images for different addresses and then
10872: compares them. Its output reflects this: first you see the output (if
10873: any) of the two Gforth invocations that produce the nonrelocatable image
10874: files, then you see the output of the comparing program: It displays the
10875: offset used for data addresses and the offset used for code addresses;
10876: moreover, for each cell that cannot be represented correctly in the
10877: image files, it displays a line like this:
10878:
10879: @example
10880: 78DC BFFFFA50 BFFFFA40
10881: @end example
10882:
10883: This means that at offset $78dc from @code{forthstart}, one input image
10884: contains $bffffa50, and the other contains $bffffa40. Since these cells
10885: cannot be represented correctly in the output image, you should examine
10886: these places in the dictionary and verify that these cells are dead
10887: (i.e., not read before they are written).
10888:
10889: @cindex --application, @code{gforthmi} option
10890: If you insert the option @code{--application} in front of the image file
10891: name, you will get an image that uses the @code{--appl-image} option
10892: instead of the @code{--image-file} option (@pxref{Invoking
10893: Gforth}). When you execute such an image on Unix (by typing the image
10894: name as command), the Gforth engine will pass all options to the image
10895: instead of trying to interpret them as engine options.
10896:
10897: If you type @file{gforthmi} with no arguments, it prints some usage
10898: instructions.
10899:
10900: @cindex @code{savesystem} during @file{gforthmi}
10901: @cindex @code{bye} during @file{gforthmi}
10902: @cindex doubly indirect threaded code
10903: @cindex environment variables
10904: @cindex @code{GFORTHD} -- environment variable
10905: @cindex @code{GFORTH} -- environment variable
10906: @cindex @code{gforth-ditc}
10907: There are a few wrinkles: After processing the passed @i{options}, the
10908: words @code{savesystem} and @code{bye} must be visible. A special doubly
10909: indirect threaded version of the @file{gforth} executable is used for
10910: creating the nonrelocatable images; you can pass the exact filename of
10911: this executable through the environment variable @code{GFORTHD}
10912: (default: @file{gforth-ditc}); if you pass a version that is not doubly
10913: indirect threaded, you will not get a fully relocatable image, but a
10914: data-relocatable image (because there is no code address offset). The
10915: normal @file{gforth} executable is used for creating the relocatable
10916: image; you can pass the exact filename of this executable through the
10917: environment variable @code{GFORTH}.
10918:
10919: @node cross.fs, , gforthmi, Fully Relocatable Image Files
10920: @subsection @file{cross.fs}
10921: @cindex @file{cross.fs}
10922: @cindex cross-compiler
10923: @cindex metacompiler
10924:
10925: You can also use @code{cross}, a batch compiler that accepts a Forth-like
10926: programming language. This @code{cross} language has to be documented
10927: yet.
10928:
10929: @cindex target compiler
10930: @code{cross} also allows you to create image files for machines with
10931: different data sizes and data formats than the one used for generating
10932: the image file. You can also use it to create an application image that
10933: does not contain a Forth compiler. These features are bought with
10934: restrictions and inconveniences in programming. E.g., addresses have to
10935: be stored in memory with special words (@code{A!}, @code{A,}, etc.) in
10936: order to make the code relocatable.
10937:
10938:
10939: @node Stack and Dictionary Sizes, Running Image Files, Fully Relocatable Image Files, Image Files
10940: @section Stack and Dictionary Sizes
10941: @cindex image file, stack and dictionary sizes
10942: @cindex dictionary size default
10943: @cindex stack size default
10944:
10945: If you invoke Gforth with a command line flag for the size
10946: (@pxref{Invoking Gforth}), the size you specify is stored in the
10947: dictionary. If you save the dictionary with @code{savesystem} or create
10948: an image with @file{gforthmi}, this size will become the default
10949: for the resulting image file. E.g., the following will create a
10950: fully relocatable version of @file{gforth.fi} with a 1MB dictionary:
10951:
10952: @example
10953: gforthmi gforth.fi -m 1M
10954: @end example
10955:
10956: In other words, if you want to set the default size for the dictionary
10957: and the stacks of an image, just invoke @file{gforthmi} with the
10958: appropriate options when creating the image.
10959:
10960: @cindex stack size, cache-friendly
10961: Note: For cache-friendly behaviour (i.e., good performance), you should
10962: make the sizes of the stacks modulo, say, 2K, somewhat different. E.g.,
10963: the default stack sizes are: data: 16k (mod 2k=0); fp: 15.5k (mod
10964: 2k=1.5k); return: 15k(mod 2k=1k); locals: 14.5k (mod 2k=0.5k).
10965:
10966: @node Running Image Files, Modifying the Startup Sequence, Stack and Dictionary Sizes, Image Files
10967: @section Running Image Files
10968: @cindex running image files
10969: @cindex invoking image files
10970: @cindex image file invocation
10971:
10972: @cindex -i, invoke image file
10973: @cindex --image file, invoke image file
10974: You can invoke Gforth with an image file @i{image} instead of the
10975: default @file{gforth.fi} with the @code{-i} flag (@pxref{Invoking Gforth}):
10976: @example
10977: gforth -i @i{image}
10978: @end example
10979:
10980: @cindex executable image file
10981: @cindex image file, executable
10982: If your operating system supports starting scripts with a line of the
10983: form @code{#! ...}, you just have to type the image file name to start
10984: Gforth with this image file (note that the file extension @code{.fi} is
10985: just a convention). I.e., to run Gforth with the image file @i{image},
10986: you can just type @i{image} instead of @code{gforth -i @i{image}}.
10987: This works because every @code{.fi} file starts with a line of this
10988: format:
10989:
10990: @example
10991: #! /usr/local/bin/gforth-0.4.0 -i
10992: @end example
10993:
10994: The file and pathname for the Gforth engine specified on this line is
10995: the specific Gforth executable that it was built against; i.e. the value
10996: of the environment variable @code{GFORTH} at the time that
10997: @file{gforthmi} was executed.
10998:
10999: You can make use of the same shell capability to make a Forth source
11000: file into an executable. For example, if you place this text in a file:
11001:
11002: @example
11003: #! /usr/local/bin/gforth
11004:
11005: ." Hello, world" CR
11006: bye
11007: @end example
11008:
11009: @noindent
11010: and then make the file executable (chmod +x in Unix), you can run it
11011: directly from the command line. The sequence @code{#!} is used in two
11012: ways; firstly, it is recognised as a ``magic sequence'' by the operating
11013: system@footnote{The Unix kernel actually recognises two types of files:
11014: executable files and files of data, where the data is processed by an
11015: interpreter that is specified on the ``interpreter line'' -- the first
11016: line of the file, starting with the sequence #!. There may be a small
11017: limit (e.g., 32) on the number of characters that may be specified on
11018: the interpreter line.} secondly it is treated as a comment character by
11019: Gforth. Because of the second usage, a space is required between
11020: @code{#!} and the path to the executable.
11021:
11022: The disadvantage of this latter technique, compared with using
11023: @file{gforthmi}, is that it is slower; the Forth source code is compiled
11024: on-the-fly, each time the program is invoked.
11025:
11026:
11027: doc-#!
11028:
11029:
11030: @node Modifying the Startup Sequence, , Running Image Files, Image Files
11031: @section Modifying the Startup Sequence
11032: @cindex startup sequence for image file
11033: @cindex image file initialization sequence
11034: @cindex initialization sequence of image file
11035:
11036: You can add your own initialization to the startup sequence through the
11037: deferred word @code{'cold}. @code{'cold} is invoked just before the
11038: image-specific command line processing (by default, loading files and
11039: evaluating (@code{-e}) strings) starts.
11040:
11041: A sequence for adding your initialization usually looks like this:
11042:
11043: @example
11044: :noname
11045: Defers 'cold \ do other initialization stuff (e.g., rehashing wordlists)
11046: ... \ your stuff
11047: ; IS 'cold
11048: @end example
11049:
11050: @cindex turnkey image files
11051: @cindex image file, turnkey applications
11052: You can make a turnkey image by letting @code{'cold} execute a word
11053: (your turnkey application) that never returns; instead, it exits Gforth
11054: via @code{bye} or @code{throw}.
11055:
11056: @cindex command-line arguments, access
11057: @cindex arguments on the command line, access
11058: You can access the (image-specific) command-line arguments through the
11059: variables @code{argc} and @code{argv}. @code{arg} provides convenient
11060: access to @code{argv}.
11061:
11062: If @code{'cold} exits normally, Gforth processes the command-line
11063: arguments as files to be loaded and strings to be evaluated. Therefore,
11064: @code{'cold} should remove the arguments it has used in this case.
11065:
11066:
11067:
11068: doc-'cold
11069: doc-argc
11070: doc-argv
11071: doc-arg
11072:
11073:
11074:
11075: @c ******************************************************************
11076: @node Engine, Binding to System Library, Image Files, Top
11077: @chapter Engine
11078: @cindex engine
11079: @cindex virtual machine
11080:
11081: Reading this chapter is not necessary for programming with Gforth. It
11082: may be helpful for finding your way in the Gforth sources.
11083:
11084: The ideas in this section have also been published in the papers
11085: @cite{ANS fig/GNU/??? Forth} (in German) by Bernd Paysan, presented at
11086: the Forth-Tagung '93 and @cite{A Portable Forth Engine} by M. Anton
11087: Ertl, presented at EuroForth '93; the latter is available at
11088: @*@url{http://www.complang.tuwien.ac.at/papers/ertl93.ps.Z}.
11089:
11090: @menu
11091: * Portability::
11092: * Threading::
11093: * Primitives::
11094: * Performance::
11095: @end menu
11096:
11097: @node Portability, Threading, Engine, Engine
11098: @section Portability
11099: @cindex engine portability
11100:
11101: An important goal of the Gforth Project is availability across a wide
11102: range of personal machines. fig-Forth, and, to a lesser extent, F83,
11103: achieved this goal by manually coding the engine in assembly language
11104: for several then-popular processors. This approach is very
11105: labor-intensive and the results are short-lived due to progress in
11106: computer architecture.
11107:
11108: @cindex C, using C for the engine
11109: Others have avoided this problem by coding in C, e.g., Mitch Bradley
11110: (cforth), Mikael Patel (TILE) and Dirk Zoller (pfe). This approach is
11111: particularly popular for UNIX-based Forths due to the large variety of
11112: architectures of UNIX machines. Unfortunately an implementation in C
11113: does not mix well with the goals of efficiency and with using
11114: traditional techniques: Indirect or direct threading cannot be expressed
11115: in C, and switch threading, the fastest technique available in C, is
11116: significantly slower. Another problem with C is that it is very
11117: cumbersome to express double integer arithmetic.
11118:
11119: @cindex GNU C for the engine
11120: @cindex long long
11121: Fortunately, there is a portable language that does not have these
11122: limitations: GNU C, the version of C processed by the GNU C compiler
11123: (@pxref{C Extensions, , Extensions to the C Language Family, gcc.info,
11124: GNU C Manual}). Its labels as values feature (@pxref{Labels as Values, ,
11125: Labels as Values, gcc.info, GNU C Manual}) makes direct and indirect
11126: threading possible, its @code{long long} type (@pxref{Long Long, ,
11127: Double-Word Integers, gcc.info, GNU C Manual}) corresponds to Forth's
11128: double numbers@footnote{Unfortunately, long longs are not implemented
11129: properly on all machines (e.g., on alpha-osf1, long longs are only 64
11130: bits, the same size as longs (and pointers), but they should be twice as
11131: long according to @pxref{Long Long, , Double-Word Integers, gcc.info, GNU
11132: C Manual}). So, we had to implement doubles in C after all. Still, on
11133: most machines we can use long longs and achieve better performance than
11134: with the emulation package.}. GNU C is available for free on all
11135: important (and many unimportant) UNIX machines, VMS, 80386s running
11136: MS-DOS, the Amiga, and the Atari ST, so a Forth written in GNU C can run
11137: on all these machines.
11138:
11139: Writing in a portable language has the reputation of producing code that
11140: is slower than assembly. For our Forth engine we repeatedly looked at
11141: the code produced by the compiler and eliminated most compiler-induced
11142: inefficiencies by appropriate changes in the source code.
11143:
11144: @cindex explicit register declarations
11145: @cindex --enable-force-reg, configuration flag
11146: @cindex -DFORCE_REG
11147: However, register allocation cannot be portably influenced by the
11148: programmer, leading to some inefficiencies on register-starved
11149: machines. We use explicit register declarations (@pxref{Explicit Reg
11150: Vars, , Variables in Specified Registers, gcc.info, GNU C Manual}) to
11151: improve the speed on some machines. They are turned on by using the
11152: configuration flag @code{--enable-force-reg} (@code{gcc} switch
11153: @code{-DFORCE_REG}). Unfortunately, this feature not only depends on the
11154: machine, but also on the compiler version: On some machines some
11155: compiler versions produce incorrect code when certain explicit register
11156: declarations are used. So by default @code{-DFORCE_REG} is not used.
11157:
11158: @node Threading, Primitives, Portability, Engine
11159: @section Threading
11160: @cindex inner interpreter implementation
11161: @cindex threaded code implementation
11162:
11163: @cindex labels as values
11164: GNU C's labels as values extension (available since @code{gcc-2.0},
11165: @pxref{Labels as Values, , Labels as Values, gcc.info, GNU C Manual})
11166: makes it possible to take the address of @i{label} by writing
11167: @code{&&@i{label}}. This address can then be used in a statement like
11168: @code{goto *@i{address}}. I.e., @code{goto *&&x} is the same as
11169: @code{goto x}.
11170:
11171: @cindex @code{NEXT}, indirect threaded
11172: @cindex indirect threaded inner interpreter
11173: @cindex inner interpreter, indirect threaded
11174: With this feature an indirect threaded @code{NEXT} looks like:
11175: @example
11176: cfa = *ip++;
11177: ca = *cfa;
11178: goto *ca;
11179: @end example
11180: @cindex instruction pointer
11181: For those unfamiliar with the names: @code{ip} is the Forth instruction
11182: pointer; the @code{cfa} (code-field address) corresponds to ANS Forths
11183: execution token and points to the code field of the next word to be
11184: executed; The @code{ca} (code address) fetched from there points to some
11185: executable code, e.g., a primitive or the colon definition handler
11186: @code{docol}.
11187:
11188: @cindex @code{NEXT}, direct threaded
11189: @cindex direct threaded inner interpreter
11190: @cindex inner interpreter, direct threaded
11191: Direct threading is even simpler:
11192: @example
11193: ca = *ip++;
11194: goto *ca;
11195: @end example
11196:
11197: Of course we have packaged the whole thing neatly in macros called
11198: @code{NEXT} and @code{NEXT1} (the part of @code{NEXT} after fetching the cfa).
11199:
11200: @menu
11201: * Scheduling::
11202: * Direct or Indirect Threaded?::
11203: * DOES>::
11204: @end menu
11205:
11206: @node Scheduling, Direct or Indirect Threaded?, Threading, Threading
11207: @subsection Scheduling
11208: @cindex inner interpreter optimization
11209:
11210: There is a little complication: Pipelined and superscalar processors,
11211: i.e., RISC and some modern CISC machines can process independent
11212: instructions while waiting for the results of an instruction. The
11213: compiler usually reorders (schedules) the instructions in a way that
11214: achieves good usage of these delay slots. However, on our first tries
11215: the compiler did not do well on scheduling primitives. E.g., for
11216: @code{+} implemented as
11217: @example
11218: n=sp[0]+sp[1];
11219: sp++;
11220: sp[0]=n;
11221: NEXT;
11222: @end example
11223: the @code{NEXT} comes strictly after the other code, i.e., there is nearly no
11224: scheduling. After a little thought the problem becomes clear: The
11225: compiler cannot know that @code{sp} and @code{ip} point to different
11226: addresses (and the version of @code{gcc} we used would not know it even
11227: if it was possible), so it could not move the load of the cfa above the
11228: store to the TOS. Indeed the pointers could be the same, if code on or
11229: very near the top of stack were executed. In the interest of speed we
11230: chose to forbid this probably unused ``feature'' and helped the compiler
11231: in scheduling: @code{NEXT} is divided into the loading part (@code{NEXT_P1})
11232: and the goto part (@code{NEXT_P2}). @code{+} now looks like:
11233: @example
11234: n=sp[0]+sp[1];
11235: sp++;
11236: NEXT_P1;
11237: sp[0]=n;
11238: NEXT_P2;
11239: @end example
11240: This can be scheduled optimally by the compiler.
11241:
11242: This division can be turned off with the switch @code{-DCISC_NEXT}. This
11243: switch is on by default on machines that do not profit from scheduling
11244: (e.g., the 80386), in order to preserve registers.
11245:
11246: @node Direct or Indirect Threaded?, DOES>, Scheduling, Threading
11247: @subsection Direct or Indirect Threaded?
11248: @cindex threading, direct or indirect?
11249:
11250: @cindex -DDIRECT_THREADED
11251: Both! After packaging the nasty details in macro definitions we
11252: realized that we could switch between direct and indirect threading by
11253: simply setting a compilation flag (@code{-DDIRECT_THREADED}) and
11254: defining a few machine-specific macros for the direct-threading case.
11255: On the Forth level we also offer access words that hide the
11256: differences between the threading methods (@pxref{Threading Words}).
11257:
11258: Indirect threading is implemented completely machine-independently.
11259: Direct threading needs routines for creating jumps to the executable
11260: code (e.g. to @code{docol} or @code{dodoes}). These routines are inherently
11261: machine-dependent, but they do not amount to many source lines. Therefore,
11262: even porting direct threading to a new machine requires little effort.
11263:
11264: @cindex --enable-indirect-threaded, configuration flag
11265: @cindex --enable-direct-threaded, configuration flag
11266: The default threading method is machine-dependent. You can enforce a
11267: specific threading method when building Gforth with the configuration
11268: flag @code{--enable-direct-threaded} or
11269: @code{--enable-indirect-threaded}. Note that direct threading is not
11270: supported on all machines.
11271:
11272: @node DOES>, , Direct or Indirect Threaded?, Threading
11273: @subsection DOES>
11274: @cindex @code{DOES>} implementation
11275:
11276: @cindex @code{dodoes} routine
11277: @cindex @code{DOES>}-code
11278: One of the most complex parts of a Forth engine is @code{dodoes}, i.e.,
11279: the chunk of code executed by every word defined by a
11280: @code{CREATE}...@code{DOES>} pair. The main problem here is: How to find
11281: the Forth code to be executed, i.e. the code after the
11282: @code{DOES>} (the @code{DOES>}-code)? There are two solutions:
11283:
11284: In fig-Forth the code field points directly to the @code{dodoes} and the
11285: @code{DOES>}-code address is stored in the cell after the code address (i.e. at
11286: @code{@i{CFA} cell+}). It may seem that this solution is illegal in
11287: the Forth-79 and all later standards, because in fig-Forth this address
11288: lies in the body (which is illegal in these standards). However, by
11289: making the code field larger for all words this solution becomes legal
11290: again. We use this approach for the indirect threaded version and for
11291: direct threading on some machines. Leaving a cell unused in most words
11292: is a bit wasteful, but on the machines we are targeting this is hardly a
11293: problem. The other reason for having a code field size of two cells is
11294: to avoid having different image files for direct and indirect threaded
11295: systems (direct threaded systems require two-cell code fields on many
11296: machines).
11297:
11298: @cindex @code{DOES>}-handler
11299: The other approach is that the code field points or jumps to the cell
11300: after @code{DOES>}. In this variant there is a jump to @code{dodoes} at
11301: this address (the @code{DOES>}-handler). @code{dodoes} can then get the
11302: @code{DOES>}-code address by computing the code address, i.e., the address of
11303: the jump to @code{dodoes}, and add the length of that jump field. A variant of
11304: this is to have a call to @code{dodoes} after the @code{DOES>}; then the
11305: return address (which can be found in the return register on RISCs) is
11306: the @code{DOES>}-code address. Since the two cells available in the code field
11307: are used up by the jump to the code address in direct threading on many
11308: architectures, we use this approach for direct threading on these
11309: architectures. We did not want to add another cell to the code field.
11310:
11311: @node Primitives, Performance, Threading, Engine
11312: @section Primitives
11313: @cindex primitives, implementation
11314: @cindex virtual machine instructions, implementation
11315:
11316: @menu
11317: * Automatic Generation::
11318: * TOS Optimization::
11319: * Produced code::
11320: @end menu
11321:
11322: @node Automatic Generation, TOS Optimization, Primitives, Primitives
11323: @subsection Automatic Generation
11324: @cindex primitives, automatic generation
11325:
11326: @cindex @file{prims2x.fs}
11327: Since the primitives are implemented in a portable language, there is no
11328: longer any need to minimize the number of primitives. On the contrary,
11329: having many primitives has an advantage: speed. In order to reduce the
11330: number of errors in primitives and to make programming them easier, we
11331: provide a tool, the primitive generator (@file{prims2x.fs}), that
11332: automatically generates most (and sometimes all) of the C code for a
11333: primitive from the stack effect notation. The source for a primitive
11334: has the following form:
11335:
11336: @cindex primitive source format
11337: @format
11338: @i{Forth-name} @i{stack-effect} @i{category} [@i{pronounc.}]
11339: [@code{""}@i{glossary entry}@code{""}]
11340: @i{C code}
11341: [@code{:}
11342: @i{Forth code}]
11343: @end format
11344:
11345: The items in brackets are optional. The category and glossary fields
11346: are there for generating the documentation, the Forth code is there
11347: for manual implementations on machines without GNU C. E.g., the source
11348: for the primitive @code{+} is:
11349: @example
11350: + n1 n2 -- n core plus
11351: n = n1+n2;
11352: @end example
11353:
11354: This looks like a specification, but in fact @code{n = n1+n2} is C
11355: code. Our primitive generation tool extracts a lot of information from
11356: the stack effect notations@footnote{We use a one-stack notation, even
11357: though we have separate data and floating-point stacks; The separate
11358: notation can be generated easily from the unified notation.}: The number
11359: of items popped from and pushed on the stack, their type, and by what
11360: name they are referred to in the C code. It then generates a C code
11361: prelude and postlude for each primitive. The final C code for @code{+}
11362: looks like this:
11363:
11364: @example
11365: I_plus: /* + ( n1 n2 -- n ) */ /* label, stack effect */
11366: /* */ /* documentation */
11367: @{
11368: DEF_CA /* definition of variable ca (indirect threading) */
11369: Cell n1; /* definitions of variables */
11370: Cell n2;
11371: Cell n;
11372: n1 = (Cell) sp[1]; /* input */
11373: n2 = (Cell) TOS;
11374: sp += 1; /* stack adjustment */
11375: NAME("+") /* debugging output (with -DDEBUG) */
11376: @{
11377: n = n1+n2; /* C code taken from the source */
11378: @}
11379: NEXT_P1; /* NEXT part 1 */
11380: TOS = (Cell)n; /* output */
11381: NEXT_P2; /* NEXT part 2 */
11382: @}
11383: @end example
11384:
11385: This looks long and inefficient, but the GNU C compiler optimizes quite
11386: well and produces optimal code for @code{+} on, e.g., the R3000 and the
11387: HP RISC machines: Defining the @code{n}s does not produce any code, and
11388: using them as intermediate storage also adds no cost.
11389:
11390: There are also other optimizations that are not illustrated by this
11391: example: assignments between simple variables are usually for free (copy
11392: propagation). If one of the stack items is not used by the primitive
11393: (e.g. in @code{drop}), the compiler eliminates the load from the stack
11394: (dead code elimination). On the other hand, there are some things that
11395: the compiler does not do, therefore they are performed by
11396: @file{prims2x.fs}: The compiler does not optimize code away that stores
11397: a stack item to the place where it just came from (e.g., @code{over}).
11398:
11399: While programming a primitive is usually easy, there are a few cases
11400: where the programmer has to take the actions of the generator into
11401: account, most notably @code{?dup}, but also words that do not (always)
11402: fall through to @code{NEXT}.
11403:
11404: @node TOS Optimization, Produced code, Automatic Generation, Primitives
11405: @subsection TOS Optimization
11406: @cindex TOS optimization for primitives
11407: @cindex primitives, keeping the TOS in a register
11408:
11409: An important optimization for stack machine emulators, e.g., Forth
11410: engines, is keeping one or more of the top stack items in
11411: registers. If a word has the stack effect @i{in1}...@i{inx} @code{--}
11412: @i{out1}...@i{outy}, keeping the top @i{n} items in registers
11413: @itemize @bullet
11414: @item
11415: is better than keeping @i{n-1} items, if @i{x>=n} and @i{y>=n},
11416: due to fewer loads from and stores to the stack.
11417: @item is slower than keeping @i{n-1} items, if @i{x<>y} and @i{x<n} and
11418: @i{y<n}, due to additional moves between registers.
11419: @end itemize
11420:
11421: @cindex -DUSE_TOS
11422: @cindex -DUSE_NO_TOS
11423: In particular, keeping one item in a register is never a disadvantage,
11424: if there are enough registers. Keeping two items in registers is a
11425: disadvantage for frequent words like @code{?branch}, constants,
11426: variables, literals and @code{i}. Therefore our generator only produces
11427: code that keeps zero or one items in registers. The generated C code
11428: covers both cases; the selection between these alternatives is made at
11429: C-compile time using the switch @code{-DUSE_TOS}. @code{TOS} in the C
11430: code for @code{+} is just a simple variable name in the one-item case,
11431: otherwise it is a macro that expands into @code{sp[0]}. Note that the
11432: GNU C compiler tries to keep simple variables like @code{TOS} in
11433: registers, and it usually succeeds, if there are enough registers.
11434:
11435: @cindex -DUSE_FTOS
11436: @cindex -DUSE_NO_FTOS
11437: The primitive generator performs the TOS optimization for the
11438: floating-point stack, too (@code{-DUSE_FTOS}). For floating-point
11439: operations the benefit of this optimization is even larger:
11440: floating-point operations take quite long on most processors, but can be
11441: performed in parallel with other operations as long as their results are
11442: not used. If the FP-TOS is kept in a register, this works. If
11443: it is kept on the stack, i.e., in memory, the store into memory has to
11444: wait for the result of the floating-point operation, lengthening the
11445: execution time of the primitive considerably.
11446:
11447: The TOS optimization makes the automatic generation of primitives a
11448: bit more complicated. Just replacing all occurrences of @code{sp[0]} by
11449: @code{TOS} is not sufficient. There are some special cases to
11450: consider:
11451: @itemize @bullet
11452: @item In the case of @code{dup ( w -- w w )} the generator must not
11453: eliminate the store to the original location of the item on the stack,
11454: if the TOS optimization is turned on.
11455: @item Primitives with stack effects of the form @code{--}
11456: @i{out1}...@i{outy} must store the TOS to the stack at the start.
11457: Likewise, primitives with the stack effect @i{in1}...@i{inx} @code{--}
11458: must load the TOS from the stack at the end. But for the null stack
11459: effect @code{--} no stores or loads should be generated.
11460: @end itemize
11461:
11462: @node Produced code, , TOS Optimization, Primitives
11463: @subsection Produced code
11464: @cindex primitives, assembly code listing
11465:
11466: @cindex @file{engine.s}
11467: To see what assembly code is produced for the primitives on your machine
11468: with your compiler and your flag settings, type @code{make engine.s} and
11469: look at the resulting file @file{engine.s}.
11470:
11471: @node Performance, , Primitives, Engine
11472: @section Performance
11473: @cindex performance of some Forth interpreters
11474: @cindex engine performance
11475: @cindex benchmarking Forth systems
11476: @cindex Gforth performance
11477:
11478: On RISCs the Gforth engine is very close to optimal; i.e., it is usually
11479: impossible to write a significantly faster engine.
11480:
11481: On register-starved machines like the 386 architecture processors
11482: improvements are possible, because @code{gcc} does not utilize the
11483: registers as well as a human, even with explicit register declarations;
11484: e.g., Bernd Beuster wrote a Forth system fragment in assembly language
11485: and hand-tuned it for the 486; this system is 1.19 times faster on the
11486: Sieve benchmark on a 486DX2/66 than Gforth compiled with
11487: @code{gcc-2.6.3} with @code{-DFORCE_REG}. The situation has improved
11488: with gcc-2.95 and gforth-0.4.9; now the most important virtual machine
11489: registers fit in real registers (and we can even afford to use the TOS
11490: optimization), resulting in a speedup of 1.14 on the sieve over the
11491: earlier results.
11492:
11493: @cindex Win32Forth performance
11494: @cindex NT Forth performance
11495: @cindex eforth performance
11496: @cindex ThisForth performance
11497: @cindex PFE performance
11498: @cindex TILE performance
11499: The potential advantage of assembly language implementations
11500: is not necessarily realized in complete Forth systems: We compared
11501: Gforth-0.4.9 (direct threaded, compiled with @code{gcc-2.95.1} and
11502: @code{-DFORCE_REG}) with Win32Forth 1.2093, LMI's NT Forth (Beta, May
11503: 1994) and Eforth (with and without peephole (aka pinhole) optimization
11504: of the threaded code); all these systems were written in assembly
11505: language. We also compared Gforth with three systems written in C:
11506: PFE-0.9.14 (compiled with @code{gcc-2.6.3} with the default
11507: configuration for Linux: @code{-O2 -fomit-frame-pointer -DUSE_REGS
11508: -DUNROLL_NEXT}), ThisForth Beta (compiled with @code{gcc-2.6.3 -O3
11509: -fomit-frame-pointer}; ThisForth employs peephole optimization of the
11510: threaded code) and TILE (compiled with @code{make opt}). We benchmarked
11511: Gforth, PFE, ThisForth and TILE on a 486DX2/66 under Linux. Kenneth
11512: O'Heskin kindly provided the results for Win32Forth and NT Forth on a
11513: 486DX2/66 with similar memory performance under Windows NT. Marcel
11514: Hendrix ported Eforth to Linux, then extended it to run the benchmarks,
11515: added the peephole optimizer, ran the benchmarks and reported the
11516: results.
11517:
11518: We used four small benchmarks: the ubiquitous Sieve; bubble-sorting and
11519: matrix multiplication come from the Stanford integer benchmarks and have
11520: been translated into Forth by Martin Fraeman; we used the versions
11521: included in the TILE Forth package, but with bigger data set sizes; and
11522: a recursive Fibonacci number computation for benchmarking calling
11523: performance. The following table shows the time taken for the benchmarks
11524: scaled by the time taken by Gforth (in other words, it shows the speedup
11525: factor that Gforth achieved over the other systems).
11526:
11527: @example
11528: relative Win32- NT eforth This-
11529: time Gforth Forth Forth eforth +opt PFE Forth TILE
11530: sieve 1.00 1.58 1.30 1.58 0.97 1.80 3.63 9.79
11531: bubble 1.00 1.55 1.67 1.75 1.04 1.78 4.59
11532: matmul 1.00 1.67 1.53 1.66 0.84 1.79 4.63
11533: fib 1.00 1.75 1.53 1.40 0.99 1.99 3.43 4.93
11534: @end example
11535:
11536: You may be quite surprised by the good performance of Gforth when
11537: compared with systems written in assembly language. One important reason
11538: for the disappointing performance of these other systems is probably
11539: that they are not written optimally for the 486 (e.g., they use the
11540: @code{lods} instruction). In addition, Win32Forth uses a comfortable,
11541: but costly method for relocating the Forth image: like @code{cforth}, it
11542: computes the actual addresses at run time, resulting in two address
11543: computations per @code{NEXT} (@pxref{Image File Background}).
11544:
11545: Only Eforth with the peephole optimizer performs comparable to
11546: Gforth. The speedups achieved with peephole optimization of threaded
11547: code are quite remarkable. Adding a peephole optimizer to Gforth should
11548: cause similar speedups.
11549:
11550: The speedup of Gforth over PFE, ThisForth and TILE can be easily
11551: explained with the self-imposed restriction of the latter systems to
11552: standard C, which makes efficient threading impossible (however, the
11553: measured implementation of PFE uses a GNU C extension: @pxref{Global Reg
11554: Vars, , Defining Global Register Variables, gcc.info, GNU C Manual}).
11555: Moreover, current C compilers have a hard time optimizing other aspects
11556: of the ThisForth and the TILE source.
11557:
11558: The performance of Gforth on 386 architecture processors varies widely
11559: with the version of @code{gcc} used. E.g., @code{gcc-2.5.8} failed to
11560: allocate any of the virtual machine registers into real machine
11561: registers by itself and would not work correctly with explicit register
11562: declarations, giving a 1.5 times slower engine (on a 486DX2/66 running
11563: the Sieve) than the one measured above.
11564:
11565: Note that there have been several releases of Win32Forth since the
11566: release presented here, so the results presented above may have little
11567: predictive value for the performance of Win32Forth today (results for
11568: the current release on an i486DX2/66 are welcome).
11569:
11570: @cindex @file{Benchres}
11571: In @cite{Translating Forth to Efficient C} by M. Anton Ertl and Martin
11572: Maierhofer (presented at EuroForth '95), an indirect threaded version of
11573: Gforth is compared with Win32Forth, NT Forth, PFE, and ThisForth; that
11574: version of Gforth is slower on a 486 than the direct threaded version
11575: used here. The paper available at
11576: @*@url{http://www.complang.tuwien.ac.at/papers/ertl&maierhofer95.ps.gz};
11577: it also contains numbers for some native code systems. You can find a
11578: newer version of these measurements at
11579: @url{http://www.complang.tuwien.ac.at/forth/performance.html}. You can
11580: find numbers for Gforth on various machines in @file{Benchres}.
11581:
11582: @c ******************************************************************
11583: @node Binding to System Library, Cross Compiler, Engine, Top
11584: @chapter Binding to System Library
11585:
11586: @node Cross Compiler, Bugs, Binding to System Library, Top
11587: @chapter Cross Compiler
11588:
11589: Cross Compiler
11590:
11591: @menu
11592: * Using the Cross Compiler::
11593: * How the Cross Compiler Works::
11594: @end menu
11595:
11596: @node Using the Cross Compiler, How the Cross Compiler Works, Cross Compiler, Cross Compiler
11597: @section Using the Cross Compiler
11598:
11599: @node How the Cross Compiler Works, , Using the Cross Compiler, Cross Compiler
11600: @section How the Cross Compiler Works
11601:
11602: @node Bugs, Origin, Cross Compiler, Top
11603: @appendix Bugs
11604: @cindex bug reporting
11605:
11606: Known bugs are described in the file @file{BUGS} in the Gforth distribution.
11607:
11608: If you find a bug, please send a bug report to
11609: @email{bug-gforth@@gnu.org}. A bug report should include this
11610: information:
11611:
11612: @itemize @bullet
11613: @item
11614: The Gforth version used (it is announced at the start of an
11615: interactive Gforth session).
11616: @item
11617: The machine and operating system (on Unix
11618: systems @code{uname -a} will report this information).
11619: @item
11620: The installation options (send the file @file{config.status}).
11621: @item
11622: A complete list of changes (if any) you (or your installer) have made to the
11623: Gforth sources.
11624: @item
11625: A program (or a sequence of keyboard commands) that reproduces the bug.
11626: @item
11627: A description of what you think constitutes the buggy behaviour.
11628: @end itemize
11629:
11630: For a thorough guide on reporting bugs read @ref{Bug Reporting, , How
11631: to Report Bugs, gcc.info, GNU C Manual}.
11632:
11633:
11634: @node Origin, Forth-related information, Bugs, Top
11635: @appendix Authors and Ancestors of Gforth
11636:
11637: @section Authors and Contributors
11638: @cindex authors of Gforth
11639: @cindex contributors to Gforth
11640:
11641: The Gforth project was started in mid-1992 by Bernd Paysan and Anton
11642: Ertl. The third major author was Jens Wilke. Lennart Benschop (who was
11643: one of Gforth's first users, in mid-1993) and Stuart Ramsden inspired us
11644: with their continuous feedback. Lennart Benshop contributed
11645: @file{glosgen.fs}, while Stuart Ramsden has been working on automatic
11646: support for calling C libraries. Helpful comments also came from Paul
11647: Kleinrubatscher, Christian Pirker, Dirk Zoller, Marcel Hendrix, John
11648: Wavrik, Barrie Stott, Marc de Groot, and Jorge Acerada. Since the
11649: release of Gforth-0.2.1 there were also helpful comments from many
11650: others; thank you all, sorry for not listing you here (but digging
11651: through my mailbox to extract your names is on my to-do list). Since the
11652: release of Gforth-0.4.0 Neal Crook worked on the manual.
11653:
11654: Gforth also owes a lot to the authors of the tools we used (GCC, CVS,
11655: and autoconf, among others), and to the creators of the Internet: Gforth
11656: was developed across the Internet, and its authors did not meet
11657: physically for the first 4 years of development.
11658:
11659: @section Pedigree
11660: @cindex pedigree of Gforth
11661:
11662: Gforth descends from bigFORTH (1993) and fig-Forth. Gforth and PFE (by
11663: Dirk Zoller) will cross-fertilize each other. Of course, a significant
11664: part of the design of Gforth was prescribed by ANS Forth.
11665:
11666: Bernd Paysan wrote bigFORTH, a descendent from TurboForth, an unreleased
11667: 32 bit native code version of VolksForth for the Atari ST, written
11668: mostly by Dietrich Weineck.
11669:
11670: VolksForth descends from F83. It was written by Klaus Schleisiek, Bernd
11671: Pennemann, Georg Rehfeld and Dietrich Weineck for the C64 (called
11672: UltraForth there) in the mid-80s and ported to the Atari ST in 1986.
11673:
11674: Henry Laxen and Mike Perry wrote F83 as a model implementation of the
11675: Forth-83 standard. !! Pedigree? When?
11676:
11677: A team led by Bill Ragsdale implemented fig-Forth on many processors in
11678: 1979. Robert Selzer and Bill Ragsdale developed the original
11679: implementation of fig-Forth for the 6502 based on microForth.
11680:
11681: The principal architect of microForth was Dean Sanderson. microForth was
11682: FORTH, Inc.'s first off-the-shelf product. It was developed in 1976 for
11683: the 1802, and subsequently implemented on the 8080, the 6800 and the
11684: Z80.
11685:
11686: All earlier Forth systems were custom-made, usually by Charles Moore,
11687: who discovered (as he puts it) Forth during the late 60s. The first full
11688: Forth existed in 1971.
11689:
11690: A part of the information in this section comes from @cite{The Evolution
11691: of Forth} by Elizabeth D. Rather, Donald R. Colburn and Charles
11692: H. Moore, presented at the HOPL-II conference and preprinted in SIGPLAN
11693: Notices 28(3), 1993. You can find more historical and genealogical
11694: information about Forth there.
11695:
11696: @node Forth-related information, Word Index, Origin, Top
11697: @appendix Other Forth-related information
11698: @cindex Forth-related information
11699:
11700: @menu
11701: * Internet resources::
11702: * Books::
11703: * The Forth Interest Group::
11704: * Conferences::
11705: @end menu
11706:
11707:
11708: @node Internet resources, Books, Forth-related information, Forth-related information
11709: @section Internet resources
11710: @cindex internet resources
11711:
11712: @cindex comp.lang.forth
11713: @cindex frequently asked questions
11714: There is an active news group (comp.lang.forth) discussing Forth and
11715: Forth-related issues. A frequently-asked-questions (FAQ) list
11716: is posted to the news group regularly, and archived at these sites:
11717:
11718: @itemize @bullet
11719: @item
11720: @url{ftp://rtfm.mit.edu/pub/usenet-by-group/comp.lang.forth/}
11721: @item
11722: @url{ftp://ftp.forth.org/pub/Forth/FAQ/}
11723: @end itemize
11724:
11725: The FAQ list should be considered mandatory reading before posting to
11726: the news group.
11727:
11728: Here are some other web sites holding Forth-related material:
11729:
11730: @itemize @bullet
11731: @item
11732: @url{http://www.taygeta.com/forth.html} -- Skip Carter's Forth pages.
11733: @item
11734: @url{http://www.jwdt.com/~paysan/gforth.html} -- the Gforth home page.
11735: @item
11736: @url{http://www.minerva.com/uathena.htm} -- home of ANS Forth Standard.
11737: @item
11738: @url{http://dec.bournemouth.ac.uk/forth/index.html} -- the Forth
11739: Research page, including links to the Journal of Forth Application and
11740: Research (JFAR) and a searchable Forth bibliography.
11741: @end itemize
11742:
11743:
11744: @node Books, The Forth Interest Group, Internet resources, Forth-related information
11745: @section Books
11746: @cindex books on Forth
11747:
11748: As the Standard is relatively new, there are not many books out yet. It
11749: is not recommended to learn Forth by using Gforth and a book that is not
11750: written for ANS Forth, as you will not know your mistakes from the
11751: deviations of the book. However, books based on the Forth-83 standard
11752: should be ok, because ANS Forth is primarily an extension of Forth-83.
11753: Refer to the Forth FAQ for details of Forth-related books.
11754:
11755: @cindex standard document for ANS Forth
11756: @cindex ANS Forth document
11757: The definite reference if you want to write ANS Forth programs is, of
11758: course, the ANS Forth document. It is available in printed form from the
11759: National Standards Institute Sales Department (Tel.: USA (212) 642-4900;
11760: Fax.: USA (212) 302-1286) as document @cite{X3.215-1994} for about
11761: $200. You can also get it from Global Engineering Documents (Tel.: USA
11762: (800) 854-7179; Fax.: (303) 843-9880) for about $300.
11763:
11764: @cite{dpANS6}, the last draft of the standard, which was then submitted
11765: to ANSI for publication is available electronically and for free in some
11766: MS Word format, and it has been converted to HTML
11767: (@url{http://www.taygeta.com/forth/dpans.html}; this HTML version also
11768: includes the answers to Requests for Interpretation (RFIs). Some
11769: pointers to these versions can be found through
11770: @*@url{http://www.complang.tuwien.ac.at/projects/forth.html}.
11771:
11772:
11773: @node The Forth Interest Group, Conferences, Books, Forth-related information
11774: @section The Forth Interest Group
11775: @cindex Forth interest group (FIG)
11776:
11777: The Forth Interest Group (FIG) is a world-wide, non-profit,
11778: member-supported organisation. It publishes a regular magazine,
11779: @var{FORTH Dimensions}, and offers other benefits of membership. You can
11780: contact the FIG through their office email address:
11781: @email{office@@forth.org} or by visiting their web site at
11782: @url{http://www.forth.org/}. This web site also includes links to FIG
11783: chapters in other countries and American cities
11784: (@url{http://www.forth.org/chapters.html}).
11785:
11786: @node Conferences, , The Forth Interest Group, Forth-related information
11787: @section Conferences
11788: @cindex Conferences
11789:
11790: There are several regular conferences related to Forth. They are all
11791: well-publicised in @var{FORTH Dimensions} and on the comp.lang.forth
11792: news group:
11793:
11794: @itemize @bullet
11795: @item
11796: FORML -- the Forth modification laboratory convenes every year near
11797: Monterey, California.
11798: @item
11799: The Rochester Forth Conference -- an annual conference traditionally
11800: held in Rochester, New York.
11801: @item
11802: EuroForth -- this European conference takes place annually.
11803: @end itemize
11804:
11805:
11806: @node Word Index, Name Index, Forth-related information, Top
11807: @unnumbered Word Index
11808:
11809: This index is a list of Forth words that have ``glossary'' entries
11810: within this manual. Each word is listed with its stack effect and
11811: wordset.
11812:
11813: @printindex fn
11814:
11815: @node Name Index, Concept Index, Word Index, Top
11816: @unnumbered Name Index
11817:
11818: This index is a list of Forth words that have ``glossary'' entries
11819: within this manual.
11820:
11821: @printindex ky
11822:
11823: @node Concept Index, , Name Index, Top
11824: @unnumbered Concept and Word Index
11825:
11826: Not all entries listed in this index are present verbatim in the
11827: text. This index also duplicates, in abbreviated form, all of the words
11828: listed in the Word Index (only the names are listed for the words here).
11829:
11830: @printindex cp
11831:
11832: @contents
11833: @bye
11834:
FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>
![[BACK]](/cvsweb/icons/back.gif){kind=icon}
Return to gforth.ds CVS log ![[TXT]](/cvsweb/icons/text.gif){kind=icon}
![[DIR]](/cvsweb/icons/dir.gif){kind=icon}
Up to [gforth] / gforth / doc