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:
126: @node Top, License, (dir), (dir)
127: @ifinfo
128: Gforth is a free implementation of ANS Forth available on many
129: personal machines. This manual corresponds to version @value{VERSION}.
130: @end ifinfo
131:
132: @menu
133: * License:: The GPL
134: * Goals:: About the Gforth Project
135: * Gforth Environment:: Starting (and exiting) Gforth
136: * Introduction:: An introduction to ANS Forth
137: * Words:: Forth words available in Gforth
138: * Error messages:: How to interpret them
139: * Tools:: Programming tools
140: * ANS conformance:: Implementation-defined options etc.
141: * Model:: The abstract machine of Gforth
142: * Integrating Gforth:: Forth as scripting language for applications
143: * Emacs and Gforth:: The Gforth Mode
144: * Image Files:: @code{.fi} files contain compiled code
145: * Engine:: The inner interpreter and the primitives
146: * Binding to System Library::
147: * Cross Compiler:: The Cross Compiler
148: * Bugs:: How to report them
149: * Origin:: Authors and ancestors of Gforth
150: * Forth-related information:: Books and places to look on the WWW
151: * Word Index:: An item for each Forth word
152: * Name Index:: Forth words, only names listed
153: * Concept Index:: A menu covering many topics
154:
155: @detailmenu
156: --- The Detailed Node Listing ---
157:
158: Goals of Gforth
159:
160: * Gforth Extensions Sinful?::
161:
162: Gforth Environment
163:
164: * Invoking Gforth:: Getting in
165: * Leaving Gforth:: Getting out
166: * Command-line editing::
167: * Upper and lower case::
168: * Environment variables:: ..that affect how Gforth starts up
169: * Gforth Files:: What gets installed and where
170:
171: An Introduction to ANS Forth
172:
173: * Introducing the Text Interpreter::
174: * Stacks and Postfix notation::
175: * Your first definition::
176: * How does that work?::
177: * Forth is written in Forth::
178: * Review - elements of a Forth system::
179: * Where to go next::
180: * Exercises::
181:
182: Forth Words
183:
184: * Notation::
185: * Comments::
186: * Boolean Flags::
187: * Arithmetic::
188: * Stack Manipulation::
189: * Memory::
190: * Control Structures::
191: * Defining Words::
192: * The Text Interpreter::
193: * Tokens for Words::
194: * Word Lists::
195: * Environmental Queries::
196: * Files::
197: * Blocks::
198: * Other I/O::
199: * Programming Tools::
200: * Assembler and Code Words::
201: * Threading Words::
202: * Locals::
203: * Structures::
204: * Object-oriented Forth::
205: * Passing Commands to the OS::
206: * Miscellaneous Words::
207:
208: Arithmetic
209:
210: * Single precision::
211: * Bitwise operations::
212: * Double precision:: Double-cell integer arithmetic
213: * Numeric comparison::
214: * Mixed precision:: Operations with single and double-cell integers
215: * Floating Point::
216:
217: Stack Manipulation
218:
219: * Data stack::
220: * Floating point stack::
221: * Return stack::
222: * Locals stack::
223: * Stack pointer manipulation::
224:
225: Memory
226:
227: * Memory model::
228: * Dictionary allocation::
229: * Heap Allocation::
230: * Memory Access::
231: * Address arithmetic::
232: * Memory Blocks::
233:
234: Control Structures
235:
236: * Selection:: IF ... ELSE ... ENDIF
237: * Simple Loops:: BEGIN ...
238: * Counted Loops:: DO
239: * Arbitrary control structures::
240: * Calls and returns::
241: * Exception Handling::
242:
243: Defining Words
244:
245: * @code{CREATE}::
246: * Variables:: Variables and user variables
247: * Constants::
248: * Values:: Initialised variables
249: * Colon Definitions::
250: * Anonymous Definitions:: Definitions without names
251: * User-defined Defining Words::
252: * Deferred words:: Allow forward references
253: * Aliases::
254: * Supplying names::
255: * Interpretation and Compilation Semantics::
256: * Combined words::
257:
258: The Text Interpreter
259:
260: * Input Sources::
261: * Number Conversion::
262: * Interpret/Compile states::
263: * Literals::
264: * Interpreter Directives::
265:
266: Word Lists
267:
268: * Why use word lists?::
269: * Word list examples::
270:
271: Files
272:
273: * Forth source files::
274: * General files::
275: * Search Paths::
276: * Forth Search Paths::
277: * General Search Paths::
278:
279: Other I/O
280:
281: * Simple numeric output:: Predefined formats
282: * Formatted numeric output:: Formatted (pictured) output
283: * String Formats:: How Forth stores strings in memory
284: * Displaying characters and strings:: Other stuff
285: * Input:: Input
286:
287: Programming Tools
288:
289: * Debugging:: Simple and quick.
290: * Assertions:: Making your programs self-checking.
291: * Singlestep Debugger:: Executing your program word by word.
292:
293: Locals
294:
295: * Gforth locals::
296: * ANS Forth locals::
297:
298: Gforth locals
299:
300: * Where are locals visible by name?::
301: * How long do locals live?::
302: * Programming Style::
303: * Implementation::
304:
305: Structures
306:
307: * Why explicit structure support?::
308: * Structure Usage::
309: * Structure Naming Convention::
310: * Structure Implementation::
311: * Structure Glossary::
312:
313: Object-oriented Forth
314:
315: * Why object-oriented programming?::
316: * Object-Oriented Terminology::
317: * Objects::
318: * OOF::
319: * Mini-OOF::
320: * Comparison with other object models::
321:
322: The @file{objects.fs} model
323:
324: * Properties of the Objects model::
325: * Basic Objects Usage::
326: * The Objects base class::
327: * Creating objects::
328: * Object-Oriented Programming Style::
329: * Class Binding::
330: * Method conveniences::
331: * Classes and Scoping::
332: * Dividing classes::
333: * Object Interfaces::
334: * Objects Implementation::
335: * Objects Glossary::
336:
337: The @file{oof.fs} model
338:
339: * Properties of the OOF model::
340: * Basic OOF Usage::
341: * The OOF base class::
342: * Class Declaration::
343: * Class Implementation::
344:
345: The @file{mini-oof.fs} model
346:
347: * Basic Mini-OOF Usage::
348: * Mini-OOF Example::
349: * Mini-OOF Implementation::
350:
351: Tools
352:
353: * ANS Report:: Report the words used, sorted by wordset.
354:
355: ANS conformance
356:
357: * The Core Words::
358: * The optional Block word set::
359: * The optional Double Number word set::
360: * The optional Exception word set::
361: * The optional Facility word set::
362: * The optional File-Access word set::
363: * The optional Floating-Point word set::
364: * The optional Locals word set::
365: * The optional Memory-Allocation word set::
366: * The optional Programming-Tools word set::
367: * The optional Search-Order word set::
368:
369: The Core Words
370:
371: * core-idef:: Implementation Defined Options
372: * core-ambcond:: Ambiguous Conditions
373: * core-other:: Other System Documentation
374:
375: The optional Block word set
376:
377: * block-idef:: Implementation Defined Options
378: * block-ambcond:: Ambiguous Conditions
379: * block-other:: Other System Documentation
380:
381: The optional Double Number word set
382:
383: * double-ambcond:: Ambiguous Conditions
384:
385: The optional Exception word set
386:
387: * exception-idef:: Implementation Defined Options
388:
389: The optional Facility word set
390:
391: * facility-idef:: Implementation Defined Options
392: * facility-ambcond:: Ambiguous Conditions
393:
394: The optional File-Access word set
395:
396: * file-idef:: Implementation Defined Options
397: * file-ambcond:: Ambiguous Conditions
398:
399: The optional Floating-Point word set
400:
401: * floating-idef:: Implementation Defined Options
402: * floating-ambcond:: Ambiguous Conditions
403:
404: The optional Locals word set
405:
406: * locals-idef:: Implementation Defined Options
407: * locals-ambcond:: Ambiguous Conditions
408:
409: The optional Memory-Allocation word set
410:
411: * memory-idef:: Implementation Defined Options
412:
413: The optional Programming-Tools word set
414:
415: * programming-idef:: Implementation Defined Options
416: * programming-ambcond:: Ambiguous Conditions
417:
418: The optional Search-Order word set
419:
420: * search-idef:: Implementation Defined Options
421: * search-ambcond:: Ambiguous Conditions
422:
423: Image Files
424:
425: * Image Licensing Issues:: Distribution terms for images.
426: * Image File Background:: Why have image files?
427: * Non-Relocatable Image Files:: don't always work.
428: * Data-Relocatable Image Files:: are better.
429: * Fully Relocatable Image Files:: better yet.
430: * Stack and Dictionary Sizes:: Setting the default sizes for an image.
431: * Running Image Files:: @code{gforth -i @i{file}} or @i{file}.
432: * Modifying the Startup Sequence:: and turnkey applications.
433:
434: Fully Relocatable Image Files
435:
436: * gforthmi:: The normal way
437: * cross.fs:: The hard way
438:
439: Engine
440:
441: * Portability::
442: * Threading::
443: * Primitives::
444: * Performance::
445:
446: Threading
447:
448: * Scheduling::
449: * Direct or Indirect Threaded?::
450: * DOES>::
451:
452: Primitives
453:
454: * Automatic Generation::
455: * TOS Optimization::
456: * Produced code::
457:
458: Cross Compiler
459:
460: * Using the Cross Compiler::
461: * How the Cross Compiler Works::
462:
463: Other Forth-related information
464:
465: * Internet resources::
466: * Books::
467: * The Forth Interest Group::
468: * Conferences::
469:
470: @end detailmenu
471: @end menu
472:
473: @node License, Goals, Top, Top
474: @unnumbered GNU GENERAL PUBLIC LICENSE
475: @center Version 2, June 1991
476:
477: @display
478: Copyright @copyright{} 1989, 1991 Free Software Foundation, Inc.
479: 675 Mass Ave, Cambridge, MA 02139, USA
480:
481: Everyone is permitted to copy and distribute verbatim copies
482: of this license document, but changing it is not allowed.
483: @end display
484:
485: @unnumberedsec Preamble
486:
487: The licenses for most software are designed to take away your
488: freedom to share and change it. By contrast, the GNU General Public
489: License is intended to guarantee your freedom to share and change free
490: software---to make sure the software is free for all its users. This
491: General Public License applies to most of the Free Software
492: Foundation's software and to any other program whose authors commit to
493: using it. (Some other Free Software Foundation software is covered by
494: the GNU Library General Public License instead.) You can apply it to
495: your programs, too.
496:
497: When we speak of free software, we are referring to freedom, not
498: price. Our General Public Licenses are designed to make sure that you
499: have the freedom to distribute copies of free software (and charge for
500: this service if you wish), that you receive source code or can get it
501: if you want it, that you can change the software or use pieces of it
502: in new free programs; and that you know you can do these things.
503:
504: To protect your rights, we need to make restrictions that forbid
505: anyone to deny you these rights or to ask you to surrender the rights.
506: These restrictions translate to certain responsibilities for you if you
507: distribute copies of the software, or if you modify it.
508:
509: For example, if you distribute copies of such a program, whether
510: gratis or for a fee, you must give the recipients all the rights that
511: you have. You must make sure that they, too, receive or can get the
512: source code. And you must show them these terms so they know their
513: rights.
514:
515: We protect your rights with two steps: (1) copyright the software, and
516: (2) offer you this license which gives you legal permission to copy,
517: distribute and/or modify the software.
518:
519: Also, for each author's protection and ours, we want to make certain
520: that everyone understands that there is no warranty for this free
521: software. If the software is modified by someone else and passed on, we
522: want its recipients to know that what they have is not the original, so
523: that any problems introduced by others will not reflect on the original
524: authors' reputations.
525:
526: Finally, any free program is threatened constantly by software
527: patents. We wish to avoid the danger that redistributors of a free
528: program will individually obtain patent licenses, in effect making the
529: program proprietary. To prevent this, we have made it clear that any
530: patent must be licensed for everyone's free use or not licensed at all.
531:
532: The precise terms and conditions for copying, distribution and
533: modification follow.
534:
535: @iftex
536: @unnumberedsec TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION
537: @end iftex
538: @ifinfo
539: @center TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION
540: @end ifinfo
541:
542: @enumerate 0
543: @item
544: This License applies to any program or other work which contains
545: a notice placed by the copyright holder saying it may be distributed
546: under the terms of this General Public License. The ``Program'', below,
547: refers to any such program or work, and a ``work based on the Program''
548: means either the Program or any derivative work under copyright law:
549: that is to say, a work containing the Program or a portion of it,
550: either verbatim or with modifications and/or translated into another
551: language. (Hereinafter, translation is included without limitation in
552: the term ``modification''.) Each licensee is addressed as ``you''.
553:
554: Activities other than copying, distribution and modification are not
555: covered by this License; they are outside its scope. The act of
556: running the Program is not restricted, and the output from the Program
557: is covered only if its contents constitute a work based on the
558: Program (independent of having been made by running the Program).
559: Whether that is true depends on what the Program does.
560:
561: @item
562: You may copy and distribute verbatim copies of the Program's
563: source code as you receive it, in any medium, provided that you
564: conspicuously and appropriately publish on each copy an appropriate
565: copyright notice and disclaimer of warranty; keep intact all the
566: notices that refer to this License and to the absence of any warranty;
567: and give any other recipients of the Program a copy of this License
568: along with the Program.
569:
570: You may charge a fee for the physical act of transferring a copy, and
571: you may at your option offer warranty protection in exchange for a fee.
572:
573: @item
574: You may modify your copy or copies of the Program or any portion
575: of it, thus forming a work based on the Program, and copy and
576: distribute such modifications or work under the terms of Section 1
577: above, provided that you also meet all of these conditions:
578:
579: @enumerate a
580: @item
581: You must cause the modified files to carry prominent notices
582: stating that you changed the files and the date of any change.
583:
584: @item
585: You must cause any work that you distribute or publish, that in
586: whole or in part contains or is derived from the Program or any
587: part thereof, to be licensed as a whole at no charge to all third
588: parties under the terms of this License.
589:
590: @item
591: If the modified program normally reads commands interactively
592: when run, you must cause it, when started running for such
593: interactive use in the most ordinary way, to print or display an
594: announcement including an appropriate copyright notice and a
595: notice that there is no warranty (or else, saying that you provide
596: a warranty) and that users may redistribute the program under
597: these conditions, and telling the user how to view a copy of this
598: License. (Exception: if the Program itself is interactive but
599: does not normally print such an announcement, your work based on
600: the Program is not required to print an announcement.)
601: @end enumerate
602:
603: These requirements apply to the modified work as a whole. If
604: identifiable sections of that work are not derived from the Program,
605: and can be reasonably considered independent and separate works in
606: themselves, then this License, and its terms, do not apply to those
607: sections when you distribute them as separate works. But when you
608: distribute the same sections as part of a whole which is a work based
609: on the Program, the distribution of the whole must be on the terms of
610: this License, whose permissions for other licensees extend to the
611: entire whole, and thus to each and every part regardless of who wrote it.
612:
613: Thus, it is not the intent of this section to claim rights or contest
614: your rights to work written entirely by you; rather, the intent is to
615: exercise the right to control the distribution of derivative or
616: collective works based on the Program.
617:
618: In addition, mere aggregation of another work not based on the Program
619: with the Program (or with a work based on the Program) on a volume of
620: a storage or distribution medium does not bring the other work under
621: the scope of this License.
622:
623: @item
624: You may copy and distribute the Program (or a work based on it,
625: under Section 2) in object code or executable form under the terms of
626: Sections 1 and 2 above provided that you also do one of the following:
627:
628: @enumerate a
629: @item
630: Accompany it with the complete corresponding machine-readable
631: source code, which must be distributed under the terms of Sections
632: 1 and 2 above on a medium customarily used for software interchange; or,
633:
634: @item
635: Accompany it with a written offer, valid for at least three
636: years, to give any third party, for a charge no more than your
637: cost of physically performing source distribution, a complete
638: machine-readable copy of the corresponding source code, to be
639: distributed under the terms of Sections 1 and 2 above on a medium
640: customarily used for software interchange; or,
641:
642: @item
643: Accompany it with the information you received as to the offer
644: to distribute corresponding source code. (This alternative is
645: allowed only for noncommercial distribution and only if you
646: received the program in object code or executable form with such
647: an offer, in accord with Subsection b above.)
648: @end enumerate
649:
650: The source code for a work means the preferred form of the work for
651: making modifications to it. For an executable work, complete source
652: code means all the source code for all modules it contains, plus any
653: associated interface definition files, plus the scripts used to
654: control compilation and installation of the executable. However, as a
655: special exception, the source code distributed need not include
656: anything that is normally distributed (in either source or binary
657: form) with the major components (compiler, kernel, and so on) of the
658: operating system on which the executable runs, unless that component
659: itself accompanies the executable.
660:
661: If distribution of executable or object code is made by offering
662: access to copy from a designated place, then offering equivalent
663: access to copy the source code from the same place counts as
664: distribution of the source code, even though third parties are not
665: compelled to copy the source along with the object code.
666:
667: @item
668: You may not copy, modify, sublicense, or distribute the Program
669: except as expressly provided under this License. Any attempt
670: otherwise to copy, modify, sublicense or distribute the Program is
671: void, and will automatically terminate your rights under this License.
672: However, parties who have received copies, or rights, from you under
673: this License will not have their licenses terminated so long as such
674: parties remain in full compliance.
675:
676: @item
677: You are not required to accept this License, since you have not
678: signed it. However, nothing else grants you permission to modify or
679: distribute the Program or its derivative works. These actions are
680: prohibited by law if you do not accept this License. Therefore, by
681: modifying or distributing the Program (or any work based on the
682: Program), you indicate your acceptance of this License to do so, and
683: all its terms and conditions for copying, distributing or modifying
684: the Program or works based on it.
685:
686: @item
687: Each time you redistribute the Program (or any work based on the
688: Program), the recipient automatically receives a license from the
689: original licensor to copy, distribute or modify the Program subject to
690: these terms and conditions. You may not impose any further
691: restrictions on the recipients' exercise of the rights granted herein.
692: You are not responsible for enforcing compliance by third parties to
693: this License.
694:
695: @item
696: If, as a consequence of a court judgment or allegation of patent
697: infringement or for any other reason (not limited to patent issues),
698: conditions are imposed on you (whether by court order, agreement or
699: otherwise) that contradict the conditions of this License, they do not
700: excuse you from the conditions of this License. If you cannot
701: distribute so as to satisfy simultaneously your obligations under this
702: License and any other pertinent obligations, then as a consequence you
703: may not distribute the Program at all. For example, if a patent
704: license would not permit royalty-free redistribution of the Program by
705: all those who receive copies directly or indirectly through you, then
706: the only way you could satisfy both it and this License would be to
707: refrain entirely from distribution of the Program.
708:
709: If any portion of this section is held invalid or unenforceable under
710: any particular circumstance, the balance of the section is intended to
711: apply and the section as a whole is intended to apply in other
712: circumstances.
713:
714: It is not the purpose of this section to induce you to infringe any
715: patents or other property right claims or to contest validity of any
716: such claims; this section has the sole purpose of protecting the
717: integrity of the free software distribution system, which is
718: implemented by public license practices. Many people have made
719: generous contributions to the wide range of software distributed
720: through that system in reliance on consistent application of that
721: system; it is up to the author/donor to decide if he or she is willing
722: to distribute software through any other system and a licensee cannot
723: impose that choice.
724:
725: This section is intended to make thoroughly clear what is believed to
726: be a consequence of the rest of this License.
727:
728: @item
729: If the distribution and/or use of the Program is restricted in
730: certain countries either by patents or by copyrighted interfaces, the
731: original copyright holder who places the Program under this License
732: may add an explicit geographical distribution limitation excluding
733: those countries, so that distribution is permitted only in or among
734: countries not thus excluded. In such case, this License incorporates
735: the limitation as if written in the body of this License.
736:
737: @item
738: The Free Software Foundation may publish revised and/or new versions
739: of the General Public License from time to time. Such new versions will
740: be similar in spirit to the present version, but may differ in detail to
741: address new problems or concerns.
742:
743: Each version is given a distinguishing version number. If the Program
744: specifies a version number of this License which applies to it and ``any
745: later version'', you have the option of following the terms and conditions
746: either of that version or of any later version published by the Free
747: Software Foundation. If the Program does not specify a version number of
748: this License, you may choose any version ever published by the Free Software
749: Foundation.
750:
751: @item
752: If you wish to incorporate parts of the Program into other free
753: programs whose distribution conditions are different, write to the author
754: to ask for permission. For software which is copyrighted by the Free
755: Software Foundation, write to the Free Software Foundation; we sometimes
756: make exceptions for this. Our decision will be guided by the two goals
757: of preserving the free status of all derivatives of our free software and
758: of promoting the sharing and reuse of software generally.
759:
760: @iftex
761: @heading NO WARRANTY
762: @end iftex
763: @ifinfo
764: @center NO WARRANTY
765: @end ifinfo
766:
767: @item
768: BECAUSE THE PROGRAM IS LICENSED FREE OF CHARGE, THERE IS NO WARRANTY
769: FOR THE PROGRAM, TO THE EXTENT PERMITTED BY APPLICABLE LAW. EXCEPT WHEN
770: OTHERWISE STATED IN WRITING THE COPYRIGHT HOLDERS AND/OR OTHER PARTIES
771: PROVIDE THE PROGRAM ``AS IS'' WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESSED
772: OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
773: MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. THE ENTIRE RISK AS
774: TO THE QUALITY AND PERFORMANCE OF THE PROGRAM IS WITH YOU. SHOULD THE
775: PROGRAM PROVE DEFECTIVE, YOU ASSUME THE COST OF ALL NECESSARY SERVICING,
776: REPAIR OR CORRECTION.
777:
778: @item
779: IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING
780: WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MAY MODIFY AND/OR
781: REDISTRIBUTE THE PROGRAM AS PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES,
782: INCLUDING ANY GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING
783: OUT OF THE USE OR INABILITY TO USE THE PROGRAM (INCLUDING BUT NOT LIMITED
784: TO LOSS OF DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY
785: YOU OR THIRD PARTIES OR A FAILURE OF THE PROGRAM TO OPERATE WITH ANY OTHER
786: PROGRAMS), EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE
787: POSSIBILITY OF SUCH DAMAGES.
788: @end enumerate
789:
790: @iftex
791: @heading END OF TERMS AND CONDITIONS
792: @end iftex
793: @ifinfo
794: @center END OF TERMS AND CONDITIONS
795: @end ifinfo
796:
797: @page
798: @unnumberedsec How to Apply These Terms to Your New Programs
799:
800: If you develop a new program, and you want it to be of the greatest
801: possible use to the public, the best way to achieve this is to make it
802: free software which everyone can redistribute and change under these terms.
803:
804: To do so, attach the following notices to the program. It is safest
805: to attach them to the start of each source file to most effectively
806: convey the exclusion of warranty; and each file should have at least
807: the ``copyright'' line and a pointer to where the full notice is found.
808:
809: @smallexample
810: @var{one line to give the program's name and a brief idea of what it does.}
811: Copyright (C) 19@var{yy} @var{name of author}
812:
813: This program is free software; you can redistribute it and/or modify
814: it under the terms of the GNU General Public License as published by
815: the Free Software Foundation; either version 2 of the License, or
816: (at your option) any later version.
817:
818: This program is distributed in the hope that it will be useful,
819: but WITHOUT ANY WARRANTY; without even the implied warranty of
820: MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
821: GNU General Public License for more details.
822:
823: You should have received a copy of the GNU General Public License
824: along with this program; if not, write to the Free Software
825: Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
826: @end smallexample
827:
828: Also add information on how to contact you by electronic and paper mail.
829:
830: If the program is interactive, make it output a short notice like this
831: when it starts in an interactive mode:
832:
833: @smallexample
834: Gnomovision version 69, Copyright (C) 19@var{yy} @var{name of author}
835: Gnomovision comes with ABSOLUTELY NO WARRANTY; for details
836: type `show w'.
837: This is free software, and you are welcome to redistribute it
838: under certain conditions; type `show c' for details.
839: @end smallexample
840:
841: The hypothetical commands @samp{show w} and @samp{show c} should show
842: the appropriate parts of the General Public License. Of course, the
843: commands you use may be called something other than @samp{show w} and
844: @samp{show c}; they could even be mouse-clicks or menu items---whatever
845: suits your program.
846:
847: You should also get your employer (if you work as a programmer) or your
848: school, if any, to sign a ``copyright disclaimer'' for the program, if
849: necessary. Here is a sample; alter the names:
850:
851: @smallexample
852: Yoyodyne, Inc., hereby disclaims all copyright interest in the program
853: `Gnomovision' (which makes passes at compilers) written by James Hacker.
854:
855: @var{signature of Ty Coon}, 1 April 1989
856: Ty Coon, President of Vice
857: @end smallexample
858:
859: This General Public License does not permit incorporating your program into
860: proprietary programs. If your program is a subroutine library, you may
861: consider it more useful to permit linking proprietary applications with the
862: library. If this is what you want to do, use the GNU Library General
863: Public License instead of this License.
864:
865: @iftex
866: @unnumbered Preface
867: @cindex Preface
868: This manual documents Gforth. Some introductory material is provided for
869: readers who are unfamiliar with Forth or who are migrating to Gforth
870: from other Forth compilers. However, this manual is primarily a
871: reference manual.
872: @end iftex
873:
874: @comment TODO much more blurb here.
875:
876: @c ******************************************************************
877: @node Goals, Gforth Environment, License, Top
878: @comment node-name, next, previous, up
879: @chapter Goals of Gforth
880: @cindex goals of the Gforth project
881: The goal of the Gforth Project is to develop a standard model for
882: ANS Forth. This can be split into several subgoals:
883:
884: @itemize @bullet
885: @item
886: Gforth should conform to the ANS Forth Standard.
887: @item
888: It should be a model, i.e. it should define all the
889: implementation-dependent things.
890: @item
891: It should become standard, i.e. widely accepted and used. This goal
892: is the most difficult one.
893: @end itemize
894:
895: To achieve these goals Gforth should be
896: @itemize @bullet
897: @item
898: Similar to previous models (fig-Forth, F83)
899: @item
900: Powerful. It should provide for all the things that are considered
901: necessary today and even some that are not yet considered necessary.
902: @item
903: Efficient. It should not get the reputation of being exceptionally
904: slow.
905: @item
906: Free.
907: @item
908: Available on many machines/easy to port.
909: @end itemize
910:
911: Have we achieved these goals? Gforth conforms to the ANS Forth
912: standard. It may be considered a model, but we have not yet documented
913: which parts of the model are stable and which parts we are likely to
914: change. It certainly has not yet become a de facto standard, but it
915: appears to be quite popular. It has some similarities to and some
916: differences from previous models. It has some powerful features, but not
917: yet everything that we envisioned. We certainly have achieved our
918: execution speed goals (@pxref{Performance}). It is free and available
919: on many machines.
920:
921: @menu
922: * Gforth Extensions Sinful?::
923: @end menu
924:
925: @node Gforth Extensions Sinful?, , Goals, Goals
926: @comment node-name, next, previous, up
927: @section Is it a Sin to use Gforth Extensions?
928: @cindex Gforth extensions
929:
930: If you've been paying attention, you will have realised that there is an
931: ANS (American National Standard) for Forth. As you read through the rest
932: of this manual, you will see documentation for @i{Standard} words, and
933: documentation for some appealing Gforth @i{extensions}. You might ask
934: yourself the question: @i{``Given that there is a standard, would I be
935: committing a sin to use (non-Standard) Gforth extensions?''}
936:
937: The answer to that question is somewhat pragmatic and somewhat
938: philosophical. Consider these points:
939:
940: @itemize @bullet
941: @item
942: A number of the Gforth extensions can be implemented in ANS Forth using
943: files provided in the @file{compat/} directory. These are mentioned in
944: the text in passing.
945: @item
946: Forth has a rich historical precedent for programmers taking advantage
947: of implementation-dependent features of their tools (for example,
948: relying on a knowledge of the dictionary structure). Sometimes these
949: techniques are necessary to extract every last bit of performance from
950: the hardware, sometimes they are just a programming shorthand.
951: @item
952: The best way to break the rules is to know what the rules are. To learn
953: the rules, there is no substitute for studying the text of the Standard
954: itself. In particular, Appendix A of the Standard (@var{Rationale})
955: provides a valuable insight into the thought processes of the technical
956: committee.
957: @item
958: The best reason to break a rule is because you have to; because it's
959: more productive to do that, because it makes your code run fast enough
960: or because you can see no Standard way to achieve what you want to
961: achieve.
962: @end itemize
963:
964: The tool @file{ans-report.fs} (@pxref{ANS Report}) makes it easy to
965: analyse your program and determine what non-Standard definitions it
966: relies upon.
967:
968:
969: @c ******************************************************************
970: @node Gforth Environment, Introduction, Goals, Top
971: @chapter Gforth Environment
972: @cindex Gforth environment
973:
974: Note: ultimately, the gforth man page will be auto-generated from the
975: material in this chapter.
976:
977: @menu
978: * Invoking Gforth:: Getting in
979: * Leaving Gforth:: Getting out
980: * Command-line editing::
981: * Upper and lower case::
982: * Environment variables:: ..that affect how Gforth starts up
983: * Gforth Files:: What gets installed and where
984: @end menu
985:
986: @xref{Image Files} for related information about the creation of images.
987:
988: @comment ----------------------------------------------
989: @node Invoking Gforth, Leaving Gforth, ,Gforth Environment
990: @section Invoking Gforth
991: @cindex invoking Gforth
992: @cindex running Gforth
993: @cindex command-line options
994: @cindex options on the command line
995: @cindex flags on the command line
996:
997: Gforth is made up of two parts; an executable ``engine'' (named
998: @file{gforth} or @file{gforth-fast}) and an image file. To start it, you
999: will usually just say @code{gforth} -- this automatically loads the
1000: default image file @file{gforth.fi}. In many other cases the default
1001: Gforth image will be invoked like this:
1002: @example
1003: gforth [file | -e forth-code] ...
1004: @end example
1005: @noindent
1006: This interprets the contents of the files and the Forth code in the order they
1007: are given.
1008:
1009: In addition to the @file{gforth} engine, there is also an engine called
1010: @file{gforth-fast}, which is faster, but gives less informative error
1011: messages (@pxref{Error messages}).
1012:
1013: In general, the command line looks like this:
1014:
1015: @example
1016: gforth[-fast] [engine options] [image options]
1017: @end example
1018:
1019: The engine options must come before the rest of the command
1020: line. They are:
1021:
1022: @table @code
1023: @cindex -i, command-line option
1024: @cindex --image-file, command-line option
1025: @item --image-file @i{file}
1026: @itemx -i @i{file}
1027: Loads the Forth image @i{file} instead of the default
1028: @file{gforth.fi} (@pxref{Image Files}).
1029:
1030: @cindex --appl-image, command-line option
1031: @item --appl-image @i{file}
1032: Loads the image @i{file} and leaves all further command-line arguments
1033: to the image (instead of processing them as options). This is useful
1034: for building executable application images on Unix, built with
1035: @code{gforthmi --application ...}.
1036:
1037: @cindex --path, command-line option
1038: @cindex -p, command-line option
1039: @item --path @i{path}
1040: @itemx -p @i{path}
1041: Uses @i{path} for searching the image file and Forth source code files
1042: instead of the default in the environment variable @code{GFORTHPATH} or
1043: the path specified at installation time (e.g.,
1044: @file{/usr/local/share/gforth/0.2.0:.}). A path is given as a list of
1045: directories, separated by @samp{:} (on Unix) or @samp{;} (on other OSs).
1046:
1047: @cindex --dictionary-size, command-line option
1048: @cindex -m, command-line option
1049: @cindex @i{size} parameters for command-line options
1050: @cindex size of the dictionary and the stacks
1051: @item --dictionary-size @i{size}
1052: @itemx -m @i{size}
1053: Allocate @i{size} space for the Forth dictionary space instead of
1054: using the default specified in the image (typically 256K). The
1055: @i{size} specification for this and subsequent options consists of
1056: an integer and a unit (e.g.,
1057: @code{4M}). The unit can be one of @code{b} (bytes), @code{e} (element
1058: size, in this case Cells), @code{k} (kilobytes), @code{M} (Megabytes),
1059: @code{G} (Gigabytes), and @code{T} (Terabytes). If no unit is specified,
1060: @code{e} is used.
1061:
1062: @cindex --data-stack-size, command-line option
1063: @cindex -d, command-line option
1064: @item --data-stack-size @i{size}
1065: @itemx -d @i{size}
1066: Allocate @i{size} space for the data stack instead of using the
1067: default specified in the image (typically 16K).
1068:
1069: @cindex --return-stack-size, command-line option
1070: @cindex -r, command-line option
1071: @item --return-stack-size @i{size}
1072: @itemx -r @i{size}
1073: Allocate @i{size} space for the return stack instead of using the
1074: default specified in the image (typically 15K).
1075:
1076: @cindex --fp-stack-size, command-line option
1077: @cindex -f, command-line option
1078: @item --fp-stack-size @i{size}
1079: @itemx -f @i{size}
1080: Allocate @i{size} space for the floating point stack instead of
1081: using the default specified in the image (typically 15.5K). In this case
1082: the unit specifier @code{e} refers to floating point numbers.
1083:
1084: @cindex --locals-stack-size, command-line option
1085: @cindex -l, command-line option
1086: @item --locals-stack-size @i{size}
1087: @itemx -l @i{size}
1088: Allocate @i{size} space for the locals stack instead of using the
1089: default specified in the image (typically 14.5K).
1090:
1091: @cindex -h, command-line option
1092: @cindex --help, command-line option
1093: @item --help
1094: @itemx -h
1095: Print a message about the command-line options
1096:
1097: @cindex -v, command-line option
1098: @cindex --version, command-line option
1099: @item --version
1100: @itemx -v
1101: Print version and exit
1102:
1103: @cindex --debug, command-line option
1104: @item --debug
1105: Print some information useful for debugging on startup.
1106:
1107: @cindex --offset-image, command-line option
1108: @item --offset-image
1109: Start the dictionary at a slightly different position than would be used
1110: otherwise (useful for creating data-relocatable images,
1111: @pxref{Data-Relocatable Image Files}).
1112:
1113: @cindex --no-offset-im, command-line option
1114: @item --no-offset-im
1115: Start the dictionary at the normal position.
1116:
1117: @cindex --clear-dictionary, command-line option
1118: @item --clear-dictionary
1119: Initialize all bytes in the dictionary to 0 before loading the image
1120: (@pxref{Data-Relocatable Image Files}).
1121:
1122: @cindex --die-on-signal, command-line-option
1123: @item --die-on-signal
1124: Normally Gforth handles most signals (e.g., the user interrupt SIGINT,
1125: or the segmentation violation SIGSEGV) by translating it into a Forth
1126: @code{THROW}. With this option, Gforth exits if it receives such a
1127: signal. This option is useful when the engine and/or the image might be
1128: severely broken (such that it causes another signal before recovering
1129: from the first); this option avoids endless loops in such cases.
1130: @end table
1131:
1132: @cindex loading files at startup
1133: @cindex executing code on startup
1134: @cindex batch processing with Gforth
1135: As explained above, the image-specific command-line arguments for the
1136: default image @file{gforth.fi} consist of a sequence of filenames and
1137: @code{-e @var{forth-code}} options that are interpreted in the sequence
1138: in which they are given. The @code{-e @var{forth-code}} or
1139: @code{--evaluate @var{forth-code}} option evaluates the Forth
1140: code. This option takes only one argument; if you want to evaluate more
1141: Forth words, you have to quote them or use @code{-e} several times. To exit
1142: after processing the command line (instead of entering interactive mode)
1143: append @code{-e bye} to the command line.
1144:
1145: @cindex versions, invoking other versions of Gforth
1146: If you have several versions of Gforth installed, @code{gforth} will
1147: invoke the version that was installed last. @code{gforth-@i{version}}
1148: invokes a specific version. You may want to use the option
1149: @code{--path}, if your environment contains the variable
1150: @code{GFORTHPATH}.
1151:
1152: Not yet implemented:
1153: On startup the system first executes the system initialization file
1154: (unless the option @code{--no-init-file} is given; note that the system
1155: resulting from using this option may not be ANS Forth conformant). Then
1156: the user initialization file @file{.gforth.fs} is executed, unless the
1157: option @code{--no-rc} is given; this file is first searched in @file{.},
1158: then in @file{~}, then in the normal path (see above).
1159:
1160:
1161:
1162: @comment ----------------------------------------------
1163: @node Leaving Gforth, Command-line editing, Invoking Gforth, Gforth Environment
1164: @section Leaving Gforth
1165: @cindex Gforth - leaving
1166: @cindex leaving Gforth
1167:
1168: You can leave Gforth by typing @code{bye} or @kbd{Ctrl-d} (at the start
1169: of a line) or (if you invoked Gforth with the @code{--die-on-signal}
1170: option) @kbd{Ctrl-c}. When you leave Gforth, all of your definitions and
1171: data are discarded. @xref{Image Files} for ways of saving the state of
1172: the system before leaving Gforth.
1173:
1174: doc-bye
1175:
1176:
1177: @comment ----------------------------------------------
1178: @node Command-line editing, Upper and lower case,Leaving Gforth,Gforth Environment
1179: @section Command-line editing
1180: @cindex command-line editing
1181:
1182: Gforth maintains a history file that records every line that you type to
1183: the text interpreter. This file is preserved between sessions, and is
1184: used to provide a command-line recall facility; if you type ctrl-P
1185: repeatedly you can recall successively older commands from this (or
1186: previous) session(s). The full list of command-line editing facilities is:
1187:
1188: @itemize @bullet
1189: @item
1190: @kbd{Ctrl-p} (``previous'') (or up-arrow) to recall successively older
1191: commands from the history buffer.
1192: @item
1193: @kbd{Ctrl-n} (``next'') (or down-arrow) to recall successively newer commands
1194: from the history buffer.
1195: @item
1196: @kbd{Ctrl-f} (or right-arrow) to move the cursor right, non-destructively.
1197: @item
1198: @kbd{Ctrl-b} (or left-arrow) to move the cursor left, non-destructively.
1199: @item
1200: @kbd{Ctrl-h} (backspace) to delete the character to the left of the cursor,
1201: closing up the line.
1202: @item
1203: @kbd{Ctrl-k} to delete (``kill'') from the cursor to the end of the line.
1204: @item
1205: @kbd{Ctrl-a} to move the cursor to the start of the line.
1206: @item
1207: @kbd{Ctrl-e} to move the cursor to the end of the line.
1208: @item
1209: @key{RET} (@kbd{Ctrl-m}) or @key{LFD} (@kbd{Ctrl-j}) to submit the current
1210: line.
1211: @item
1212: @key{TAB} to step through all possible full-word completions of the word
1213: currently being typed.
1214: @item
1215: @kbd{Ctrl-d} at the start of the line to terminate Gforth (gracefully,
1216: using @code{bye}).
1217: @end itemize
1218:
1219: When editing, displayable characters are inserted to the left of the
1220: cursor position; the line is always in ``insert'' (as opposed to
1221: ``overstrike'') mode.
1222:
1223: @cindex history file
1224: @cindex @file{.gforth-history}
1225: On Unix systems, the history file is @file{~/.gforth-history} by
1226: default@footnote{i.e. it is stored in the user's home directory.}. You
1227: can find out the name and location of your history file using:
1228:
1229: @example
1230: history-file type \ Unix-class systems
1231:
1232: history-file type \ Other systems
1233: history-dir type
1234: @end example
1235:
1236: If you enter long definitions by hand, you can use a text editor to
1237: paste them out of the history file into a Forth source file for reuse at
1238: a later time.
1239:
1240: Gforth never trims the size of the history file, so you should do this
1241: periodically, if necessary.
1242:
1243: @comment this is all defined in history.fs
1244: @comment NAC TODO the ctrl-D behaviour can either do a bye or a beep.. how is that option
1245: @comment chosen?
1246:
1247:
1248:
1249: @comment ----------------------------------------------
1250: @node Upper and lower case, Environment variables,Command-line editing,Gforth Environment
1251: @section Upper and lower case
1252: @cindex case-sensitivity
1253: @cindex upper and lower case
1254:
1255: Gforth is case-insensitive; you can enter definitions and invoke
1256: Standard words using upper, lower or mixed case (however,
1257: @pxref{core-idef, Implementation-defined options, Implementation-defined
1258: options}).
1259:
1260: ANS Forth only @i{requires} implementations to recognise Standard words
1261: when they are typed entirely in upper case. Therefore, a Standard
1262: program must use upper case for all Standard words. You can use whatever
1263: case you like for words that you define, but in a standard program you
1264: have to use the words in the same case that you defined them.
1265:
1266: Gforth supports case sensitivity through @code{table}s (case-sensitive
1267: wordlists, @pxref{Word Lists}).
1268:
1269: Two people have asked how to convert Gforth to case sensitivity; while
1270: we think this is a bad idea, you can change all wordlists into tables
1271: like this:
1272:
1273: @example
1274: ' table-find forth-wordlist wordlist-map @ !
1275: @end example
1276:
1277: Note that you now have to type the predefined words in the same case
1278: that we defined them, which are varying. You may want to convert them
1279: to your favourite case before doing this operation (I won't explain how,
1280: because if you are even contemplating to do this, you'd better have
1281: enough knowledge of Forth systems to know this already).
1282:
1283: @comment ----------------------------------------------
1284: @node Environment variables, Gforth Files, Upper and lower case,Gforth Environment
1285: @section Environment variables
1286: @cindex environment variables
1287:
1288: Gforth uses these environment variables:
1289:
1290: @itemize @bullet
1291: @item
1292: @cindex @code{GFORTHHIST} -- environment variable
1293: @code{GFORTHHIST} -- (Unix systems only) specifies the directory in which to
1294: open/create the history file, @file{.gforth-history}. Default:
1295: @code{$HOME}.
1296:
1297: @item
1298: @cindex @code{GFORTHPATH} -- environment variable
1299: @code{GFORTHPATH} -- specifies the path used when searching for the gforth image file and
1300: for Forth source-code files.
1301:
1302: @item
1303: @cindex @code{GFORTH} -- environment variable
1304: @code{GFORTH} -- used by @file{gforthmi} @xref{gforthmi}.
1305:
1306: @item
1307: @cindex @code{GFORTHD} -- environment variable
1308: @code{GFORTHD} -- used by @file{gforthmi} @xref{gforthmi}.
1309:
1310: @item
1311: @cindex @code{TMP}, @code{TEMP} - environment variable
1312: @code{TMP}, @code{TEMP} - (non-Unix systems only) used as a potential
1313: location for the history file.
1314: @end itemize
1315:
1316: @comment also POSIXELY_CORRECT LINES COLUMNS HOME but no interest in
1317: @comment mentioning these.
1318:
1319: All the Gforth environment variables default to sensible values if they
1320: are not set.
1321:
1322:
1323: @comment ----------------------------------------------
1324: @node Gforth Files, ,Environment variables,Gforth Environment
1325: @section Gforth files
1326: @cindex Gforth files
1327:
1328: When you install Gforth on a Unix system, it installs files in these
1329: locations by default:
1330:
1331: @itemize @bullet
1332: @item
1333: @file{/usr/local/bin/gforth}
1334: @item
1335: @file{/usr/local/bin/gforthmi}
1336: @item
1337: @file{/usr/local/man/man1/gforth.1} - man page.
1338: @item
1339: @file{/usr/local/info} - the Info version of this manual.
1340: @item
1341: @file{/usr/local/lib/gforth/<version>/...} - Gforth @file{.fi} files.
1342: @item
1343: @file{/usr/local/share/gforth/<version>/TAGS} - Emacs TAGS file.
1344: @item
1345: @file{/usr/local/share/gforth/<version>/...} - Gforth source files.
1346: @item
1347: @file{.../emacs/site-lisp/gforth.el} - Emacs gforth mode.
1348: @end itemize
1349:
1350: You can select different places for installation by using
1351: @code{configure} options (listed with @code{configure --help}).
1352:
1353: @c ******************************************************************
1354: @node Introduction, Words, Gforth Environment, Top
1355: @comment node-name, next, previous, up
1356: @chapter An Introduction to ANS Forth
1357: @cindex Forth - an introduction
1358:
1359: The primary purpose of this manual is to document Gforth. However, since
1360: Forth is not a widely-known language and there is a lack of up-to-date
1361: teaching material, it seems worthwhile to provide some introductory
1362: material. @xref{Forth-related information} for other sources of Forth-related
1363: information.
1364:
1365: The examples in this section should work on any ANS Forth; the
1366: output shown was produced using Gforth. Each example attempts to
1367: reproduce the exact output that Gforth produces. If you try out the
1368: examples (and you should), what you should type is shown @kbd{like this}
1369: and Gforth's response is shown @code{like this}. The single exception is
1370: that, where the example shows @key{RET} it means that you should
1371: press the ``carriage return'' key. Unfortunately, some output formats for
1372: this manual cannot show the difference between @kbd{this} and
1373: @code{this} which will make trying out the examples harder (but not
1374: impossible).
1375:
1376: Forth is an unusual language. It provides an interactive development
1377: environment which includes both an interpreter and compiler. Forth
1378: programming style encourages you to break a problem down into many
1379: @cindex factoring
1380: small fragments (@dfn{factoring}), and then to develop and test each
1381: fragment interactively. Forth advocates assert that breaking the
1382: edit-compile-test cycle used by conventional programming languages can
1383: lead to great productivity improvements.
1384:
1385: @menu
1386: * Introducing the Text Interpreter::
1387: * Stacks and Postfix notation::
1388: * Your first definition::
1389: * How does that work?::
1390: * Forth is written in Forth::
1391: * Review - elements of a Forth system::
1392: * Where to go next::
1393: * Exercises::
1394: @end menu
1395:
1396: @comment ----------------------------------------------
1397: @node Introducing the Text Interpreter, Stacks and Postfix notation, Introduction, Introduction
1398: @section Introducing the Text Interpreter
1399: @cindex text interpreter
1400: @cindex outer interpreter
1401:
1402: @c IMO this is too detailed and the pace is too slow for
1403: @c an introduction. If you know German, take a look at
1404: @c http://www.complang.tuwien.ac.at/anton/lvas/skriptum-stack.html
1405: @c to see how I do it - anton
1406:
1407: @c nac-> Where I have accepted your comments 100% and modified the text
1408: @c accordingly, I have deleted your comments. Elsewhere I have added a
1409: @c response like this to attempt to rationalise what I have done. Of
1410: @c course, this is a very clumsy mechanism for something that would be
1411: @c done far more efficiently over a beer. Please delete any dialogue
1412: @c you consider closed.
1413:
1414: When you invoke the Forth image, you will see a startup banner printed
1415: and nothing else (if you have Gforth installed on your system, try
1416: invoking it now, by typing @kbd{gforth@key{RET}}). Forth is now running
1417: its command line interpreter, which is called the @dfn{Text Interpreter}
1418: (also known as the @dfn{Outer Interpreter}). (You will learn a lot
1419: about the text interpreter as you read through this chapter, but
1420: @pxref{The Text Interpreter} for more detail).
1421:
1422: Although it's not obvious, Forth is actually waiting for your
1423: input. Type a number and press the @key{RET} key:
1424:
1425: @example
1426: @kbd{45@key{RET}} ok
1427: @end example
1428:
1429: Rather than give you a prompt to invite you to input something, the text
1430: interpreter prints a status message @i{after} it has processed a line
1431: of input. The status message in this case (``@code{ ok}'' followed by
1432: carriage-return) indicates that the text interpreter was able to process
1433: all of your input successfully. Now type something illegal:
1434:
1435: @example
1436: @kbd{qwer341@key{RET}}
1437: :1: Undefined word
1438: qwer341
1439: ^^^^^^^
1440: $400D2BA8 Bounce
1441: $400DBDA8 no.extensions
1442: @end example
1443:
1444: The exact text, other than the ``Undefined word'' may differ slightly on
1445: your system, but the effect is the same; when the text interpreter
1446: detects an error, it discards any remaining text on a line, resets
1447: certain internal state and prints an error message. @xref{Error
1448: messages} for a detailed description of error messages.
1449:
1450: The text interpreter waits for you to press carriage-return, and then
1451: processes your input line. Starting at the beginning of the line, it
1452: breaks the line into groups of characters separated by spaces. For each
1453: group of characters in turn, it makes two attempts to do something:
1454:
1455: @itemize @bullet
1456: @item
1457: @cindex name dictionary
1458: It tries to treat it as a command. It does this by searching a @dfn{name
1459: dictionary}. If the group of characters matches an entry in the name
1460: dictionary, the name dictionary provides the text interpreter with
1461: information that allows the text interpreter perform some actions. In
1462: Forth jargon, we say that the group
1463: @cindex word
1464: @cindex definition
1465: @cindex execution token
1466: @cindex xt
1467: of characters names a @dfn{word}, that the dictionary search returns an
1468: @dfn{execution token (xt)} corresponding to the @dfn{definition} of the
1469: word, and that the text interpreter executes the xt. Often, the terms
1470: @dfn{word} and @dfn{definition} are used interchangeably.
1471: @item
1472: If the text interpreter fails to find a match in the name dictionary, it
1473: tries to treat the group of characters as a number in the current number
1474: base (when you start up Forth, the current number base is base 10). If
1475: the group of characters legitimately represents a number, the text
1476: interpreter pushes the number onto a stack (we'll learn more about that
1477: in the next section).
1478: @end itemize
1479:
1480: If the text interpreter is unable to do either of these things with any
1481: group of characters, it discards the group of characters and the rest of
1482: the line, then prints an error message. If the text interpreter reaches
1483: the end of the line without error, it prints the status message ``@code{ ok}''
1484: followed by carriage-return.
1485:
1486: This is the simplest command we can give to the text interpreter:
1487:
1488: @example
1489: @key{RET} ok
1490: @end example
1491:
1492: The text interpreter did everything we asked it to do (nothing) without
1493: an error, so it said that everything is ``@code{ ok}''. Try a slightly longer
1494: command:
1495:
1496: @example
1497: @kbd{12 dup fred dup@key{RET}}
1498: :1: Undefined word
1499: 12 dup fred dup
1500: ^^^^
1501: $400D2BA8 Bounce
1502: $400DBDA8 no.extensions
1503: @end example
1504:
1505: When you press the carriage-return key, the text interpreter starts to
1506: work its way along the line:
1507:
1508: @itemize @bullet
1509: @item
1510: When it gets to the space after the @code{2}, it takes the group of
1511: characters @code{12} and looks them up in the name
1512: dictionary@footnote{We can't tell if it found them or not, but assume
1513: for now that it did not}. There is no match for this group of characters
1514: in the name dictionary, so it tries to treat them as a number. It is
1515: able to do this successfully, so it puts the number, 12, ``on the stack''
1516: (whatever that means).
1517: @item
1518: The text interpreter resumes scanning the line and gets the next group
1519: of characters, @code{dup}. It looks it up in the name dictionary and
1520: (you'll have to take my word for this) finds it, and executes the word
1521: @code{dup} (whatever that means).
1522: @item
1523: Once again, the text interpreter resumes scanning the line and gets the
1524: group of characters @code{fred}. It looks them up in the name
1525: dictionary, but can't find them. It tries to treat them as a number, but
1526: they don't represent any legal number.
1527: @end itemize
1528:
1529: At this point, the text interpreter gives up and prints an error
1530: message. The error message shows exactly how far the text interpreter
1531: got in processing the line. In particular, it shows that the text
1532: interpreter made no attempt to do anything with the final character
1533: group, @code{dup}, even though we have good reason to believe that the
1534: text interpreter would have no problem looking that word up and
1535: executing it a second time.
1536:
1537:
1538: @comment ----------------------------------------------
1539: @node Stacks and Postfix notation, Your first definition, Introducing the Text Interpreter, Introduction
1540: @section Stacks, postfix notation and parameter passing
1541: @cindex text interpreter
1542: @cindex outer interpreter
1543:
1544: In procedural programming languages (like C and Pascal), the
1545: building-block of programs is the @dfn{function} or @dfn{procedure}. These
1546: functions or procedures are called with @dfn{explicit parameters}. For
1547: example, in C we might write:
1548:
1549: @example
1550: total = total + new_volume(length,height,depth);
1551: @end example
1552:
1553: @noindent
1554: where new_volume is a function-call to another piece of code, and total,
1555: length, height and depth are all variables. length, height and depth are
1556: parameters to the function-call.
1557:
1558: In Forth, the equivalent of the function or procedure is the
1559: @dfn{definition} and parameters are implicitly passed between
1560: definitions using a shared stack that is visible to the
1561: programmer. Although Forth does support variables, the existence of the
1562: stack means that they are used far less often than in most other
1563: programming languages. When the text interpreter encounters a number, it
1564: will place (@dfn{push}) it on the stack. There are several stacks (the
1565: actual number is implementation-dependent ...) and the particular stack
1566: used for any operation is implied unambiguously by the operation being
1567: performed. The stack used for all integer operations is called the @dfn{data
1568: stack} and, since this is the stack used most commonly, references to
1569: ``the data stack'' are often abbreviated to ``the stack''.
1570:
1571: The stacks have a last-in, first-out (LIFO) organisation. If you type:
1572:
1573: @example
1574: @kbd{1 2 3@key{RET}} ok
1575: @end example
1576:
1577: Then this instructs the text interpreter to placed three numbers on the
1578: (data) stack. An analogy for the behaviour of the stack is to take a
1579: pack of playing cards and deal out the ace (1), 2 and 3 into a pile on
1580: the table. The 3 was the last card onto the pile (``last-in'') and if
1581: you take a card off the pile then, unless you're prepared to fiddle a
1582: bit, the card that you take off will be the 3 (``first-out''). The
1583: number that will be first-out of the stack is called the @dfn{top of
1584: stack}, which
1585: @cindex TOS definition
1586: is often abbreviated to @dfn{TOS}.
1587:
1588: To understand how parameters are passed in Forth, consider the
1589: behaviour of the definition @code{+} (pronounced ``plus''). You will not
1590: be surprised to learn that this definition performs addition. More
1591: precisely, it adds two number together and produces a result. Where does
1592: it get the two numbers from? It takes the top two numbers off the
1593: stack. Where does it place the result? On the stack. You can act-out the
1594: behaviour of @code{+} with your playing cards like this:
1595:
1596: @itemize @bullet
1597: @item
1598: Pick up two cards from the stack on the table
1599: @item
1600: Stare at them intently and ask yourself ``what @i{is} the sum of these two
1601: numbers''
1602: @item
1603: Decide that the answer is 5
1604: @item
1605: Shuffle the two cards back into the pack and find a 5
1606: @item
1607: Put a 5 on the remaining ace that's on the table.
1608: @end itemize
1609:
1610: If you don't have a pack of cards handy but you do have Forth running,
1611: you can use the definition @code{.s} to show the current state of the stack,
1612: without affecting the stack. Type:
1613:
1614: @example
1615: @kbd{clearstack 1 2 3@key{RET}} ok
1616: @kbd{.s@key{RET}} <3> 1 2 3 ok
1617: @end example
1618:
1619: The text interpreter looks up the word @code{clearstack} and executes
1620: it; it tidies up the stack and removes any entries that may have been
1621: left on it by earlier examples. The text interpreter pushes each of the
1622: three numbers in turn onto the stack. Finally, the text interpreter
1623: looks up the word @code{.s} and executes it. The effect of executing
1624: @code{.s} is to print the ``<3>'' (the total number of items on the stack)
1625: followed by a list of all the items on the stack; the item on the far
1626: right-hand side is the TOS.
1627:
1628: You can now type:
1629:
1630: @example
1631: @kbd{+ .s@key{RET}} <2> 1 5 ok
1632: @end example
1633:
1634: @noindent
1635: which is correct; there are now 2 items on the stack and the result of
1636: the addition is 5.
1637:
1638: If you're playing with cards, try doing a second addition: pick up the
1639: two cards, work out that their sum is 6, shuffle them into the pack,
1640: look for a 6 and place that on the table. You now have just one item on
1641: the stack. What happens if you try to do a third addition? Pick up the
1642: first card, pick up the second card -- ah! There is no second card. This
1643: is called a @dfn{stack underflow} and consitutes an error. If you try to
1644: do the same thing with Forth it will report an error (probably a Stack
1645: Underflow or an Invalid Memory Address error).
1646:
1647: The opposite situation to a stack underflow is a @dfn{stack overflow},
1648: which simply accepts that there is a finite amount of storage space
1649: reserved for the stack. To stretch the playing card analogy, if you had
1650: enough packs of cards and you piled the cards up on the table, you would
1651: eventually be unable to add another card; you'd hit the ceiling. Gforth
1652: allows you to set the maximum size of the stacks. In general, the only
1653: time that you will get a stack overflow is because a definition has a
1654: bug in it and is generating data on the stack uncontrollably.
1655:
1656: There's one final use for the playing card analogy. If you model your
1657: stack using a pack of playing cards, the maximum number of items on
1658: your stack will be 52 (I assume you didn't use the Joker). The maximum
1659: @i{value} of any item on the stack is 13 (the King). In fact, the only
1660: possible numbers are positive integer numbers 1 through 13; you can't
1661: have (for example) 0 or 27 or 3.52 or -2. If you change the way you
1662: think about some of the cards, you can accommodate different
1663: numbers. For example, you could think of the Jack as representing 0,
1664: the Queen as representing -1 and the King as representing -2. Your
1665: *range* remains unchanged (you can still only represent a total of 13
1666: numbers) but the numbers that you can represent are -2 through 10.
1667:
1668: In that analogy, the limit was the amount of information that a single
1669: stack entry could hold, and Forth has a similar limit. In Forth, the
1670: size of a stack entry is called a @dfn{cell}. The actual size of a cell is
1671: implementation dependent and affects the maximum value that a stack
1672: entry can hold. A Standard Forth provides a cell size of at least
1673: 16-bits, and most desktop systems use a cell size of 32-bits.
1674:
1675: Forth does not do any type checking for you, so you are free to
1676: manipulate and combine stack items in any way you wish. A convenient way
1677: of treating stack items is as 2's complement signed integers, and that
1678: is what Standard words like @code{+} do. Therefore you can type:
1679:
1680: @example
1681: @kbd{-5 12 + .s@key{RET}} <1> 7 ok
1682: @end example
1683:
1684: If you use numbers and definitions like @code{+} in order to turn Forth
1685: into a great big pocket calculator, you will realise that it's rather
1686: different from a normal calculator. Rather than typing 2 + 3 = you had
1687: to type 2 3 + (ignore the fact that you had to use @code{.s} to see the
1688: result). The terminology used to describe this difference is to say that
1689: your calculator uses @dfn{Infix Notation} (parameters and operators are
1690: mixed) whilst Forth uses @dfn{Postfix Notation} (parameters and
1691: operators are separate), also called @dfn{Reverse Polish Notation}.
1692:
1693: Whilst postfix notation might look confusing to begin with, it has
1694: several important advantages:
1695:
1696: @itemize @bullet
1697: @item
1698: it is unambiguous
1699: @item
1700: it is more concise
1701: @item
1702: it fits naturally with a stack-based system
1703: @end itemize
1704:
1705: To examine these claims in more detail, consider these sums:
1706:
1707: @example
1708: 6 + 5 * 4 =
1709: 4 * 5 + 6 =
1710: @end example
1711:
1712: If you're just learning maths or your maths is very rusty, you will
1713: probably come up with the answer 44 for the first and 26 for the
1714: second. If you are a bit of a whizz at maths you will remember the
1715: @i{convention} that multiplication takes precendence over addition, and
1716: you'd come up with the answer 26 both times. To explain the answer 26
1717: to someone who got the answer 44, you'd probably rewrite the first sum
1718: like this:
1719:
1720: @example
1721: 6 + (5 * 4) =
1722: @end example
1723:
1724: If what you really wanted was to perform the addition before the
1725: multiplication, you would have to use parentheses to force it.
1726:
1727: If you did the first two sums on a pocket calculator you would probably
1728: get the right answers, unless you were very cautious and entered them using
1729: these keystroke sequences:
1730:
1731: 6 + 5 = * 4 =
1732: 4 * 5 = + 6 =
1733:
1734: Postfix notation is unambiguous because the order that the operators
1735: are applied is always explicit; that also means that parentheses are
1736: never required. The operators are @i{active} (the act of quoting the
1737: operator makes the operation occur) which removes the need for ``=''.
1738:
1739: The sum 6 + 5 * 4 can be written (in postfix notation) in two
1740: equivalent ways:
1741:
1742: @example
1743: 6 5 4 * + or:
1744: 5 4 * 6 +
1745: @end example
1746:
1747: An important thing that you should notice about this notation is that
1748: the @i{order} of the numbers does not change; if you want to subtract
1749: 2 from 10 you type @code{10 2 -}.
1750:
1751: The reason that Forth uses postfix notation is very simple to explain: it
1752: makes the implementation extremely simple, and it follows naturally from
1753: using the stack as a mechanism for passing parameters. Another way of
1754: thinking about this is to realise that all Forth definitions are
1755: @i{active}; they execute as they are encountered by the text
1756: interpreter. The result of this is that the syntax of Forth is trivially
1757: simple.
1758:
1759:
1760:
1761: @comment ----------------------------------------------
1762: @node Your first definition, How does that work?, Stacks and Postfix notation, Introduction
1763: @section Your first Forth definition
1764: @cindex first definition
1765:
1766: Until now, the examples we've seen have been trivial; we've just been
1767: using Forth as a bigger-than-pocket calculator. Also, each calculation
1768: we've shown has been a ``one-off'' -- to repeat it we'd need to type it in
1769: again@footnote{That's not quite true. If you press the up-arrow key on
1770: your keyboard you should be able to scroll back to any earlier command,
1771: edit it and re-enter it.} In this section we'll see how to add new
1772: words to Forth's vocabulary.
1773:
1774: The easiest way to create a new word is to use a @dfn{colon
1775: definition}. We'll define a few and try them out before worrying too
1776: much about how they work. Try typing in these examples; be careful to
1777: copy the spaces accurately:
1778:
1779: @example
1780: : add-two 2 + . ;
1781: : greet ." Hello and welcome" ;
1782: : demo 5 add-two ;
1783: @end example
1784:
1785: @noindent
1786: Now try them out:
1787:
1788: @example
1789: @kbd{greet@key{RET}} Hello and welcome ok
1790: @kbd{greet greet@key{RET}} Hello and welcomeHello and welcome ok
1791: @kbd{4 add-two@key{RET}} 6 ok
1792: @kbd{demo@key{RET}} 7 ok
1793: @kbd{9 greet demo add-two@key{RET}} Hello and welcome7 11 ok
1794: @end example
1795:
1796: The first new thing that we've introduced here is the pair of words
1797: @code{:} and @code{;}. These are used to start and terminate a new
1798: definition, respectively. The first word after the @code{:} is the name
1799: for the new definition.
1800:
1801: As you can see from the examples, a definition is built up of words that
1802: have already been defined; Forth makes no distinction between
1803: definitions that existed when you started the system up, and those that
1804: you define yourself.
1805:
1806: The examples also introduce the words @code{.} (dot), @code{."}
1807: (dot-quote) and @code{dup} (dewp). Dot takes the value from the top of
1808: the stack and displays it. It's like @code{.s} except that it only
1809: displays the top item of the stack and it is destructive; after it has
1810: executed, the number is no longer on the stack. There is always one
1811: space printed after the number, and no spaces before it. Dot-quote
1812: defines a string (a sequence of characters) that will be printed when
1813: the word is executed. The string can contain any printable characters
1814: except @code{"}. A @code{"} has a special function; it is not a Forth
1815: word but it acts as a delimiter (the way that delimiters work is
1816: described in the next section). Finally, @code{dup} duplicates the value
1817: at the top of the stack. Try typing @code{5 dup .s} to see what it does.
1818:
1819: We already know that the text interpreter searches through the
1820: dictionary to locate names. If you've followed the examples earlier, you
1821: will already have a definition called @code{add-two}. Lets try modifying
1822: it by typing in a new definition:
1823:
1824: @example
1825: @kbd{: add-two dup . ." + 2 =" 2 + . ;@key{RET}} redefined add-two ok
1826: @end example
1827:
1828: Forth recognised that we were defining a word that already exists, and
1829: printed a message to warn us of that fact. Let's try out the new
1830: definition:
1831:
1832: @example
1833: @kbd{9 add-two@key{RET}} 9 + 2 =11 ok
1834: @end example
1835:
1836: @noindent
1837: All that we've actually done here, though, is to create a new
1838: definition, with a particular name. The fact that there was already a
1839: definition with the same name did not make any difference to the way
1840: that the new definition was created (except that Forth printed a warning
1841: message). The old definition of add-two still exists (try @code{demo}
1842: again to see that this is true). Any new definition will use the new
1843: definition of @code{add-two}, but old definitions continue to use the
1844: version that already existed at the time that they were @code{compiled}.
1845:
1846: Before you go on to the next section, try defining and redefining some
1847: words of your own.
1848:
1849: @comment ----------------------------------------------
1850: @node How does that work?, Forth is written in Forth, Your first definition, Introduction
1851: @section How does that work?
1852: @cindex parsing words
1853:
1854: @c That's pretty deep (IMO way too deep) for an introduction. - anton
1855:
1856: @c Is it a good idea to talk about the interpretation semantics of a
1857: @c number? We don't have an xt to go along with it. - anton
1858:
1859: @c Now that I have eliminated execution semantics, I wonder if it would not
1860: @c be better to keep them (or add run-time semantics), to make it easier to
1861: @c explain what compilation semantics usually does. - anton
1862:
1863: @c nac-> I removed the term ``default compilation sematics'' from the
1864: @c introductory chapter. Removing ``execution semantics'' was making
1865: @c everything simpler to explain, then I think the use of this term made
1866: @c everything more complex again. I replaced it with ``default
1867: @c semantics'' (which is used elsewhere in the manual) by which I mean
1868: @c ``a definition that has neither the immediate nor the compile-only
1869: @c flag set''. I reworded big chunks of the ``how does that work''
1870: @c section (and, unusually for me, I think I even made it shorter!). See
1871: @c what you think -- I know I have not addressed your primary concern
1872: @c that it is too heavy-going for an introduction. From what I understood
1873: @c of your course notes it looks as though they might be a good framework.
1874: @c Things that I've tried to capture here are some things that came as a
1875: @c great revelation here when I first understood them. Also, I like the
1876: @c fact that a very simple code example shows up almost all of the issues
1877: @c that you need to understand to see how Forth works. That's unique and
1878: @c worthwhile to emphasise.
1879:
1880: Now we're going to take another look at the definition of @code{add-two}
1881: from the previous section. From our knowledge of the way that the text
1882: interpreter works, we would have expected this result when we tried to
1883: define @code{add-two}:
1884:
1885: @example
1886: @kbd{: add-two 2 + . ;@key{RET}}
1887: ^^^^^^^
1888: Error: Undefined word
1889: @end example
1890:
1891: The reason that this didn't happen is bound up in the way that @code{:}
1892: works. The word @code{:} does two special things. The first special
1893: thing that it does prevents the text interpreter from ever seeing the
1894: characters @code{add-two}. The text interpreter uses a variable called
1895: @cindex modifying >IN
1896: @code{>IN} (pronounced ``to-in'') to keep track of where it is in the
1897: input line. When it encounters the word @code{:} it behaves in exactly
1898: the same way as it does for any other word; it looks it up in the name
1899: dictionary, finds its xt and executes it. When @code{:} executes, it
1900: looks at the input buffer, finds the word @code{add-two} and advances the
1901: value of @code{>IN} to point past it. It then does some other stuff
1902: associated with creating the new definition (including creating an entry
1903: for @code{add-two} in the name dictionary). When the execution of @code{:}
1904: completes, control returns to the text interpreter, which is oblivious
1905: to the fact that it has been tricked into ignoring part of the input
1906: line.
1907:
1908: @cindex parsing words
1909: Words like @code{:} -- words that advance the value of @code{>IN} and so
1910: prevent the text interpreter from acting on the whole of the input line
1911: -- are called @dfn{parsing words}.
1912:
1913: @cindex @code{state} - effect on the text interpreter
1914: @cindex text interpreter - effect of state
1915: The second special thing that @code{:} does is change the value of a
1916: variable called @code{state}, which affects the way that the text
1917: interpreter behaves. When Gforth starts up, @code{state} has the value
1918: 0, and the text interpreter is said to be @dfn{interpreting}. During a
1919: colon definition (started with @code{:}), @code{state} is set to -1 and
1920: the text interpreter is said to be @dfn{compiling}.
1921:
1922: In this example, the text interpreter is compiling when it processes the
1923: string ``@code{2 + . ;}''. It still breaks the string down into
1924: character sequences in the same way. However, instead of pushing the
1925: number @code{2} onto the stack, it lays down (@dfn{compiles}) some magic
1926: into the definition of @code{add-two} that will make the number @code{2} get
1927: pushed onto the stack when @code{add-two} is @dfn{executed}. Similarly,
1928: the behaviours of @code{+} and @code{.} are also compiled into the
1929: definition.
1930:
1931: One category of words don't get compiled. These so-called @dfn{immediate
1932: words} get executed (performed @i{now}) regardless of whether the text
1933: interpreter is interpreting or compiling. The word @code{;} is an
1934: immediate word. Rather than being compiled into the definition, it
1935: executes. Its effect is to terminate the current definition, which
1936: includes changing the value of @code{state} back to 0.
1937:
1938: When you execute @code{add-two}, it has a @dfn{run-time effect} that is
1939: exactly the same as if you had typed @code{2 + . @key{RET}} outside of a
1940: definition.
1941:
1942: In Forth, every word or number can be described in terms of two
1943: properties:
1944:
1945: @itemize @bullet
1946: @item
1947: @cindex interpretation semantics
1948: Its @dfn{interpretation semantics} describe how it will behave when the
1949: text interpreter encounters it in @dfn{interpret} state. The
1950: interpretation semantics of a word are represented by an @dfn{execution
1951: token}.
1952: @item
1953: @cindex compilation semantics
1954: Its @dfn{compilation semantics} describe how it will behave when the
1955: text interpreter encounters it in @dfn{compile} state. The compilation
1956: semantics of a word are represented in an implementation-dependent way;
1957: Gforth uses a @dfn{compilation token}.
1958: @end itemize
1959:
1960: @noindent
1961: Numbers are always treated in a fixed way:
1962:
1963: @itemize @bullet
1964: @item
1965: When the number is @dfn{interpreted}, its behaviour is to push the
1966: number onto the stack.
1967: @item
1968: When the number is @dfn{compiled}, a piece of code is appended to the
1969: current definition that pushes the number when it runs. (In other words,
1970: the compilation semantics of a number are to postpone its interpretation
1971: semantics until the run-time of the definition that it is being compiled
1972: into.)
1973: @end itemize
1974:
1975: Words don't behave in such a regular way, but most have @i{default
1976: semantics} which means that they behave like this:
1977:
1978: @itemize @bullet
1979: @item
1980: The @dfn{interpretation semantics} of the word are to do something useful.
1981: @item
1982: The @dfn{compilation semantics} of the word are to append its
1983: @dfn{interpretation semantics} to the current definition (so that its
1984: run-time behaviour is to do something useful).
1985: @end itemize
1986:
1987: @cindex immediate words
1988: The actual behaviour of any particular word can be controlled by using
1989: the words @code{immediate} and @code{compile-only} when the word is
1990: defined. These words set flags in the name dictionary entry of the most
1991: recently defined word, and these flags are retrieved by the text
1992: interpreter when it finds the word in the name dictionary.
1993:
1994: A word that is marked as @dfn{immediate} has compilation semantics that
1995: are identical to its interpretation semantics. In other words, it
1996: behaves like this:
1997:
1998: @itemize @bullet
1999: @item
2000: The @dfn{interpretation semantics} of the word are to do something useful.
2001: @item
2002: The @dfn{compilation semantics} of the word are to do something useful
2003: (and actually the same thing); i.e., it is executed during compilation.
2004: @end itemize
2005:
2006: Marking a word as @dfn{compile-only} prohibits the text interpreter from
2007: performing the interpretation semantics of the word directly; an attempt
2008: to do so will generate an error. It is never necessary to use
2009: @code{compile-only} (and it is not even part of ANS Forth, though it is
2010: provided by many implementations) but it is good etiquette to apply it
2011: to a word that will not behave correctly (and might have unexpected
2012: side-effects) in interpret state. For example, it is only legal to use
2013: the conditional word @code{IF} within a definition. If you forget this
2014: and try to use it elsewhere, the fact that (in Gforth) it is marked as
2015: @code{compile-only} allows the text interpreter to generate a helpful
2016: error message rather than subjecting you to the consequences of your
2017: folly.
2018:
2019: This example shows the difference between an immediate and a
2020: non-immediate word:
2021:
2022: @example
2023: : show-state state @@ . ;
2024: : show-state-now show-state ; immediate
2025: : word1 show-state ;
2026: : word2 show-state-now ;
2027: @end example
2028:
2029: The word @code{immediate} after the definition of @code{show-state-now}
2030: makes that word an immediate word. These definitions introduce a new
2031: word: @code{@@} (pronounced ``fetch''). This word fetches the value of a
2032: variable, and leaves it on the stack. Therefore, the behaviour of
2033: @code{show-state} is to print a number that represents the current value
2034: of @code{state}.
2035:
2036: When you execute @code{word1}, it prints the number 0, indicating that
2037: the system is interpreting. When the text interpreter compiled the
2038: definition of @code{word1}, it encountered @code{show-state} whose
2039: compilation semantics are to append its interpretation semantics to the
2040: current definition. When you execute @code{word1}, it performs the
2041: interpretation semantics of @code{show-state}. At the time that @code{word1}
2042: (and therefore @code{show-state}) are executed, the system is
2043: interpreting.
2044:
2045: When you pressed @key{RET} after entering the definition of @code{word2},
2046: you should have seen the number -1 printed, followed by ``@code{
2047: ok}''. When the text interpreter compiled the definition of
2048: @code{word2}, it encountered @code{show-state-now}, an immediate word,
2049: whose compilation semantics are therefore to perform its interpretation
2050: semantics. It is executed straight away (even before the text
2051: interpreter has moved on to process another group of characters; the
2052: @code{;} in this example). The effect of executing it are to display the
2053: value of @code{state} @i{at the time that the definition of}
2054: @code{word2} @i{is being defined}. Printing -1 demonstrates that the
2055: system is compiling at this time. If you execute @code{word2} it does
2056: nothing at all.
2057:
2058: @cindex @code{."}, how it works
2059: Before leaving the subject of immediate words, consider the behaviour of
2060: @code{."} in the definition of @code{greet}, in the previous
2061: section. This word is both a parsing word and an immediate word. Notice
2062: that there is a space between @code{."} and the start of the text
2063: @code{Hello and welcome}, but that there is no space between the last
2064: letter of @code{welcome} and the @code{"} character. The reason for this
2065: is that @code{."} is a Forth word; it must have a space after it so that
2066: the text interpreter can identify it. The @code{"} is not a Forth word;
2067: it is a @dfn{delimiter}. The examples earlier show that, when the string
2068: is displayed, there is neither a space before the @code{H} nor after the
2069: @code{e}. Since @code{."} is an immediate word, it executes at the time
2070: that @code{greet} is defined. When it executes, its behaviour is to
2071: search forward in the input line looking for the delimiter. When it
2072: finds the delimiter, it updates @code{>IN} to point past the
2073: delimiter. It also compiles some magic code into the definition of
2074: @code{greet}; the xt of a run-time routine that prints a text string. It
2075: compiles the string @code{Hello and welcome} into memory so that it is
2076: available to be printed later. When the text interpreter gains control,
2077: the next word it finds in the input stream is @code{;} and so it
2078: terminates the definition of @code{greet}.
2079:
2080:
2081: @comment ----------------------------------------------
2082: @node Forth is written in Forth, Review - elements of a Forth system, How does that work?, Introduction
2083: @section Forth is written in Forth
2084: @cindex structure of Forth programs
2085:
2086: When you start up a Forth compiler, a large number of definitions
2087: already exist. In Forth, you develop a new application using bottom-up
2088: programming techniques to create new definitions that are defined in
2089: terms of existing definitions. As you create each definition you can
2090: test and debug it interactively.
2091:
2092: If you have tried out the examples in this section, you will probably
2093: have typed them in by hand; when you leave Gforth, your definitions will
2094: be lost. You can avoid this by using a text editor to enter Forth source
2095: code into a file, and then loading code from the file using
2096: @code{include} (@xref{Forth source files}). A Forth source file is
2097: processed by the text interpreter, just as though you had typed it in by
2098: hand@footnote{Actually, there are some subtle differences -- see
2099: @ref{The Text Interpreter}.}.
2100:
2101: Gforth also supports the traditional Forth alternative to using text
2102: files for program entry (@xref{Blocks}).
2103:
2104: In common with many, if not most, Forth compilers, most of Gforth is
2105: actually written in Forth. All of the @file{.fs} files in the
2106: installation directory@footnote{For example,
2107: @file{/usr/local/share/gforth...}} are Forth source files, which you can
2108: study to see examples of Forth programming.
2109:
2110: Gforth maintains a history file that records every line that you type to
2111: the text interpreter. This file is preserved between sessions, and is
2112: used to provide a command-line recall facility. If you enter long
2113: definitions by hand, you can use a text editor to paste them out of the
2114: history file into a Forth source file for reuse at a later time
2115: (@pxref{Command-line editing} for more information).
2116:
2117:
2118: @comment ----------------------------------------------
2119: @node Review - elements of a Forth system, Where to go next, Forth is written in Forth, Introduction
2120: @section Review - elements of a Forth system
2121: @cindex elements of a Forth system
2122:
2123: To summarise this chapter:
2124:
2125: @itemize @bullet
2126: @item
2127: Forth programs use @dfn{factoring} to break a problem down into small
2128: fragments called @dfn{words} or @dfn{definitions}.
2129: @item
2130: Forth program development is an interactive process.
2131: @item
2132: The main command loop that accepts input, and controls both
2133: interpretation and compilation, is called the @dfn{text interpreter}
2134: (also known as the @dfn{outer interpreter}).
2135: @item
2136: Forth has a very simple syntax, consisting of words and numbers
2137: separated by spaces or carriage-return characters. Any additional syntax
2138: is imposed by @dfn{parsing words}.
2139: @item
2140: Forth uses a stack to pass parameters between words. As a result, it
2141: uses postfix notation.
2142: @item
2143: To use a word that has previously been defined, the text interpreter
2144: searches for the word in the @dfn{name dictionary}.
2145: @item
2146: Words have @dfn{interpretation semantics} and @dfn{compilation semantics}.
2147: @item
2148: The text interpreter uses the value of @code{state} to select between
2149: the use of the @dfn{interpretation semantics} and the @dfn{compilation
2150: semantics} of a word that it encounters.
2151: @item
2152: The relationship between the @dfn{interpretation semantics} and
2153: @dfn{compilation semantics} for a word
2154: depend upon the way in which the word was defined (for example, whether
2155: it is an @dfn{immediate} word).
2156: @item
2157: Forth definitions can be implemented in Forth (called @dfn{high-level
2158: definitions}) or in some other way (usually a lower-level language and
2159: as a result often called @dfn{low-level definitions}, @dfn{code
2160: definitions} or @dfn{primitives}).
2161: @item
2162: Many Forth systems are implemented mainly in Forth.
2163: @end itemize
2164:
2165:
2166: @comment ----------------------------------------------
2167: @node Where to go next,Exercises,Review - elements of a Forth system, Introduction
2168: @section Where To Go Next
2169: @cindex where to go next
2170:
2171: Amazing as it may seem, if you have read (and understood) this far, you
2172: know almost all the fundamentals about the inner workings of a Forth
2173: system. You certainly know enough to be able to read and understand the
2174: rest of this manual and the ANS Forth document, to learn more about the
2175: facilities that Forth in general and Gforth in particular provide. Even
2176: scarier, you know almost enough to implement your own Forth system.
2177: However, that's not a good idea just yet... better to try writing some
2178: programs in Gforth.
2179:
2180: Forth has such a rich vocabulary that it can be hard to know where to
2181: start in learning it. This section suggests a few sets of words that are
2182: enough to write small but useful programs. Use the word index in this
2183: document to learn more about each word, then try it out and try to write
2184: small definitions using it. Start by experimenting with these words:
2185:
2186: @itemize @bullet
2187: @item
2188: Arithmetic: @code{+ - * / /MOD */ ABS INVERT}
2189: @item
2190: Comparison: @code{MIN MAX =}
2191: @item
2192: Logic: @code{AND OR XOR NOT}
2193: @item
2194: Stack manipulation: @code{DUP DROP SWAP OVER}
2195: @item
2196: Loops and decisions: @code{IF ELSE ENDIF ?DO I LOOP}
2197: @item
2198: Input/Output: @code{. ." EMIT CR KEY}
2199: @item
2200: Defining words: @code{: ; CREATE}
2201: @item
2202: Memory allocation words: @code{ALLOT ,}
2203: @item
2204: Tools: @code{SEE WORDS .S MARKER}
2205: @end itemize
2206:
2207: When you have mastered those, go on to:
2208:
2209: @itemize @bullet
2210: @item
2211: More defining words: @code{VARIABLE CONSTANT VALUE TO CREATE DOES>}
2212: @item
2213: Memory access: @code{@@ !}
2214: @end itemize
2215:
2216: When you have mastered these, there's nothing for it but to read through
2217: the whole of this manual and find out what you've missed.
2218:
2219: @comment ----------------------------------------------
2220: @node Exercises, ,Where to go next, Introduction
2221: @section Exercises
2222: @cindex exercises
2223:
2224: TODO: provide a set of programming excercises linked into the stuff done
2225: already and into other sections of the manual. Provide solutions to all
2226: the exercises in a .fs file in the distribution.
2227:
2228: @c Get some inspiration from Starting Forth and Kelly&Spies.
2229:
2230: @c excercises:
2231: @c 1. take inches and convert to feet and inches.
2232: @c 2. take temperature and convert from fahrenheight to celcius;
2233: @c may need to care about symmetric vs floored??
2234: @c 3. take input line and do character substitution
2235: @c to encipher or decipher
2236: @c 4. as above but work on a file for in and out
2237: @c 5. take input line and convert to pig-latin
2238: @c
2239: @c thing of sets of things to exercise then come up with
2240: @c problems that need those things.
2241:
2242:
2243: @c ******************************************************************
2244: @node Words, Error messages, Introduction, Top
2245: @chapter Forth Words
2246: @cindex words
2247:
2248: @menu
2249: * Notation::
2250: * Comments::
2251: * Boolean Flags::
2252: * Arithmetic::
2253: * Stack Manipulation::
2254: * Memory::
2255: * Control Structures::
2256: * Defining Words::
2257: * The Text Interpreter::
2258: * Tokens for Words::
2259: * Word Lists::
2260: * Environmental Queries::
2261: * Files::
2262: * Blocks::
2263: * Other I/O::
2264: * Programming Tools::
2265: * Assembler and Code Words::
2266: * Threading Words::
2267: * Locals::
2268: * Structures::
2269: * Object-oriented Forth::
2270: * Passing Commands to the OS::
2271: * Miscellaneous Words::
2272: @end menu
2273:
2274: @node Notation, Comments, Words, Words
2275: @section Notation
2276: @cindex notation of glossary entries
2277: @cindex format of glossary entries
2278: @cindex glossary notation format
2279: @cindex word glossary entry format
2280:
2281: The Forth words are described in this section in the glossary notation
2282: that has become a de-facto standard for Forth texts, i.e.,
2283:
2284: @format
2285: @i{word} @i{Stack effect} @i{wordset} @i{pronunciation}
2286: @end format
2287: @i{Description}
2288:
2289: @table @var
2290: @item word
2291: The name of the word.
2292:
2293: @item Stack effect
2294: @cindex stack effect
2295: The stack effect is written in the notation @code{@i{before} --
2296: @i{after}}, where @i{before} and @i{after} describe the top of
2297: stack entries before and after the execution of the word. The rest of
2298: the stack is not touched by the word. The top of stack is rightmost,
2299: i.e., a stack sequence is written as it is typed in. Note that Gforth
2300: uses a separate floating point stack, but a unified stack
2301: notation. Also, return stack effects are not shown in @i{stack
2302: effect}, but in @i{Description}. The name of a stack item describes
2303: the type and/or the function of the item. See below for a discussion of
2304: the types.
2305:
2306: All words have two stack effects: A compile-time stack effect and a
2307: run-time stack effect. The compile-time stack-effect of most words is
2308: @i{ -- }. If the compile-time stack-effect of a word deviates from
2309: this standard behaviour, or the word does other unusual things at
2310: compile time, both stack effects are shown; otherwise only the run-time
2311: stack effect is shown.
2312:
2313: @cindex pronounciation of words
2314: @item pronunciation
2315: How the word is pronounced.
2316:
2317: @cindex wordset
2318: @item wordset
2319: The ANS Forth standard is divided into several word sets. A standard
2320: system need not support all of them. Therefore, in theory, the fewer
2321: word sets your program uses the more portable it will be. However, we
2322: suspect that most ANS Forth systems on personal machines will feature
2323: all word sets. Words that are not defined in ANS Forth have
2324: @code{gforth} or @code{gforth-internal} as word set. @code{gforth}
2325: describes words that will work in future releases of Gforth;
2326: @code{gforth-internal} words are more volatile. Environmental query
2327: strings are also displayed like words; you can recognize them by the
2328: @code{environment} in the word set field.
2329:
2330: @item Description
2331: A description of the behaviour of the word.
2332: @end table
2333:
2334: @cindex types of stack items
2335: @cindex stack item types
2336: The type of a stack item is specified by the character(s) the name
2337: starts with:
2338:
2339: @table @code
2340: @item f
2341: @cindex @code{f}, stack item type
2342: Boolean flags, i.e. @code{false} or @code{true}.
2343: @item c
2344: @cindex @code{c}, stack item type
2345: Char
2346: @item w
2347: @cindex @code{w}, stack item type
2348: Cell, can contain an integer or an address
2349: @item n
2350: @cindex @code{n}, stack item type
2351: signed integer
2352: @item u
2353: @cindex @code{u}, stack item type
2354: unsigned integer
2355: @item d
2356: @cindex @code{d}, stack item type
2357: double sized signed integer
2358: @item ud
2359: @cindex @code{ud}, stack item type
2360: double sized unsigned integer
2361: @item r
2362: @cindex @code{r}, stack item type
2363: Float (on the FP stack)
2364: @item a-
2365: @cindex @code{a_}, stack item type
2366: Cell-aligned address
2367: @item c-
2368: @cindex @code{c_}, stack item type
2369: Char-aligned address (note that a Char may have two bytes in Windows NT)
2370: @item f-
2371: @cindex @code{f_}, stack item type
2372: Float-aligned address
2373: @item df-
2374: @cindex @code{df_}, stack item type
2375: Address aligned for IEEE double precision float
2376: @item sf-
2377: @cindex @code{sf_}, stack item type
2378: Address aligned for IEEE single precision float
2379: @item xt
2380: @cindex @code{xt}, stack item type
2381: Execution token, same size as Cell
2382: @item wid
2383: @cindex @code{wid}, stack item type
2384: Word list ID, same size as Cell
2385: @item f83name
2386: @cindex @code{f83name}, stack item type
2387: Pointer to a name structure
2388: @item "
2389: @cindex @code{"}, stack item type
2390: string in the input stream (not on the stack). The terminating character
2391: is a blank by default. If it is not a blank, it is shown in @code{<>}
2392: quotes.
2393: @end table
2394:
2395: @node Comments, Boolean Flags, Notation, Words
2396: @section Comments
2397: @cindex comments
2398:
2399: Forth supports two styles of comment; the traditional @i{in-line} comment,
2400: @code{(} and its modern cousin, the @i{comment to end of line}; @code{\}.
2401:
2402:
2403: doc-(
2404: doc-\
2405: doc-\G
2406:
2407:
2408: @node Boolean Flags, Arithmetic, Comments, Words
2409: @section Boolean Flags
2410: @cindex Boolean flags
2411:
2412: A Boolean flag is cell-sized. A cell with all bits clear represents the
2413: flag @code{false} and a flag with all bits set represents the flag
2414: @code{true}. Words that check a flag (for example, @code{IF}) will treat
2415: a cell that has @i{any} bit set as @code{true}.
2416:
2417:
2418: doc-true
2419: doc-false
2420: doc-on
2421: doc-off
2422:
2423:
2424: @node Arithmetic, Stack Manipulation, Boolean Flags, Words
2425: @section Arithmetic
2426: @cindex arithmetic words
2427:
2428: @cindex division with potentially negative operands
2429: Forth arithmetic is not checked, i.e., you will not hear about integer
2430: overflow on addition or multiplication, you may hear about division by
2431: zero if you are lucky. The operator is written after the operands, but
2432: the operands are still in the original order. I.e., the infix @code{2-1}
2433: corresponds to @code{2 1 -}. Forth offers a variety of division
2434: operators. If you perform division with potentially negative operands,
2435: you do not want to use @code{/} or @code{/mod} with its undefined
2436: behaviour, but rather @code{fm/mod} or @code{sm/mod} (probably the
2437: former, @pxref{Mixed precision}).
2438: @comment TODO discuss the different division forms and the std approach
2439:
2440: @menu
2441: * Single precision::
2442: * Bitwise operations::
2443: * Double precision:: Double-cell integer arithmetic
2444: * Numeric comparison::
2445: * Mixed precision:: Operations with single and double-cell integers
2446: * Floating Point::
2447: @end menu
2448:
2449: @node Single precision, Bitwise operations, Arithmetic, Arithmetic
2450: @subsection Single precision
2451: @cindex single precision arithmetic words
2452:
2453: By default, numbers in Forth are single-precision integers that are 1
2454: cell in size. They can be signed or unsigned, depending upon how you
2455: treat them. @xref{Number Conversion} for the rules used by the text
2456: interpreter for recognising single-precision integers.
2457:
2458:
2459: doc-+
2460: doc-1+
2461: doc--
2462: doc-1-
2463: doc-*
2464: doc-/
2465: doc-mod
2466: doc-/mod
2467: doc-negate
2468: doc-abs
2469: doc-min
2470: doc-max
2471: doc-d>s
2472: doc-floored
2473:
2474:
2475: @node Bitwise operations, Double precision, Single precision, Arithmetic
2476: @subsection Bitwise operations
2477: @cindex bitwise operation words
2478:
2479:
2480: doc-and
2481: doc-or
2482: doc-xor
2483: doc-invert
2484: doc-lshift
2485: doc-rshift
2486: doc-2*
2487: doc-d2*
2488: doc-2/
2489: doc-d2/
2490:
2491:
2492: @node Double precision, Numeric comparison, Bitwise operations, Arithmetic
2493: @subsection Double precision
2494: @cindex double precision arithmetic words
2495:
2496: @xref{Number Conversion} for the rules used by the text interpreter for
2497: recognising double-precision integers.
2498:
2499: A double precision number is represented by a cell pair, with the most
2500: significant cell at the TOS. It is trivial to convert an unsigned
2501: single to an (unsigned) double; simply push a @code{0} onto the
2502: TOS. Since numbers are represented by Gforth using 2's complement
2503: arithmetic, converting a signed single to a (signed) double requires
2504: sign-extension across the most significant cell. This can be achieved
2505: using @code{s>d}. The moral of the story is that you cannot convert a
2506: number without knowing whether it represents an unsigned or a
2507: signed number.
2508:
2509:
2510: doc-s>d
2511: doc-d+
2512: doc-d-
2513: doc-dnegate
2514: doc-dabs
2515: doc-dmin
2516: doc-dmax
2517:
2518:
2519: @node Numeric comparison, Mixed precision, Double precision, Arithmetic
2520: @subsection Numeric comparison
2521: @cindex numeric comparison words
2522:
2523:
2524: doc-<
2525: doc-<=
2526: doc-<>
2527: doc-=
2528: doc->
2529: doc->=
2530:
2531: doc-0<
2532: doc-0<=
2533: doc-0<>
2534: doc-0=
2535: doc-0>
2536: doc-0>=
2537:
2538: doc-u<
2539: doc-u<=
2540: @c u<> and u= exist but are the same as <> and =
2541: @c doc-u<>
2542: @c doc-u=
2543: doc-u>
2544: doc-u>=
2545:
2546: doc-within
2547:
2548: doc-d<
2549: doc-d<=
2550: doc-d<>
2551: doc-d=
2552: doc-d>
2553: doc-d>=
2554:
2555: doc-d0<
2556: doc-d0<=
2557: doc-d0<>
2558: doc-d0=
2559: doc-d0>
2560: doc-d0>=
2561:
2562: doc-du<
2563: doc-du<=
2564: @c du<> and du= exist but are the same as d<> and d=
2565: @c doc-du<>
2566: @c doc-du=
2567: doc-du>
2568: doc-du>=
2569:
2570:
2571: @node Mixed precision, Floating Point, Numeric comparison, Arithmetic
2572: @subsection Mixed precision
2573: @cindex mixed precision arithmetic words
2574:
2575:
2576: doc-m+
2577: doc-*/
2578: doc-*/mod
2579: doc-m*
2580: doc-um*
2581: doc-m*/
2582: doc-um/mod
2583: doc-fm/mod
2584: doc-sm/rem
2585:
2586:
2587: @node Floating Point, , Mixed precision, Arithmetic
2588: @subsection Floating Point
2589: @cindex floating point arithmetic words
2590:
2591: @xref{Number Conversion} for the rules used by the text interpreter for
2592: recognising floating-point numbers.
2593:
2594: Gforth has a separate floating point
2595: stack, but the documentation uses the unified notation.
2596:
2597: @cindex floating-point arithmetic, pitfalls
2598: Floating point numbers have a number of unpleasant surprises for the
2599: unwary (e.g., floating point addition is not associative) and even a few
2600: for the wary. You should not use them unless you know what you are doing
2601: or you don't care that the results you get are totally bogus. If you
2602: want to learn about the problems of floating point numbers (and how to
2603: avoid them), you might start with @cite{David Goldberg, What Every
2604: Computer Scientist Should Know About Floating-Point Arithmetic, ACM
2605: Computing Surveys 23(1):5@minus{}48, March 1991}
2606: (@url{http://www.validgh.com/goldberg/paper.ps}).
2607:
2608:
2609: doc-d>f
2610: doc-f>d
2611: doc-f+
2612: doc-f-
2613: doc-f*
2614: doc-f/
2615: doc-fnegate
2616: doc-fabs
2617: doc-fmax
2618: doc-fmin
2619: doc-floor
2620: doc-fround
2621: doc-f**
2622: doc-fsqrt
2623: doc-fexp
2624: doc-fexpm1
2625: doc-fln
2626: doc-flnp1
2627: doc-flog
2628: doc-falog
2629: doc-f2*
2630: doc-f2/
2631: doc-1/f
2632: doc-precision
2633: doc-set-precision
2634:
2635: @cindex angles in trigonometric operations
2636: @cindex trigonometric operations
2637: Angles in floating point operations are given in radians (a full circle
2638: has 2 pi radians).
2639:
2640: doc-fsin
2641: doc-fcos
2642: doc-fsincos
2643: doc-ftan
2644: doc-fasin
2645: doc-facos
2646: doc-fatan
2647: doc-fatan2
2648: doc-fsinh
2649: doc-fcosh
2650: doc-ftanh
2651: doc-fasinh
2652: doc-facosh
2653: doc-fatanh
2654: doc-pi
2655:
2656: @cindex equality of floats
2657: @cindex floating-point comparisons
2658: One particular problem with floating-point arithmetic is that comparison
2659: for equality often fails when you would expect it to succeed. For this
2660: reason approximate equality is often preferred (but you still have to
2661: know what you are doing). The comparison words are:
2662:
2663: doc-f~rel
2664: doc-f~abs
2665: doc-f=
2666: doc-f~
2667: doc-f<>
2668:
2669: doc-f<
2670: doc-f<=
2671: doc-f>
2672: doc-f>=
2673:
2674: doc-f0<
2675: doc-f0<=
2676: doc-f0<>
2677: doc-f0=
2678: doc-f0>
2679: doc-f0>=
2680:
2681:
2682: @node Stack Manipulation, Memory, Arithmetic, Words
2683: @section Stack Manipulation
2684: @cindex stack manipulation words
2685:
2686: @cindex floating-point stack in the standard
2687: Gforth maintains a number of separate stacks:
2688:
2689: @cindex data stack
2690: @cindex parameter stack
2691: @itemize @bullet
2692: @item
2693: A data stack (also known as the @dfn{parameter stack}) -- for
2694: characters, cells, addresses, and double cells.
2695:
2696: @cindex floating-point stack
2697: @item
2698: A floating point stack -- for holding floating point (FP) numbers.
2699:
2700: @cindex return stack
2701: @item
2702: A return stack -- for holding the return addresses of colon
2703: definitions and other (non-FP) data.
2704:
2705: @cindex locals stack
2706: @item
2707: A locals stack -- for holding local variables.
2708: @end itemize
2709:
2710: @menu
2711: * Data stack::
2712: * Floating point stack::
2713: * Return stack::
2714: * Locals stack::
2715: * Stack pointer manipulation::
2716: @end menu
2717:
2718: @node Data stack, Floating point stack, Stack Manipulation, Stack Manipulation
2719: @subsection Data stack
2720: @cindex data stack manipulation words
2721: @cindex stack manipulations words, data stack
2722:
2723:
2724: doc-drop
2725: doc-nip
2726: doc-dup
2727: doc-over
2728: doc-tuck
2729: doc-swap
2730: doc-pick
2731: doc-rot
2732: doc--rot
2733: doc-?dup
2734: doc-roll
2735: doc-2drop
2736: doc-2nip
2737: doc-2dup
2738: doc-2over
2739: doc-2tuck
2740: doc-2swap
2741: doc-2rot
2742:
2743:
2744: @node Floating point stack, Return stack, Data stack, Stack Manipulation
2745: @subsection Floating point stack
2746: @cindex floating-point stack manipulation words
2747: @cindex stack manipulation words, floating-point stack
2748:
2749: Whilst every sane Forth has a separate floating-point stack, it is not
2750: strictly required; an ANS Forth system could theoretically keep
2751: floating-point numbers on the data stack. As an additional difficulty,
2752: you don't know how many cells a floating-point number takes. It is
2753: reportedly possible to write words in a way that they work also for a
2754: unified stack model, but we do not recommend trying it. Instead, just
2755: say that your program has an environmental dependency on a separate
2756: floating-point stack.
2757:
2758: doc-floating-stack
2759:
2760: doc-fdrop
2761: doc-fnip
2762: doc-fdup
2763: doc-fover
2764: doc-ftuck
2765: doc-fswap
2766: doc-fpick
2767: doc-frot
2768:
2769:
2770: @node Return stack, Locals stack, Floating point stack, Stack Manipulation
2771: @subsection Return stack
2772: @cindex return stack manipulation words
2773: @cindex stack manipulation words, return stack
2774:
2775: @cindex return stack and locals
2776: @cindex locals and return stack
2777: A Forth system is allowed to keep local variables on the
2778: return stack. This is reasonable, as local variables usually eliminate
2779: the need to use the return stack explicitly. So, if you want to produce
2780: a standard compliant program and you are using local variables in a
2781: word, forget about return stack manipulations in that word (refer to the
2782: standard document for the exact rules).
2783:
2784: doc->r
2785: doc-r>
2786: doc-r@
2787: doc-rdrop
2788: doc-2>r
2789: doc-2r>
2790: doc-2r@
2791: doc-2rdrop
2792:
2793:
2794: @node Locals stack, Stack pointer manipulation, Return stack, Stack Manipulation
2795: @subsection Locals stack
2796:
2797: @comment TODO
2798:
2799: @node Stack pointer manipulation, , Locals stack, Stack Manipulation
2800: @subsection Stack pointer manipulation
2801: @cindex stack pointer manipulation words
2802:
2803: @c removed s0 r0 l0 -- they are obsolete aliases for sp0 rp0 lp0
2804: doc-sp0
2805: doc-sp@
2806: doc-sp!
2807: doc-fp0
2808: doc-fp@
2809: doc-fp!
2810: doc-rp0
2811: doc-rp@
2812: doc-rp!
2813: doc-lp0
2814: doc-lp@
2815: doc-lp!
2816:
2817:
2818: @node Memory, Control Structures, Stack Manipulation, Words
2819: @section Memory
2820: @cindex memory words
2821:
2822: @menu
2823: * Memory model::
2824: * Dictionary allocation::
2825: * Heap Allocation::
2826: * Memory Access::
2827: * Address arithmetic::
2828: * Memory Blocks::
2829: @end menu
2830:
2831: @node Memory model, Dictionary allocation, Memory, Memory
2832: @subsection ANS Forth and Gforth memory models
2833:
2834: @c The ANS Forth description is a mess (e.g., is the heap part of
2835: @c the dictionary?), so let's not stick to closely with it.
2836:
2837: ANS Forth considers a Forth system as consisting of several memories, of
2838: which only @dfn{data space} is managed and accessible with the memory
2839: words. Memory not necessarily in data space includes the stacks, the
2840: code (called code space) and the headers (called name space). In Gforth
2841: everything is in data space, but the code for the primitives is usually
2842: read-only.
2843:
2844: Data space is divided into a number of areas: The (data space portion of
2845: the) dictionary@footnote{Sometimes, the term @dfn{dictionary} is used to
2846: refer to the search data structure embodied in word lists and headers,
2847: because it is used for looking up names, just as you would in a
2848: conventional dictionary.}, the heap, and a number of system-allocated
2849: buffers.
2850:
2851: In ANS Forth data space is also divided into contiguous regions. You
2852: can only use address arithmetic within a contiguous region, not between
2853: them. Usually each allocation gives you one contiguous region, but the
2854: dictionary allocation words have additional rules (@pxref{Dictionary
2855: allocation}).
2856:
2857: Gforth provides one big address space, and address arithmetic can be
2858: performed between any addresses. However, in the dictionary headers or
2859: code are interleaved with data, so almost the only contiguous data space
2860: regions there are those described by ANS Forth as contiguous; but you
2861: can be sure that the dictionary is allocated towards increasing
2862: addresses even between contiguous regions. The memory order of
2863: allocations in the heap is platform-dependent (and possibly different
2864: from one run to the next).
2865:
2866: @subsubsection ANS Forth dictionary details
2867:
2868: This section is just informative, you can skip it if you are in a hurry.
2869:
2870: When you create a colon definition, the text interpreter compiles the
2871: code for the definition into the code space and compiles the name
2872: of the definition into the header space, together with other
2873: information about the definition (such as its execution token).
2874:
2875: When you create a variable, the execution of @code{Variable} will
2876: compile some code, assign one cell in data space, and compile the name
2877: of the variable into the header space.
2878:
2879: @cindex memory regions - relationship between them
2880: ANS Forth does not specify the relationship between the three memory
2881: regions, and specifies that a Standard program must not access code or
2882: data space directly -- it may only access data space directly. In
2883: addition, the Standard defines what relationships you may and may not
2884: rely on when allocating regions in data space. These constraints are
2885: simply a reflection of the many diverse techniques that are used to
2886: implement Forth systems; understanding and following the requirements of
2887: the Standard allows you to write portable programs -- programs that run
2888: in the same way on any of these diverse systems. Another way of looking
2889: at this is to say that ANS Forth was designed to permit compliant Forth
2890: systems to be implemented in many diverse ways.
2891:
2892: @cindex memory regions - how they are assigned
2893: Here are some examples of ways in which name, code and data spaces
2894: might be assigned in different Forth implementations:
2895:
2896: @itemize @bullet
2897: @item
2898: For a Forth system that runs from RAM under a general-purpose operating
2899: system, it can be convenient to interleave name, code and data spaces in
2900: a single contiguous memory region. This organisation can be
2901: memory-efficient (for example, because the relationship between the name
2902: dictionary entry and the associated code space entry can be
2903: implicit, rather than requiring an explicit memory pointer to reference
2904: from the header space and the code space). This is the
2905: organisation used by Gforth, as this example@footnote{The addresses
2906: in the example have been truncated to fit it onto the page, and the
2907: addresses and data shown will not match the output from your system} shows:
2908: @example
2909: hex
2910: variable fred 123456 fred !
2911: variable jim abcd jim !
2912: : foo + / - ;
2913: ' fred 10 - 50 dump
2914: ..80: 5C 46 0E 40 84 66 72 65 - 64 20 20 20 20 20 20 20 \F.@.fred
2915: ..90: D0 9B 04 08 00 00 00 00 - 56 34 12 00 80 46 0E 40 ........V4...F.@
2916: ..A0: 83 6A 69 6D 20 20 20 20 - D0 9B 04 08 00 00 00 00 .jim ........
2917: ..B0: CD AB 00 00 9C 46 0E 40 - 83 66 6F 6F 20 20 20 20 .....F.@.foo
2918: ..C0: 80 9B 04 08 00 00 00 00 - E4 2E 05 08 0C 2F 05 08 ............./..
2919: @end example
2920:
2921: @item
2922: For a high-performance system running on a modern RISC processor with a
2923: modified Harvard architecture (one that has a unified main memory but
2924: separate instruction and data caches), it is desirable to separate
2925: processor instructions from processor data. This encourages a high cache
2926: density and therefore a high cache hit rate. The Forth code space
2927: is not necessarily made up entirely of processor instructions; its
2928: nature is dependent upon the Forth implementation.
2929:
2930: @item
2931: A Forth compiler that runs on a segmented 8086 processor could be
2932: designed to interleave the name, code and data spaces within a single
2933: 64Kbyte segment. A more common implementation choice is to use a
2934: separate 64Kbyte segment for each region, which provides more memory
2935: overall but provides an address map in which only the data space is
2936: accessible.
2937:
2938: @item
2939: Microprocessors exist that run Forth (or many of the primitives required
2940: to implement the Forth virtual machine efficiently) directly. On these
2941: processors, the relationship between name, code and data spaces may be
2942: imposed as a side-effect of the architecture of the processor.
2943:
2944: @item
2945: A Forth compiler that executes from ROM on an embedded system needs its
2946: data space separated from the name and code spaces so that the data
2947: space can be mapped to a RAM area.
2948:
2949: @item
2950: A Forth compiler that runs on an embedded system may have a requirement
2951: for a small memory footprint. On such a system it can be useful to
2952: separate the header space from the data and code spaces; once the
2953: application has been compiled, the header space is no longer
2954: required@footnote{more strictly speaking, most applications can be
2955: designed so that this is the case}. The header space can be deleted
2956: entirely, or could be stored in memory on a remote @i{host} system for
2957: debug and development purposes. In the latter case, the compiler running
2958: on the @i{target} system could implement a protocol across a
2959: communication link that would allow it to interrogate the header space.
2960: @end itemize
2961:
2962: @node Dictionary allocation, Heap Allocation, Memory model, Memory
2963: @subsection Dictionary allocation
2964: @cindex reserving data space
2965: @cindex data space - reserving some
2966:
2967: Dictionary allocation is a stack-oriented allocation scheme, i.e., if
2968: you want to deallocate X, you also deallocate everything
2969: allocated after X.
2970:
2971: The allocations using the words below are contiguous and grow the region
2972: towards increasing addresses. Other words that allocate dictionary
2973: memory of any kind (i.e., defining words including @code{:noname}) end
2974: the contiguous region and start a new one.
2975:
2976: In ANS Forth only @code{create}d words are guaranteed to produce an
2977: address that is the start of the following contiguous region. In
2978: particular, the cell allocated by @code{variable} is not guaranteed to
2979: be contiguous with following @code{allot}ed memory.
2980:
2981: You can deallocate memory by using @code{allot} with a negative argument
2982: (with some restrictions, see @code{allot}). For larger deallocations use
2983: @code{marker}.
2984:
2985:
2986: doc-here
2987: doc-unused
2988: doc-allot
2989: doc-c,
2990: doc-f,
2991: doc-,
2992: doc-2,
2993: @cindex user space
2994: doc-udp
2995: doc-uallot
2996:
2997: Memory accesses have to be aligned (@pxref{Address arithmetic}). So of
2998: course you should allocate memory in an aligned way, too. I.e., before
2999: allocating allocating a cell, @code{here} must be cell-aligned, etc.
3000: The words below align @code{here} if it is not already. Basically it is
3001: only already aligned for a type, if the last allocation was a multiple
3002: of the size of this type and if @code{here} was aligned for this type
3003: before.
3004:
3005: After freshly @code{create}ing a word, @code{here} is @code{align}ed in
3006: ANS Forth (@code{maxalign}ed in Gforth).
3007:
3008: doc-align
3009: doc-falign
3010: doc-sfalign
3011: doc-dfalign
3012: doc-maxalign
3013: doc-cfalign
3014:
3015:
3016: @node Heap Allocation, Memory Access, Dictionary allocation, Memory
3017: @subsection Heap allocation
3018: @cindex heap allocation
3019: @cindex dynamic allocation of memory
3020: @cindex memory-allocation word set
3021:
3022: Heap allocation supports deallocation of allocated memory in any
3023: order. Dictionary allocation is not affected by it (i.e., it does not
3024: end a contiguous region). In Gforth, these words are implemented using
3025: the standard C library calls malloc(), free() and resize().
3026:
3027: doc-allocate
3028: doc-free
3029: doc-resize
3030:
3031:
3032: @node Memory Access, Address arithmetic, Heap Allocation, Memory
3033: @subsection Memory Access
3034: @cindex memory access words
3035:
3036:
3037: doc-@
3038: doc-!
3039: doc-+!
3040: doc-c@
3041: doc-c!
3042: doc-2@
3043: doc-2!
3044: doc-f@
3045: doc-f!
3046: doc-sf@
3047: doc-sf!
3048: doc-df@
3049: doc-df!
3050:
3051: @node Address arithmetic, Memory Blocks, Memory Access, Memory
3052: @subsection Address arithmetic
3053: @cindex address arithmetic words
3054:
3055: Address arithmetic is the foundation on which data structures like
3056: arrays, records (@pxref{Structures}) and objects (@pxref{Object-oriented
3057: Forth}) are built.
3058:
3059: ANS Forth does not specify the sizes of the data types. Instead, it
3060: offers a number of words for computing sizes and doing address
3061: arithmetic. Address arithmetic is performed in terms of address units
3062: (aus); on most systems the address unit is one byte. Note that a
3063: character may have more than one au, so @code{chars} is no noop (on
3064: systems where it is a noop, it compiles to nothing).
3065:
3066: @cindex alignment of addresses for types
3067: ANS Forth also defines words for aligning addresses for specific
3068: types. Many computers require that accesses to specific data types
3069: must only occur at specific addresses; e.g., that cells may only be
3070: accessed at addresses divisible by 4. Even if a machine allows unaligned
3071: accesses, it can usually perform aligned accesses faster.
3072:
3073: For the performance-conscious: alignment operations are usually only
3074: necessary during the definition of a data structure, not during the
3075: (more frequent) accesses to it.
3076:
3077: ANS Forth defines no words for character-aligning addresses. This is not
3078: an oversight, but reflects the fact that addresses that are not
3079: char-aligned have no use in the standard and therefore will not be
3080: created.
3081:
3082: @cindex @code{CREATE} and alignment
3083: ANS Forth guarantees that addresses returned by @code{CREATE}d words
3084: are cell-aligned; in addition, Gforth guarantees that these addresses
3085: are aligned for all purposes.
3086:
3087: Note that the ANS Forth word @code{char} has nothing to do with address
3088: arithmetic.
3089:
3090:
3091: doc-chars
3092: doc-char+
3093: doc-cells
3094: doc-cell+
3095: doc-cell
3096: doc-aligned
3097: doc-floats
3098: doc-float+
3099: doc-float
3100: doc-faligned
3101: doc-sfloats
3102: doc-sfloat+
3103: doc-sfaligned
3104: doc-dfloats
3105: doc-dfloat+
3106: doc-dfaligned
3107: doc-maxaligned
3108: doc-cfaligned
3109: doc-address-unit-bits
3110:
3111:
3112: @node Memory Blocks, , Address arithmetic, Memory
3113: @subsection Memory Blocks
3114: @cindex memory block words
3115: @cindex character strings - moving and copying
3116:
3117: Memory blocks often represent character strings; @xref{String Formats}
3118: for ways of storing character strings in memory. @xref{Displaying
3119: characters and strings} for other string-processing words.
3120:
3121: Some of these words work on address units. Others work on character
3122: units (increments of @code{CHAR}), and expect a @code{CHAR}-aligned
3123: address. Choose the correct operation depending upon your data type.
3124:
3125: When copying characters between overlapping memory regions, choose
3126: carefully between @code{cmove} and @code{cmove>}.
3127:
3128: You can only use any of these words @i{portably} to access data space.
3129:
3130: @comment TODO - think the naming of the arguments is wrong for move
3131: @comment well, really it seems to be the Standard that's wrong; it
3132: @comment describes MOVE as a word that requires a CELL-aligned source
3133: @comment and destination address but a xtranfer count that need not
3134: @comment be a multiple of CELL.
3135:
3136: doc-move
3137: doc-erase
3138: doc-cmove
3139: doc-cmove>
3140: doc-fill
3141: doc-blank
3142: doc-compare
3143: doc-search
3144: doc--trailing
3145: doc-/string
3146:
3147:
3148: @comment TODO examples
3149:
3150:
3151: @node Control Structures, Defining Words, Memory, Words
3152: @section Control Structures
3153: @cindex control structures
3154:
3155: Control structures in Forth cannot be used interpretively, only in a
3156: colon definition@footnote{To be precise, they have no interpretation
3157: semantics (@pxref{Interpretation and Compilation Semantics}).}. We do
3158: not like this limitation, but have not seen a satisfying way around it
3159: yet, although many schemes have been proposed.
3160:
3161: @menu
3162: * Selection:: IF ... ELSE ... ENDIF
3163: * Simple Loops:: BEGIN ...
3164: * Counted Loops:: DO
3165: * Arbitrary control structures::
3166: * Calls and returns::
3167: * Exception Handling::
3168: @end menu
3169:
3170: @node Selection, Simple Loops, Control Structures, Control Structures
3171: @subsection Selection
3172: @cindex selection control structures
3173: @cindex control structures for selection
3174:
3175: @c what's the purpose of all these @i? Maybe we should define a macro
3176: @c so we can produce logical markup. - anton
3177:
3178: @c nac-> When I started working on the manual, a mixture of @i and @var
3179: @c were used inconsistently in code examples and \Glossary entries. These
3180: @c two behave differently in info format so I decided to standardize on @i.
3181: @c Logical markup would be better but texi isn't really upto it, and
3182: @c texi2html just ignores macros.
3183:
3184: @cindex @code{IF} control structure
3185: @example
3186: @i{flag}
3187: IF
3188: @i{code}
3189: ENDIF
3190: @end example
3191: @noindent
3192:
3193: If @i{flag} is non-zero (as far as @code{IF} etc. are concerned, a cell
3194: with any bit set represents truth) @i{code} is executed.
3195:
3196: @example
3197: @i{flag}
3198: IF
3199: @i{code1}
3200: ELSE
3201: @i{code2}
3202: ENDIF
3203: @end example
3204:
3205: If @var{flag} is true, @i{code1} is executed, otherwise @i{code2} is
3206: executed.
3207:
3208: You can use @code{THEN} instead of @code{ENDIF}. Indeed, @code{THEN} is
3209: standard, and @code{ENDIF} is not, although it is quite popular. We
3210: recommend using @code{ENDIF}, because it is less confusing for people
3211: who also know other languages (and is not prone to reinforcing negative
3212: prejudices against Forth in these people). Adding @code{ENDIF} to a
3213: system that only supplies @code{THEN} is simple:
3214: @example
3215: : ENDIF POSTPONE THEN ; immediate
3216: @end example
3217:
3218: [According to @cite{Webster's New Encyclopedic Dictionary}, @dfn{then
3219: (adv.)} has the following meanings:
3220: @quotation
3221: ... 2b: following next after in order ... 3d: as a necessary consequence
3222: (if you were there, then you saw them).
3223: @end quotation
3224: Forth's @code{THEN} has the meaning 2b, whereas @code{THEN} in Pascal
3225: and many other programming languages has the meaning 3d.]
3226:
3227: Gforth also provides the words @code{?DUP-IF} and @code{?DUP-0=-IF}, so
3228: you can avoid using @code{?dup}. Using these alternatives is also more
3229: efficient than using @code{?dup}. Definitions in ANS Forth
3230: for @code{ENDIF}, @code{?DUP-IF} and @code{?DUP-0=-IF} are provided in
3231: @file{compat/control.fs}.
3232:
3233: @cindex @code{CASE} control structure
3234: @example
3235: @i{n}
3236: CASE
3237: @i{n1} OF @i{code1} ENDOF
3238: @i{n2} OF @i{code2} ENDOF
3239: @dots{}
3240: ENDCASE
3241: @end example
3242:
3243: Executes the first @i{codei}, where the @i{ni} is equal to
3244: @i{n}. A default case can be added by simply writing the code after
3245: the last @code{ENDOF}. It may use @i{n}, which is on top of the stack,
3246: but must not consume it.
3247:
3248: @node Simple Loops, Counted Loops, Selection, Control Structures
3249: @subsection Simple Loops
3250: @cindex simple loops
3251: @cindex loops without count
3252:
3253: @cindex @code{WHILE} loop
3254: @example
3255: BEGIN
3256: @i{code1}
3257: @i{flag}
3258: WHILE
3259: @i{code2}
3260: REPEAT
3261: @end example
3262:
3263: @i{code1} is executed and @i{flag} is computed. If it is true,
3264: @i{code2} is executed and the loop is restarted; If @i{flag} is
3265: false, execution continues after the @code{REPEAT}.
3266:
3267: @cindex @code{UNTIL} loop
3268: @example
3269: BEGIN
3270: @i{code}
3271: @i{flag}
3272: UNTIL
3273: @end example
3274:
3275: @i{code} is executed. The loop is restarted if @code{flag} is false.
3276:
3277: @cindex endless loop
3278: @cindex loops, endless
3279: @example
3280: BEGIN
3281: @i{code}
3282: AGAIN
3283: @end example
3284:
3285: This is an endless loop.
3286:
3287: @node Counted Loops, Arbitrary control structures, Simple Loops, Control Structures
3288: @subsection Counted Loops
3289: @cindex counted loops
3290: @cindex loops, counted
3291: @cindex @code{DO} loops
3292:
3293: The basic counted loop is:
3294: @example
3295: @i{limit} @i{start}
3296: ?DO
3297: @i{body}
3298: LOOP
3299: @end example
3300:
3301: This performs one iteration for every integer, starting from @i{start}
3302: and up to, but excluding @i{limit}. The counter, or @i{index}, can be
3303: accessed with @code{i}. For example, the loop:
3304: @example
3305: 10 0 ?DO
3306: i .
3307: LOOP
3308: @end example
3309: @noindent
3310: prints @code{0 1 2 3 4 5 6 7 8 9}
3311:
3312: The index of the innermost loop can be accessed with @code{i}, the index
3313: of the next loop with @code{j}, and the index of the third loop with
3314: @code{k}.
3315:
3316:
3317: doc-i
3318: doc-j
3319: doc-k
3320:
3321:
3322: The loop control data are kept on the return stack, so there are some
3323: restrictions on mixing return stack accesses and counted loop words. In
3324: particuler, if you put values on the return stack outside the loop, you
3325: cannot read them inside the loop@footnote{well, not in a way that is
3326: portable.}. If you put values on the return stack within a loop, you
3327: have to remove them before the end of the loop and before accessing the
3328: index of the loop.
3329:
3330: There are several variations on the counted loop:
3331:
3332: @itemize @bullet
3333: @item
3334: @code{LEAVE} leaves the innermost counted loop immediately; execution
3335: continues after the associated @code{LOOP} or @code{NEXT}. For example:
3336:
3337: @example
3338: 10 0 ?DO i DUP . 3 = IF LEAVE THEN LOOP
3339: @end example
3340: prints @code{0 1 2 3}
3341:
3342:
3343: @item
3344: @code{UNLOOP} prepares for an abnormal loop exit, e.g., via
3345: @code{EXIT}. @code{UNLOOP} removes the loop control parameters from the
3346: return stack so @code{EXIT} can get to its return address. For example:
3347:
3348: @example
3349: : demo 10 0 ?DO i DUP . 3 = IF UNLOOP EXIT THEN LOOP ." Done" ;
3350: @end example
3351: prints @code{0 1 2 3}
3352:
3353:
3354: @item
3355: If @i{start} is greater than @i{limit}, a @code{?DO} loop is entered
3356: (and @code{LOOP} iterates until they become equal by wrap-around
3357: arithmetic). This behaviour is usually not what you want. Therefore,
3358: Gforth offers @code{+DO} and @code{U+DO} (as replacements for
3359: @code{?DO}), which do not enter the loop if @i{start} is greater than
3360: @i{limit}; @code{+DO} is for signed loop parameters, @code{U+DO} for
3361: unsigned loop parameters.
3362:
3363: @item
3364: @code{?DO} can be replaced by @code{DO}. @code{DO} always enters
3365: the loop, independent of the loop parameters. Do not use @code{DO}, even
3366: if you know that the loop is entered in any case. Such knowledge tends
3367: to become invalid during maintenance of a program, and then the
3368: @code{DO} will make trouble.
3369:
3370: @item
3371: @code{LOOP} can be replaced with @code{@i{n} +LOOP}; this updates the
3372: index by @i{n} instead of by 1. The loop is terminated when the border
3373: between @i{limit-1} and @i{limit} is crossed. E.g.:
3374:
3375: @example
3376: 4 0 +DO i . 2 +LOOP
3377: @end example
3378: @noindent
3379: prints @code{0 2}
3380:
3381: @example
3382: 4 1 +DO i . 2 +LOOP
3383: @end example
3384: @noindent
3385: prints @code{1 3}
3386:
3387:
3388: @cindex negative increment for counted loops
3389: @cindex counted loops with negative increment
3390: The behaviour of @code{@i{n} +LOOP} is peculiar when @i{n} is negative:
3391:
3392: @example
3393: -1 0 ?DO i . -1 +LOOP
3394: @end example
3395: @noindent
3396: prints @code{0 -1}
3397:
3398: @example
3399: 0 0 ?DO i . -1 +LOOP
3400: @end example
3401: prints nothing.
3402:
3403: Therefore we recommend avoiding @code{@i{n} +LOOP} with negative
3404: @i{n}. One alternative is @code{@i{u} -LOOP}, which reduces the
3405: index by @i{u} each iteration. The loop is terminated when the border
3406: between @i{limit+1} and @i{limit} is crossed. Gforth also provides
3407: @code{-DO} and @code{U-DO} for down-counting loops. E.g.:
3408:
3409: @example
3410: -2 0 -DO i . 1 -LOOP
3411: @end example
3412: @noindent
3413: prints @code{0 -1}
3414:
3415: @example
3416: -1 0 -DO i . 1 -LOOP
3417: @end example
3418: @noindent
3419: prints @code{0}
3420:
3421: @example
3422: 0 0 -DO i . 1 -LOOP
3423: @end example
3424: @noindent
3425: prints nothing.
3426:
3427: @end itemize
3428:
3429: Unfortunately, @code{+DO}, @code{U+DO}, @code{-DO}, @code{U-DO} and
3430: @code{-LOOP} are not defined in ANS Forth. However, an implementation
3431: for these words that uses only standard words is provided in
3432: @file{compat/loops.fs}.
3433:
3434:
3435: @cindex @code{FOR} loops
3436: Another counted loop is:
3437: @example
3438: @i{n}
3439: FOR
3440: @i{body}
3441: NEXT
3442: @end example
3443: This is the preferred loop of native code compiler writers who are too
3444: lazy to optimize @code{?DO} loops properly. This loop structure is not
3445: defined in ANS Forth. In Gforth, this loop iterates @i{n+1} times;
3446: @code{i} produces values starting with @i{n} and ending with 0. Other
3447: Forth systems may behave differently, even if they support @code{FOR}
3448: loops. To avoid problems, don't use @code{FOR} loops.
3449:
3450: @node Arbitrary control structures, Calls and returns, Counted Loops, Control Structures
3451: @subsection Arbitrary control structures
3452: @cindex control structures, user-defined
3453:
3454: @cindex control-flow stack
3455: ANS Forth permits and supports using control structures in a non-nested
3456: way. Information about incomplete control structures is stored on the
3457: control-flow stack. This stack may be implemented on the Forth data
3458: stack, and this is what we have done in Gforth.
3459:
3460: @cindex @code{orig}, control-flow stack item
3461: @cindex @code{dest}, control-flow stack item
3462: An @i{orig} entry represents an unresolved forward branch, a @i{dest}
3463: entry represents a backward branch target. A few words are the basis for
3464: building any control structure possible (except control structures that
3465: need storage, like calls, coroutines, and backtracking).
3466:
3467:
3468: doc-if
3469: doc-ahead
3470: doc-then
3471: doc-begin
3472: doc-until
3473: doc-again
3474: doc-cs-pick
3475: doc-cs-roll
3476:
3477:
3478: The Standard words @code{CS-PICK} and @code{CS-ROLL} allow you to
3479: manipulate the control-flow stack in a portable way. Without them, you
3480: would need to know how many stack items are occupied by a control-flow
3481: entry (many systems use one cell. In Gforth they currently take three,
3482: but this may change in the future).
3483:
3484: Some standard control structure words are built from these words:
3485:
3486:
3487: doc-else
3488: doc-while
3489: doc-repeat
3490:
3491:
3492: @noindent
3493: Gforth adds some more control-structure words:
3494:
3495:
3496: doc-endif
3497: doc-?dup-if
3498: doc-?dup-0=-if
3499:
3500:
3501: @noindent
3502: Counted loop words constitute a separate group of words:
3503:
3504:
3505: doc-?do
3506: doc-+do
3507: doc-u+do
3508: doc--do
3509: doc-u-do
3510: doc-do
3511: doc-for
3512: doc-loop
3513: doc-+loop
3514: doc--loop
3515: doc-next
3516: doc-leave
3517: doc-?leave
3518: doc-unloop
3519: doc-done
3520:
3521:
3522: The standard does not allow using @code{CS-PICK} and @code{CS-ROLL} on
3523: @i{do-sys}. Gforth allows it, but it's your job to ensure that for
3524: every @code{?DO} etc. there is exactly one @code{UNLOOP} on any path
3525: through the definition (@code{LOOP} etc. compile an @code{UNLOOP} on the
3526: fall-through path). Also, you have to ensure that all @code{LEAVE}s are
3527: resolved (by using one of the loop-ending words or @code{DONE}).
3528:
3529: @noindent
3530: Another group of control structure words are:
3531:
3532:
3533: doc-case
3534: doc-endcase
3535: doc-of
3536: doc-endof
3537:
3538:
3539: @i{case-sys} and @i{of-sys} cannot be processed using @code{CS-PICK} and
3540: @code{CS-ROLL}.
3541:
3542: @subsubsection Programming Style
3543:
3544: In order to ensure readability we recommend that you do not create
3545: arbitrary control structures directly, but define new control structure
3546: words for the control structure you want and use these words in your
3547: program. For example, instead of writing:
3548:
3549: @example
3550: BEGIN
3551: ...
3552: IF [ 1 CS-ROLL ]
3553: ...
3554: AGAIN THEN
3555: @end example
3556:
3557: @noindent
3558: we recommend defining control structure words, e.g.,
3559:
3560: @example
3561: : WHILE ( DEST -- ORIG DEST )
3562: POSTPONE IF
3563: 1 CS-ROLL ; immediate
3564:
3565: : REPEAT ( orig dest -- )
3566: POSTPONE AGAIN
3567: POSTPONE THEN ; immediate
3568: @end example
3569:
3570: @noindent
3571: and then using these to create the control structure:
3572:
3573: @example
3574: BEGIN
3575: ...
3576: WHILE
3577: ...
3578: REPEAT
3579: @end example
3580:
3581: That's much easier to read, isn't it? Of course, @code{REPEAT} and
3582: @code{WHILE} are predefined, so in this example it would not be
3583: necessary to define them.
3584:
3585: @node Calls and returns, Exception Handling, Arbitrary control structures, Control Structures
3586: @subsection Calls and returns
3587: @cindex calling a definition
3588: @cindex returning from a definition
3589:
3590: @cindex recursive definitions
3591: A definition can be called simply be writing the name of the definition
3592: to be called. Normally a definition is invisible during its own
3593: definition. If you want to write a directly recursive definition, you
3594: can use @code{recursive} to make the current definition visible, or
3595: @code{recurse} to call the current definition directly.
3596:
3597:
3598: doc-recursive
3599: doc-recurse
3600:
3601:
3602: @comment TODO add example of the two recursion methods
3603: @quotation
3604: @progstyle
3605: I prefer using @code{recursive} to @code{recurse}, because calling the
3606: definition by name is more descriptive (if the name is well-chosen) than
3607: the somewhat cryptic @code{recurse}. E.g., in a quicksort
3608: implementation, it is much better to read (and think) ``now sort the
3609: partitions'' than to read ``now do a recursive call''.
3610: @end quotation
3611:
3612: For mutual recursion, use @code{Defer}red words, like this:
3613:
3614: @example
3615: Defer foo
3616:
3617: : bar ( ... -- ... )
3618: ... foo ... ;
3619:
3620: :noname ( ... -- ... )
3621: ... bar ... ;
3622: IS foo
3623: @end example
3624:
3625: Deferred words are discussed in more detail in @ref{Deferred words}.
3626:
3627: The current definition returns control to the calling definition when
3628: the end of the definition is reached or @code{EXIT} is encountered.
3629:
3630: doc-exit
3631: doc-;s
3632:
3633:
3634: @node Exception Handling, , Calls and returns, Control Structures
3635: @subsection Exception Handling
3636: @cindex exceptions
3637:
3638: If your program detects a fatal error condition, the simplest action
3639: that it can take is to @code{quit}. This resets the return stack and
3640: restarts the text interpreter, but does not print any error message.
3641:
3642: The next stage in severity is to execute @code{abort}, which has the
3643: same effect as @code{quit}, with the addition that it resets the data
3644: stack.
3645:
3646: A slightly more sophisticated approach is use use @code{abort"}, which
3647: compiles a string to be used as an error message and does a conditional
3648: @code{abort} at run-time. For example:
3649:
3650: @example
3651: @kbd{: checker abort" That flag was true" ." A false flag" ;@key{RET}} ok
3652: @kbd{0 checker@key{RET}} A false flag ok
3653: @kbd{1 checker@key{RET}}
3654: :1: That flag was true
3655: 1 checker
3656: ^^^^^^^
3657: $400D1648 throw
3658: $400E4660
3659: @end example
3660:
3661: These simple techniques allow a program to react to a fatal error
3662: condition, but they are not exactly user-friendly. The ANS Forth
3663: Exception word set provides the pair of words @code{throw} and
3664: @code{catch}, which can be used to provide sophisticated error-handling.
3665:
3666: @code{catch} has a similar behaviour to @code{execute}, in that it takes
3667: an @i{xt} as a parameter and starts execution of the xt. However,
3668: before passing control to the xt, @code{catch} pushes an
3669: @dfn{exception frame} onto the @dfn{exception stack}. This exception
3670: frame is used to restore the system to a known state if a detected error
3671: occurs during the execution of the xt. A typical way to use @code{catch}
3672: would be:
3673:
3674: @example
3675: ... ['] foo catch IF ...
3676: @end example
3677:
3678: @c TOS is undefined. - anton
3679:
3680: @c nac-> TODO -- I need to look at this example again.
3681:
3682: Whilst @code{foo} executes, it can call other words to any level of
3683: nesting, as usual. If @code{foo} (and all the words that it calls)
3684: execute successfully, control will ultimately pass to the word following
3685: the @code{catch}, and there will be a 0 at TOS. However, if any word
3686: detects an error, it can terminate the execution of @code{foo} by
3687: pushing a non-zero error code onto the stack and then performing a
3688: @code{throw}. The execution of @code{throw} will pass control to the
3689: word following the @code{catch}, but this time the TOS will hold the
3690: error code. Therefore, the @code{IF} in the example can be used to
3691: determine whether @code{foo} executed successfully.
3692:
3693: This simple example shows how you can use @code{throw} and @code{catch}
3694: to ``take over'' exception handling from the system:
3695: @example
3696: : my-div ['] / catch if ." DIVIDE ERROR" else ." OK.. " . then ;
3697: @end example
3698:
3699: The next example is more sophisticated and shows a multi-level
3700: @code{throw} and @code{catch}. To understand this example, start at the
3701: definition of @code{top-level} and work backwards:
3702:
3703: @example
3704: : lowest-level ( -- c )
3705: key dup 27 = if
3706: 1 throw \ ESCAPE key pressed
3707: else
3708: ." lowest-level successful" CR
3709: then
3710: ;
3711:
3712: : lower-level ( -- c )
3713: lowest-level
3714: \ at this level consider a CTRL-U to be a fatal error
3715: dup 21 = if \ CTRL-U
3716: 2 throw
3717: else
3718: ." lower-level successful" CR
3719: then
3720: ;
3721:
3722: : low-level ( -- c )
3723: ['] lower-level catch
3724: ?dup if
3725: \ error occurred - do we recognise it?
3726: dup 1 = if
3727: \ ESCAPE key pressed.. pretend it was an E
3728: [char] E
3729: else throw \ propogate the error upwards
3730: then
3731: then
3732: ." low-level successfull" CR
3733: ;
3734:
3735: : top-level ( -- )
3736: CR ['] low-level catch \ CATCH is used like EXECUTE
3737: ?dup if \ error occurred..
3738: ." Error " . ." occurred - contact your supplier"
3739: else
3740: ." The '" emit ." ' key was pressed" CR
3741: then
3742: ;
3743: @end example
3744:
3745: The ANS Forth document assigns @code{throw} codes thus:
3746:
3747: @itemize @bullet
3748: @item
3749: codes in the range -1 -- -255 are reserved to be assigned by the
3750: Standard. Assignments for codes in the range -1 -- -58 are currently
3751: documented in the Standard. In particular, @code{-1 throw} is equivalent
3752: to @code{abort} and @code{-2 throw} is equivalent to @code{abort"}.
3753: @item
3754: codes in the range -256 -- -4095 are reserved to be assigned by the system.
3755: @item
3756: all other codes may be assigned by programs.
3757: @end itemize
3758:
3759: Gforth provides the word @code{exception} as a mechanism for assigning
3760: system throw codes to applications. This allows multiple applications to
3761: co-exist in memory without any clash of @code{throw} codes. A definition
3762: of @code{exception} in ANS Forth is provided in
3763: @file{compat/exception.fs}.
3764:
3765:
3766: doc-quit
3767: doc-abort
3768: doc-abort"
3769:
3770: doc-catch
3771: doc-throw
3772: doc---exception-exception
3773:
3774:
3775:
3776: @c -------------------------------------------------------------
3777: @node Defining Words, The Text Interpreter, Control Structures, Words
3778: @section Defining Words
3779: @cindex defining words
3780:
3781: @menu
3782: * CREATE::
3783: * Variables:: Variables and user variables
3784: * Constants::
3785: * Values:: Initialised variables
3786: * Colon Definitions::
3787: * Anonymous Definitions:: Definitions without names
3788: * User-defined Defining Words::
3789: * Deferred words:: Allow forward references
3790: * Aliases::
3791: * Supplying names::
3792: * Interpretation and Compilation Semantics::
3793: * Combined words::
3794: @end menu
3795:
3796: @node CREATE, Variables, Defining Words, Defining Words
3797: @subsection @code{CREATE}
3798: @cindex simple defining words
3799: @cindex defining words, simple
3800:
3801: Defining words are used to create new entries in the dictionary. The
3802: simplest defining word is @code{CREATE}. @code{CREATE} is used like
3803: this:
3804:
3805: @example
3806: CREATE new-word1
3807: @end example
3808:
3809: @code{CREATE} is a parsing word that generates a dictionary entry for
3810: @code{new-word1}. When @code{new-word1} is executed, all that it does is
3811: leave an address on the stack. The address represents the value of
3812: the data space pointer (@code{HERE}) at the time that @code{new-word1}
3813: was defined. Therefore, @code{CREATE} is a way of associating a name
3814: with the address of a region of memory.
3815:
3816: doc-create
3817:
3818: By extending this example to reserve some memory in data space, we end
3819: up with a @i{variable}. Here are two different ways to do it:
3820:
3821: @example
3822: CREATE new-word2 1 cells allot \ reserve 1 cell - initial value undefined
3823: CREATE new-word3 4 , \ reserve 1 cell and initialise it (to 4)
3824: @end example
3825:
3826: The variable can be examined and modified using @code{@@} (``fetch'') and
3827: @code{!} (``store'') like this:
3828:
3829: @example
3830: new-word2 @@ . \ get address, fetch from it and display
3831: 1234 new-word2 ! \ new value, get address, store to it
3832: @end example
3833:
3834: @cindex arrays
3835: A similar mechanism can be used to create arrays. For example, an
3836: 80-character text input buffer:
3837:
3838: @example
3839: CREATE text-buf 80 chars allot
3840:
3841: text-buf 0 chars c@@ \ the 1st character (offset 0)
3842: text-buf 3 chars c@@ \ the 4th character (offset 3)
3843: @end example
3844:
3845: You can build arbitrarily complex data structures by allocating
3846: appropriate areas of memory. @xref{Structures} for further discussions
3847: of this, and to learn about some Gforth tools that make it easier.
3848:
3849:
3850: @node Variables, Constants, CREATE, Defining Words
3851: @subsection Variables
3852: @cindex variables
3853:
3854: The previous section showed how a sequence of commands could be used to
3855: generate a variable. As a final refinement, the whole code sequence can
3856: be wrapped up in a defining word (pre-empting the subject of the next
3857: section), making it easier to create new variables:
3858:
3859: @example
3860: : myvariableX ( "name" -- a-addr ) CREATE 1 cells allot ;
3861: : myvariable0 ( "name" -- a-addr ) CREATE 0 , ;
3862:
3863: myvariableX foo \ variable foo starts off with an unknown value
3864: myvariable0 joe \ whilst joe is initialised to 0
3865:
3866: 45 3 * foo ! \ set foo to 135
3867: 1234 joe ! \ set joe to 1234
3868: 3 joe +! \ increment joe by 3.. to 1237
3869: @end example
3870:
3871: Not surprisingly, there is no need to define @code{myvariable}, since
3872: Forth already has a definition @code{Variable}. ANS Forth does not
3873: require a @code{Variable} to be initialised when it is created (i.e., it
3874: behaves like @code{myvariableX}). In contrast, Gforth's @code{Variable}
3875: initialises the variable to 0 (i.e., it behaves exactly like
3876: @code{myvariable0}). Forth also provides @code{2Variable} and
3877: @code{fvariable} for double and floating-point variables,
3878: respectively -- both are initialised to 0 in Gforth.
3879:
3880: doc-variable
3881: doc-2variable
3882: doc-fvariable
3883:
3884:
3885: @cindex user variables
3886: @cindex user space
3887: The defining word @code{User} behaves in the same way as @code{Variable}.
3888: The difference is that it reserves space in @i{user (data) space} rather
3889: than normal data space. In a Forth system that has a multi-tasker, each
3890: task has its own set of user variables.
3891:
3892: doc-user
3893:
3894: @comment TODO is that stuff about user variables strictly correct? Is it
3895: @comment just terminal tasks that have user variables?
3896: @comment should document tasker.fs (with some examples) elsewhere
3897: @comment in this manual, then expand on user space and user variables.
3898:
3899:
3900: @node Constants, Values, Variables, Defining Words
3901: @subsection Constants
3902: @cindex constants
3903:
3904: @code{Constant} allows you to declare a fixed value and refer to it by
3905: name. For example:
3906:
3907: @example
3908: 12 Constant INCHES-PER-FOOT
3909: 3E+08 fconstant SPEED-O-LIGHT
3910: @end example
3911:
3912: A @code{Variable} can be both read and written, so its run-time
3913: behaviour is to supply an address through which its current value can be
3914: manipulated. In contrast, the value of a @code{Constant} cannot be
3915: changed once it has been declared@footnote{Well, often it can be -- but
3916: not in a Standard, portable way. It's safer to use a @code{Value} (read
3917: on).} so it's not necessary to supply the address -- it is more
3918: efficient to return the value of the constant directly. That's exactly
3919: what happens; the run-time effect of a constant is to put its value on
3920: the top of the stack (@ref{User-defined Defining Words} describes one
3921: way of implementing @code{Constant}).
3922:
3923: Gforth also provides @code{2Constant} and @code{fconstant} for defining
3924: double and floating-point constants, respectively.
3925:
3926: doc-constant
3927: doc-2constant
3928: doc-fconstant
3929:
3930: @c that's too deep, and it's not necessarily true for all ANS Forths. - anton
3931: @c nac-> How could that not be true in an ANS Forth? You can't define a
3932: @c constant, use it and then delete the definition of the constant..
3933: @c I agree that it's rather deep, but IMO it is an important difference
3934: @c relative to other programming languages.. often it's annoying: it
3935: @c certainly changes my programming style relative to C.
3936:
3937: Constants in Forth behave differently from their equivalents in other
3938: programming languages. In other languages, a constant (such as an EQU in
3939: assembler or a #define in C) only exists at compile-time; in the
3940: executable program the constant has been translated into an absolute
3941: number and, unless you are using a symbolic debugger, it's impossible to
3942: know what abstract thing that number represents. In Forth a constant has
3943: an entry in the header space and remains there after the code that uses
3944: it has been defined. In fact, it must remain in the dictionary since it
3945: has run-time duties to perform. For example:
3946:
3947: @example
3948: 12 Constant INCHES-PER-FOOT
3949: : FEET-TO-INCHES ( n1 -- n2 ) INCHES-PER-FOOT * ;
3950: @end example
3951:
3952: @cindex in-lining of constants
3953: When @code{FEET-TO-INCHES} is executed, it will in turn execute the xt
3954: associated with the constant @code{INCHES-PER-FOOT}. If you use
3955: @code{see} to decompile the definition of @code{FEET-TO-INCHES}, you can
3956: see that it makes a call to @code{INCHES-PER-FOOT}. Some Forth compilers
3957: attempt to optimise constants by in-lining them where they are used. You
3958: can force Gforth to in-line a constant like this:
3959:
3960: @example
3961: : FEET-TO-INCHES ( n1 -- n2 ) [ INCHES-PER-FOOT ] LITERAL * ;
3962: @end example
3963:
3964: If you use @code{see} to decompile @i{this} version of
3965: @code{FEET-TO-INCHES}, you can see that @code{INCHES-PER-FOOT} is no
3966: longer present. @xref{Interpret/Compile states} and @ref{Literals} on
3967: how this works.
3968:
3969: In-lining constants in this way might improve execution time
3970: fractionally, and can ensure that a constant is now only referenced at
3971: compile-time. However, the definition of the constant still remains in
3972: the dictionary. Some Forth compilers provide a mechanism for controlling
3973: a second dictionary for holding transient words such that this second
3974: dictionary can be deleted later in order to recover memory
3975: space. However, there is no standard way of doing this.
3976:
3977:
3978: @node Values, Colon Definitions, Constants, Defining Words
3979: @subsection Values
3980: @cindex values
3981:
3982: A @code{Value} is like a @code{Variable} but with two important
3983: differences:
3984:
3985: @itemize @bullet
3986: @item
3987: A @code{Value} is initialised when it is declared; like a
3988: @code{Constant} but unlike a @code{Variable}.
3989: @item
3990: A @code{Value} returns its value rather than its address when it is
3991: executed; i.e., it has the same run-time behaviour as @code{Constant}.
3992: @end itemize
3993:
3994: A @code{Value} needs an additional word, @code{TO} to allow its value to
3995: be changed. Here are some examples:
3996:
3997: @example
3998: 12 Value APPLES \ Define APPLES with an initial value of 12
3999: 34 TO APPLES \ Change the value of APPLES. TO is a parsing word
4000: APPLES \ puts 34 on the top of the stack.
4001: @end example
4002:
4003: doc-value
4004: doc-to
4005:
4006:
4007: @node Colon Definitions, Anonymous Definitions, Values, Defining Words
4008: @subsection Colon Definitions
4009: @cindex colon definitions
4010:
4011: @example
4012: : name ( ... -- ... )
4013: word1 word2 word3 ;
4014: @end example
4015:
4016: @noindent
4017: Creates a word called @code{name} that, upon execution, executes
4018: @code{word1 word2 word3}. @code{name} is a @dfn{(colon) definition}.
4019:
4020: The explanation above is somewhat superficial. @xref{Your first
4021: definition} for simple examples of colon definitions, then
4022: @xref{Interpretation and Compilation Semantics} for an in-depth
4023: discussion of some of the issues involved.
4024:
4025: doc-:
4026: doc-;
4027:
4028:
4029: @node Anonymous Definitions, User-defined Defining Words, Colon Definitions, Defining Words
4030: @subsection Anonymous Definitions
4031: @cindex colon definitions
4032: @cindex defining words without name
4033:
4034: Sometimes you want to define an @dfn{anonymous word}; a word without a
4035: name. You can do this with:
4036:
4037: doc-:noname
4038:
4039: This leaves the execution token for the word on the stack after the
4040: closing @code{;}. Here's an example in which a deferred word is
4041: initialised with an @code{xt} from an anonymous colon definition:
4042:
4043: @example
4044: Defer deferred
4045: :noname ( ... -- ... )
4046: ... ;
4047: IS deferred
4048: @end example
4049:
4050: @noindent
4051: Gforth provides an alternative way of doing this, using two separate
4052: words:
4053:
4054: doc-noname
4055: @cindex execution token of last defined word
4056: doc-lastxt
4057:
4058: @noindent
4059: The previous example can be rewritten using @code{noname} and
4060: @code{lastxt}:
4061:
4062: @example
4063: Defer deferred
4064: noname : ( ... -- ... )
4065: ... ;
4066: lastxt IS deferred
4067: @end example
4068:
4069: @noindent
4070: @code{noname} works with any defining word, not just @code{:}.
4071:
4072: @code{lastxt} also works when the last word was not defined as
4073: @code{noname}. It also has the useful property that is is valid as soon
4074: as the header for a definition has been built. Thus:
4075:
4076: @example
4077: lastxt . : foo [ lastxt . ] ; ' foo .
4078: @end example
4079:
4080: @noindent
4081: prints 3 numbers; the last two are the same.
4082:
4083:
4084: @node User-defined Defining Words, Deferred words, Anonymous Definitions, Defining Words
4085: @subsection User-defined Defining Words
4086: @cindex user-defined defining words
4087: @cindex defining words, user-defined
4088:
4089: You can create a new defining word by wrapping defining-time code around
4090: an existing defining word and putting the sequence in a colon
4091: definition. For example, suppose that you have a word @code{stats} that
4092: gathers statistics about colon definitions given the @i{xt} of the
4093: definition, and you want every colon definition in your application to
4094: make a call to @code{stats}. You can define and use a new version of
4095: @code{:} like this:
4096:
4097: @example
4098: : stats ( xt -- ) DUP ." (Gathering statistics for " . ." )"
4099: ... ; \ other code
4100:
4101: : my: : lastxt postpone literal ['] stats compile, ;
4102:
4103: my: foo + - ;
4104: @end example
4105:
4106: When @code{foo} is defined using @code{my:} these steps occur:
4107:
4108: @itemize @bullet
4109: @item
4110: @code{my:} is executed.
4111: @item
4112: The @code{:} within the definition (the one between @code{my:} and
4113: @code{lastxt}) is executed, and does just what it always does; it parses
4114: the input stream for a name, builds a dictionary header for the name
4115: @code{foo} and switches @code{state} from interpret to compile.
4116: @item
4117: The word @code{lastxt} is executed. It puts the @i{xt} for the word that is
4118: being defined -- @code{foo} -- onto the stack.
4119: @item
4120: The code that was produced by @code{postpone literal} is executed; this
4121: causes the value on the stack to be compiled as a literal in the code
4122: area of @code{foo}.
4123: @item
4124: The code @code{['] stats} compiles a literal into the definition of
4125: @code{my:}. When @code{compile,} is executed, that literal -- the
4126: execution token for @code{stats} -- is layed down in the code area of
4127: @code{foo} , following the literal@footnote{Strictly speaking, the
4128: mechanism that @code{compile,} uses to convert an @i{xt} into something
4129: in the code area is implementation-dependent. A threaded implementation
4130: might spit out the execution token directly whilst another
4131: implementation might spit out a native code sequence.}.
4132: @item
4133: At this point, the execution of @code{my:} is complete, and control
4134: returns to the text interpreter. The text interpreter is in compile
4135: state, so subsequent text @code{+ -} is compiled into the definition of
4136: @code{foo} and the @code{;} terminates the definition as always.
4137: @end itemize
4138:
4139: You can use @code{see} to decompile a word that was defined using
4140: @code{my:} and see how it is different from a normal @code{:}
4141: definition. For example:
4142:
4143: @example
4144: : bar + - ; \ like foo but using : rather than my:
4145: see bar
4146: : bar
4147: + - ;
4148: see foo
4149: : foo
4150: 107645672 stats + - ;
4151:
4152: \ use ' stats . to show that 107645672 is the xt for stats
4153: @end example
4154:
4155: You can use techniques like this to make new defining words in terms of
4156: @i{any} existing defining word.
4157:
4158:
4159: @cindex defining defining words
4160: @cindex @code{CREATE} ... @code{DOES>}
4161: If you want the words defined with your defining words to behave
4162: differently from words defined with standard defining words, you can
4163: write your defining word like this:
4164:
4165: @example
4166: : def-word ( "name" -- )
4167: CREATE @i{code1}
4168: DOES> ( ... -- ... )
4169: @i{code2} ;
4170:
4171: def-word name
4172: @end example
4173:
4174: @cindex child words
4175: This fragment defines a @dfn{defining word} @code{def-word} and then
4176: executes it. When @code{def-word} executes, it @code{CREATE}s a new
4177: word, @code{name}, and executes the code @i{code1}. The code @i{code2}
4178: is not executed at this time. The word @code{name} is sometimes called a
4179: @dfn{child} of @code{def-word}.
4180:
4181: When you execute @code{name}, the address of the body of @code{name} is
4182: put on the data stack and @i{code2} is executed (the address of the body
4183: of @code{name} is the address @code{HERE} returns immediately after the
4184: @code{CREATE}).
4185:
4186: @cindex atavism in child words
4187: You can use @code{def-word} to define a set of child words that behave
4188: differently, though atavistically; they all have a common run-time
4189: behaviour determined by @i{code2}. Typically, the @i{code1} sequence
4190: builds a data area in the body of the child word. The structure of the
4191: data is common to all children of @code{def-word}, but the data values
4192: are specific -- and private -- to each child word. When a child word is
4193: executed, the address of its private data area is passed as a parameter
4194: on TOS to be used and manipulated@footnote{It is legitimate both to read
4195: and write to this data area.} by @i{code2}.
4196:
4197: The two fragments of code that make up the defining words act (are
4198: executed) at two completely separate times:
4199:
4200: @itemize @bullet
4201: @item
4202: At @i{define time}, the defining word executes @i{code1} to generate a
4203: child word
4204: @item
4205: At @i{child execution time}, when a child word is invoked, @i{code2}
4206: is executed, using parameters (data) that are private and specific to
4207: the child word.
4208: @end itemize
4209:
4210: Another way of understanding the behaviour of @code{def-word} and
4211: @code{name} is to say that, if you make the following definitions:
4212: @example
4213: : def-word1 ( "name" -- )
4214: CREATE @i{code1} ;
4215:
4216: : action1 ( ... -- ... )
4217: @i{code2} ;
4218:
4219: def-word1 name1
4220: @end example
4221:
4222: @noindent
4223: Then using @code{name1 action1} is equivalent to using @code{name}.
4224:
4225: The classic example is that you can define @code{CONSTANT} in this way:
4226:
4227: @example
4228: : CONSTANT ( w "name" -- )
4229: CREATE ,
4230: DOES> ( -- w )
4231: @@ ;
4232: @end example
4233:
4234: @comment There is a beautiful description of how this works and what
4235: @comment it does in the Forthwrite 100th edition.. as well as an elegant
4236: @comment commentary on the Counting Fruits problem.
4237:
4238: When you create a constant with @code{5 CONSTANT five}, a set of
4239: define-time actions take place; first a new word @code{five} is created,
4240: then the value 5 is laid down in the body of @code{five} with
4241: @code{,}. When @code{five} is executed, the address of the body is put on
4242: the stack, and @code{@@} retrieves the value 5. The word @code{five} has
4243: no code of its own; it simply contains a data field and a pointer to the
4244: code that follows @code{DOES>} in its defining word. That makes words
4245: created in this way very compact.
4246:
4247: The final example in this section is intended to remind you that space
4248: reserved in @code{CREATE}d words is @i{data} space and therefore can be
4249: both read and written by a Standard program@footnote{Exercise: use this
4250: example as a starting point for your own implementation of @code{Value}
4251: and @code{TO} -- if you get stuck, investigate the behaviour of @code{'} and
4252: @code{[']}.}:
4253:
4254: @example
4255: : foo ( "name" -- )
4256: CREATE -1 ,
4257: DOES> ( -- )
4258: @@ . ;
4259:
4260: foo first-word
4261: foo second-word
4262:
4263: 123 ' first-word >BODY !
4264: @end example
4265:
4266: If @code{first-word} had been a @code{CREATE}d word, we could simply
4267: have executed it to get the address of its data field. However, since it
4268: was defined to have @code{DOES>} actions, its execution semantics are to
4269: perform those @code{DOES>} actions. To get the address of its data field
4270: it's necessary to use @code{'} to get its xt, then @code{>BODY} to
4271: translate the xt into the address of the data field. When you execute
4272: @code{first-word}, it will display @code{123}. When you execute
4273: @code{second-word} it will display @code{-1}.
4274:
4275: @cindex stack effect of @code{DOES>}-parts
4276: @cindex @code{DOES>}-parts, stack effect
4277: In the examples above the stack comment after the @code{DOES>} specifies
4278: the stack effect of the defined words, not the stack effect of the
4279: following code (the following code expects the address of the body on
4280: the top of stack, which is not reflected in the stack comment). This is
4281: the convention that I use and recommend (it clashes a bit with using
4282: locals declarations for stack effect specification, though).
4283:
4284: @subsubsection Applications of @code{CREATE..DOES>}
4285: @cindex @code{CREATE} ... @code{DOES>}, applications
4286:
4287: You may wonder how to use this feature. Here are some usage patterns:
4288:
4289: @cindex factoring similar colon definitions
4290: When you see a sequence of code occurring several times, and you can
4291: identify a meaning, you will factor it out as a colon definition. When
4292: you see similar colon definitions, you can factor them using
4293: @code{CREATE..DOES>}. E.g., an assembler usually defines several words
4294: that look very similar:
4295: @example
4296: : ori, ( reg-target reg-source n -- )
4297: 0 asm-reg-reg-imm ;
4298: : andi, ( reg-target reg-source n -- )
4299: 1 asm-reg-reg-imm ;
4300: @end example
4301:
4302: @noindent
4303: This could be factored with:
4304: @example
4305: : reg-reg-imm ( op-code -- )
4306: CREATE ,
4307: DOES> ( reg-target reg-source n -- )
4308: @@ asm-reg-reg-imm ;
4309:
4310: 0 reg-reg-imm ori,
4311: 1 reg-reg-imm andi,
4312: @end example
4313:
4314: @cindex currying
4315: Another view of @code{CREATE..DOES>} is to consider it as a crude way to
4316: supply a part of the parameters for a word (known as @dfn{currying} in
4317: the functional language community). E.g., @code{+} needs two
4318: parameters. Creating versions of @code{+} with one parameter fixed can
4319: be done like this:
4320: @example
4321: : curry+ ( n1 -- )
4322: CREATE ,
4323: DOES> ( n2 -- n1+n2 )
4324: @@ + ;
4325:
4326: 3 curry+ 3+
4327: -2 curry+ 2-
4328: @end example
4329:
4330: @subsubsection The gory details of @code{CREATE..DOES>}
4331: @cindex @code{CREATE} ... @code{DOES>}, details
4332:
4333: doc-does>
4334:
4335: @cindex @code{DOES>} in a separate definition
4336: This means that you need not use @code{CREATE} and @code{DOES>} in the
4337: same definition; you can put the @code{DOES>}-part in a separate
4338: definition. This allows us to, e.g., select among different @code{DOES>}-parts:
4339: @example
4340: : does1
4341: DOES> ( ... -- ... )
4342: ... ;
4343:
4344: : does2
4345: DOES> ( ... -- ... )
4346: ... ;
4347:
4348: : def-word ( ... -- ... )
4349: create ...
4350: IF
4351: does1
4352: ELSE
4353: does2
4354: ENDIF ;
4355: @end example
4356:
4357: In this example, the selection of whether to use @code{does1} or
4358: @code{does2} is made at compile-time; at the time that the child word is
4359: @code{CREATE}d.
4360:
4361: @cindex @code{DOES>} in interpretation state
4362: In a standard program you can apply a @code{DOES>}-part only if the last
4363: word was defined with @code{CREATE}. In Gforth, the @code{DOES>}-part
4364: will override the behaviour of the last word defined in any case. In a
4365: standard program, you can use @code{DOES>} only in a colon
4366: definition. In Gforth, you can also use it in interpretation state, in a
4367: kind of one-shot mode; for example:
4368: @example
4369: CREATE name ( ... -- ... )
4370: @i{initialization}
4371: DOES>
4372: @i{code} ;
4373: @end example
4374:
4375: @noindent
4376: is equivalent to the standard:
4377: @example
4378: :noname
4379: DOES>
4380: @i{code} ;
4381: CREATE name EXECUTE ( ... -- ... )
4382: @i{initialization}
4383: @end example
4384:
4385:
4386: doc->body
4387:
4388:
4389: @node Deferred words, Aliases, User-defined Defining Words, Defining Words
4390: @subsection Deferred words
4391: @cindex deferred words
4392:
4393: The defining word @code{Defer} allows you to define a word by name
4394: without defining its behaviour; the definition of its behaviour is
4395: deferred. Here are two situation where this can be useful:
4396:
4397: @itemize @bullet
4398: @item
4399: Where you want to allow the behaviour of a word to be altered later, and
4400: for all precompiled references to the word to change when its behaviour
4401: is changed.
4402: @item
4403: For mutual recursion; @xref{Calls and returns}.
4404: @end itemize
4405:
4406: In the following example, @code{foo} always invokes the version of
4407: @code{greet} that prints ``@code{Good morning}'' whilst @code{bar}
4408: always invokes the version that prints ``@code{Hello}''. There is no way
4409: of getting @code{foo} to use the later version without re-ordering the
4410: source code and recompiling it.
4411:
4412: @example
4413: : greet ." Good morning" ;
4414: : foo ... greet ... ;
4415: : greet ." Hello" ;
4416: : bar ... greet ... ;
4417: @end example
4418:
4419: This problem can be solved by defining @code{greet} as a @code{Defer}red
4420: word. The behaviour of a @code{Defer}red word can be defined and
4421: redefined at any time by using @code{IS} to associate the xt of a
4422: previously-defined word with it. The previous example becomes:
4423:
4424: @example
4425: Defer greet
4426: : foo ... greet ... ;
4427: : bar ... greet ... ;
4428: : greet1 ." Good morning" ;
4429: : greet2 ." Hello" ;
4430: ' greet2 <IS> greet \ make greet behave like greet2
4431: @end example
4432:
4433: A deferred word can be used to improve the statistics-gathering example
4434: from @ref{User-defined Defining Words}; rather than edit the
4435: application's source code to change every @code{:} to a @code{my:}, do
4436: this:
4437:
4438: @example
4439: : real: : ; \ retain access to the original
4440: defer : \ redefine as a deferred word
4441: ' my: IS : \ use special version of :
4442: \
4443: \ load application here
4444: \
4445: ' real: IS : \ go back to the original
4446: @end example
4447:
4448:
4449: One thing to note is that @code{<IS>} consumes its name when it is
4450: executed. If you want to specify the name at compile time, use
4451: @code{[IS]}:
4452:
4453: @example
4454: : set-greet ( xt -- )
4455: [IS] greet ;
4456:
4457: ' greet1 set-greet
4458: @end example
4459:
4460: A deferred word can only inherit default semantics from the xt (because
4461: that is all that an xt can represent -- @pxref{Tokens for Words} for
4462: more discussion of this). However, the semantics of the deferred word
4463: itself can be modified at the time that it is defined. For example:
4464:
4465: @example
4466: : bar .... ; compile-only
4467: Defer fred immediate
4468: Defer jim
4469:
4470: ' bar <IS> jim \ jim has default semantics
4471: ' bar <IS> fred \ fred is immediate
4472: @end example
4473:
4474: doc-defer
4475: doc-<is>
4476: doc-[is]
4477: doc-is
4478: @comment TODO document these: what's defers [is]
4479: doc-what's
4480: doc-defers
4481:
4482: @c Use @code{words-deferred} to see a list of deferred words.
4483:
4484: Definitions in ANS Forth for @code{defer}, @code{<is>} and @code{[is]}
4485: are provided in @file{compat/defer.fs}.
4486:
4487:
4488: @node Aliases, Supplying names, Deferred words, Defining Words
4489: @subsection Aliases
4490: @cindex aliases
4491:
4492: The defining word @code{Alias} allows you to define a word by name that
4493: has the same behaviour as some other word. Here are two situation where
4494: this can be useful:
4495:
4496: @itemize @bullet
4497: @item
4498: When you want access to a word's definition from a different word list
4499: (for an example of this, see the definition of the @code{Root} word list
4500: in the Gforth source).
4501: @item
4502: When you want to create a synonym; a definition that can be known by
4503: either of two names (for example, @code{THEN} and @code{ENDIF} are
4504: aliases).
4505: @end itemize
4506:
4507: The word whose behaviour the alias is to inherit is represented by an
4508: xt. Therefore, the alias only inherits default semantics from its
4509: ancestor. The semantics of the alias itself can be modified at the time
4510: that it is defined. For example:
4511:
4512: @example
4513: : foo ... ; immediate
4514:
4515: ' foo Alias bar \ bar is not an immediate word
4516: ' foo Alias fooby immediate \ fooby is an immediate word
4517: @end example
4518:
4519: Words that are aliases have the same xt, different headers in the
4520: dictionary, and consequently different name tokens (@pxref{Tokens for
4521: Words}) and possibly different immediate flags. An alias can only have
4522: default or immediate compilation semantics; you can define aliases for
4523: combined words with @code{interpret/compile:} -- see @ref{Combined words}.
4524:
4525: doc-alias
4526:
4527:
4528: @node Supplying names, Interpretation and Compilation Semantics, Aliases, Defining Words
4529: @subsection Supplying the name of a defined word
4530: @cindex names for defined words
4531: @cindex defining words, name given in a string
4532:
4533: By default, a defining word takes the name for the defined word from the
4534: input stream. Sometimes you want to supply the name from a string. You
4535: can do this with:
4536:
4537: doc-nextname
4538:
4539: For example:
4540:
4541: @example
4542: s" foo" nextname create
4543: @end example
4544:
4545: @noindent
4546: is equivalent to:
4547:
4548: @example
4549: create foo
4550: @end example
4551:
4552: @noindent
4553: @code{nextname} works with any defining word, not just @code{:}.
4554:
4555:
4556: @node Interpretation and Compilation Semantics, Combined words, Supplying names, Defining Words
4557: @subsection Interpretation and Compilation Semantics
4558: @cindex semantics, interpretation and compilation
4559:
4560: @cindex interpretation semantics
4561: The @dfn{interpretation semantics} of a word are what the text
4562: interpreter does when it encounters the word in interpret state. It also
4563: appears in some other contexts, e.g., the execution token returned by
4564: @code{' @i{word}} identifies the interpretation semantics of
4565: @i{word} (in other words, @code{' @i{word} execute} is equivalent to
4566: interpret-state text interpretation of @code{@i{word}}).
4567:
4568: @cindex compilation semantics
4569: The @dfn{compilation semantics} of a word are what the text interpreter
4570: does when it encounters the word in compile state. It also appears in
4571: other contexts, e.g, @code{POSTPONE @i{word}} compiles@footnote{In
4572: standard terminology, ``appends to the current definition''.} the
4573: compilation semantics of @i{word}.
4574:
4575: @cindex execution semantics
4576: The standard also talks about @dfn{execution semantics}. They are used
4577: only for defining the interpretation and compilation semantics of many
4578: words. By default, the interpretation semantics of a word are to
4579: @code{execute} its execution semantics, and the compilation semantics of
4580: a word are to @code{compile,} its execution semantics.@footnote{In
4581: standard terminology: The default interpretation semantics are its
4582: execution semantics; the default compilation semantics are to append its
4583: execution semantics to the execution semantics of the current
4584: definition.}
4585:
4586: @comment TODO expand, make it co-operate with new sections on text interpreter.
4587:
4588: @cindex immediate words
4589: @cindex compile-only words
4590: You can change the semantics of the most-recently defined word:
4591:
4592:
4593: doc-immediate
4594: doc-compile-only
4595: doc-restrict
4596:
4597:
4598: Note that ticking (@code{'}) a compile-only word gives an error
4599: (``Interpreting a compile-only word'').
4600:
4601:
4602: @node Combined words, ,Interpretation and Compilation Semantics, Defining Words
4603: @subsection Combined Words
4604: @cindex combined words
4605:
4606: Gforth allows you to define @dfn{combined words} -- words that have an
4607: arbitrary combination of interpretation and compilation semantics.
4608:
4609:
4610: doc-interpret/compile:
4611:
4612:
4613: This feature was introduced for implementing @code{TO} and @code{S"}. I
4614: recommend that you do not define such words, as cute as they may be:
4615: they make it hard to get at both parts of the word in some contexts.
4616: E.g., assume you want to get an execution token for the compilation
4617: part. Instead, define two words, one that embodies the interpretation
4618: part, and one that embodies the compilation part. Once you have done
4619: that, you can define a combined word with @code{interpret/compile:} for
4620: the convenience of your users.
4621:
4622: You might try to use this feature to provide an optimizing
4623: implementation of the default compilation semantics of a word. For
4624: example, by defining:
4625: @example
4626: :noname
4627: foo bar ;
4628: :noname
4629: POSTPONE foo POSTPONE bar ;
4630: interpret/compile: opti-foobar
4631: @end example
4632:
4633: @noindent
4634: as an optimizing version of:
4635:
4636: @example
4637: : foobar
4638: foo bar ;
4639: @end example
4640:
4641: Unfortunately, this does not work correctly with @code{[compile]},
4642: because @code{[compile]} assumes that the compilation semantics of all
4643: @code{interpret/compile:} words are non-default. I.e., @code{[compile]
4644: opti-foobar} would compile compilation semantics, whereas
4645: @code{[compile] foobar} would compile interpretation semantics.
4646:
4647: @cindex state-smart words (are a bad idea)
4648: Some people try to use @dfn{state-smart} words to emulate the feature provided
4649: by @code{interpret/compile:} (words are state-smart if they check
4650: @code{STATE} during execution). E.g., they would try to code
4651: @code{foobar} like this:
4652:
4653: @example
4654: : foobar
4655: STATE @@
4656: IF ( compilation state )
4657: POSTPONE foo POSTPONE bar
4658: ELSE
4659: foo bar
4660: ENDIF ; immediate
4661: @end example
4662:
4663: Although this works if @code{foobar} is only processed by the text
4664: interpreter, it does not work in other contexts (like @code{'} or
4665: @code{POSTPONE}). E.g., @code{' foobar} will produce an execution token
4666: for a state-smart word, not for the interpretation semantics of the
4667: original @code{foobar}; when you execute this execution token (directly
4668: with @code{EXECUTE} or indirectly through @code{COMPILE,}) in compile
4669: state, the result will not be what you expected (i.e., it will not
4670: perform @code{foo bar}). State-smart words are a bad idea. Simply don't
4671: write them@footnote{For a more detailed discussion of this topic, see
4672: @cite{@code{State}-smartness -- Why it is Evil and How to Exorcise it} by Anton
4673: Ertl; presented at EuroForth '98 and available from
4674: @url{http://www.complang.tuwien.ac.at/papers/ertl98.ps.gz}}!
4675:
4676: @cindex defining words with arbitrary semantics combinations
4677: It is also possible to write defining words that define words with
4678: arbitrary combinations of interpretation and compilation semantics. In
4679: general, they look like this:
4680:
4681: @example
4682: : def-word
4683: create-interpret/compile
4684: @i{code1}
4685: interpretation>
4686: @i{code2}
4687: <interpretation
4688: compilation>
4689: @i{code3}
4690: <compilation ;
4691: @end example
4692:
4693: For a @i{word} defined with @code{def-word}, the interpretation
4694: semantics are to push the address of the body of @i{word} and perform
4695: @i{code2}, and the compilation semantics are to push the address of
4696: the body of @i{word} and perform @i{code3}. E.g., @code{constant}
4697: can also be defined like this (except that the defined constants don't
4698: behave correctly when @code{[compile]}d):
4699:
4700: @example
4701: : constant ( n "name" -- )
4702: create-interpret/compile
4703: ,
4704: interpretation> ( -- n )
4705: @@
4706: <interpretation
4707: compilation> ( compilation. -- ; run-time. -- n )
4708: @@ postpone literal
4709: <compilation ;
4710: @end example
4711:
4712:
4713: doc-create-interpret/compile
4714: doc-interpretation>
4715: doc-<interpretation
4716: doc-compilation>
4717: doc-<compilation
4718:
4719:
4720: Words defined with @code{interpret/compile:} and
4721: @code{create-interpret/compile} have an extended header structure that
4722: differs from other words; however, unless you try to access them with
4723: plain address arithmetic, you should not notice this. Words for
4724: accessing the header structure usually know how to deal with this; e.g.,
4725: @code{'} @i{word} @code{>body} also gives you the body of a word created
4726: with @code{create-interpret/compile}.
4727:
4728:
4729: doc-postpone
4730:
4731: @comment TODO -- expand glossary text for POSTPONE
4732:
4733: @c ----------------------------------------------------------
4734: @node The Text Interpreter, Tokens for Words, Defining Words, Words
4735: @section The Text Interpreter
4736: @cindex interpreter - outer
4737: @cindex text interpreter
4738: @cindex outer interpreter
4739:
4740: @c Should we really describe all these ugly details? IMO the text
4741: @c interpreter should be much cleaner, but that may not be possible within
4742: @c ANS Forth. - anton
4743: @c nac-> I wanted to explain how it works to show how you can exploit
4744: @c it in your own programs. When I was writing a cross-compiler, figuring out
4745: @c some of these gory details was very helpful to me. None of the textbooks
4746: @c I've seen cover it, and the most modern Forth textbook -- Forth Inc's,
4747: @c seems to positively avoid going into too much detail for some of
4748: @c the internals.
4749:
4750: The text interpreter@footnote{This is an expanded version of the
4751: material in @ref{Introducing the Text Interpreter}.} is an endless loop
4752: that processes input from the current input device. It is also called
4753: the outer interpreter, in contrast to the inner interpreter
4754: (@pxref{Engine}) which executes the compiled Forth code on interpretive
4755: implementations.
4756:
4757: @cindex interpret state
4758: @cindex compile state
4759: The text interpreter operates in one of two states: @dfn{interpret
4760: state} and @dfn{compile state}. The current state is defined by the
4761: aptly-named variable, @code{state}.
4762:
4763: This section starts by describing how the text interpreter behaves when
4764: it is in interpret state, processing input from the user input device --
4765: the keyboard. This is the mode that a Forth system is in after it starts
4766: up.
4767:
4768: @cindex input buffer
4769: @cindex terminal input buffer
4770: The text interpreter works from an area of memory called the @dfn{input
4771: buffer}@footnote{When the text interpreter is processing input from the
4772: keyboard, this area of memory is called the @dfn{terminal input buffer}
4773: (TIB) and is addressed by the (obsolescent) words @code{TIB} and
4774: @code{#TIB}.}, which stores your keyboard input when you press the
4775: @key{RET} key. Starting at the beginning of the input buffer, it skips
4776: leading spaces (called @dfn{delimiters}) then parses a string (a
4777: sequence of non-space characters) until it reaches either a space
4778: character or the end of the buffer. Having parsed a string, it makes two
4779: attempts to process it:
4780:
4781: @cindex dictionary
4782: @itemize @bullet
4783: @item
4784: It looks for the string in a @dfn{dictionary} of definitions. If the
4785: string is found, the string names a @dfn{definition} (also known as a
4786: @dfn{word}) and the dictionary search returns information that allows
4787: the text interpreter to perform the word's @dfn{interpretation
4788: semantics}. In most cases, this simply means that the word will be
4789: executed.
4790: @item
4791: If the string is not found in the dictionary, the text interpreter
4792: attempts to treat it as a number, using the rules described in
4793: @ref{Number Conversion}. If the string represents a legal number in the
4794: current radix, the number is pushed onto a parameter stack (the data
4795: stack for integers, the floating-point stack for floating-point
4796: numbers).
4797: @end itemize
4798:
4799: If both attempts fail, or if the word is found in the dictionary but has
4800: no interpretation semantics@footnote{This happens if the word was
4801: defined as @code{COMPILE-ONLY}.} the text interpreter discards the
4802: remainder of the input buffer, issues an error message and waits for
4803: more input. If one of the attempts succeeds, the text interpreter
4804: repeats the parsing process until the whole of the input buffer has been
4805: processed, at which point it prints the status message ``@code{ ok}''
4806: and waits for more input.
4807:
4808: @cindex parse area
4809: The text interpreter keeps track of its position in the input buffer by
4810: updating a variable called @code{>IN} (pronounced ``to-in''). The value
4811: of @code{>IN} starts out as 0, indicating an offset of 0 from the start
4812: of the input buffer. The region from offset @code{>IN @@} to the end of
4813: the input buffer is called the @dfn{parse area}@footnote{In other words,
4814: the text interpreter processes the contents of the input buffer by
4815: parsing strings from the parse area until the parse area is empty.}.
4816: This example shows how @code{>IN} changes as the text interpreter parses
4817: the input buffer:
4818:
4819: @example
4820: : remaining >IN @@ SOURCE 2 PICK - -ROT + SWAP
4821: CR ." ->" TYPE ." <-" ; IMMEDIATE
4822:
4823: 1 2 3 remaining + remaining .
4824:
4825: : foo 1 2 3 remaining SWAP remaining ;
4826: @end example
4827:
4828: @noindent
4829: The result is:
4830:
4831: @example
4832: ->+ remaining .<-
4833: ->.<-5 ok
4834:
4835: ->SWAP remaining ;-<
4836: ->;<- ok
4837: @end example
4838:
4839: @cindex parsing words
4840: The value of @code{>IN} can also be modified by a word in the input
4841: buffer that is executed by the text interpreter. This means that a word
4842: can ``trick'' the text interpreter into either skipping a section of the
4843: input buffer@footnote{This is how parsing words work.} or into parsing a
4844: section twice. For example:
4845:
4846: @example
4847: : lat ." <<lat>>" ;
4848: : flat ." <<flat>>" >IN DUP @@ 3 - SWAP ! ;
4849: @end example
4850:
4851: @noindent
4852: When @code{flat} is executed, this output is produced@footnote{Exercise
4853: for the reader: what would happen if the @code{3} were replaced with
4854: @code{4}?}:
4855:
4856: @example
4857: <<flat>><<lat>>
4858: @end example
4859:
4860: @noindent
4861: Two important notes about the behaviour of the text interpreter:
4862:
4863: @itemize @bullet
4864: @item
4865: It processes each input string to completion before parsing additional
4866: characters from the input buffer.
4867: @item
4868: It treats the input buffer as a read-only region (and so must your code).
4869: @end itemize
4870:
4871: @noindent
4872: When the text interpreter is in compile state, its behaviour changes in
4873: these ways:
4874:
4875: @itemize @bullet
4876: @item
4877: If a parsed string is found in the dictionary, the text interpreter will
4878: perform the word's @dfn{compilation semantics}. In most cases, this
4879: simply means that the execution semantics of the word will be appended
4880: to the current definition.
4881: @item
4882: When a number is encountered, it is compiled into the current definition
4883: (as a literal) rather than being pushed onto a parameter stack.
4884: @item
4885: If an error occurs, @code{state} is modified to put the text interpreter
4886: back into interpret state.
4887: @item
4888: Each time a line is entered from the keyboard, Gforth prints
4889: ``@code{ compiled}'' rather than `` @code{ok}''.
4890: @end itemize
4891:
4892: @cindex text interpreter - input sources
4893: When the text interpreter is using an input device other than the
4894: keyboard, its behaviour changes in these ways:
4895:
4896: @itemize @bullet
4897: @item
4898: When the parse area is empty, the text interpreter attempts to refill
4899: the input buffer from the input source. When the input source is
4900: exhausted, the input source is set back to the user input device.
4901: @item
4902: It doesn't print out ``@code{ ok}'' or ``@code{ compiled}'' messages each
4903: time the parse area is emptied.
4904: @item
4905: If an error occurs, the input source is set back to the user input
4906: device.
4907: @end itemize
4908:
4909: @ref{Input Sources} describes this in more detail.
4910:
4911:
4912: doc->in
4913: doc-source
4914:
4915: doc-tib
4916: doc-#tib
4917:
4918:
4919: @menu
4920: * Input Sources::
4921: * Number Conversion::
4922: * Interpret/Compile states::
4923: * Literals::
4924: * Interpreter Directives::
4925: @end menu
4926:
4927: @node Input Sources, Number Conversion, The Text Interpreter, The Text Interpreter
4928: @subsection Input Sources
4929: @cindex input sources
4930: @cindex text interpreter - input sources
4931:
4932: By default, the text interpreter processes input from the user input
4933: device (the keyboard) when Forth starts up. The text interpreter can
4934: process input from any of these sources:
4935:
4936: @itemize @bullet
4937: @item
4938: The user input device -- the keyboard.
4939: @item
4940: A file, using the words described in @ref{Forth source files}.
4941: @item
4942: A block, using the words described in @ref{Blocks}.
4943: @item
4944: A text string, using @code{evaluate}.
4945: @end itemize
4946:
4947: A program can identify the current input device from the values of
4948: @code{source-id} and @code{blk}.
4949:
4950:
4951: doc-source-id
4952: doc-blk
4953:
4954: doc-save-input
4955: doc-restore-input
4956:
4957: doc-evaluate
4958:
4959:
4960:
4961: @node Number Conversion, Interpret/Compile states, Input Sources, The Text Interpreter
4962: @subsection Number Conversion
4963: @cindex number conversion
4964: @cindex double-cell numbers, input format
4965: @cindex input format for double-cell numbers
4966: @cindex single-cell numbers, input format
4967: @cindex input format for single-cell numbers
4968: @cindex floating-point numbers, input format
4969: @cindex input format for floating-point numbers
4970:
4971: This section describes the rules that the text interpreter uses when it
4972: tries to convert a string into a number.
4973:
4974: Let <digit> represent any character that is a legal digit in the current
4975: number base@footnote{For example, 0-9 when the number base is decimal or
4976: 0-9, A-F when the number base is hexadecimal.}.
4977:
4978: Let <decimal digit> represent any character in the range 0-9.
4979:
4980: Let @{@i{a b}@} represent the @i{optional} presence of any of the characters
4981: in the braces (@i{a} or @i{b} or neither).
4982:
4983: Let * represent any number of instances of the previous character
4984: (including none).
4985:
4986: Let any other character represent itself.
4987:
4988: @noindent
4989: Now, the conversion rules are:
4990:
4991: @itemize @bullet
4992: @item
4993: A string of the form <digit><digit>* is treated as a single-precision
4994: (cell-sized) positive integer. Examples are 0 123 6784532 32343212343456 42
4995: @item
4996: A string of the form -<digit><digit>* is treated as a single-precision
4997: (cell-sized) negative integer, and is represented using 2's-complement
4998: arithmetic. Examples are -45 -5681 -0
4999: @item
5000: A string of the form <digit><digit>*.<digit>* is treated as a double-precision
5001: (double-cell-sized) positive integer. Examples are 3465. 3.465 34.65
5002: (all three of these represent the same number).
5003: @item
5004: A string of the form -<digit><digit>*.<digit>* is treated as a
5005: double-precision (double-cell-sized) negative integer, and is
5006: represented using 2's-complement arithmetic. Examples are -3465. -3.465
5007: -34.65 (all three of these represent the same number).
5008: @item
5009: A string of the form @{+ -@}<decimal digit>@{.@}<decimal digit>*@{e
5010: E@}@{+ -@}<decimal digit><decimal digit>* is treated as a floating-point
5011: number. Examples are 1e 1e0 1.e 1.e0 +1e+0 (which all represent the same
5012: number) +12.E-4
5013: @end itemize
5014:
5015: By default, the number base used for integer number conversion is given
5016: by the contents of the variable @code{base}. Note that a lot of
5017: confusion can result from unexpected values of @code{base}. If you
5018: change @code{base} anywhere, make sure to save the old value and restore
5019: it afterwards. In general I recommend keeping @code{base} decimal, and
5020: using the prefixes described below for the popular non-decimal bases.
5021:
5022: doc-dpl
5023: doc-base
5024: doc-hex
5025: doc-decimal
5026:
5027:
5028: @cindex '-prefix for character strings
5029: @cindex &-prefix for decimal numbers
5030: @cindex %-prefix for binary numbers
5031: @cindex $-prefix for hexadecimal numbers
5032: Gforth allows you to override the value of @code{base} by using a
5033: prefix@footnote{Some Forth implementations provide a similar scheme by
5034: implementing @code{$} etc. as parsing words that process the subsequent
5035: number in the input stream and push it onto the stack. For example, see
5036: @cite{Number Conversion and Literals}, by Wil Baden; Forth Dimensions
5037: 20(3) pages 26--27. In such implementations, unlike in Gforth, a space
5038: is required between the prefix and the number.} before the first digit
5039: of an (integer) number. Four prefixes are supported:
5040:
5041: @itemize @bullet
5042: @item
5043: @code{&} -- decimal
5044: @item
5045: @code{%} -- binary
5046: @item
5047: @code{$} -- hexadecimal
5048: @item
5049: @code{'} -- base @code{max-char+1}
5050: @end itemize
5051:
5052: Here are some examples, with the equivalent decimal number shown after
5053: in braces:
5054:
5055: -$41 (-65), %1001101 (205), %1001.0001 (145 - a double-precision number),
5056: 'AB (16706; ascii A is 65, ascii B is 66, number is 65*256 + 66),
5057: 'ab (24930; ascii a is 97, ascii B is 98, number is 97*256 + 98),
5058: &905 (905), $abc (2478), $ABC (2478).
5059:
5060: @cindex number conversion - traps for the unwary
5061: @noindent
5062: Number conversion has a number of traps for the unwary:
5063:
5064: @itemize @bullet
5065: @item
5066: You cannot determine the current number base using the code sequence
5067: @code{base @@ .} -- the number base is always 10 in the current number
5068: base. Instead, use something like @code{base @@ dec.}
5069: @item
5070: If the number base is set to a value greater than 14 (for example,
5071: hexadecimal), the number 123E4 is ambiguous; the conversion rules allow
5072: it to be intepreted as either a single-precision integer or a
5073: floating-point number (Gforth treats it as an integer). The ambiguity
5074: can be resolved by explicitly stating the sign of the mantissa and/or
5075: exponent: 123E+4 or +123E4 -- if the number base is decimal, no
5076: ambiguity arises; either representation will be treated as a
5077: floating-point number.
5078: @item
5079: There is a word @code{bin} but it does @i{not} set the number base!
5080: It is used to specify file types.
5081: @item
5082: ANS Forth requires the @code{.} of a double-precision number to
5083: be the final character in the string. Allowing the @code{.} to be
5084: anywhere after the first digit is a Gforth extension.
5085: @item
5086: The number conversion process does not check for overflow.
5087: @item
5088: In Gforth, number conversion to floating-point numbers always use base
5089: 10, irrespective of the value of @code{base}. In ANS Forth,
5090: conversion to floating-point numbers whilst the value of
5091: @code{base} is not 10 is an ambiguous condition.
5092: @end itemize
5093:
5094: @ref{Input} describes words that you can use to read numbers into your
5095: programs.
5096:
5097: @node Interpret/Compile states, Literals, Number Conversion, The Text Interpreter
5098: @subsection Interpret/Compile states
5099: @cindex Interpret/Compile states
5100:
5101: A standard program is not permitted to change @code{state}
5102: explicitly. However, it can change @code{state} implicitly, using the
5103: words @code{[} and @code{]}. When @code{[} is executed it switches
5104: @code{state} to interpret state, and therefore the text interpreter
5105: starts interpreting. When @code{]} is executed it switches @code{state}
5106: to compile state and therefore the text interpreter starts
5107: compiling. The most common usage for these words is for switching into
5108: interpret state and back from within a colon definition; this technique
5109: can be used to compile a literal (@pxref{Literals} for an example) or
5110: for conditional compilation (@pxref{Interpreter Directives} for an
5111: example).
5112:
5113:
5114: @c This is a bad example: It's non-standard, and it's not necessary.
5115: @c However, I can't think of a good example for switching into compile
5116: @c state when there is no current word (@code{state}-smart words are not a
5117: @c good reason). So maybe we should use an example for switching into
5118: @c interpret @code{state} in a colon def. - anton
5119: @c nac-> I agree. I started out by putting in the example, then realised
5120: @c that it was non-ANS, so wrote more words around it. I hope this
5121: @c re-written version is acceptable to you. I do want to keep the example
5122: @c as it is helpful for showing what is and what is not portable, particularly
5123: @c where it outlaws a style in common use.
5124:
5125:
5126: @code{[} and @code{]} also give you the ability to switch into compile
5127: state and back, but we cannot think of any useful Standard application
5128: for this ability. Pre-ANS Forth textbooks have examples like this:
5129:
5130: @example
5131: : AA ." this is A" ;
5132: : BB ." this is B" ;
5133: : CC ." this is C" ;
5134:
5135: create table ] aa bb cc [
5136:
5137: : go ( n -- ) \ n is offset into table.. 0 for 1st entry
5138: cells table + @ execute ;
5139: @end example
5140:
5141: This example builds a jump table; @code{0 go} will display ``@code{this
5142: is A}''. Using @code{[} and @code{]} in this example is equivalent to
5143: defining @code{table} like this:
5144:
5145: @example
5146: create table ' aa COMPILE, ' bb COMPILE, ' cc COMPILE,
5147: @end example
5148:
5149: The problem with this code is that the definition of @code{table} is not
5150: portable -- it @i{compile}s execution tokens into code space. Whilst it
5151: @i{may} work on systems where code space and data space co-incide, the
5152: Standard only allows data space to be assigned for a @code{CREATE}d
5153: word. In addition, the Standard only allows @code{@@} to access data
5154: space, whilst this example is using it to access code space. The only
5155: portable, Standard way to build this table is to build it in data space,
5156: like this:
5157:
5158: @example
5159: create table ' aa , ' bb , ' cc ,
5160: @end example
5161:
5162: doc-state
5163: doc-[
5164: doc-]
5165:
5166:
5167: @node Literals, Interpreter Directives, Interpret/Compile states, The Text Interpreter
5168: @subsection Literals
5169: @cindex Literals
5170:
5171: Often, you want to use a number within a colon definition. When you do
5172: this, the text interpreter automatically compiles the number as a
5173: @i{literal}. A literal is a number whose run-time effect is to be pushed
5174: onto the stack. If you had to do some maths to generate the number, you
5175: might write it like this:
5176:
5177: @example
5178: : HOUR-TO-SEC ( n1 -- n2 )
5179: 60 * \ to minutes
5180: 60 * ; \ to seconds
5181: @end example
5182:
5183: It is very clear what this definition is doing, but it's inefficient
5184: since it is performing 2 multiples at run-time. An alternative would be
5185: to write:
5186:
5187: @example
5188: : HOUR-TO-SEC ( n1 -- n2 )
5189: 3600 * ; \ to seconds
5190: @end example
5191:
5192: Which does the same thing, and has the advantage of using a single
5193: multiply. Ideally, we'd like the efficiency of the second with the
5194: readability of the first.
5195:
5196: @code{Literal} allows us to achieve that. It takes a number from the
5197: stack and lays it down in the current definition just as though the
5198: number had been typed directly into the definition. Our first attempt
5199: might look like this:
5200:
5201: @example
5202: 60 \ mins per hour
5203: 60 * \ seconds per minute
5204: : HOUR-TO-SEC ( n1 -- n2 )
5205: Literal * ; \ to seconds
5206: @end example
5207:
5208: But this produces the error message @code{unstructured}. What happened?
5209: The stack notation for @code{:} is (@i{ -- colon-sys}) and the size of
5210: @i{colon-sys} is implementation-defined. In other words, once we start a
5211: colon definition we can't portably access anything that was on the stack
5212: before the definition began@footnote{@cite{Two Problems in ANS Forth},
5213: by Thomas Worthington; Forth Dimensions 20(2) pages 32--34 describes
5214: some situations where you might want to access stack items above
5215: colon-sys, and provides a solution to the problem.}. The correct way of
5216: solving this problem in this instance is to use @code{[ ]} like this:
5217:
5218: @example
5219: : HOUR-TO-SEC ( n1 -- n2 )
5220: [ 60 \ minutes per hour
5221: 60 * ] \ seconds per minute
5222: LITERAL * ; \ to seconds
5223: @end example
5224:
5225:
5226: doc-literal
5227: doc-]L
5228: doc-2literal
5229: doc-fliteral
5230:
5231:
5232: @node Interpreter Directives, , Literals, The Text Interpreter
5233: @subsection Interpreter Directives
5234: @cindex interpreter directives
5235:
5236: These words are usually used in interpret state; typically to control
5237: which parts of a source file are processed by the text
5238: interpreter. There are only a few ANS Forth Standard words, but Gforth
5239: supplements these with a rich set of immediate control structure words
5240: to compensate for the fact that the non-immediate versions can only be
5241: used in compile state (@pxref{Control Structures}). Typical usages:
5242:
5243: @example
5244: FALSE Constant ASSEMBLER
5245: .
5246: .
5247: ASSEMBLER [IF]
5248: : ASSEMBLER-FEATURE
5249: ...
5250: ;
5251: [ENDIF]
5252: .
5253: .
5254: : SEE
5255: ... \ general-purpose SEE code
5256: [ ASSEMBLER [IF] ]
5257: ... \ assembler-specific SEE code
5258: [ [ENDIF] ]
5259: ;
5260: @end example
5261:
5262:
5263: doc-[IF]
5264: doc-[ELSE]
5265: doc-[THEN]
5266: doc-[ENDIF]
5267:
5268: doc-[IFDEF]
5269: doc-[IFUNDEF]
5270:
5271: doc-[?DO]
5272: doc-[DO]
5273: doc-[FOR]
5274: doc-[LOOP]
5275: doc-[+LOOP]
5276: doc-[NEXT]
5277:
5278: doc-[BEGIN]
5279: doc-[UNTIL]
5280: doc-[AGAIN]
5281: doc-[WHILE]
5282: doc-[REPEAT]
5283:
5284:
5285:
5286:
5287: @c -------------------------------------------------------------
5288: @node Tokens for Words, Word Lists, The Text Interpreter, Words
5289: @section Tokens for Words
5290: @cindex tokens for words
5291:
5292: This section describes the creation and use of tokens that represent
5293: words.
5294:
5295: Named words have information stored in their header space entries to
5296: indicate any non-default semantics (@pxref{Interpretation and
5297: Compilation Semantics}). The semantics can be modified, using
5298: @code{immediate} and/or @code{compile-only}, at the time that the words
5299: are defined. Unnamed words have (by definition) no header space
5300: entry, and therefore must have default semantics.
5301:
5302: Named words have interpretation and compilation semantics. Unnamed words
5303: just have execution semantics.
5304:
5305: @cindex xt
5306: @cindex execution token
5307: The execution semantics of an unnamed word are represented by an
5308: @dfn{execution token} (@i{xt}). As explained in @ref{Supplying names},
5309: the execution token of the last word defined can be produced with
5310: @code{lastxt}.
5311:
5312: The interpretation semantics of a named word are also represented by an
5313: execution token. You can produce the execution token using @code{'} or
5314: @code{[']}. A simple example shows the difference between the two:
5315:
5316: @example
5317: : greet ( -- ) ." Hello" ;
5318: : foo ( -- xt ) ['] greet execute ; \ ['] parses greet at compile-time
5319: : bar ( -- ) ' execute ; \ ' parses at run-time
5320:
5321: \ the next four lines all do the same thing
5322: foo
5323: bar greet
5324: greet
5325: ' greet EXECUTE
5326: @end example
5327:
5328: An execution token occupies one cell.
5329: @cindex code field address
5330: @cindex CFA
5331: In Gforth, the abstract data type @i{execution token} is implemented
5332: as a code field address (CFA).
5333: @comment TODO note that the standard does not say what it represents..
5334: @comment and you cannot necessarily compile it in all Forths (eg native
5335: @comment compilers?).
5336:
5337: For literals, use @code{'} in interpreted code and @code{[']} in
5338: compiled code. Gforth's @code{'} and @code{[']} behave somewhat
5339: unusually by complaining about compile-only words. To get the execution
5340: token for a compile-only word @i{name}, use @code{COMP' @i{name} DROP}
5341: or @code{[COMP'] @i{name} DROP}.
5342:
5343: @cindex compilation token
5344: The compilation semantics of a named word are represented by a
5345: @dfn{compilation token} consisting of two cells: @i{w xt}. The top cell
5346: @i{xt} is an execution token. The compilation semantics represented by
5347: the compilation token can be performed with @code{execute}, which
5348: consumes the whole compilation token, with an additional stack effect
5349: determined by the represented compilation semantics.
5350:
5351: At present, the @i{w} part of a compilation token is an execution token,
5352: and the @i{xt} part represents either @code{execute} or
5353: @code{compile,}@footnote{Depending upon the compilation semantics of the
5354: word. If the word has default compilation semantics, the @i{xt} will
5355: represent @code{compile,}. Otherwise (e.g., for immediate words), the
5356: @i{xt} will represent @code{execute}.}. However, don't rely on that
5357: knowledge, unless necessary; future versions of Gforth may introduce
5358: unusual compilation tokens (e.g., a compilation token that represents
5359: the compilation semantics of a literal).
5360:
5361: You can compile the compilation semantics with @code{postpone,}. I.e.,
5362: @code{COMP' @i{word} postpone,} is equivalent to @code{postpone
5363: @i{word}}.
5364:
5365: @cindex name token
5366: @cindex name field address
5367: @cindex NFA
5368: Named words are also represented by the @dfn{name token}, (@i{nt}). In
5369: Gforth, the abstract data type @emph{name token} is implemented as a
5370: name field address (NFA).
5371:
5372:
5373: doc-execute
5374: doc-compile,
5375: doc-[']
5376: doc-'
5377: doc-[comp']
5378: doc-comp'
5379: doc-postpone,
5380:
5381: doc-find-name
5382: doc-name>int
5383: doc-name?int
5384: doc-name>comp
5385: doc-name>string
5386:
5387:
5388: @c -------------------------------------------------------------
5389: @node Word Lists, Environmental Queries, Tokens for Words, Words
5390: @section Word Lists
5391: @cindex word lists
5392: @cindex header space
5393:
5394: A wordlist is a list of named words; you can add new words and look up
5395: words by name (and you can remove words in a restricted way with
5396: markers). Every named (and @code{reveal}ed) word is in one wordlist.
5397:
5398: @cindex search order stack
5399: The text interpreter searches the wordlists present in the search order
5400: (a stack of wordlists), from the top to the bottom. Within each
5401: wordlist, the search starts conceptually at the newest word; i.e., if
5402: two words in a wordlist have the same name, the newer word is found.
5403:
5404: @cindex compilation word list
5405: New words are added to the @dfn{compilation wordlist} (aka current
5406: wordlist).
5407:
5408: @cindex wid
5409: A word list is identified by a cell-sized word list identifier (@i{wid})
5410: in much the same way as a file is identified by a file handle. The
5411: numerical value of the wid has no (portable) meaning, and might change
5412: from session to session.
5413:
5414: The ANS Forth ``Search order'' word set is intended to provide a set of
5415: low-level tools that allow various different schemes to be
5416: implemented. Gforth provides @code{vocabulary}, a traditional Forth
5417: word. @file{compat/vocabulary.fs} provides an implementation in ANS
5418: Standard Forth.
5419:
5420: @comment TODO: locals section refers to here, saying that every word list (aka
5421: @comment vocabulary) has its own methods for searching etc. Need to document that.
5422:
5423: @comment the thisone- prefix is used to pick out the true definition of a
5424: @comment word from the source files, rather than some alias.
5425:
5426: doc-forth-wordlist
5427: doc-definitions
5428: doc-get-current
5429: doc-set-current
5430: doc-get-order
5431: doc---thisone-set-order
5432: doc-wordlist
5433: doc-table
5434: doc-push-order
5435: doc-previous
5436: doc-also
5437: doc---thisone-forth
5438: doc-only
5439: doc---thisone-order
5440:
5441: doc-find
5442: doc-search-wordlist
5443:
5444: doc-words
5445: doc-vlist
5446: @c doc-words-deferred
5447:
5448: doc-mappedwordlist
5449: doc-root
5450: doc-vocabulary
5451: doc-seal
5452: doc-vocs
5453: doc-current
5454: doc-context
5455:
5456:
5457: @menu
5458: * Why use word lists?::
5459: * Word list examples::
5460: @end menu
5461:
5462: @node Why use word lists?, Word list examples, Word Lists, Word Lists
5463: @subsection Why use word lists?
5464: @cindex word lists - why use them?
5465:
5466: Here are some reasons for using multiple word lists:
5467:
5468: @itemize @bullet
5469: @item
5470: To improve compilation speed by reducing the number of header space
5471: entries that must be searched. This is achieved by creating a new
5472: word list that contains all of the definitions that are used in the
5473: definition of a Forth system but which would not usually be used by
5474: programs running on that system. That word list would be on the search
5475: list when the Forth system was compiled but would be removed from the
5476: search list for normal operation. This can be a useful technique for
5477: low-performance systems (for example, 8-bit processors in embedded
5478: systems) but is unlikely to be necessary in high-performance desktop
5479: systems.
5480: @item
5481: To prevent a set of words from being used outside the context in which
5482: they are valid. Two classic examples of this are an integrated editor
5483: (all of the edit commands are defined in a separate word list; the
5484: search order is set to the editor word list when the editor is invoked;
5485: the old search order is restored when the editor is terminated) and an
5486: integrated assembler (the op-codes for the machine are defined in a
5487: separate word list which is used when a @code{CODE} word is defined).
5488: @item
5489: To prevent a name-space clash between multiple definitions with the same
5490: name. For example, when building a cross-compiler you might have a word
5491: @code{IF} that generates conditional code for your target system. By
5492: placing this definition in a different word list you can control whether
5493: the host system's @code{IF} or the target system's @code{IF} get used in
5494: any particular context by controlling the order of the word lists on the
5495: search order stack.
5496: @end itemize
5497:
5498: @node Word list examples, ,Why use word lists?, Word Lists
5499: @subsection Word list examples
5500: @cindex word lists - examples
5501:
5502: Here is an example of creating and using a new wordlist using ANS
5503: Forth Standard words:
5504:
5505: @example
5506: wordlist constant my-new-words-wordlist
5507: : my-new-words get-order nip my-new-words-wordlist swap set-order ;
5508:
5509: \ add it to the search order
5510: also my-new-words
5511:
5512: \ alternatively, add it to the search order and make it
5513: \ the compilation word list
5514: also my-new-words definitions
5515: \ type "order" to see the problem
5516: @end example
5517:
5518: The problem with this example is that @code{order} has no way to
5519: associate the name @code{my-new-words} with the wid of the word list (in
5520: Gforth, @code{order} and @code{vocs} will display @code{???} for a wid
5521: that has no associated name). There is no Standard way of associating a
5522: name with a wid.
5523:
5524: In Gforth, this example can be re-coded using @code{vocabulary}, which
5525: associates a name with a wid:
5526:
5527: @example
5528: vocabulary my-new-words
5529:
5530: \ add it to the search order
5531: my-new-words
5532:
5533: \ alternatively, add it to the search order and make it
5534: \ the compilation word list
5535: my-new-words definitions
5536: \ type "order" to see that the problem is solved
5537: @end example
5538:
5539: @c -------------------------------------------------------------
5540: @node Environmental Queries, Files, Word Lists, Words
5541: @section Environmental Queries
5542: @cindex environmental queries
5543:
5544: ANS Forth introduced the idea of ``environmental queries'' as a way
5545: for a program running on a system to determine certain characteristics of the system.
5546: The Standard specifies a number of strings that might be recognised by a system.
5547:
5548: The Standard requires that the header space used for environmental queries
5549: be distinct from the header space used for definitions.
5550:
5551: Typically, environmental queries are supported by creating a set of
5552: definitions in a word list that is @i{only} used during environmental
5553: queries; that is what Gforth does. There is no Standard way of adding
5554: definitions to the set of recognised environmental queries, but any
5555: implementation that supports the loading of optional word sets must have
5556: some mechanism for doing this (after loading the word set, the
5557: associated environmental query string must return @code{true}). In
5558: Gforth, the word list used to honour environmental queries can be
5559: manipulated just like any other word list.
5560:
5561:
5562: doc-environment?
5563: doc-environment-wordlist
5564:
5565: doc-gforth
5566: doc-os-class
5567:
5568:
5569: Note that, whilst the documentation for (e.g.) @code{gforth} shows it
5570: returning two items on the stack, querying it using @code{environment?}
5571: will return an additional item; the @code{true} flag that shows that the
5572: string was recognised.
5573:
5574: @comment TODO Document the standard strings or note where they are documented herein
5575:
5576: Here are some examples of using environmental queries:
5577:
5578: @example
5579: s" address-unit-bits" environment? 0=
5580: [IF]
5581: cr .( environmental attribute address-units-bits unknown... ) cr
5582: [THEN]
5583:
5584: s" block" environment? [IF] DROP include block.fs [THEN]
5585:
5586: s" gforth" environment? [IF] 2DROP include compat/vocabulary.fs [THEN]
5587:
5588: s" gforth" environment? [IF] .( Gforth version ) TYPE
5589: [ELSE] .( Not Gforth..) [THEN]
5590: @end example
5591:
5592:
5593: Here is an example of adding a definition to the environment word list:
5594:
5595: @example
5596: get-current environment-wordlist set-current
5597: true constant block
5598: true constant block-ext
5599: set-current
5600: @end example
5601:
5602: You can see what definitions are in the environment word list like this:
5603:
5604: @example
5605: get-order 1+ environment-wordlist swap set-order words previous
5606: @end example
5607:
5608:
5609: @c -------------------------------------------------------------
5610: @node Files, Blocks, Environmental Queries, Words
5611: @section Files
5612: @cindex files
5613: @cindex I/O - file-handling
5614:
5615: Gforth provides facilities for accessing files that are stored in the
5616: host operating system's file-system. Files that are processed by Gforth
5617: can be divided into two categories:
5618:
5619: @itemize @bullet
5620: @item
5621: Files that are processed by the Text Interpreter (@dfn{Forth source files}).
5622: @item
5623: Files that are processed by some other program (@dfn{general files}).
5624: @end itemize
5625:
5626: @menu
5627: * Forth source files::
5628: * General files::
5629: * Search Paths::
5630: * Forth Search Paths::
5631: * General Search Paths::
5632: @end menu
5633:
5634:
5635: @c -------------------------------------------------------------
5636: @node Forth source files, General files, Files, Files
5637: @subsection Forth source files
5638: @cindex including files
5639: @cindex Forth source files
5640:
5641: The simplest way to interpret the contents of a file is to use one of
5642: these two formats:
5643:
5644: @example
5645: include mysource.fs
5646: s" mysource.fs" included
5647: @end example
5648:
5649: Sometimes you want to include a file only if it is not included already
5650: (by, say, another source file). In that case, you can use one of these
5651: fomats:
5652:
5653: @example
5654: require mysource.fs
5655: needs mysource.fs
5656: s" mysource.fs" required
5657: @end example
5658:
5659: @cindex stack effect of included files
5660: @cindex including files, stack effect
5661: I recommend that you write your source files such that interpreting them
5662: does not change the stack. This allows using these files with
5663: @code{required} and friends without complications. For example:
5664:
5665: @example
5666: 1 require foo.fs drop
5667: @end example
5668:
5669:
5670: doc-include-file
5671: doc-included
5672: doc-included?
5673: doc-include
5674: doc-required
5675: doc-require
5676: doc-needs
5677: doc-init-included-files
5678:
5679:
5680: A definition in ANS Forth for @code{required} is provided in
5681: @file{compat/required.fs}.
5682:
5683: @c -------------------------------------------------------------
5684: @node General files, Search Paths, Forth source files, Files
5685: @subsection General files
5686: @cindex general files
5687: @cindex file-handling
5688:
5689: Files are opened/created by name and type. The following types are
5690: recognised:
5691:
5692:
5693: doc-r/o
5694: doc-r/w
5695: doc-w/o
5696: doc-bin
5697:
5698:
5699: When a file is opened/created, it returns a file identifier,
5700: @i{wfileid} that is used for all other file commands. All file
5701: commands also return a status value, @i{wior}, that is 0 for a
5702: successful operation and an implementation-defined non-zero value in the
5703: case of an error.
5704:
5705:
5706: doc-open-file
5707: doc-create-file
5708:
5709: doc-close-file
5710: doc-delete-file
5711: doc-rename-file
5712: doc-read-file
5713: doc-read-line
5714: doc-write-file
5715: doc-write-line
5716: doc-emit-file
5717: doc-flush-file
5718:
5719: doc-file-status
5720: doc-file-position
5721: doc-reposition-file
5722: doc-file-size
5723: doc-resize-file
5724:
5725:
5726: @c ---------------------------------------------------------
5727: @node Search Paths, Forth Search Paths, General files, Files
5728: @subsection Search Paths
5729: @cindex path for @code{included}
5730: @cindex file search path
5731: @cindex @code{include} search path
5732: @cindex search path for files
5733:
5734: If you specify an absolute filename (i.e., a filename starting with
5735: @file{/} or @file{~}, or with @file{:} in the second position (as in
5736: @samp{C:...})) for @code{included} and friends, that file is included
5737: just as you would expect.
5738:
5739: For relative filenames, Gforth uses a search path similar to Forth's
5740: search order (@pxref{Word Lists}). It tries to find the given filename
5741: in the directories present in the path, and includes the first one it
5742: finds. There are separate search paths for Forth source files and
5743: general files.
5744:
5745: If the search path contains the directory @file{.} (as it should), this
5746: refers to the directory that the present file was @code{included}
5747: from. This allows files to include other files relative to their own
5748: position (irrespective of the current working directory or the absolute
5749: position). This feature is essential for libraries consisting of
5750: several files, where a file may include other files from the library.
5751: It corresponds to @code{#include "..."} in C. If the current input
5752: source is not a file, @file{.} refers to the directory of the innermost
5753: file being included, or, if there is no file being included, to the
5754: current working directory.
5755:
5756: Use @file{~+} to refer to the current working directory (as in the
5757: @code{bash}).
5758:
5759: If the filename starts with @file{./}, the search path is not searched
5760: (just as with absolute filenames), and the @file{.} has the same meaning
5761: as described above.
5762:
5763: @c ---------------------------------------------------------
5764: @node Forth Search Paths, General Search Paths, Search Paths, Files
5765: @subsubsection Forth Search Paths
5766: @cindex search path control - Forth
5767:
5768: The search path is initialized when you start Gforth (@pxref{Invoking
5769: Gforth}). You can display it and change it using these words:
5770:
5771:
5772: doc-.fpath
5773: doc-fpath+
5774: doc-fpath=
5775: doc-open-fpath-file
5776:
5777:
5778: @noindent
5779: Here is an example of using @code{fpath} and @code{require}:
5780:
5781: @example
5782: fpath= /usr/lib/forth/|./
5783: require timer.fs
5784: @end example
5785:
5786: @c ---------------------------------------------------------
5787: @node General Search Paths, , Forth Search Paths, Files
5788: @subsubsection General Search Paths
5789: @cindex search path control - for user applications
5790:
5791: Your application may need to search files in several directories, like
5792: @code{included} does. To facilitate this, Gforth allows you to define
5793: and use your own search paths, by providing generic equivalents of the
5794: Forth search path words:
5795:
5796:
5797: doc-.path
5798: doc-path+
5799: doc-path=
5800: doc-open-path-file
5801:
5802:
5803: Here's an example of creating a search path:
5804:
5805: @example
5806: \ Make a buffer for the path:
5807: create mypath 100 chars , \ maximum length (is checked)
5808: 0 , \ real len
5809: 100 chars allot \ space for path
5810: @end example
5811:
5812: @c -------------------------------------------------------------
5813: @node Blocks, Other I/O, Files, Words
5814: @section Blocks
5815: @cindex I/O - blocks
5816: @cindex blocks
5817:
5818: When you run Gforth on a modern desk-top computer, it runs under the
5819: control of an operating system which provides certain services. One of
5820: these services is @var{file services}, which allows Forth source code
5821: and data to be stored in files and read into Gforth (@pxref{Files}).
5822:
5823: Traditionally, Forth has been an important programming language on
5824: systems where it has interfaced directly to the underlying hardware with
5825: no intervening operating system. Forth provides a mechanism, called
5826: @dfn{blocks}, for accessing mass storage on such systems.
5827:
5828: A block is a 1024-byte data area, which can be used to hold data or
5829: Forth source code. No structure is imposed on the contents of the
5830: block. A block is identified by its number; blocks are numbered
5831: contiguously from 1 to an implementation-defined maximum.
5832:
5833: A typical system that used blocks but no operating system might use a
5834: single floppy-disk drive for mass storage, with the disks formatted to
5835: provide 256-byte sectors. Blocks would be implemented by assigning the
5836: first four sectors of the disk to block 1, the second four sectors to
5837: block 2 and so on, up to the limit of the capacity of the disk. The disk
5838: would not contain any file system information, just the set of blocks.
5839:
5840: @cindex blocks file
5841: On systems that do provide file services, blocks are typically
5842: implemented by storing a sequence of blocks within a single @dfn{blocks
5843: file}. The size of the blocks file will be an exact multiple of 1024
5844: bytes, corresponding to the number of blocks it contains. This is the
5845: mechanism that Gforth uses.
5846:
5847: @cindex @file{blocks.fb}
5848: Only 1 blocks file can be open at a time. If you use block words without
5849: having specified a blocks file, Gforth defaults to the blocks file
5850: @file{blocks.fb}. Gforth uses the Forth search path when attempting to
5851: locate a blocks file (@pxref{Forth Search Paths}).
5852:
5853: @cindex block buffers
5854: When you read and write blocks under program control, Gforth uses a
5855: number of @dfn{block buffers} as intermediate storage. These buffers are
5856: not used when you use @code{load} to interpret the contents of a block.
5857:
5858: The behaviour of the block buffers is directly analagous to that of a
5859: cache. Each block buffer has three states:
5860:
5861: @itemize @bullet
5862: @item
5863: Unassigned
5864: @item
5865: Assigned-clean
5866: @item
5867: Assigned-dirty
5868: @end itemize
5869:
5870: Initially, all block buffers are @i{unassigned}. In order to access a
5871: block, the block (specified by its block number) must be assigned to a
5872: block buffer.
5873:
5874: The assignment of a block to a block buffer is performed by @code{block}
5875: or @code{buffer}. Use @code{block} when you wish to modify the existing
5876: contents of a block. Use @code{buffer} when you don't care about the
5877: existing contents of the block@footnote{The ANS Forth definition of
5878: @code{buffer} is intended not to cause disk I/O; if the data associated
5879: with the particular block is already stored in a block buffer due to an
5880: earlier @code{block} command, @code{buffer} will return that block
5881: buffer and the existing contents of the block will be
5882: available. Otherwise, @code{buffer} will simply assign a new, empty
5883: block buffer for the block.}.
5884:
5885: Once a block has been assigned to a block buffer, the block buffer state
5886: becomes @i{assigned-clean}. Data can now be manipulated within the
5887: block buffer.
5888:
5889: When the contents of a block buffer is changed it is necessary,
5890: @i{before calling} @code{block} @i{or} @code{buffer} @i{again}, to
5891: either abandon the changes (by doing nothing) or commit the changes,
5892: using @code{update}. Using @code{update} does not change the blocks
5893: file; it simply changes a block buffer's state to @i{assigned-dirty}.
5894:
5895: The word @code{flush} causes all @i{assigned-dirty} blocks to be
5896: written back to the blocks file on disk. Leaving Gforth using @code{bye}
5897: also causes a @code{flush} to be performed.
5898:
5899: In Gforth, @code{block} and @code{buffer} use a @i{direct-mapped}
5900: algorithm to assign a block buffer to a block. That means that any
5901: particular block can only be assigned to one specific block buffer,
5902: called (for the particular operation) the @i{victim buffer}. If the
5903: victim buffer is @i{unassigned} or @i{assigned-clean} it can be
5904: allocated to the new block immediately. If it is @i{assigned-dirty}
5905: its current contents must be written out to disk before it can be
5906: allocated to the new block.
5907:
5908: Although no structure is imposed on the contents of a block, it is
5909: traditional to display the contents as 16 lines each of 64 characters. A
5910: block provides a single, continuous stream of input (for example, it
5911: acts as a single parse area) -- there are no end-of-line characters
5912: within a block, and no end-of-file character at the end of a
5913: block. There are two consequences of this:
5914:
5915: @itemize @bullet
5916: @item
5917: The last character of one line wraps straight into the first character
5918: of the following line
5919: @item
5920: The word @code{\} -- comment to end of line -- requires special
5921: treatment; in the context of a block it causes all characters until the
5922: end of the current 64-character ``line'' to be ignored.
5923: @end itemize
5924:
5925: In Gforth, when you use @code{block} with a non-existent block number,
5926: the current block file will be extended to the appropriate size and the
5927: block buffer will be initialised with spaces.
5928:
5929: Gforth doesn't encourage the use of blocks; the mechanism is only
5930: provided for backward compatibility -- ANS Forth requires blocks to be
5931: available when files are.
5932:
5933: Common techniques that are used when working with blocks include:
5934:
5935: @itemize @bullet
5936: @item
5937: A screen editor that allows you to edit blocks without leaving the Forth
5938: environment.
5939: @item
5940: Shadow screens; where every code block has an associated block
5941: containing comments (for example: code in odd block numbers, comments in
5942: even block numbers). Typically, the block editor provides a convenient
5943: mechanism to toggle between code and comments.
5944: @item
5945: Load blocks; a single block (typically block 1) contains a number of
5946: @code{thru} commands which @code{load} the whole of the application.
5947: @end itemize
5948:
5949: See Frank Sergeant's Pygmy Forth to see just how well blocks can be
5950: integrated into a Forth programming environment.
5951:
5952: @comment TODO what about errors on open-blocks?
5953:
5954: doc-open-blocks
5955: doc-use
5956: doc-get-block-fid
5957: doc-block-position
5958:
5959: doc-scr
5960: doc-list
5961:
5962: doc---block-block
5963: doc-buffer
5964:
5965: doc-update
5966: doc-updated?
5967: doc-save-buffers
5968: doc-empty-buffers
5969: doc-empty-buffer
5970: doc-flush
5971:
5972: doc-load
5973: doc-thru
5974: doc-+load
5975: doc-+thru
5976: doc---gforth--->
5977: doc-block-included
5978:
5979:
5980: @c -------------------------------------------------------------
5981: @node Other I/O, Programming Tools, Blocks, Words
5982: @section Other I/O
5983: @cindex I/O - keyboard and display
5984:
5985: @menu
5986: * Simple numeric output:: Predefined formats
5987: * Formatted numeric output:: Formatted (pictured) output
5988: * String Formats:: How Forth stores strings in memory
5989: * Displaying characters and strings:: Other stuff
5990: * Input:: Input
5991: @end menu
5992:
5993: @node Simple numeric output, Formatted numeric output, Other I/O, Other I/O
5994: @subsection Simple numeric output
5995: @cindex numeric output - simple/free-format
5996:
5997: The simplest output functions are those that display numbers from the
5998: data or floating-point stacks. Floating-point output is always displayed
5999: using base 10. Numbers displayed from the data stack use the value stored
6000: in @code{base}.
6001:
6002:
6003: doc-.
6004: doc-dec.
6005: doc-hex.
6006: doc-u.
6007: doc-.r
6008: doc-u.r
6009: doc-d.
6010: doc-ud.
6011: doc-d.r
6012: doc-ud.r
6013: doc-f.
6014: doc-fe.
6015: doc-fs.
6016:
6017:
6018: Examples of printing the number 1234.5678E23 in the different floating-point output
6019: formats are shown below:
6020:
6021: @example
6022: f. 123456779999999000000000000.
6023: fe. 123.456779999999E24
6024: fs. 1.23456779999999E26
6025: @end example
6026:
6027:
6028: @node Formatted numeric output, String Formats, Simple numeric output, Other I/O
6029: @subsection Formatted numeric output
6030: @cindex formatted numeric output
6031: @cindex pictured numeric output
6032: @cindex numeric output - formatted
6033:
6034: Forth traditionally uses a technique called @dfn{pictured numeric
6035: output} for formatted printing of integers. In this technique, digits
6036: are extracted from the number (using the current output radix defined by
6037: @code{base}), converted to ASCII codes and appended to a string that is
6038: built in a scratch-pad area of memory (@pxref{core-idef,
6039: Implementation-defined options, Implementation-defined
6040: options}). Arbitrary characters can be appended to the string during the
6041: extraction process. The completed string is specified by an address
6042: and length and can be manipulated (@code{TYPE}ed, copied, modified)
6043: under program control.
6044:
6045: All of the words described in the previous section for simple numeric
6046: output are implemented in Gforth using pictured numeric output.
6047:
6048: Three important things to remember about Pictured Numeric Output:
6049:
6050: @itemize @bullet
6051: @item
6052: It always operates on double-precision numbers; to display a
6053: single-precision number, convert it first (@pxref{Double precision} for
6054: ways of doing this).
6055: @item
6056: It always treats the double-precision number as though it were
6057: unsigned. The examples below show ways of printing signed numbers.
6058: @item
6059: The string is built up from right to left; least significant digit first.
6060: @end itemize
6061:
6062:
6063: doc-<#
6064: doc-#
6065: doc-#s
6066: doc-hold
6067: doc-sign
6068: doc-#>
6069:
6070: doc-represent
6071:
6072:
6073: @noindent
6074: Here are some examples of using pictured numeric output:
6075:
6076: @example
6077: : my-u. ( u -- )
6078: \ Simplest use of pns.. behaves like Standard u.
6079: 0 \ convert to unsigned double
6080: <# \ start conversion
6081: #s \ convert all digits
6082: #> \ complete conversion
6083: TYPE SPACE ; \ display, with trailing space
6084:
6085: : cents-only ( u -- )
6086: 0 \ convert to unsigned double
6087: <# \ start conversion
6088: # # \ convert two least-significant digits
6089: #> \ complete conversion, discard other digits
6090: TYPE SPACE ; \ display, with trailing space
6091:
6092: : dollars-and-cents ( u -- )
6093: 0 \ convert to unsigned double
6094: <# \ start conversion
6095: # # \ convert two least-significant digits
6096: [char] . hold \ insert decimal point
6097: #s \ convert remaining digits
6098: [char] $ hold \ append currency symbol
6099: #> \ complete conversion
6100: TYPE SPACE ; \ display, with trailing space
6101:
6102: : my-. ( n -- )
6103: \ handling negatives.. behaves like Standard .
6104: s>d \ convert to signed double
6105: swap over dabs \ leave sign byte followed by unsigned double
6106: <# \ start conversion
6107: #s \ convert all digits
6108: rot sign \ get at sign byte, append "-" if needed
6109: #> \ complete conversion
6110: TYPE SPACE ; \ display, with trailing space
6111:
6112: : account. ( n -- )
6113: \ accountants don't like minus signs, they use braces
6114: \ for negative numbers
6115: s>d \ convert to signed double
6116: swap over dabs \ leave sign byte followed by unsigned double
6117: <# \ start conversion
6118: 2 pick \ get copy of sign byte
6119: 0< IF [char] ) hold THEN \ right-most character of output
6120: #s \ convert all digits
6121: rot \ get at sign byte
6122: 0< IF [char] ( hold THEN
6123: #> \ complete conversion
6124: TYPE SPACE ; \ display, with trailing space
6125: @end example
6126:
6127: Here are some examples of using these words:
6128:
6129: @example
6130: 1 my-u. 1
6131: hex -1 my-u. decimal FFFFFFFF
6132: 1 cents-only 01
6133: 1234 cents-only 34
6134: 2 dollars-and-cents $0.02
6135: 1234 dollars-and-cents $12.34
6136: 123 my-. 123
6137: -123 my. -123
6138: 123 account. 123
6139: -456 account. (456)
6140: @end example
6141:
6142:
6143: @node String Formats, Displaying characters and strings, Formatted numeric output, Other I/O
6144: @subsection String Formats
6145: @cindex strings - see character strings
6146: @cindex character strings - formats
6147: @cindex I/O - see character strings
6148:
6149: Forth commonly uses two different methods for representing character
6150: strings:
6151:
6152: @itemize @bullet
6153: @item
6154: @cindex address of counted string
6155: As a @dfn{counted string}, represented by a @i{c-addr}. The char
6156: addressed by @i{c-addr} contains a character-count, @i{n}, of the
6157: string and the string occupies the subsequent @i{n} char addresses in
6158: memory.
6159: @item
6160: As cell pair on the stack; @i{c-addr u}, where @i{u} is the length
6161: of the string in characters, and @i{c-addr} is the address of the
6162: first byte of the string.
6163: @end itemize
6164:
6165: ANS Forth encourages the use of the second format when representing
6166: strings on the stack, whilst conceeding that the counted string format
6167: remains useful as a way of storing strings in memory.
6168:
6169:
6170: doc-count
6171:
6172:
6173: @xref{Memory Blocks} for words that move, copy and search
6174: for strings. @xref{Displaying characters and strings,} for words that
6175: display characters and strings.
6176:
6177:
6178: @node Displaying characters and strings, Input, String Formats, Other I/O
6179: @subsection Displaying characters and strings
6180: @cindex characters - compiling and displaying
6181: @cindex character strings - compiling and displaying
6182:
6183: This section starts with a glossary of Forth words and ends with a set
6184: of examples.
6185:
6186:
6187: doc-bl
6188: doc-space
6189: doc-spaces
6190: doc-emit
6191: doc-toupper
6192: doc-."
6193: doc-.(
6194: doc-type
6195: doc-typewhite
6196: doc-cr
6197: @cindex cursor control
6198: doc-at-xy
6199: doc-page
6200: doc-s"
6201: doc-c"
6202: doc-char
6203: doc-[char]
6204: doc-sliteral
6205:
6206:
6207: @noindent
6208: As an example, consider the following text, stored in a file @file{test.fs}:
6209:
6210: @example
6211: .( text-1)
6212: : my-word
6213: ." text-2" cr
6214: .( text-3)
6215: ;
6216:
6217: ." text-4"
6218:
6219: : my-char
6220: [char] ALPHABET emit
6221: char emit
6222: ;
6223: @end example
6224:
6225: When you load this code into Gforth, the following output is generated:
6226:
6227: @example
6228: @kbd{include test.fs @key{RET}} text-1text-3text-4 ok
6229: @end example
6230:
6231: @itemize @bullet
6232: @item
6233: Messages @code{text-1} and @code{text-3} are displayed because @code{.(}
6234: is an immediate word; it behaves in the same way whether it is used inside
6235: or outside a colon definition.
6236: @item
6237: Message @code{text-4} is displayed because of Gforth's added interpretation
6238: semantics for @code{."}.
6239: @item
6240: Message @code{text-2} is @i{not} displayed, because the text interpreter
6241: performs the compilation semantics for @code{."} within the definition of
6242: @code{my-word}.
6243: @end itemize
6244:
6245: Here are some examples of executing @code{my-word} and @code{my-char}:
6246:
6247: @example
6248: @kbd{my-word @key{RET}} text-2
6249: ok
6250: @kbd{my-char fred @key{RET}} Af ok
6251: @kbd{my-char jim @key{RET}} Aj ok
6252: @end example
6253:
6254: @itemize @bullet
6255: @item
6256: Message @code{text-2} is displayed because of the run-time behaviour of
6257: @code{."}.
6258: @item
6259: @code{[char]} compiles the ``A'' from ``ALPHABET'' and puts its display code
6260: on the stack at run-time. @code{emit} always displays the character
6261: when @code{my-char} is executed.
6262: @item
6263: @code{char} parses a string at run-time and the second @code{emit} displays
6264: the first character of the string.
6265: @item
6266: If you type @code{see my-char} you can see that @code{[char]} discarded
6267: the text ``LPHABET'' and only compiled the display code for ``A'' into the
6268: definition of @code{my-char}.
6269: @end itemize
6270:
6271:
6272:
6273: @node Input, , Displaying characters and strings, Other I/O
6274: @subsection Input
6275: @cindex input
6276: @cindex I/O - see input
6277: @cindex parsing a string
6278:
6279: @xref{String Formats} for ways of storing character strings in memory.
6280:
6281: @comment TODO examples for >number >float accept key key? pad parse word refill
6282: @comment then index them
6283:
6284:
6285: doc-key
6286: doc-key?
6287: doc->number
6288: doc->float
6289: doc-accept
6290: doc-pad
6291: doc-parse
6292: doc-word
6293: doc-sword
6294: doc-(name)
6295: doc-refill
6296: @comment obsolescent words..
6297: doc-convert
6298: doc-query
6299: doc-expect
6300: doc-span
6301:
6302:
6303:
6304: @c -------------------------------------------------------------
6305: @node Programming Tools, Assembler and Code Words, Other I/O, Words
6306: @section Programming Tools
6307: @cindex programming tools
6308:
6309: @menu
6310: * Debugging:: Simple and quick.
6311: * Assertions:: Making your programs self-checking.
6312: * Singlestep Debugger:: Executing your program word by word.
6313: @end menu
6314:
6315: @node Debugging, Assertions, Programming Tools, Programming Tools
6316: @subsection Debugging
6317: @cindex debugging
6318:
6319: Languages with a slow edit/compile/link/test development loop tend to
6320: require sophisticated tracing/stepping debuggers to facilate
6321: productive debugging.
6322:
6323: A much better (faster) way in fast-compiling languages is to add
6324: printing code at well-selected places, let the program run, look at
6325: the output, see where things went wrong, add more printing code, etc.,
6326: until the bug is found.
6327:
6328: The simple debugging aids provided in @file{debugs.fs}
6329: are meant to support this style of debugging. In addition, there are
6330: words for non-destructively inspecting the stack and memory:
6331:
6332:
6333: doc-.s
6334: doc-f.s
6335:
6336:
6337: There is a word @code{.r} but it does @i{not} display the return
6338: stack! It is used for formatted numeric output.
6339:
6340:
6341: doc-depth
6342: doc-fdepth
6343: doc-clearstack
6344: doc-?
6345: doc-dump
6346:
6347:
6348: The word @code{~~} prints debugging information (by default the source
6349: location and the stack contents). It is easy to insert. If you use Emacs
6350: it is also easy to remove (@kbd{C-x ~} in the Emacs Forth mode to
6351: query-replace them with nothing). The deferred words
6352: @code{printdebugdata} and @code{printdebugline} control the output of
6353: @code{~~}. The default source location output format works well with
6354: Emacs' compilation mode, so you can step through the program at the
6355: source level using @kbd{C-x `} (the advantage over a stepping debugger
6356: is that you can step in any direction and you know where the crash has
6357: happened or where the strange data has occurred).
6358:
6359: The default actions of @code{~~} clobber the contents of the pictured
6360: numeric output string, so you should not use @code{~~}, e.g., between
6361: @code{<#} and @code{#>}.
6362:
6363:
6364: doc-~~
6365: doc-printdebugdata
6366: doc-printdebugline
6367:
6368: doc-see
6369: doc-marker
6370:
6371:
6372: Here's an example of using @code{marker} at the start of a source file
6373: that you are debugging; it ensures that you only ever have one copy of
6374: the file's definitions compiled at any time:
6375:
6376: @example
6377: [IFDEF] my-code
6378: my-code
6379: [ENDIF]
6380:
6381: marker my-code
6382: init-included-files
6383:
6384: \ .. definitions start here
6385: \ .
6386: \ .
6387: \ end
6388: @end example
6389:
6390:
6391:
6392: @node Assertions, Singlestep Debugger, Debugging, Programming Tools
6393: @subsection Assertions
6394: @cindex assertions
6395:
6396: It is a good idea to make your programs self-checking, especially if you
6397: make an assumption that may become invalid during maintenance (for
6398: example, that a certain field of a data structure is never zero). Gforth
6399: supports @dfn{assertions} for this purpose. They are used like this:
6400:
6401: @example
6402: assert( @i{flag} )
6403: @end example
6404:
6405: The code between @code{assert(} and @code{)} should compute a flag, that
6406: should be true if everything is alright and false otherwise. It should
6407: not change anything else on the stack. The overall stack effect of the
6408: assertion is @code{( -- )}. E.g.
6409:
6410: @example
6411: assert( 1 1 + 2 = ) \ what we learn in school
6412: assert( dup 0<> ) \ assert that the top of stack is not zero
6413: assert( false ) \ this code should not be reached
6414: @end example
6415:
6416: The need for assertions is different at different times. During
6417: debugging, we want more checking, in production we sometimes care more
6418: for speed. Therefore, assertions can be turned off, i.e., the assertion
6419: becomes a comment. Depending on the importance of an assertion and the
6420: time it takes to check it, you may want to turn off some assertions and
6421: keep others turned on. Gforth provides several levels of assertions for
6422: this purpose:
6423:
6424:
6425: doc-assert0(
6426: doc-assert1(
6427: doc-assert2(
6428: doc-assert3(
6429: doc-assert(
6430: doc-)
6431:
6432:
6433: The variable @code{assert-level} specifies the highest assertions that
6434: are turned on. I.e., at the default @code{assert-level} of one,
6435: @code{assert0(} and @code{assert1(} assertions perform checking, while
6436: @code{assert2(} and @code{assert3(} assertions are treated as comments.
6437:
6438: The value of @code{assert-level} is evaluated at compile-time, not at
6439: run-time. Therefore you cannot turn assertions on or off at run-time;
6440: you have to set the @code{assert-level} appropriately before compiling a
6441: piece of code. You can compile different pieces of code at different
6442: @code{assert-level}s (e.g., a trusted library at level 1 and
6443: newly-written code at level 3).
6444:
6445:
6446: doc-assert-level
6447:
6448:
6449: If an assertion fails, a message compatible with Emacs' compilation mode
6450: is produced and the execution is aborted (currently with @code{ABORT"}.
6451: If there is interest, we will introduce a special throw code. But if you
6452: intend to @code{catch} a specific condition, using @code{throw} is
6453: probably more appropriate than an assertion).
6454:
6455: Definitions in ANS Forth for these assertion words are provided
6456: in @file{compat/assert.fs}.
6457:
6458:
6459: @node Singlestep Debugger, , Assertions, Programming Tools
6460: @subsection Singlestep Debugger
6461: @cindex singlestep Debugger
6462: @cindex debugging Singlestep
6463: @cindex @code{dbg}
6464: @cindex @code{BREAK:}
6465: @cindex @code{BREAK"}
6466:
6467: When you create a new word there's often the need to check whether it
6468: behaves correctly or not. You can do this by typing @code{dbg
6469: badword}. A debug session might look like this:
6470:
6471: @example
6472: : badword 0 DO i . LOOP ; ok
6473: 2 dbg badword
6474: : badword
6475: Scanning code...
6476:
6477: Nesting debugger ready!
6478:
6479: 400D4738 8049BC4 0 -> [ 2 ] 00002 00000
6480: 400D4740 8049F68 DO -> [ 0 ]
6481: 400D4744 804A0C8 i -> [ 1 ] 00000
6482: 400D4748 400C5E60 . -> 0 [ 0 ]
6483: 400D474C 8049D0C LOOP -> [ 0 ]
6484: 400D4744 804A0C8 i -> [ 1 ] 00001
6485: 400D4748 400C5E60 . -> 1 [ 0 ]
6486: 400D474C 8049D0C LOOP -> [ 0 ]
6487: 400D4758 804B384 ; -> ok
6488: @end example
6489:
6490: Each line displayed is one step. You always have to hit return to
6491: execute the next word that is displayed. If you don't want to execute
6492: the next word in a whole, you have to type @kbd{n} for @code{nest}. Here is
6493: an overview what keys are available:
6494:
6495: @table @i
6496:
6497: @item @key{RET}
6498: Next; Execute the next word.
6499:
6500: @item n
6501: Nest; Single step through next word.
6502:
6503: @item u
6504: Unnest; Stop debugging and execute rest of word. If we got to this word
6505: with nest, continue debugging with the calling word.
6506:
6507: @item d
6508: Done; Stop debugging and execute rest.
6509:
6510: @item s
6511: Stop; Abort immediately.
6512:
6513: @end table
6514:
6515: Debugging large application with this mechanism is very difficult, because
6516: you have to nest very deeply into the program before the interesting part
6517: begins. This takes a lot of time.
6518:
6519: To do it more directly put a @code{BREAK:} command into your source code.
6520: When program execution reaches @code{BREAK:} the single step debugger is
6521: invoked and you have all the features described above.
6522:
6523: If you have more than one part to debug it is useful to know where the
6524: program has stopped at the moment. You can do this by the
6525: @code{BREAK" string"} command. This behaves like @code{BREAK:} except that
6526: string is typed out when the ``breakpoint'' is reached.
6527:
6528:
6529: doc-dbg
6530: doc-BREAK:
6531: doc-BREAK"
6532:
6533:
6534:
6535: @c -------------------------------------------------------------
6536: @node Assembler and Code Words, Threading Words, Programming Tools, Words
6537: @section Assembler and Code Words
6538: @cindex assembler
6539: @cindex code words
6540:
6541: Gforth provides some words for defining primitives (words written in
6542: machine code), and for defining the machine-code equivalent of
6543: @code{DOES>}-based defining words. However, the machine-independent
6544: nature of Gforth poses a few problems: First of all, Gforth runs on
6545: several architectures, so it can provide no standard assembler. What's
6546: worse is that the register allocation not only depends on the processor,
6547: but also on the @code{gcc} version and options used.
6548:
6549: The words that Gforth offers encapsulate some system dependences (e.g.,
6550: the header structure), so a system-independent assembler may be used in
6551: Gforth. If you do not have an assembler, you can compile machine code
6552: directly with @code{,} and @code{c,}@footnote{This isn't portable,
6553: because these words emit stuff in @i{data} space; it works because
6554: Gforth has unified code/data spaces. Assembler isn't likely to be
6555: portable anyway.}.
6556:
6557:
6558: doc-assembler
6559: doc-code
6560: doc-end-code
6561: doc-;code
6562: doc-flush-icache
6563:
6564:
6565: If @code{flush-icache} does not work correctly, @code{code} words
6566: etc. will not work (reliably), either.
6567:
6568: The typical usage of these @code{code} words can be shown most easily by
6569: analogy to the equivalent high-level defining words:
6570:
6571: @example
6572: : foo code foo
6573: <high-level Forth words> <assembler>
6574: ; end-code
6575:
6576: : bar : bar
6577: <high-level Forth words> <high-level Forth words>
6578: CREATE CREATE
6579: <high-level Forth words> <high-level Forth words>
6580: DOES> ;code
6581: <high-level Forth words> <assembler>
6582: ; end-code
6583: @end example
6584:
6585: @code{flush-icache} is always present. The other words are rarely used
6586: and reside in @code{code.fs}, which is usually not loaded. You can load
6587: it with @code{require code.fs}.
6588:
6589: @cindex registers of the inner interpreter
6590: In the assembly code you will want to refer to the inner interpreter's
6591: registers (e.g., the data stack pointer) and you may want to use other
6592: registers for temporary storage. Unfortunately, the register allocation
6593: is installation-dependent.
6594:
6595: The easiest solution is to use explicit register declarations
6596: (@pxref{Explicit Reg Vars, , Variables in Specified Registers, gcc.info,
6597: GNU C Manual}) for all of the inner interpreter's registers: You have to
6598: compile Gforth with @code{-DFORCE_REG} (configure option
6599: @code{--enable-force-reg}) and the appropriate declarations must be
6600: present in the @code{machine.h} file (see @code{mips.h} for an example;
6601: you can find a full list of all declarable register symbols with
6602: @code{grep register engine.c}). If you give explicit registers to all
6603: variables that are declared at the beginning of @code{engine()}, you
6604: should be able to use the other caller-saved registers for temporary
6605: storage. Alternatively, you can use the @code{gcc} option
6606: @code{-ffixed-REG} (@pxref{Code Gen Options, , Options for Code
6607: Generation Conventions, gcc.info, GNU C Manual}) to reserve a register
6608: (however, this restriction on register allocation may slow Gforth
6609: significantly).
6610:
6611: If this solution is not viable (e.g., because @code{gcc} does not allow
6612: you to explicitly declare all the registers you need), you have to find
6613: out by looking at the code where the inner interpreter's registers
6614: reside and which registers can be used for temporary storage. You can
6615: get an assembly listing of the engine's code with @code{make engine.s}.
6616:
6617: In any case, it is good practice to abstract your assembly code from the
6618: actual register allocation. E.g., if the data stack pointer resides in
6619: register @code{$17}, create an alias for this register called @code{sp},
6620: and use that in your assembly code.
6621:
6622: @cindex code words, portable
6623: Another option for implementing normal and defining words efficiently
6624: is to add the desired functionality to the source of Gforth. For normal
6625: words you just have to edit @file{primitives} (@pxref{Automatic
6626: Generation}). Defining words (equivalent to @code{;CODE} words, for fast
6627: defined words) may require changes in @file{engine.c}, @file{kernel.fs},
6628: @file{prims2x.fs}, and possibly @file{cross.fs}.
6629:
6630:
6631: @c -------------------------------------------------------------
6632: @node Threading Words, Locals, Assembler and Code Words, Words
6633: @section Threading Words
6634: @cindex threading words
6635:
6636: @cindex code address
6637: These words provide access to code addresses and other threading stuff
6638: in Gforth (and, possibly, other interpretive Forths). It more or less
6639: abstracts away the differences between direct and indirect threading
6640: (and, for direct threading, the machine dependences). However, at
6641: present this wordset is still incomplete. It is also pretty low-level;
6642: some day it will hopefully be made unnecessary by an internals wordset
6643: that abstracts implementation details away completely.
6644:
6645:
6646: doc-threading-method
6647: doc->code-address
6648: doc->does-code
6649: doc-code-address!
6650: doc-does-code!
6651: doc-does-handler!
6652: doc-/does-handler
6653:
6654:
6655: The code addresses produced by various defining words are produced by
6656: the following words:
6657:
6658:
6659: doc-docol:
6660: doc-docon:
6661: doc-dovar:
6662: doc-douser:
6663: doc-dodefer:
6664: doc-dofield:
6665:
6666:
6667: You can recognize words defined by a @code{CREATE}...@code{DOES>} word
6668: with @code{>does-code}. If the word was defined in that way, the value
6669: returned is non-zero and identifies the @code{DOES>} used by the
6670: defining word.
6671: @comment TODO should that be ``identifies the xt of the DOES> ??''
6672:
6673: @c -------------------------------------------------------------
6674: @node Locals, Structures, Threading Words, Words
6675: @section Locals
6676: @cindex locals
6677:
6678: Local variables can make Forth programming more enjoyable and Forth
6679: programs easier to read. Unfortunately, the locals of ANS Forth are
6680: laden with restrictions. Therefore, we provide not only the ANS Forth
6681: locals wordset, but also our own, more powerful locals wordset (we
6682: implemented the ANS Forth locals wordset through our locals wordset).
6683:
6684: The ideas in this section have also been published in the paper
6685: @cite{Automatic Scoping of Local Variables} by M. Anton Ertl, presented
6686: at EuroForth '94; it is available at
6687: @*@url{http://www.complang.tuwien.ac.at/papers/ertl94l.ps.gz}.
6688:
6689: @menu
6690: * Gforth locals::
6691: * ANS Forth locals::
6692: @end menu
6693:
6694: @node Gforth locals, ANS Forth locals, Locals, Locals
6695: @subsection Gforth locals
6696: @cindex Gforth locals
6697: @cindex locals, Gforth style
6698:
6699: Locals can be defined with
6700:
6701: @example
6702: @{ local1 local2 ... -- comment @}
6703: @end example
6704: or
6705: @example
6706: @{ local1 local2 ... @}
6707: @end example
6708:
6709: E.g.,
6710: @example
6711: : max @{ n1 n2 -- n3 @}
6712: n1 n2 > if
6713: n1
6714: else
6715: n2
6716: endif ;
6717: @end example
6718:
6719: The similarity of locals definitions with stack comments is intended. A
6720: locals definition often replaces the stack comment of a word. The order
6721: of the locals corresponds to the order in a stack comment and everything
6722: after the @code{--} is really a comment.
6723:
6724: This similarity has one disadvantage: It is too easy to confuse locals
6725: declarations with stack comments, causing bugs and making them hard to
6726: find. However, this problem can be avoided by appropriate coding
6727: conventions: Do not use both notations in the same program. If you do,
6728: they should be distinguished using additional means, e.g. by position.
6729:
6730: @cindex types of locals
6731: @cindex locals types
6732: The name of the local may be preceded by a type specifier, e.g.,
6733: @code{F:} for a floating point value:
6734:
6735: @example
6736: : CX* @{ F: Ar F: Ai F: Br F: Bi -- Cr Ci @}
6737: \ complex multiplication
6738: Ar Br f* Ai Bi f* f-
6739: Ar Bi f* Ai Br f* f+ ;
6740: @end example
6741:
6742: @cindex flavours of locals
6743: @cindex locals flavours
6744: @cindex value-flavoured locals
6745: @cindex variable-flavoured locals
6746: Gforth currently supports cells (@code{W:}, @code{W^}), doubles
6747: (@code{D:}, @code{D^}), floats (@code{F:}, @code{F^}) and characters
6748: (@code{C:}, @code{C^}) in two flavours: a value-flavoured local (defined
6749: with @code{W:}, @code{D:} etc.) produces its value and can be changed
6750: with @code{TO}. A variable-flavoured local (defined with @code{W^} etc.)
6751: produces its address (which becomes invalid when the variable's scope is
6752: left). E.g., the standard word @code{emit} can be defined in terms of
6753: @code{type} like this:
6754:
6755: @example
6756: : emit @{ C^ char* -- @}
6757: char* 1 type ;
6758: @end example
6759:
6760: @cindex default type of locals
6761: @cindex locals, default type
6762: A local without type specifier is a @code{W:} local. Both flavours of
6763: locals are initialized with values from the data or FP stack.
6764:
6765: Currently there is no way to define locals with user-defined data
6766: structures, but we are working on it.
6767:
6768: Gforth allows defining locals everywhere in a colon definition. This
6769: poses the following questions:
6770:
6771: @menu
6772: * Where are locals visible by name?::
6773: * How long do locals live?::
6774: * Programming Style::
6775: * Implementation::
6776: @end menu
6777:
6778: @node Where are locals visible by name?, How long do locals live?, Gforth locals, Gforth locals
6779: @subsubsection Where are locals visible by name?
6780: @cindex locals visibility
6781: @cindex visibility of locals
6782: @cindex scope of locals
6783:
6784: Basically, the answer is that locals are visible where you would expect
6785: it in block-structured languages, and sometimes a little longer. If you
6786: want to restrict the scope of a local, enclose its definition in
6787: @code{SCOPE}...@code{ENDSCOPE}.
6788:
6789:
6790: doc-scope
6791: doc-endscope
6792:
6793:
6794: These words behave like control structure words, so you can use them
6795: with @code{CS-PICK} and @code{CS-ROLL} to restrict the scope in
6796: arbitrary ways.
6797:
6798: If you want a more exact answer to the visibility question, here's the
6799: basic principle: A local is visible in all places that can only be
6800: reached through the definition of the local@footnote{In compiler
6801: construction terminology, all places dominated by the definition of the
6802: local.}. In other words, it is not visible in places that can be reached
6803: without going through the definition of the local. E.g., locals defined
6804: in @code{IF}...@code{ENDIF} are visible until the @code{ENDIF}, locals
6805: defined in @code{BEGIN}...@code{UNTIL} are visible after the
6806: @code{UNTIL} (until, e.g., a subsequent @code{ENDSCOPE}).
6807:
6808: The reasoning behind this solution is: We want to have the locals
6809: visible as long as it is meaningful. The user can always make the
6810: visibility shorter by using explicit scoping. In a place that can
6811: only be reached through the definition of a local, the meaning of a
6812: local name is clear. In other places it is not: How is the local
6813: initialized at the control flow path that does not contain the
6814: definition? Which local is meant, if the same name is defined twice in
6815: two independent control flow paths?
6816:
6817: This should be enough detail for nearly all users, so you can skip the
6818: rest of this section. If you really must know all the gory details and
6819: options, read on.
6820:
6821: In order to implement this rule, the compiler has to know which places
6822: are unreachable. It knows this automatically after @code{AHEAD},
6823: @code{AGAIN}, @code{EXIT} and @code{LEAVE}; in other cases (e.g., after
6824: most @code{THROW}s), you can use the word @code{UNREACHABLE} to tell the
6825: compiler that the control flow never reaches that place. If
6826: @code{UNREACHABLE} is not used where it could, the only consequence is
6827: that the visibility of some locals is more limited than the rule above
6828: says. If @code{UNREACHABLE} is used where it should not (i.e., if you
6829: lie to the compiler), buggy code will be produced.
6830:
6831:
6832: doc-unreachable
6833:
6834:
6835: Another problem with this rule is that at @code{BEGIN}, the compiler
6836: does not know which locals will be visible on the incoming
6837: back-edge. All problems discussed in the following are due to this
6838: ignorance of the compiler (we discuss the problems using @code{BEGIN}
6839: loops as examples; the discussion also applies to @code{?DO} and other
6840: loops). Perhaps the most insidious example is:
6841: @example
6842: AHEAD
6843: BEGIN
6844: x
6845: [ 1 CS-ROLL ] THEN
6846: @{ x @}
6847: ...
6848: UNTIL
6849: @end example
6850:
6851: This should be legal according to the visibility rule. The use of
6852: @code{x} can only be reached through the definition; but that appears
6853: textually below the use.
6854:
6855: From this example it is clear that the visibility rules cannot be fully
6856: implemented without major headaches. Our implementation treats common
6857: cases as advertised and the exceptions are treated in a safe way: The
6858: compiler makes a reasonable guess about the locals visible after a
6859: @code{BEGIN}; if it is too pessimistic, the
6860: user will get a spurious error about the local not being defined; if the
6861: compiler is too optimistic, it will notice this later and issue a
6862: warning. In the case above the compiler would complain about @code{x}
6863: being undefined at its use. You can see from the obscure examples in
6864: this section that it takes quite unusual control structures to get the
6865: compiler into trouble, and even then it will often do fine.
6866:
6867: If the @code{BEGIN} is reachable from above, the most optimistic guess
6868: is that all locals visible before the @code{BEGIN} will also be
6869: visible after the @code{BEGIN}. This guess is valid for all loops that
6870: are entered only through the @code{BEGIN}, in particular, for normal
6871: @code{BEGIN}...@code{WHILE}...@code{REPEAT} and
6872: @code{BEGIN}...@code{UNTIL} loops and it is implemented in our
6873: compiler. When the branch to the @code{BEGIN} is finally generated by
6874: @code{AGAIN} or @code{UNTIL}, the compiler checks the guess and
6875: warns the user if it was too optimistic:
6876: @example
6877: IF
6878: @{ x @}
6879: BEGIN
6880: \ x ?
6881: [ 1 cs-roll ] THEN
6882: ...
6883: UNTIL
6884: @end example
6885:
6886: Here, @code{x} lives only until the @code{BEGIN}, but the compiler
6887: optimistically assumes that it lives until the @code{THEN}. It notices
6888: this difference when it compiles the @code{UNTIL} and issues a
6889: warning. The user can avoid the warning, and make sure that @code{x}
6890: is not used in the wrong area by using explicit scoping:
6891: @example
6892: IF
6893: SCOPE
6894: @{ x @}
6895: ENDSCOPE
6896: BEGIN
6897: [ 1 cs-roll ] THEN
6898: ...
6899: UNTIL
6900: @end example
6901:
6902: Since the guess is optimistic, there will be no spurious error messages
6903: about undefined locals.
6904:
6905: If the @code{BEGIN} is not reachable from above (e.g., after
6906: @code{AHEAD} or @code{EXIT}), the compiler cannot even make an
6907: optimistic guess, as the locals visible after the @code{BEGIN} may be
6908: defined later. Therefore, the compiler assumes that no locals are
6909: visible after the @code{BEGIN}. However, the user can use
6910: @code{ASSUME-LIVE} to make the compiler assume that the same locals are
6911: visible at the BEGIN as at the point where the top control-flow stack
6912: item was created.
6913:
6914:
6915: doc-assume-live
6916:
6917:
6918: @noindent
6919: E.g.,
6920: @example
6921: @{ x @}
6922: AHEAD
6923: ASSUME-LIVE
6924: BEGIN
6925: x
6926: [ 1 CS-ROLL ] THEN
6927: ...
6928: UNTIL
6929: @end example
6930:
6931: Other cases where the locals are defined before the @code{BEGIN} can be
6932: handled by inserting an appropriate @code{CS-ROLL} before the
6933: @code{ASSUME-LIVE} (and changing the control-flow stack manipulation
6934: behind the @code{ASSUME-LIVE}).
6935:
6936: Cases where locals are defined after the @code{BEGIN} (but should be
6937: visible immediately after the @code{BEGIN}) can only be handled by
6938: rearranging the loop. E.g., the ``most insidious'' example above can be
6939: arranged into:
6940: @example
6941: BEGIN
6942: @{ x @}
6943: ... 0=
6944: WHILE
6945: x
6946: REPEAT
6947: @end example
6948:
6949: @node How long do locals live?, Programming Style, Where are locals visible by name?, Gforth locals
6950: @subsubsection How long do locals live?
6951: @cindex locals lifetime
6952: @cindex lifetime of locals
6953:
6954: The right answer for the lifetime question would be: A local lives at
6955: least as long as it can be accessed. For a value-flavoured local this
6956: means: until the end of its visibility. However, a variable-flavoured
6957: local could be accessed through its address far beyond its visibility
6958: scope. Ultimately, this would mean that such locals would have to be
6959: garbage collected. Since this entails un-Forth-like implementation
6960: complexities, I adopted the same cowardly solution as some other
6961: languages (e.g., C): The local lives only as long as it is visible;
6962: afterwards its address is invalid (and programs that access it
6963: afterwards are erroneous).
6964:
6965: @node Programming Style, Implementation, How long do locals live?, Gforth locals
6966: @subsubsection Programming Style
6967: @cindex locals programming style
6968: @cindex programming style, locals
6969:
6970: The freedom to define locals anywhere has the potential to change
6971: programming styles dramatically. In particular, the need to use the
6972: return stack for intermediate storage vanishes. Moreover, all stack
6973: manipulations (except @code{PICK}s and @code{ROLL}s with run-time
6974: determined arguments) can be eliminated: If the stack items are in the
6975: wrong order, just write a locals definition for all of them; then
6976: write the items in the order you want.
6977:
6978: This seems a little far-fetched and eliminating stack manipulations is
6979: unlikely to become a conscious programming objective. Still, the number
6980: of stack manipulations will be reduced dramatically if local variables
6981: are used liberally (e.g., compare @code{max} in @ref{Gforth locals} with
6982: a traditional implementation of @code{max}).
6983:
6984: This shows one potential benefit of locals: making Forth programs more
6985: readable. Of course, this benefit will only be realized if the
6986: programmers continue to honour the principle of factoring instead of
6987: using the added latitude to make the words longer.
6988:
6989: @cindex single-assignment style for locals
6990: Using @code{TO} can and should be avoided. Without @code{TO},
6991: every value-flavoured local has only a single assignment and many
6992: advantages of functional languages apply to Forth. I.e., programs are
6993: easier to analyse, to optimize and to read: It is clear from the
6994: definition what the local stands for, it does not turn into something
6995: different later.
6996:
6997: E.g., a definition using @code{TO} might look like this:
6998: @example
6999: : strcmp @{ addr1 u1 addr2 u2 -- n @}
7000: u1 u2 min 0
7001: ?do
7002: addr1 c@@ addr2 c@@ -
7003: ?dup-if
7004: unloop exit
7005: then
7006: addr1 char+ TO addr1
7007: addr2 char+ TO addr2
7008: loop
7009: u1 u2 - ;
7010: @end example
7011: Here, @code{TO} is used to update @code{addr1} and @code{addr2} at
7012: every loop iteration. @code{strcmp} is a typical example of the
7013: readability problems of using @code{TO}. When you start reading
7014: @code{strcmp}, you think that @code{addr1} refers to the start of the
7015: string. Only near the end of the loop you realize that it is something
7016: else.
7017:
7018: This can be avoided by defining two locals at the start of the loop that
7019: are initialized with the right value for the current iteration.
7020: @example
7021: : strcmp @{ addr1 u1 addr2 u2 -- n @}
7022: addr1 addr2
7023: u1 u2 min 0
7024: ?do @{ s1 s2 @}
7025: s1 c@@ s2 c@@ -
7026: ?dup-if
7027: unloop exit
7028: then
7029: s1 char+ s2 char+
7030: loop
7031: 2drop
7032: u1 u2 - ;
7033: @end example
7034: Here it is clear from the start that @code{s1} has a different value
7035: in every loop iteration.
7036:
7037: @node Implementation, , Programming Style, Gforth locals
7038: @subsubsection Implementation
7039: @cindex locals implementation
7040: @cindex implementation of locals
7041:
7042: @cindex locals stack
7043: Gforth uses an extra locals stack. The most compelling reason for
7044: this is that the return stack is not float-aligned; using an extra stack
7045: also eliminates the problems and restrictions of using the return stack
7046: as locals stack. Like the other stacks, the locals stack grows toward
7047: lower addresses. A few primitives allow an efficient implementation:
7048:
7049:
7050: doc-@local#
7051: doc-f@local#
7052: doc-laddr#
7053: doc-lp+!#
7054: doc-lp!
7055: doc->l
7056: doc-f>l
7057:
7058:
7059: In addition to these primitives, some specializations of these
7060: primitives for commonly occurring inline arguments are provided for
7061: efficiency reasons, e.g., @code{@@local0} as specialization of
7062: @code{@@local#} for the inline argument 0. The following compiling words
7063: compile the right specialized version, or the general version, as
7064: appropriate:
7065:
7066:
7067: doc-compile-@local
7068: doc-compile-f@local
7069: doc-compile-lp+!
7070:
7071:
7072: Combinations of conditional branches and @code{lp+!#} like
7073: @code{?branch-lp+!#} (the locals pointer is only changed if the branch
7074: is taken) are provided for efficiency and correctness in loops.
7075:
7076: A special area in the dictionary space is reserved for keeping the
7077: local variable names. @code{@{} switches the dictionary pointer to this
7078: area and @code{@}} switches it back and generates the locals
7079: initializing code. @code{W:} etc.@ are normal defining words. This
7080: special area is cleared at the start of every colon definition.
7081:
7082: @cindex word list for defining locals
7083: A special feature of Gforth's dictionary is used to implement the
7084: definition of locals without type specifiers: every word list (aka
7085: vocabulary) has its own methods for searching
7086: etc. (@pxref{Word Lists}). For the present purpose we defined a word list
7087: with a special search method: When it is searched for a word, it
7088: actually creates that word using @code{W:}. @code{@{} changes the search
7089: order to first search the word list containing @code{@}}, @code{W:} etc.,
7090: and then the word list for defining locals without type specifiers.
7091:
7092: The lifetime rules support a stack discipline within a colon
7093: definition: The lifetime of a local is either nested with other locals
7094: lifetimes or it does not overlap them.
7095:
7096: At @code{BEGIN}, @code{IF}, and @code{AHEAD} no code for locals stack
7097: pointer manipulation is generated. Between control structure words
7098: locals definitions can push locals onto the locals stack. @code{AGAIN}
7099: is the simplest of the other three control flow words. It has to
7100: restore the locals stack depth of the corresponding @code{BEGIN}
7101: before branching. The code looks like this:
7102: @format
7103: @code{lp+!#} current-locals-size @minus{} dest-locals-size
7104: @code{branch} <begin>
7105: @end format
7106:
7107: @code{UNTIL} is a little more complicated: If it branches back, it
7108: must adjust the stack just like @code{AGAIN}. But if it falls through,
7109: the locals stack must not be changed. The compiler generates the
7110: following code:
7111: @format
7112: @code{?branch-lp+!#} <begin> current-locals-size @minus{} dest-locals-size
7113: @end format
7114: The locals stack pointer is only adjusted if the branch is taken.
7115:
7116: @code{THEN} can produce somewhat inefficient code:
7117: @format
7118: @code{lp+!#} current-locals-size @minus{} orig-locals-size
7119: <orig target>:
7120: @code{lp+!#} orig-locals-size @minus{} new-locals-size
7121: @end format
7122: The second @code{lp+!#} adjusts the locals stack pointer from the
7123: level at the @i{orig} point to the level after the @code{THEN}. The
7124: first @code{lp+!#} adjusts the locals stack pointer from the current
7125: level to the level at the orig point, so the complete effect is an
7126: adjustment from the current level to the right level after the
7127: @code{THEN}.
7128:
7129: @cindex locals information on the control-flow stack
7130: @cindex control-flow stack items, locals information
7131: In a conventional Forth implementation a dest control-flow stack entry
7132: is just the target address and an orig entry is just the address to be
7133: patched. Our locals implementation adds a word list to every orig or dest
7134: item. It is the list of locals visible (or assumed visible) at the point
7135: described by the entry. Our implementation also adds a tag to identify
7136: the kind of entry, in particular to differentiate between live and dead
7137: (reachable and unreachable) orig entries.
7138:
7139: A few unusual operations have to be performed on locals word lists:
7140:
7141:
7142: doc-common-list
7143: doc-sub-list?
7144: doc-list-size
7145:
7146:
7147: Several features of our locals word list implementation make these
7148: operations easy to implement: The locals word lists are organised as
7149: linked lists; the tails of these lists are shared, if the lists
7150: contain some of the same locals; and the address of a name is greater
7151: than the address of the names behind it in the list.
7152:
7153: Another important implementation detail is the variable
7154: @code{dead-code}. It is used by @code{BEGIN} and @code{THEN} to
7155: determine if they can be reached directly or only through the branch
7156: that they resolve. @code{dead-code} is set by @code{UNREACHABLE},
7157: @code{AHEAD}, @code{EXIT} etc., and cleared at the start of a colon
7158: definition, by @code{BEGIN} and usually by @code{THEN}.
7159:
7160: Counted loops are similar to other loops in most respects, but
7161: @code{LEAVE} requires special attention: It performs basically the same
7162: service as @code{AHEAD}, but it does not create a control-flow stack
7163: entry. Therefore the information has to be stored elsewhere;
7164: traditionally, the information was stored in the target fields of the
7165: branches created by the @code{LEAVE}s, by organizing these fields into a
7166: linked list. Unfortunately, this clever trick does not provide enough
7167: space for storing our extended control flow information. Therefore, we
7168: introduce another stack, the leave stack. It contains the control-flow
7169: stack entries for all unresolved @code{LEAVE}s.
7170:
7171: Local names are kept until the end of the colon definition, even if
7172: they are no longer visible in any control-flow path. In a few cases
7173: this may lead to increased space needs for the locals name area, but
7174: usually less than reclaiming this space would cost in code size.
7175:
7176:
7177: @node ANS Forth locals, , Gforth locals, Locals
7178: @subsection ANS Forth locals
7179: @cindex locals, ANS Forth style
7180:
7181: The ANS Forth locals wordset does not define a syntax for locals, but
7182: words that make it possible to define various syntaxes. One of the
7183: possible syntaxes is a subset of the syntax we used in the Gforth locals
7184: wordset, i.e.:
7185:
7186: @example
7187: @{ local1 local2 ... -- comment @}
7188: @end example
7189: @noindent
7190: or
7191: @example
7192: @{ local1 local2 ... @}
7193: @end example
7194:
7195: The order of the locals corresponds to the order in a stack comment. The
7196: restrictions are:
7197:
7198: @itemize @bullet
7199: @item
7200: Locals can only be cell-sized values (no type specifiers are allowed).
7201: @item
7202: Locals can be defined only outside control structures.
7203: @item
7204: Locals can interfere with explicit usage of the return stack. For the
7205: exact (and long) rules, see the standard. If you don't use return stack
7206: accessing words in a definition using locals, you will be all right. The
7207: purpose of this rule is to make locals implementation on the return
7208: stack easier.
7209: @item
7210: The whole definition must be in one line.
7211: @end itemize
7212:
7213: Locals defined in this way behave like @code{VALUE}s
7214: (@xref{Values}). I.e., they are initialized from the stack. Using their
7215: name produces their value. Their value can be changed using @code{TO}.
7216:
7217: Since this syntax is supported by Gforth directly, you need not do
7218: anything to use it. If you want to port a program using this syntax to
7219: another ANS Forth system, use @file{compat/anslocal.fs} to implement the
7220: syntax on the other system.
7221:
7222: Note that a syntax shown in the standard, section A.13 looks
7223: similar, but is quite different in having the order of locals
7224: reversed. Beware!
7225:
7226: The ANS Forth locals wordset itself consists of a word:
7227:
7228:
7229: doc-(local)
7230:
7231:
7232: The ANS Forth locals extension wordset defines a syntax using @code{locals|}, but it is so
7233: awful that we strongly recommend not to use it. We have implemented this
7234: syntax to make porting to Gforth easy, but do not document it here. The
7235: problem with this syntax is that the locals are defined in an order
7236: reversed with respect to the standard stack comment notation, making
7237: programs harder to read, and easier to misread and miswrite. The only
7238: merit of this syntax is that it is easy to implement using the ANS Forth
7239: locals wordset.
7240:
7241:
7242: @c ----------------------------------------------------------
7243: @node Structures, Object-oriented Forth, Locals, Words
7244: @section Structures
7245: @cindex structures
7246: @cindex records
7247:
7248: This section presents the structure package that comes with Gforth. A
7249: version of the package implemented in ANS Forth is available in
7250: @file{compat/struct.fs}. This package was inspired by a posting on
7251: comp.lang.forth in 1989 (unfortunately I don't remember, by whom;
7252: possibly John Hayes). A version of this section has been published in
7253: ???. Marcel Hendrix provided helpful comments.
7254:
7255: @menu
7256: * Why explicit structure support?::
7257: * Structure Usage::
7258: * Structure Naming Convention::
7259: * Structure Implementation::
7260: * Structure Glossary::
7261: @end menu
7262:
7263: @node Why explicit structure support?, Structure Usage, Structures, Structures
7264: @subsection Why explicit structure support?
7265:
7266: @cindex address arithmetic for structures
7267: @cindex structures using address arithmetic
7268: If we want to use a structure containing several fields, we could simply
7269: reserve memory for it, and access the fields using address arithmetic
7270: (@pxref{Address arithmetic}). As an example, consider a structure with
7271: the following fields
7272:
7273: @table @code
7274: @item a
7275: is a float
7276: @item b
7277: is a cell
7278: @item c
7279: is a float
7280: @end table
7281:
7282: Given the (float-aligned) base address of the structure we get the
7283: address of the field
7284:
7285: @table @code
7286: @item a
7287: without doing anything further.
7288: @item b
7289: with @code{float+}
7290: @item c
7291: with @code{float+ cell+ faligned}
7292: @end table
7293:
7294: It is easy to see that this can become quite tiring.
7295:
7296: Moreover, it is not very readable, because seeing a
7297: @code{cell+} tells us neither which kind of structure is
7298: accessed nor what field is accessed; we have to somehow infer the kind
7299: of structure, and then look up in the documentation, which field of
7300: that structure corresponds to that offset.
7301:
7302: Finally, this kind of address arithmetic also causes maintenance
7303: troubles: If you add or delete a field somewhere in the middle of the
7304: structure, you have to find and change all computations for the fields
7305: afterwards.
7306:
7307: So, instead of using @code{cell+} and friends directly, how
7308: about storing the offsets in constants:
7309:
7310: @example
7311: 0 constant a-offset
7312: 0 float+ constant b-offset
7313: 0 float+ cell+ faligned c-offset
7314: @end example
7315:
7316: Now we can get the address of field @code{x} with @code{x-offset
7317: +}. This is much better in all respects. Of course, you still
7318: have to change all later offset definitions if you add a field. You can
7319: fix this by declaring the offsets in the following way:
7320:
7321: @example
7322: 0 constant a-offset
7323: a-offset float+ constant b-offset
7324: b-offset cell+ faligned constant c-offset
7325: @end example
7326:
7327: Since we always use the offsets with @code{+}, we could use a defining
7328: word @code{cfield} that includes the @code{+} in the action of the
7329: defined word:
7330:
7331: @example
7332: : cfield ( n "name" -- )
7333: create ,
7334: does> ( name execution: addr1 -- addr2 )
7335: @@ + ;
7336:
7337: 0 cfield a
7338: 0 a float+ cfield b
7339: 0 b cell+ faligned cfield c
7340: @end example
7341:
7342: Instead of @code{x-offset +}, we now simply write @code{x}.
7343:
7344: The structure field words now can be used quite nicely. However,
7345: their definition is still a bit cumbersome: We have to repeat the
7346: name, the information about size and alignment is distributed before
7347: and after the field definitions etc. The structure package presented
7348: here addresses these problems.
7349:
7350: @node Structure Usage, Structure Naming Convention, Why explicit structure support?, Structures
7351: @subsection Structure Usage
7352: @cindex structure usage
7353:
7354: @cindex @code{field} usage
7355: @cindex @code{struct} usage
7356: @cindex @code{end-struct} usage
7357: You can define a structure for a (data-less) linked list with:
7358: @example
7359: struct
7360: cell% field list-next
7361: end-struct list%
7362: @end example
7363:
7364: With the address of the list node on the stack, you can compute the
7365: address of the field that contains the address of the next node with
7366: @code{list-next}. E.g., you can determine the length of a list
7367: with:
7368:
7369: @example
7370: : list-length ( list -- n )
7371: \ "list" is a pointer to the first element of a linked list
7372: \ "n" is the length of the list
7373: 0 BEGIN ( list1 n1 )
7374: over
7375: WHILE ( list1 n1 )
7376: 1+ swap list-next @@ swap
7377: REPEAT
7378: nip ;
7379: @end example
7380:
7381: You can reserve memory for a list node in the dictionary with
7382: @code{list% %allot}, which leaves the address of the list node on the
7383: stack. For the equivalent allocation on the heap you can use @code{list%
7384: %alloc} (or, for an @code{allocate}-like stack effect (i.e., with ior),
7385: use @code{list% %allocate}). You can get the the size of a list
7386: node with @code{list% %size} and its alignment with @code{list%
7387: %alignment}.
7388:
7389: Note that in ANS Forth the body of a @code{create}d word is
7390: @code{aligned} but not necessarily @code{faligned};
7391: therefore, if you do a:
7392: @example
7393: create @emph{name} foo% %allot
7394: @end example
7395:
7396: @noindent
7397: then the memory alloted for @code{foo%} is
7398: guaranteed to start at the body of @code{@emph{name}} only if
7399: @code{foo%} contains only character, cell and double fields.
7400:
7401: @cindex strcutures containing structures
7402: You can include a structure @code{foo%} as a field of
7403: another structure, like this:
7404: @example
7405: struct
7406: ...
7407: foo% field ...
7408: ...
7409: end-struct ...
7410: @end example
7411:
7412: @cindex structure extension
7413: @cindex extended records
7414: Instead of starting with an empty structure, you can extend an
7415: existing structure. E.g., a plain linked list without data, as defined
7416: above, is hardly useful; You can extend it to a linked list of integers,
7417: like this:@footnote{This feature is also known as @emph{extended
7418: records}. It is the main innovation in the Oberon language; in other
7419: words, adding this feature to Modula-2 led Wirth to create a new
7420: language, write a new compiler etc. Adding this feature to Forth just
7421: required a few lines of code.}
7422:
7423: @example
7424: list%
7425: cell% field intlist-int
7426: end-struct intlist%
7427: @end example
7428:
7429: @code{intlist%} is a structure with two fields:
7430: @code{list-next} and @code{intlist-int}.
7431:
7432: @cindex structures containing arrays
7433: You can specify an array type containing @emph{n} elements of
7434: type @code{foo%} like this:
7435:
7436: @example
7437: foo% @emph{n} *
7438: @end example
7439:
7440: You can use this array type in any place where you can use a normal
7441: type, e.g., when defining a @code{field}, or with
7442: @code{%allot}.
7443:
7444: @cindex first field optimization
7445: The first field is at the base address of a structure and the word
7446: for this field (e.g., @code{list-next}) actually does not change
7447: the address on the stack. You may be tempted to leave it away in the
7448: interest of run-time and space efficiency. This is not necessary,
7449: because the structure package optimizes this case and compiling such
7450: words does not generate any code. So, in the interest of readability
7451: and maintainability you should include the word for the field when
7452: accessing the field.
7453:
7454: @node Structure Naming Convention, Structure Implementation, Structure Usage, Structures
7455: @subsection Structure Naming Convention
7456: @cindex structure naming convention
7457:
7458: The field names that come to (my) mind are often quite generic, and,
7459: if used, would cause frequent name clashes. E.g., many structures
7460: probably contain a @code{counter} field. The structure names
7461: that come to (my) mind are often also the logical choice for the names
7462: of words that create such a structure.
7463:
7464: Therefore, I have adopted the following naming conventions:
7465:
7466: @itemize @bullet
7467: @cindex field naming convention
7468: @item
7469: The names of fields are of the form
7470: @code{@emph{struct}-@emph{field}}, where
7471: @code{@emph{struct}} is the basic name of the structure, and
7472: @code{@emph{field}} is the basic name of the field. You can
7473: think of field words as converting the (address of the)
7474: structure into the (address of the) field.
7475:
7476: @cindex structure naming convention
7477: @item
7478: The names of structures are of the form
7479: @code{@emph{struct}%}, where
7480: @code{@emph{struct}} is the basic name of the structure.
7481: @end itemize
7482:
7483: This naming convention does not work that well for fields of extended
7484: structures; e.g., the integer list structure has a field
7485: @code{intlist-int}, but has @code{list-next}, not
7486: @code{intlist-next}.
7487:
7488: @node Structure Implementation, Structure Glossary, Structure Naming Convention, Structures
7489: @subsection Structure Implementation
7490: @cindex structure implementation
7491: @cindex implementation of structures
7492:
7493: The central idea in the implementation is to pass the data about the
7494: structure being built on the stack, not in some global
7495: variable. Everything else falls into place naturally once this design
7496: decision is made.
7497:
7498: The type description on the stack is of the form @emph{align
7499: size}. Keeping the size on the top-of-stack makes dealing with arrays
7500: very simple.
7501:
7502: @code{field} is a defining word that uses @code{Create}
7503: and @code{DOES>}. The body of the field contains the offset
7504: of the field, and the normal @code{DOES>} action is simply:
7505:
7506: @example
7507: @ +
7508: @end example
7509:
7510: @noindent
7511: i.e., add the offset to the address, giving the stack effect
7512: @i{addr1 -- addr2} for a field.
7513:
7514: @cindex first field optimization, implementation
7515: This simple structure is slightly complicated by the optimization
7516: for fields with offset 0, which requires a different
7517: @code{DOES>}-part (because we cannot rely on there being
7518: something on the stack if such a field is invoked during
7519: compilation). Therefore, we put the different @code{DOES>}-parts
7520: in separate words, and decide which one to invoke based on the
7521: offset. For a zero offset, the field is basically a noop; it is
7522: immediate, and therefore no code is generated when it is compiled.
7523:
7524: @node Structure Glossary, , Structure Implementation, Structures
7525: @subsection Structure Glossary
7526: @cindex structure glossary
7527:
7528:
7529: doc-%align
7530: doc-%alignment
7531: doc-%alloc
7532: doc-%allocate
7533: doc-%allot
7534: doc-cell%
7535: doc-char%
7536: doc-dfloat%
7537: doc-double%
7538: doc-end-struct
7539: doc-field
7540: doc-float%
7541: doc-naligned
7542: doc-sfloat%
7543: doc-%size
7544: doc-struct
7545:
7546:
7547: @c -------------------------------------------------------------
7548: @node Object-oriented Forth, Passing Commands to the OS, Structures, Words
7549: @section Object-oriented Forth
7550:
7551: Gforth comes with three packages for object-oriented programming:
7552: @file{objects.fs}, @file{oof.fs}, and @file{mini-oof.fs}; none of them
7553: is preloaded, so you have to @code{include} them before use. The most
7554: important differences between these packages (and others) are discussed
7555: in @ref{Comparison with other object models}. All packages are written
7556: in ANS Forth and can be used with any other ANS Forth.
7557:
7558: @menu
7559: * Why object-oriented programming?::
7560: * Object-Oriented Terminology::
7561: * Objects::
7562: * OOF::
7563: * Mini-OOF::
7564: * Comparison with other object models::
7565: @end menu
7566:
7567:
7568: @node Why object-oriented programming?, Object-Oriented Terminology, , Object-oriented Forth
7569: @subsubsection Why object-oriented programming?
7570: @cindex object-oriented programming motivation
7571: @cindex motivation for object-oriented programming
7572:
7573: Often we have to deal with several data structures (@emph{objects}),
7574: that have to be treated similarly in some respects, but differently in
7575: others. Graphical objects are the textbook example: circles, triangles,
7576: dinosaurs, icons, and others, and we may want to add more during program
7577: development. We want to apply some operations to any graphical object,
7578: e.g., @code{draw} for displaying it on the screen. However, @code{draw}
7579: has to do something different for every kind of object.
7580: @comment TODO add some other operations eg perimeter, area
7581: @comment and tie in to concrete examples later..
7582:
7583: We could implement @code{draw} as a big @code{CASE}
7584: control structure that executes the appropriate code depending on the
7585: kind of object to be drawn. This would be not be very elegant, and,
7586: moreover, we would have to change @code{draw} every time we add
7587: a new kind of graphical object (say, a spaceship).
7588:
7589: What we would rather do is: When defining spaceships, we would tell
7590: the system: ``Here's how you @code{draw} a spaceship; you figure
7591: out the rest''.
7592:
7593: This is the problem that all systems solve that (rightfully) call
7594: themselves object-oriented; the object-oriented packages presented here
7595: solve this problem (and not much else).
7596: @comment TODO ?list properties of oo systems.. oo vs o-based?
7597:
7598: @node Object-Oriented Terminology, Objects, Why object-oriented programming?, Object-oriented Forth
7599: @subsubsection Object-Oriented Terminology
7600: @cindex object-oriented terminology
7601: @cindex terminology for object-oriented programming
7602:
7603: This section is mainly for reference, so you don't have to understand
7604: all of it right away. The terminology is mainly Smalltalk-inspired. In
7605: short:
7606:
7607: @table @emph
7608: @cindex class
7609: @item class
7610: a data structure definition with some extras.
7611:
7612: @cindex object
7613: @item object
7614: an instance of the data structure described by the class definition.
7615:
7616: @cindex instance variables
7617: @item instance variables
7618: fields of the data structure.
7619:
7620: @cindex selector
7621: @cindex method selector
7622: @cindex virtual function
7623: @item selector
7624: (or @emph{method selector}) a word (e.g.,
7625: @code{draw}) that performs an operation on a variety of data
7626: structures (classes). A selector describes @emph{what} operation to
7627: perform. In C++ terminology: a (pure) virtual function.
7628:
7629: @cindex method
7630: @item method
7631: the concrete definition that performs the operation
7632: described by the selector for a specific class. A method specifies
7633: @emph{how} the operation is performed for a specific class.
7634:
7635: @cindex selector invocation
7636: @cindex message send
7637: @cindex invoking a selector
7638: @item selector invocation
7639: a call of a selector. One argument of the call (the TOS (top-of-stack))
7640: is used for determining which method is used. In Smalltalk terminology:
7641: a message (consisting of the selector and the other arguments) is sent
7642: to the object.
7643:
7644: @cindex receiving object
7645: @item receiving object
7646: the object used for determining the method executed by a selector
7647: invocation. In the @file{objects.fs} model, it is the object that is on
7648: the TOS when the selector is invoked. (@emph{Receiving} comes from
7649: the Smalltalk @emph{message} terminology.)
7650:
7651: @cindex child class
7652: @cindex parent class
7653: @cindex inheritance
7654: @item child class
7655: a class that has (@emph{inherits}) all properties (instance variables,
7656: selectors, methods) from a @emph{parent class}. In Smalltalk
7657: terminology: The subclass inherits from the superclass. In C++
7658: terminology: The derived class inherits from the base class.
7659:
7660: @end table
7661:
7662: @c If you wonder about the message sending terminology, it comes from
7663: @c a time when each object had it's own task and objects communicated via
7664: @c message passing; eventually the Smalltalk developers realized that
7665: @c they can do most things through simple (indirect) calls. They kept the
7666: @c terminology.
7667:
7668:
7669: @node Objects, OOF, Object-Oriented Terminology, Object-oriented Forth
7670: @subsection The @file{objects.fs} model
7671: @cindex objects
7672: @cindex object-oriented programming
7673:
7674: @cindex @file{objects.fs}
7675: @cindex @file{oof.fs}
7676:
7677: This section describes the @file{objects.fs} package. This material also
7678: has been published in @cite{Yet Another Forth Objects Package} by Anton
7679: Ertl and appeared in Forth Dimensions 19(2), pages 37--43
7680: (@url{http://www.complang.tuwien.ac.at/forth/objects/objects.html}).
7681: @c McKewan's and Zsoter's packages
7682:
7683: This section assumes that you have read @ref{Structures}.
7684:
7685: The techniques on which this model is based have been used to implement
7686: the parser generator, Gray, and have also been used in Gforth for
7687: implementing the various flavours of word lists (hashed or not,
7688: case-sensitive or not, special-purpose word lists for locals etc.).
7689:
7690:
7691: @menu
7692: * Properties of the Objects model::
7693: * Basic Objects Usage::
7694: * The Objects base class::
7695: * Creating objects::
7696: * Object-Oriented Programming Style::
7697: * Class Binding::
7698: * Method conveniences::
7699: * Classes and Scoping::
7700: * Dividing classes::
7701: * Object Interfaces::
7702: * Objects Implementation::
7703: * Objects Glossary::
7704: @end menu
7705:
7706: Marcel Hendrix provided helpful comments on this section. Andras Zsoter
7707: and Bernd Paysan helped me with the related works section.
7708:
7709: @node Properties of the Objects model, Basic Objects Usage, Objects, Objects
7710: @subsubsection Properties of the @file{objects.fs} model
7711: @cindex @file{objects.fs} properties
7712:
7713: @itemize @bullet
7714: @item
7715: It is straightforward to pass objects on the stack. Passing
7716: selectors on the stack is a little less convenient, but possible.
7717:
7718: @item
7719: Objects are just data structures in memory, and are referenced by their
7720: address. You can create words for objects with normal defining words
7721: like @code{constant}. Likewise, there is no difference between instance
7722: variables that contain objects and those that contain other data.
7723:
7724: @item
7725: Late binding is efficient and easy to use.
7726:
7727: @item
7728: It avoids parsing, and thus avoids problems with state-smartness
7729: and reduced extensibility; for convenience there are a few parsing
7730: words, but they have non-parsing counterparts. There are also a few
7731: defining words that parse. This is hard to avoid, because all standard
7732: defining words parse (except @code{:noname}); however, such
7733: words are not as bad as many other parsing words, because they are not
7734: state-smart.
7735:
7736: @item
7737: It does not try to incorporate everything. It does a few things and does
7738: them well (IMO). In particular, this model was not designed to support
7739: information hiding (although it has features that may help); you can use
7740: a separate package for achieving this.
7741:
7742: @item
7743: It is layered; you don't have to learn and use all features to use this
7744: model. Only a few features are necessary (@xref{Basic Objects Usage},
7745: @xref{The Objects base class}, @xref{Creating objects}.), the others
7746: are optional and independent of each other.
7747:
7748: @item
7749: An implementation in ANS Forth is available.
7750:
7751: @end itemize
7752:
7753:
7754: @node Basic Objects Usage, The Objects base class, Properties of the Objects model, Objects
7755: @subsubsection Basic @file{objects.fs} Usage
7756: @cindex basic objects usage
7757: @cindex objects, basic usage
7758:
7759: You can define a class for graphical objects like this:
7760:
7761: @cindex @code{class} usage
7762: @cindex @code{end-class} usage
7763: @cindex @code{selector} usage
7764: @example
7765: object class \ "object" is the parent class
7766: selector draw ( x y graphical -- )
7767: end-class graphical
7768: @end example
7769:
7770: This code defines a class @code{graphical} with an
7771: operation @code{draw}. We can perform the operation
7772: @code{draw} on any @code{graphical} object, e.g.:
7773:
7774: @example
7775: 100 100 t-rex draw
7776: @end example
7777:
7778: @noindent
7779: where @code{t-rex} is a word (say, a constant) that produces a
7780: graphical object.
7781:
7782: @comment TODO add a 2nd operation eg perimeter.. and use for
7783: @comment a concrete example
7784:
7785: @cindex abstract class
7786: How do we create a graphical object? With the present definitions,
7787: we cannot create a useful graphical object. The class
7788: @code{graphical} describes graphical objects in general, but not
7789: any concrete graphical object type (C++ users would call it an
7790: @emph{abstract class}); e.g., there is no method for the selector
7791: @code{draw} in the class @code{graphical}.
7792:
7793: For concrete graphical objects, we define child classes of the
7794: class @code{graphical}, e.g.:
7795:
7796: @cindex @code{overrides} usage
7797: @cindex @code{field} usage in class definition
7798: @example
7799: graphical class \ "graphical" is the parent class
7800: cell% field circle-radius
7801:
7802: :noname ( x y circle -- )
7803: circle-radius @@ draw-circle ;
7804: overrides draw
7805:
7806: :noname ( n-radius circle -- )
7807: circle-radius ! ;
7808: overrides construct
7809:
7810: end-class circle
7811: @end example
7812:
7813: Here we define a class @code{circle} as a child of @code{graphical},
7814: with field @code{circle-radius} (which behaves just like a field
7815: (@pxref{Structures}); it defines (using @code{overrides}) new methods
7816: for the selectors @code{draw} and @code{construct} (@code{construct} is
7817: defined in @code{object}, the parent class of @code{graphical}).
7818:
7819: Now we can create a circle on the heap (i.e.,
7820: @code{allocate}d memory) with:
7821:
7822: @cindex @code{heap-new} usage
7823: @example
7824: 50 circle heap-new constant my-circle
7825: @end example
7826:
7827: @noindent
7828: @code{heap-new} invokes @code{construct}, thus
7829: initializing the field @code{circle-radius} with 50. We can draw
7830: this new circle at (100,100) with:
7831:
7832: @example
7833: 100 100 my-circle draw
7834: @end example
7835:
7836: @cindex selector invocation, restrictions
7837: @cindex class definition, restrictions
7838: Note: You can only invoke a selector if the object on the TOS
7839: (the receiving object) belongs to the class where the selector was
7840: defined or one of its descendents; e.g., you can invoke
7841: @code{draw} only for objects belonging to @code{graphical}
7842: or its descendents (e.g., @code{circle}). Immediately before
7843: @code{end-class}, the search order has to be the same as
7844: immediately after @code{class}.
7845:
7846: @node The Objects base class, Creating objects, Basic Objects Usage, Objects
7847: @subsubsection The @file{object.fs} base class
7848: @cindex @code{object} class
7849:
7850: When you define a class, you have to specify a parent class. So how do
7851: you start defining classes? There is one class available from the start:
7852: @code{object}. It is ancestor for all classes and so is the
7853: only class that has no parent. It has two selectors: @code{construct}
7854: and @code{print}.
7855:
7856: @node Creating objects, Object-Oriented Programming Style, The Objects base class, Objects
7857: @subsubsection Creating objects
7858: @cindex creating objects
7859: @cindex object creation
7860: @cindex object allocation options
7861:
7862: @cindex @code{heap-new} discussion
7863: @cindex @code{dict-new} discussion
7864: @cindex @code{construct} discussion
7865: You can create and initialize an object of a class on the heap with
7866: @code{heap-new} ( ... class -- object ) and in the dictionary
7867: (allocation with @code{allot}) with @code{dict-new} (
7868: ... class -- object ). Both words invoke @code{construct}, which
7869: consumes the stack items indicated by "..." above.
7870:
7871: @cindex @code{init-object} discussion
7872: @cindex @code{class-inst-size} discussion
7873: If you want to allocate memory for an object yourself, you can get its
7874: alignment and size with @code{class-inst-size 2@@} ( class --
7875: align size ). Once you have memory for an object, you can initialize
7876: it with @code{init-object} ( ... class object -- );
7877: @code{construct} does only a part of the necessary work.
7878:
7879: @node Object-Oriented Programming Style, Class Binding, Creating objects, Objects
7880: @subsubsection Object-Oriented Programming Style
7881: @cindex object-oriented programming style
7882:
7883: This section is not exhaustive.
7884:
7885: @cindex stack effects of selectors
7886: @cindex selectors and stack effects
7887: In general, it is a good idea to ensure that all methods for the
7888: same selector have the same stack effect: when you invoke a selector,
7889: you often have no idea which method will be invoked, so, unless all
7890: methods have the same stack effect, you will not know the stack effect
7891: of the selector invocation.
7892:
7893: One exception to this rule is methods for the selector
7894: @code{construct}. We know which method is invoked, because we
7895: specify the class to be constructed at the same place. Actually, I
7896: defined @code{construct} as a selector only to give the users a
7897: convenient way to specify initialization. The way it is used, a
7898: mechanism different from selector invocation would be more natural
7899: (but probably would take more code and more space to explain).
7900:
7901: @node Class Binding, Method conveniences, Object-Oriented Programming Style, Objects
7902: @subsubsection Class Binding
7903: @cindex class binding
7904: @cindex early binding
7905:
7906: @cindex late binding
7907: Normal selector invocations determine the method at run-time depending
7908: on the class of the receiving object. This run-time selection is called
7909: @i{late binding}.
7910:
7911: Sometimes it's preferable to invoke a different method. For example,
7912: you might want to use the simple method for @code{print}ing
7913: @code{object}s instead of the possibly long-winded @code{print} method
7914: of the receiver class. You can achieve this by replacing the invocation
7915: of @code{print} with:
7916:
7917: @cindex @code{[bind]} usage
7918: @example
7919: [bind] object print
7920: @end example
7921:
7922: @noindent
7923: in compiled code or:
7924:
7925: @cindex @code{bind} usage
7926: @example
7927: bind object print
7928: @end example
7929:
7930: @cindex class binding, alternative to
7931: @noindent
7932: in interpreted code. Alternatively, you can define the method with a
7933: name (e.g., @code{print-object}), and then invoke it through the
7934: name. Class binding is just a (often more convenient) way to achieve
7935: the same effect; it avoids name clutter and allows you to invoke
7936: methods directly without naming them first.
7937:
7938: @cindex superclass binding
7939: @cindex parent class binding
7940: A frequent use of class binding is this: When we define a method
7941: for a selector, we often want the method to do what the selector does
7942: in the parent class, and a little more. There is a special word for
7943: this purpose: @code{[parent]}; @code{[parent]
7944: @emph{selector}} is equivalent to @code{[bind] @emph{parent
7945: selector}}, where @code{@emph{parent}} is the parent
7946: class of the current class. E.g., a method definition might look like:
7947:
7948: @cindex @code{[parent]} usage
7949: @example
7950: :noname
7951: dup [parent] foo \ do parent's foo on the receiving object
7952: ... \ do some more
7953: ; overrides foo
7954: @end example
7955:
7956: @cindex class binding as optimization
7957: In @cite{Object-oriented programming in ANS Forth} (Forth Dimensions,
7958: March 1997), Andrew McKewan presents class binding as an optimization
7959: technique. I recommend not using it for this purpose unless you are in
7960: an emergency. Late binding is pretty fast with this model anyway, so the
7961: benefit of using class binding is small; the cost of using class binding
7962: where it is not appropriate is reduced maintainability.
7963:
7964: While we are at programming style questions: You should bind
7965: selectors only to ancestor classes of the receiving object. E.g., say,
7966: you know that the receiving object is of class @code{foo} or its
7967: descendents; then you should bind only to @code{foo} and its
7968: ancestors.
7969:
7970: @node Method conveniences, Classes and Scoping, Class Binding, Objects
7971: @subsubsection Method conveniences
7972: @cindex method conveniences
7973:
7974: In a method you usually access the receiving object pretty often. If
7975: you define the method as a plain colon definition (e.g., with
7976: @code{:noname}), you may have to do a lot of stack
7977: gymnastics. To avoid this, you can define the method with @code{m:
7978: ... ;m}. E.g., you could define the method for
7979: @code{draw}ing a @code{circle} with
7980:
7981: @cindex @code{this} usage
7982: @cindex @code{m:} usage
7983: @cindex @code{;m} usage
7984: @example
7985: m: ( x y circle -- )
7986: ( x y ) this circle-radius @@ draw-circle ;m
7987: @end example
7988:
7989: @cindex @code{exit} in @code{m: ... ;m}
7990: @cindex @code{exitm} discussion
7991: @cindex @code{catch} in @code{m: ... ;m}
7992: When this method is executed, the receiver object is removed from the
7993: stack; you can access it with @code{this} (admittedly, in this
7994: example the use of @code{m: ... ;m} offers no advantage). Note
7995: that I specify the stack effect for the whole method (i.e. including
7996: the receiver object), not just for the code between @code{m:}
7997: and @code{;m}. You cannot use @code{exit} in
7998: @code{m:...;m}; instead, use
7999: @code{exitm}.@footnote{Moreover, for any word that calls
8000: @code{catch} and was defined before loading
8001: @code{objects.fs}, you have to redefine it like I redefined
8002: @code{catch}: @code{: catch this >r catch r> to-this ;}}
8003:
8004: @cindex @code{inst-var} usage
8005: You will frequently use sequences of the form @code{this
8006: @emph{field}} (in the example above: @code{this
8007: circle-radius}). If you use the field only in this way, you can
8008: define it with @code{inst-var} and eliminate the
8009: @code{this} before the field name. E.g., the @code{circle}
8010: class above could also be defined with:
8011:
8012: @example
8013: graphical class
8014: cell% inst-var radius
8015:
8016: m: ( x y circle -- )
8017: radius @@ draw-circle ;m
8018: overrides draw
8019:
8020: m: ( n-radius circle -- )
8021: radius ! ;m
8022: overrides construct
8023:
8024: end-class circle
8025: @end example
8026:
8027: @code{radius} can only be used in @code{circle} and its
8028: descendent classes and inside @code{m:...;m}.
8029:
8030: @cindex @code{inst-value} usage
8031: You can also define fields with @code{inst-value}, which is
8032: to @code{inst-var} what @code{value} is to
8033: @code{variable}. You can change the value of such a field with
8034: @code{[to-inst]}. E.g., we could also define the class
8035: @code{circle} like this:
8036:
8037: @example
8038: graphical class
8039: inst-value radius
8040:
8041: m: ( x y circle -- )
8042: radius draw-circle ;m
8043: overrides draw
8044:
8045: m: ( n-radius circle -- )
8046: [to-inst] radius ;m
8047: overrides construct
8048:
8049: end-class circle
8050: @end example
8051:
8052: Finally, you can define named methods with @code{:m}. One use of this
8053: feature is the definition of words that occur only in one class and are
8054: not intended to be overridden, but which still need method context
8055: (e.g., for accessing @code{inst-var}s). Another use is for methods that
8056: would be bound frequently, if defined anonymously.
8057:
8058:
8059: @node Classes and Scoping, Dividing classes, Method conveniences, Objects
8060: @subsubsection Classes and Scoping
8061: @cindex classes and scoping
8062: @cindex scoping and classes
8063:
8064: Inheritance is frequent, unlike structure extension. This exacerbates
8065: the problem with the field name convention (@pxref{Structure Naming
8066: Convention}): One always has to remember in which class the field was
8067: originally defined; changing a part of the class structure would require
8068: changes for renaming in otherwise unaffected code.
8069:
8070: @cindex @code{inst-var} visibility
8071: @cindex @code{inst-value} visibility
8072: To solve this problem, I added a scoping mechanism (which was not in my
8073: original charter): A field defined with @code{inst-var} (or
8074: @code{inst-value}) is visible only in the class where it is defined and in
8075: the descendent classes of this class. Using such fields only makes
8076: sense in @code{m:}-defined methods in these classes anyway.
8077:
8078: This scoping mechanism allows us to use the unadorned field name,
8079: because name clashes with unrelated words become much less likely.
8080:
8081: @cindex @code{protected} discussion
8082: @cindex @code{private} discussion
8083: Once we have this mechanism, we can also use it for controlling the
8084: visibility of other words: All words defined after
8085: @code{protected} are visible only in the current class and its
8086: descendents. @code{public} restores the compilation
8087: (i.e. @code{current}) word list that was in effect before. If you
8088: have several @code{protected}s without an intervening
8089: @code{public} or @code{set-current}, @code{public}
8090: will restore the compilation word list in effect before the first of
8091: these @code{protected}s.
8092:
8093: @node Dividing classes, Object Interfaces, Classes and Scoping, Objects
8094: @subsubsection Dividing classes
8095: @cindex Dividing classes
8096: @cindex @code{methods}...@code{end-methods}
8097:
8098: You may want to do the definition of methods separate from the
8099: definition of the class, its selectors, fields, and instance variables,
8100: i.e., separate the implementation from the definition. You can do this
8101: in the following way:
8102:
8103: @example
8104: graphical class
8105: inst-value radius
8106: end-class circle
8107:
8108: ... \ do some other stuff
8109:
8110: circle methods \ now we are ready
8111:
8112: m: ( x y circle -- )
8113: radius draw-circle ;m
8114: overrides draw
8115:
8116: m: ( n-radius circle -- )
8117: [to-inst] radius ;m
8118: overrides construct
8119:
8120: end-methods
8121: @end example
8122:
8123: You can use several @code{methods}...@code{end-methods} sections. The
8124: only things you can do to the class in these sections are: defining
8125: methods, and overriding the class's selectors. You must not define new
8126: selectors or fields.
8127:
8128: Note that you often have to override a selector before using it. In
8129: particular, you usually have to override @code{construct} with a new
8130: method before you can invoke @code{heap-new} and friends. E.g., you
8131: must not create a circle before the @code{overrides construct} sequence
8132: in the example above.
8133:
8134: @node Object Interfaces, Objects Implementation, Dividing classes, Objects
8135: @subsubsection Object Interfaces
8136: @cindex object interfaces
8137: @cindex interfaces for objects
8138:
8139: In this model you can only call selectors defined in the class of the
8140: receiving objects or in one of its ancestors. If you call a selector
8141: with a receiving object that is not in one of these classes, the
8142: result is undefined; if you are lucky, the program crashes
8143: immediately.
8144:
8145: @cindex selectors common to hardly-related classes
8146: Now consider the case when you want to have a selector (or several)
8147: available in two classes: You would have to add the selector to a
8148: common ancestor class, in the worst case to @code{object}. You
8149: may not want to do this, e.g., because someone else is responsible for
8150: this ancestor class.
8151:
8152: The solution for this problem is interfaces. An interface is a
8153: collection of selectors. If a class implements an interface, the
8154: selectors become available to the class and its descendents. A class
8155: can implement an unlimited number of interfaces. For the problem
8156: discussed above, we would define an interface for the selector(s), and
8157: both classes would implement the interface.
8158:
8159: As an example, consider an interface @code{storage} for
8160: writing objects to disk and getting them back, and a class
8161: @code{foo} that implements it. The code would look like this:
8162:
8163: @cindex @code{interface} usage
8164: @cindex @code{end-interface} usage
8165: @cindex @code{implementation} usage
8166: @example
8167: interface
8168: selector write ( file object -- )
8169: selector read1 ( file object -- )
8170: end-interface storage
8171:
8172: bar class
8173: storage implementation
8174:
8175: ... overrides write
8176: ... overrides read1
8177: ...
8178: end-class foo
8179: @end example
8180:
8181: @noindent
8182: (I would add a word @code{read} @i{( file -- object )} that uses
8183: @code{read1} internally, but that's beyond the point illustrated
8184: here.)
8185:
8186: Note that you cannot use @code{protected} in an interface; and
8187: of course you cannot define fields.
8188:
8189: In the Neon model, all selectors are available for all classes;
8190: therefore it does not need interfaces. The price you pay in this model
8191: is slower late binding, and therefore, added complexity to avoid late
8192: binding.
8193:
8194: @node Objects Implementation, Objects Glossary, Object Interfaces, Objects
8195: @subsubsection @file{objects.fs} Implementation
8196: @cindex @file{objects.fs} implementation
8197:
8198: @cindex @code{object-map} discussion
8199: An object is a piece of memory, like one of the data structures
8200: described with @code{struct...end-struct}. It has a field
8201: @code{object-map} that points to the method map for the object's
8202: class.
8203:
8204: @cindex method map
8205: @cindex virtual function table
8206: The @emph{method map}@footnote{This is Self terminology; in C++
8207: terminology: virtual function table.} is an array that contains the
8208: execution tokens (@i{xt}s) of the methods for the object's class. Each
8209: selector contains an offset into a method map.
8210:
8211: @cindex @code{selector} implementation, class
8212: @code{selector} is a defining word that uses
8213: @code{CREATE} and @code{DOES>}. The body of the
8214: selector contains the offset; the @code{DOES>} action for a
8215: class selector is, basically:
8216:
8217: @example
8218: ( object addr ) @@ over object-map @@ + @@ execute
8219: @end example
8220:
8221: Since @code{object-map} is the first field of the object, it
8222: does not generate any code. As you can see, calling a selector has a
8223: small, constant cost.
8224:
8225: @cindex @code{current-interface} discussion
8226: @cindex class implementation and representation
8227: A class is basically a @code{struct} combined with a method
8228: map. During the class definition the alignment and size of the class
8229: are passed on the stack, just as with @code{struct}s, so
8230: @code{field} can also be used for defining class
8231: fields. However, passing more items on the stack would be
8232: inconvenient, so @code{class} builds a data structure in memory,
8233: which is accessed through the variable
8234: @code{current-interface}. After its definition is complete, the
8235: class is represented on the stack by a pointer (e.g., as parameter for
8236: a child class definition).
8237:
8238: A new class starts off with the alignment and size of its parent,
8239: and a copy of the parent's method map. Defining new fields extends the
8240: size and alignment; likewise, defining new selectors extends the
8241: method map. @code{overrides} just stores a new @i{xt} in the method
8242: map at the offset given by the selector.
8243:
8244: @cindex class binding, implementation
8245: Class binding just gets the @i{xt} at the offset given by the selector
8246: from the class's method map and @code{compile,}s (in the case of
8247: @code{[bind]}) it.
8248:
8249: @cindex @code{this} implementation
8250: @cindex @code{catch} and @code{this}
8251: @cindex @code{this} and @code{catch}
8252: I implemented @code{this} as a @code{value}. At the
8253: start of an @code{m:...;m} method the old @code{this} is
8254: stored to the return stack and restored at the end; and the object on
8255: the TOS is stored @code{TO this}. This technique has one
8256: disadvantage: If the user does not leave the method via
8257: @code{;m}, but via @code{throw} or @code{exit},
8258: @code{this} is not restored (and @code{exit} may
8259: crash). To deal with the @code{throw} problem, I have redefined
8260: @code{catch} to save and restore @code{this}; the same
8261: should be done with any word that can catch an exception. As for
8262: @code{exit}, I simply forbid it (as a replacement, there is
8263: @code{exitm}).
8264:
8265: @cindex @code{inst-var} implementation
8266: @code{inst-var} is just the same as @code{field}, with
8267: a different @code{DOES>} action:
8268: @example
8269: @@ this +
8270: @end example
8271: Similar for @code{inst-value}.
8272:
8273: @cindex class scoping implementation
8274: Each class also has a word list that contains the words defined with
8275: @code{inst-var} and @code{inst-value}, and its protected
8276: words. It also has a pointer to its parent. @code{class} pushes
8277: the word lists of the class and all its ancestors onto the search order stack,
8278: and @code{end-class} drops them.
8279:
8280: @cindex interface implementation
8281: An interface is like a class without fields, parent and protected
8282: words; i.e., it just has a method map. If a class implements an
8283: interface, its method map contains a pointer to the method map of the
8284: interface. The positive offsets in the map are reserved for class
8285: methods, therefore interface map pointers have negative
8286: offsets. Interfaces have offsets that are unique throughout the
8287: system, unlike class selectors, whose offsets are only unique for the
8288: classes where the selector is available (invokable).
8289:
8290: This structure means that interface selectors have to perform one
8291: indirection more than class selectors to find their method. Their body
8292: contains the interface map pointer offset in the class method map, and
8293: the method offset in the interface method map. The
8294: @code{does>} action for an interface selector is, basically:
8295:
8296: @example
8297: ( object selector-body )
8298: 2dup selector-interface @@ ( object selector-body object interface-offset )
8299: swap object-map @@ + @@ ( object selector-body map )
8300: swap selector-offset @@ + @@ execute
8301: @end example
8302:
8303: where @code{object-map} and @code{selector-offset} are
8304: first fields and generate no code.
8305:
8306: As a concrete example, consider the following code:
8307:
8308: @example
8309: interface
8310: selector if1sel1
8311: selector if1sel2
8312: end-interface if1
8313:
8314: object class
8315: if1 implementation
8316: selector cl1sel1
8317: cell% inst-var cl1iv1
8318:
8319: ' m1 overrides construct
8320: ' m2 overrides if1sel1
8321: ' m3 overrides if1sel2
8322: ' m4 overrides cl1sel2
8323: end-class cl1
8324:
8325: create obj1 object dict-new drop
8326: create obj2 cl1 dict-new drop
8327: @end example
8328:
8329: The data structure created by this code (including the data structure
8330: for @code{object}) is shown in the <a
8331: href="objects-implementation.eps">figure</a>, assuming a cell size of 4.
8332: @comment TODO add this diagram..
8333:
8334: @node Objects Glossary, , Objects Implementation, Objects
8335: @subsubsection @file{objects.fs} Glossary
8336: @cindex @file{objects.fs} Glossary
8337:
8338:
8339: doc---objects-bind
8340: doc---objects-<bind>
8341: doc---objects-bind'
8342: doc---objects-[bind]
8343: doc---objects-class
8344: doc---objects-class->map
8345: doc---objects-class-inst-size
8346: doc---objects-class-override!
8347: doc---objects-construct
8348: doc---objects-current'
8349: doc---objects-[current]
8350: doc---objects-current-interface
8351: doc---objects-dict-new
8352: doc---objects-drop-order
8353: doc---objects-end-class
8354: doc---objects-end-class-noname
8355: doc---objects-end-interface
8356: doc---objects-end-interface-noname
8357: doc---objects-end-methods
8358: doc---objects-exitm
8359: doc---objects-heap-new
8360: doc---objects-implementation
8361: doc---objects-init-object
8362: doc---objects-inst-value
8363: doc---objects-inst-var
8364: doc---objects-interface
8365: doc---objects-m:
8366: doc---objects-:m
8367: doc---objects-;m
8368: doc---objects-method
8369: doc---objects-methods
8370: doc---objects-object
8371: doc---objects-overrides
8372: doc---objects-[parent]
8373: doc---objects-print
8374: doc---objects-protected
8375: doc---objects-public
8376: doc---objects-push-order
8377: doc---objects-selector
8378: doc---objects-this
8379: doc---objects-<to-inst>
8380: doc---objects-[to-inst]
8381: doc---objects-to-this
8382: doc---objects-xt-new
8383:
8384:
8385: @c -------------------------------------------------------------
8386: @node OOF, Mini-OOF, Objects, Object-oriented Forth
8387: @subsection The @file{oof.fs} model
8388: @cindex oof
8389: @cindex object-oriented programming
8390:
8391: @cindex @file{objects.fs}
8392: @cindex @file{oof.fs}
8393:
8394: This section describes the @file{oof.fs} package.
8395:
8396: The package described in this section has been used in bigFORTH since 1991, and
8397: used for two large applications: a chromatographic system used to
8398: create new medicaments, and a graphic user interface library (MINOS).
8399:
8400: You can find a description (in German) of @file{oof.fs} in @cite{Object
8401: oriented bigFORTH} by Bernd Paysan, published in @cite{Vierte Dimension}
8402: 10(2), 1994.
8403:
8404: @menu
8405: * Properties of the OOF model::
8406: * Basic OOF Usage::
8407: * The OOF base class::
8408: * Class Declaration::
8409: * Class Implementation::
8410: @end menu
8411:
8412: @node Properties of the OOF model, Basic OOF Usage, OOF, OOF
8413: @subsubsection Properties of the @file{oof.fs} model
8414: @cindex @file{oof.fs} properties
8415:
8416: @itemize @bullet
8417: @item
8418: This model combines object oriented programming with information
8419: hiding. It helps you writing large application, where scoping is
8420: necessary, because it provides class-oriented scoping.
8421:
8422: @item
8423: Named objects, object pointers, and object arrays can be created,
8424: selector invocation uses the ``object selector'' syntax. Selector invocation
8425: to objects and/or selectors on the stack is a bit less convenient, but
8426: possible.
8427:
8428: @item
8429: Selector invocation and instance variable usage of the active object is
8430: straightforward, since both make use of the active object.
8431:
8432: @item
8433: Late binding is efficient and easy to use.
8434:
8435: @item
8436: State-smart objects parse selectors. However, extensibility is provided
8437: using a (parsing) selector @code{postpone} and a selector @code{'}.
8438:
8439: @item
8440: An implementation in ANS Forth is available.
8441:
8442: @end itemize
8443:
8444:
8445: @node Basic OOF Usage, The OOF base class, Properties of the OOF model, OOF
8446: @subsubsection Basic @file{oof.fs} Usage
8447: @cindex @file{oof.fs} usage
8448:
8449: This section uses the same example as for @code{objects} (@pxref{Basic Objects Usage}).
8450:
8451: You can define a class for graphical objects like this:
8452:
8453: @cindex @code{class} usage
8454: @cindex @code{class;} usage
8455: @cindex @code{method} usage
8456: @example
8457: object class graphical \ "object" is the parent class
8458: method draw ( x y graphical -- )
8459: class;
8460: @end example
8461:
8462: This code defines a class @code{graphical} with an
8463: operation @code{draw}. We can perform the operation
8464: @code{draw} on any @code{graphical} object, e.g.:
8465:
8466: @example
8467: 100 100 t-rex draw
8468: @end example
8469:
8470: @noindent
8471: where @code{t-rex} is an object or object pointer, created with e.g.
8472: @code{graphical : t-rex}.
8473:
8474: @cindex abstract class
8475: How do we create a graphical object? With the present definitions,
8476: we cannot create a useful graphical object. The class
8477: @code{graphical} describes graphical objects in general, but not
8478: any concrete graphical object type (C++ users would call it an
8479: @emph{abstract class}); e.g., there is no method for the selector
8480: @code{draw} in the class @code{graphical}.
8481:
8482: For concrete graphical objects, we define child classes of the
8483: class @code{graphical}, e.g.:
8484:
8485: @example
8486: graphical class circle \ "graphical" is the parent class
8487: cell var circle-radius
8488: how:
8489: : draw ( x y -- )
8490: circle-radius @@ draw-circle ;
8491:
8492: : init ( n-radius -- (
8493: circle-radius ! ;
8494: class;
8495: @end example
8496:
8497: Here we define a class @code{circle} as a child of @code{graphical},
8498: with a field @code{circle-radius}; it defines new methods for the
8499: selectors @code{draw} and @code{init} (@code{init} is defined in
8500: @code{object}, the parent class of @code{graphical}).
8501:
8502: Now we can create a circle in the dictionary with:
8503:
8504: @example
8505: 50 circle : my-circle
8506: @end example
8507:
8508: @noindent
8509: @code{:} invokes @code{init}, thus initializing the field
8510: @code{circle-radius} with 50. We can draw this new circle at (100,100)
8511: with:
8512:
8513: @example
8514: 100 100 my-circle draw
8515: @end example
8516:
8517: @cindex selector invocation, restrictions
8518: @cindex class definition, restrictions
8519: Note: You can only invoke a selector if the receiving object belongs to
8520: the class where the selector was defined or one of its descendents;
8521: e.g., you can invoke @code{draw} only for objects belonging to
8522: @code{graphical} or its descendents (e.g., @code{circle}). The scoping
8523: mechanism will check if you try to invoke a selector that is not
8524: defined in this class hierarchy, so you'll get an error at compilation
8525: time.
8526:
8527:
8528: @node The OOF base class, Class Declaration, Basic OOF Usage, OOF
8529: @subsubsection The @file{oof.fs} base class
8530: @cindex @file{oof.fs} base class
8531:
8532: When you define a class, you have to specify a parent class. So how do
8533: you start defining classes? There is one class available from the start:
8534: @code{object}. You have to use it as ancestor for all classes. It is the
8535: only class that has no parent. Classes are also objects, except that
8536: they don't have instance variables; class manipulation such as
8537: inheritance or changing definitions of a class is handled through
8538: selectors of the class @code{object}.
8539:
8540: @code{object} provides a number of selectors:
8541:
8542: @itemize @bullet
8543: @item
8544: @code{class} for subclassing, @code{definitions} to add definitions
8545: later on, and @code{class?} to get type informations (is the class a
8546: subclass of the class passed on the stack?).
8547:
8548: doc---object-class
8549: doc---object-definitions
8550: doc---object-class?
8551:
8552:
8553: @item
8554: @code{init} and @code{dispose} as constructor and destructor of the
8555: object. @code{init} is invocated after the object's memory is allocated,
8556: while @code{dispose} also handles deallocation. Thus if you redefine
8557: @code{dispose}, you have to call the parent's dispose with @code{super
8558: dispose}, too.
8559:
8560: doc---object-init
8561: doc---object-dispose
8562:
8563:
8564: @item
8565: @code{new}, @code{new[]}, @code{:}, @code{ptr}, @code{asptr}, and
8566: @code{[]} to create named and unnamed objects and object arrays or
8567: object pointers.
8568:
8569: doc---object-new
8570: doc---object-new[]
8571: doc---object-:
8572: doc---object-ptr
8573: doc---object-asptr
8574: doc---object-[]
8575:
8576:
8577: @item
8578: @code{::} and @code{super} for explicit scoping. You should use explicit
8579: scoping only for super classes or classes with the same set of instance
8580: variables. Explicitly-scoped selectors use early binding.
8581:
8582: doc---object-::
8583: doc---object-super
8584:
8585:
8586: @item
8587: @code{self} to get the address of the object
8588:
8589: doc---object-self
8590:
8591:
8592: @item
8593: @code{bind}, @code{bound}, @code{link}, and @code{is} to assign object
8594: pointers and instance defers.
8595:
8596: doc---object-bind
8597: doc---object-bound
8598: doc---object-link
8599: doc---object-is
8600:
8601:
8602: @item
8603: @code{'} to obtain selector tokens, @code{send} to invocate selectors
8604: form the stack, and @code{postpone} to generate selector invocation code.
8605:
8606: doc---object-'
8607: doc---object-postpone
8608:
8609:
8610: @item
8611: @code{with} and @code{endwith} to select the active object from the
8612: stack, and enable its scope. Using @code{with} and @code{endwith}
8613: also allows you to create code using selector @code{postpone} without being
8614: trapped by the state-smart objects.
8615:
8616: doc---object-with
8617: doc---object-endwith
8618:
8619:
8620: @end itemize
8621:
8622: @node Class Declaration, Class Implementation, The OOF base class, OOF
8623: @subsubsection Class Declaration
8624: @cindex class declaration
8625:
8626: @itemize @bullet
8627: @item
8628: Instance variables
8629:
8630: doc---oof-var
8631:
8632:
8633: @item
8634: Object pointers
8635:
8636: doc---oof-ptr
8637: doc---oof-asptr
8638:
8639:
8640: @item
8641: Instance defers
8642:
8643: doc---oof-defer
8644:
8645:
8646: @item
8647: Method selectors
8648:
8649: doc---oof-early
8650: doc---oof-method
8651:
8652:
8653: @item
8654: Class-wide variables
8655:
8656: doc---oof-static
8657:
8658:
8659: @item
8660: End declaration
8661:
8662: doc---oof-how:
8663: doc---oof-class;
8664:
8665:
8666: @end itemize
8667:
8668: @c -------------------------------------------------------------
8669: @node Class Implementation, , Class Declaration, OOF
8670: @subsubsection Class Implementation
8671: @cindex class implementation
8672:
8673: @c -------------------------------------------------------------
8674: @node Mini-OOF, Comparison with other object models, OOF, Object-oriented Forth
8675: @subsection The @file{mini-oof.fs} model
8676: @cindex mini-oof
8677:
8678: Gforth's third object oriented Forth package is a 12-liner. It uses a
8679: mixture of the @file{object.fs} and the @file{oof.fs} syntax,
8680: and reduces to the bare minimum of features. This is based on a posting
8681: of Bernd Paysan in comp.arch.
8682:
8683: @menu
8684: * Basic Mini-OOF Usage::
8685: * Mini-OOF Example::
8686: * Mini-OOF Implementation::
8687: @end menu
8688:
8689: @c -------------------------------------------------------------
8690: @node Basic Mini-OOF Usage, Mini-OOF Example, , Mini-OOF
8691: @subsubsection Basic @file{mini-oof.fs} Usage
8692: @cindex mini-oof usage
8693:
8694: There is a base class (@code{class}, which allocates one cell for the
8695: object pointer) plus seven other words: to define a method, a variable,
8696: a class; to end a class, to resolve binding, to allocate an object and
8697: to compile a class method.
8698: @comment TODO better description of the last one
8699:
8700:
8701: doc-object
8702: doc-method
8703: doc-var
8704: doc-class
8705: doc-end-class
8706: doc-defines
8707: doc-new
8708: doc-::
8709:
8710:
8711:
8712: @c -------------------------------------------------------------
8713: @node Mini-OOF Example, Mini-OOF Implementation, Basic Mini-OOF Usage, Mini-OOF
8714: @subsubsection Mini-OOF Example
8715: @cindex mini-oof example
8716:
8717: A short example shows how to use this package. This example, in slightly
8718: extended form, is supplied as @file{moof-exm.fs}
8719: @comment TODO could flesh this out with some comments from the Forthwrite article
8720:
8721: @example
8722: object class
8723: method init
8724: method draw
8725: end-class graphical
8726: @end example
8727:
8728: This code defines a class @code{graphical} with an
8729: operation @code{draw}. We can perform the operation
8730: @code{draw} on any @code{graphical} object, e.g.:
8731:
8732: @example
8733: 100 100 t-rex draw
8734: @end example
8735:
8736: where @code{t-rex} is an object or object pointer, created with e.g.
8737: @code{graphical new Constant t-rex}.
8738:
8739: For concrete graphical objects, we define child classes of the
8740: class @code{graphical}, e.g.:
8741:
8742: @example
8743: graphical class
8744: cell var circle-radius
8745: end-class circle \ "graphical" is the parent class
8746:
8747: :noname ( x y -- )
8748: circle-radius @@ draw-circle ; circle defines draw
8749: :noname ( r -- )
8750: circle-radius ! ; circle defines init
8751: @end example
8752:
8753: There is no implicit init method, so we have to define one. The creation
8754: code of the object now has to call init explicitely.
8755:
8756: @example
8757: circle new Constant my-circle
8758: 50 my-circle init
8759: @end example
8760:
8761: It is also possible to add a function to create named objects with
8762: automatic call of @code{init}, given that all objects have @code{init}
8763: on the same place:
8764:
8765: @example
8766: : new: ( .. o "name" -- )
8767: new dup Constant init ;
8768: 80 circle new: large-circle
8769: @end example
8770:
8771: We can draw this new circle at (100,100) with:
8772:
8773: @example
8774: 100 100 my-circle draw
8775: @end example
8776:
8777: @node Mini-OOF Implementation, , Mini-OOF Example, Mini-OOF
8778: @subsubsection @file{mini-oof.fs} Implementation
8779:
8780: Object-oriented systems with late binding typically use a
8781: ``vtable''-approach: the first variable in each object is a pointer to a
8782: table, which contains the methods as function pointers. The vtable
8783: may also contain other information.
8784:
8785: So first, let's declare methods:
8786:
8787: @example
8788: : method ( m v -- m' v ) Create over , swap cell+ swap
8789: DOES> ( ... o -- ... ) @ over @ + @ execute ;
8790: @end example
8791:
8792: During method declaration, the number of methods and instance
8793: variables is on the stack (in address units). @code{method} creates
8794: one method and increments the method number. To execute a method, it
8795: takes the object, fetches the vtable pointer, adds the offset, and
8796: executes the @i{xt} stored there. Each method takes the object it is
8797: invoked from as top of stack parameter. The method itself should
8798: consume that object.
8799:
8800: Now, we also have to declare instance variables
8801:
8802: @example
8803: : var ( m v size -- m v' ) Create over , +
8804: DOES> ( o -- addr ) @ + ;
8805: @end example
8806:
8807: As before, a word is created with the current offset. Instance
8808: variables can have different sizes (cells, floats, doubles, chars), so
8809: all we do is take the size and add it to the offset. If your machine
8810: has alignment restrictions, put the proper @code{aligned} or
8811: @code{faligned} before the variable, to adjust the variable
8812: offset. That's why it is on the top of stack.
8813:
8814: We need a starting point (the base object) and some syntactic sugar:
8815:
8816: @example
8817: Create object 1 cells , 2 cells ,
8818: : class ( class -- class methods vars ) dup 2@ ;
8819: @end example
8820:
8821: For inheritance, the vtable of the parent object has to be
8822: copied when a new, derived class is declared. This gives all the
8823: methods of the parent class, which can be overridden, though.
8824:
8825: @example
8826: : end-class ( class methods vars -- )
8827: Create here >r , dup , 2 cells ?DO ['] noop , 1 cells +LOOP
8828: cell+ dup cell+ r> rot @ 2 cells /string move ;
8829: @end example
8830:
8831: The first line creates the vtable, initialized with
8832: @code{noop}s. The second line is the inheritance mechanism, it
8833: copies the xts from the parent vtable.
8834:
8835: We still have no way to define new methods, let's do that now:
8836:
8837: @example
8838: : defines ( xt class -- ) ' >body @ + ! ;
8839: @end example
8840:
8841: To allocate a new object, we need a word, too:
8842:
8843: @example
8844: : new ( class -- o ) here over @ allot swap over ! ;
8845: @end example
8846:
8847: Sometimes derived classes want to access the method of the
8848: parent object. There are two ways to achieve this with Mini-OOF:
8849: first, you could use named words, and second, you could look up the
8850: vtable of the parent object.
8851:
8852: @example
8853: : :: ( class "name" -- ) ' >body @ + @ compile, ;
8854: @end example
8855:
8856:
8857: Nothing can be more confusing than a good example, so here is
8858: one. First let's declare a text object (called
8859: @code{button}), that stores text and position:
8860:
8861: @example
8862: object class
8863: cell var text
8864: cell var len
8865: cell var x
8866: cell var y
8867: method init
8868: method draw
8869: end-class button
8870: @end example
8871:
8872: @noindent
8873: Now, implement the two methods, @code{draw} and @code{init}:
8874:
8875: @example
8876: :noname ( o -- )
8877: >r r@ x @ r@ y @ at-xy r@ text @ r> len @ type ;
8878: button defines draw
8879: :noname ( addr u o -- )
8880: >r 0 r@ x ! 0 r@ y ! r@ len ! r> text ! ;
8881: button defines init
8882: @end example
8883:
8884: @noindent
8885: To demonstrate inheritance, we define a class @code{bold-button}, with no
8886: new data and no new methods:
8887:
8888: @example
8889: button class
8890: end-class bold-button
8891:
8892: : bold 27 emit ." [1m" ;
8893: : normal 27 emit ." [0m" ;
8894: @end example
8895:
8896: @noindent
8897: The class @code{bold-button} has a different draw method to
8898: @code{button}, but the new method is defined in terms of the draw method
8899: for @code{button}:
8900:
8901: @example
8902: :noname bold [ button :: draw ] normal ; bold-button defines draw
8903: @end example
8904:
8905: @noindent
8906: Finally, create two objects and apply methods:
8907:
8908: @example
8909: button new Constant foo
8910: s" thin foo" foo init
8911: page
8912: foo draw
8913: bold-button new Constant bar
8914: s" fat bar" bar init
8915: 1 bar y !
8916: bar draw
8917: @end example
8918:
8919:
8920: @node Comparison with other object models, , Mini-OOF, Object-oriented Forth
8921: @subsubsection Comparison with other object models
8922: @cindex comparison of object models
8923: @cindex object models, comparison
8924:
8925: Many object-oriented Forth extensions have been proposed (@cite{A survey
8926: of object-oriented Forths} (SIGPLAN Notices, April 1996) by Bradford
8927: J. Rodriguez and W. F. S. Poehlman lists 17). This section discusses the
8928: relation of the object models described here to two well-known and two
8929: closely-related (by the use of method maps) models.
8930:
8931: @cindex Neon model
8932: The most popular model currently seems to be the Neon model (see
8933: @cite{Object-oriented programming in ANS Forth} (Forth Dimensions, March
8934: 1997) by Andrew McKewan) but this model has a number of limitations
8935: @footnote{A longer version of this critique can be
8936: found in @cite{On Standardizing Object-Oriented Forth Extensions} (Forth
8937: Dimensions, May 1997) by Anton Ertl.}:
8938:
8939: @itemize @bullet
8940: @item
8941: It uses a @code{@emph{selector
8942: object}} syntax, which makes it unnatural to pass objects on the
8943: stack.
8944:
8945: @item
8946: It requires that the selector parses the input stream (at
8947: compile time); this leads to reduced extensibility and to bugs that are+
8948: hard to find.
8949:
8950: @item
8951: It allows using every selector to every object;
8952: this eliminates the need for classes, but makes it harder to create
8953: efficient implementations.
8954: @end itemize
8955:
8956: @cindex Pountain's object-oriented model
8957: Another well-known publication is @cite{Object-Oriented Forth} (Academic
8958: Press, London, 1987) by Dick Pountain. However, it is not really about
8959: object-oriented programming, because it hardly deals with late
8960: binding. Instead, it focuses on features like information hiding and
8961: overloading that are characteristic of modular languages like Ada (83).
8962:
8963: @cindex Zsoter's object-oriented model
8964: In @cite{Does late binding have to be slow?} (Forth Dimensions 18(1) 1996, pages 31-35)
8965: Andras Zsoter describes a model that makes heavy use of an active object
8966: (like @code{this} in @file{objects.fs}): The active object is not only
8967: used for accessing all fields, but also specifies the receiving object
8968: of every selector invocation; you have to change the active object
8969: explicitly with @code{@{ ... @}}, whereas in @file{objects.fs} it
8970: changes more or less implicitly at @code{m: ... ;m}. Such a change at
8971: the method entry point is unnecessary with the Zsoter's model, because
8972: the receiving object is the active object already. On the other hand, the explicit
8973: change is absolutely necessary in that model, because otherwise no one
8974: could ever change the active object. An ANS Forth implementation of this
8975: model is available at @url{http://www.forth.org/fig/oopf.html}.
8976:
8977: @cindex @file{oof.fs}, differences to other models
8978: The @file{oof.fs} model combines information hiding and overloading
8979: resolution (by keeping names in various word lists) with object-oriented
8980: programming. It sets the active object implicitly on method entry, but
8981: also allows explicit changing (with @code{>o...o>} or with
8982: @code{with...endwith}). It uses parsing and state-smart objects and
8983: classes for resolving overloading and for early binding: the object or
8984: class parses the selector and determines the method from this. If the
8985: selector is not parsed by an object or class, it performs a call to the
8986: selector for the active object (late binding), like Zsoter's model.
8987: Fields are always accessed through the active object. The big
8988: disadvantage of this model is the parsing and the state-smartness, which
8989: reduces extensibility and increases the opportunities for subtle bugs;
8990: essentially, you are only safe if you never tick or @code{postpone} an
8991: object or class (Bernd disagrees, but I (Anton) am not convinced).
8992:
8993: @cindex @file{mini-oof.fs}, differences to other models
8994: The @file{mini-oof.fs} model is quite similar to a very stripped-down version of
8995: the @file{objects.fs} model, but syntactically it is a mixture of the @file{objects.fs} and
8996: @file{oof.fs} models.
8997:
8998: @c -------------------------------------------------------------
8999: @node Passing Commands to the OS, Miscellaneous Words, Object-oriented Forth, Words
9000: @section Passing Commands to the Operating System
9001: @cindex operating system - passing commands
9002: @cindex shell commands
9003:
9004: Gforth allows you to pass an arbitrary string to the host operating
9005: system shell (if such a thing exists) for execution.
9006:
9007:
9008: doc-sh
9009: doc-system
9010: doc-$?
9011: doc-getenv
9012:
9013:
9014: @c -------------------------------------------------------------
9015: @node Miscellaneous Words, , Passing Commands to the OS, Words
9016: @section Miscellaneous Words
9017: @cindex miscellaneous words
9018:
9019: @comment TODO find homes for these
9020:
9021: These section lists the ANS Forth words that are not documented
9022: elsewhere in this manual. Ultimately, they all need proper homes.
9023:
9024:
9025: doc-ms
9026: doc-time&date
9027:
9028:
9029:
9030: doc-[compile]
9031:
9032:
9033: The following ANS Forth words are not currently supported by Gforth
9034: (@pxref{ANS conformance}):
9035:
9036: @code{EDITOR}
9037: @code{EMIT?}
9038: @code{FORGET}
9039:
9040: @c ******************************************************************
9041: @node Error messages, Tools, Words, Top
9042: @chapter Error messages
9043: @cindex error messages
9044: @cindex backtrace
9045:
9046: A typical Gforth error message looks like this:
9047:
9048: @example
9049: in file included from :-1
9050: in file included from ./yyy.fs:1
9051: ./xxx.fs:4: Invalid memory address
9052: bar
9053: ^^^
9054: $400E664C @@
9055: $400E6664 foo
9056: @end example
9057:
9058: The message identifying the error is @code{Invalid memory address}. The
9059: error happened when text-interpreting line 4 of the file
9060: @file{./xxx.fs}. This line is given (it contains @code{bar}), and the
9061: word on the line where the error happened, is pointed out (with
9062: @code{^^^}).
9063:
9064: The file containing the error was included in line 1 of @file{./yyy.fs},
9065: and @file{yyy.fs} was included from a non-file (in this case, by giving
9066: @file{yyy.fs} as command-line parameter to Gforth).
9067:
9068: At the end of the error message you find a return stack dump that can be
9069: interpreted as a backtrace (possibly empty). On top you find the top of
9070: the return stack when the @code{throw} happened, and at the bottom you
9071: find the return stack entry just above the return stack of the topmost
9072: text interpreter.
9073:
9074: To the right of most return stack entries you see a guess for the word
9075: that pushed that return stack entry as its return address. This gives a
9076: backtrace. In our case we see that @code{bar} called @code{foo}, and
9077: @code{foo} called @code{@@} (and @code{@@} had an @emph{Invalid memory
9078: address} exception).
9079:
9080: Note that the backtrace is not perfect: We don't know which return stack
9081: entries are return addresses (so we may get false positives); and in
9082: some cases (e.g., for @code{abort"}) we cannot determine from the return
9083: address the word that pushed the return address, so for some return
9084: addresses you see no names in the return stack dump.
9085:
9086: @cindex @code{catch} and backtraces
9087: The return stack dump represents the return stack at the time when a
9088: specific @code{throw} was executed. In programs that make use of
9089: @code{catch}, it is not necessarily clear which @code{throw} should be
9090: used for the return stack dump (e.g., consider one @code{throw} that
9091: indicates an error, which is caught, and during recovery another error
9092: happens; which @code{throw} should be used for the stack dump?). Gforth
9093: presents the return stack dump for the first @code{throw} after the last
9094: executed (not returned-to) @code{catch}; this works well in the usual
9095: case.
9096:
9097: @cindex @code{gforth-fast} and backtraces
9098: @cindex @code{gforth-fast}, difference from @code{gforth}
9099: @cindex backtraces with @code{gforth-fast}
9100: @cindex return stack dump with @code{gforth-fast}
9101: @code{gforth} is able to do a return stack dump for throws generated
9102: from primitives (e.g., invalid memory address, stack empty etc.);
9103: @code{gforth-fast} is only able to do a return stack dump from a
9104: directly called @code{throw} (including @code{abort} etc.). This is the
9105: only difference (apart from a speed factor of between 1.15 (K6-2) and
9106: 1.6 (21164A)) between @code{gforth} and @code{gforth-fast}. Given an
9107: exception caused by a primitive in @code{gforth-fast}, you will
9108: typically see no return stack dump at all; however, if the exception is
9109: caught by @code{catch} (e.g., for restoring some state), and then
9110: @code{throw}n again, the return stack dump will be for the first such
9111: @code{throw}.
9112:
9113: @c ******************************************************************
9114: @node Tools, ANS conformance, Error messages, Top
9115: @chapter Tools
9116:
9117: @menu
9118: * ANS Report:: Report the words used, sorted by wordset.
9119: @end menu
9120:
9121: See also @ref{Emacs and Gforth}.
9122:
9123: @node ANS Report, , Tools, Tools
9124: @section @file{ans-report.fs}: Report the words used, sorted by wordset
9125: @cindex @file{ans-report.fs}
9126: @cindex report the words used in your program
9127: @cindex words used in your program
9128:
9129: If you want to label a Forth program as ANS Forth Program, you must
9130: document which wordsets the program uses; for extension wordsets, it is
9131: helpful to list the words the program requires from these wordsets
9132: (because Forth systems are allowed to provide only some words of them).
9133:
9134: The @file{ans-report.fs} tool makes it easy for you to determine which
9135: words from which wordset and which non-ANS words your application
9136: uses. You simply have to include @file{ans-report.fs} before loading the
9137: program you want to check. After loading your program, you can get the
9138: report with @code{print-ans-report}. A typical use is to run this as
9139: batch job like this:
9140: @example
9141: gforth ans-report.fs myprog.fs -e "print-ans-report bye"
9142: @end example
9143:
9144: The output looks like this (for @file{compat/control.fs}):
9145: @example
9146: The program uses the following words
9147: from CORE :
9148: : POSTPONE THEN ; immediate ?dup IF 0=
9149: from BLOCK-EXT :
9150: \
9151: from FILE :
9152: (
9153: @end example
9154:
9155: @subsection Caveats
9156:
9157: Note that @file{ans-report.fs} just checks which words are used, not whether
9158: they are used in an ANS Forth conforming way!
9159:
9160: Some words are defined in several wordsets in the
9161: standard. @file{ans-report.fs} reports them for only one of the
9162: wordsets, and not necessarily the one you expect. It depends on usage
9163: which wordset is the right one to specify. E.g., if you only use the
9164: compilation semantics of @code{S"}, it is a Core word; if you also use
9165: its interpretation semantics, it is a File word.
9166:
9167: @c ******************************************************************
9168: @node ANS conformance, Model, Tools, Top
9169: @chapter ANS conformance
9170: @cindex ANS conformance of Gforth
9171:
9172: To the best of our knowledge, Gforth is an
9173:
9174: ANS Forth System
9175: @itemize @bullet
9176: @item providing the Core Extensions word set
9177: @item providing the Block word set
9178: @item providing the Block Extensions word set
9179: @item providing the Double-Number word set
9180: @item providing the Double-Number Extensions word set
9181: @item providing the Exception word set
9182: @item providing the Exception Extensions word set
9183: @item providing the Facility word set
9184: @item providing @code{EKEY}, @code{EKEY>CHAR}, @code{EKEY?}, @code{MS} and @code{TIME&DATE} from the Facility Extensions word set
9185: @item providing the File Access word set
9186: @item providing the File Access Extensions word set
9187: @item providing the Floating-Point word set
9188: @item providing the Floating-Point Extensions word set
9189: @item providing the Locals word set
9190: @item providing the Locals Extensions word set
9191: @item providing the Memory-Allocation word set
9192: @item providing the Memory-Allocation Extensions word set (that one's easy)
9193: @item providing the Programming-Tools word set
9194: @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
9195: @item providing the Search-Order word set
9196: @item providing the Search-Order Extensions word set
9197: @item providing the String word set
9198: @item providing the String Extensions word set (another easy one)
9199: @end itemize
9200:
9201: @cindex system documentation
9202: In addition, ANS Forth systems are required to document certain
9203: implementation choices. This chapter tries to meet these
9204: requirements. In many cases it gives a way to ask the system for the
9205: information instead of providing the information directly, in
9206: particular, if the information depends on the processor, the operating
9207: system or the installation options chosen, or if they are likely to
9208: change during the maintenance of Gforth.
9209:
9210: @comment The framework for the rest has been taken from pfe.
9211:
9212: @menu
9213: * The Core Words::
9214: * The optional Block word set::
9215: * The optional Double Number word set::
9216: * The optional Exception word set::
9217: * The optional Facility word set::
9218: * The optional File-Access word set::
9219: * The optional Floating-Point word set::
9220: * The optional Locals word set::
9221: * The optional Memory-Allocation word set::
9222: * The optional Programming-Tools word set::
9223: * The optional Search-Order word set::
9224: @end menu
9225:
9226:
9227: @c =====================================================================
9228: @node The Core Words, The optional Block word set, ANS conformance, ANS conformance
9229: @comment node-name, next, previous, up
9230: @section The Core Words
9231: @c =====================================================================
9232: @cindex core words, system documentation
9233: @cindex system documentation, core words
9234:
9235: @menu
9236: * core-idef:: Implementation Defined Options
9237: * core-ambcond:: Ambiguous Conditions
9238: * core-other:: Other System Documentation
9239: @end menu
9240:
9241: @c ---------------------------------------------------------------------
9242: @node core-idef, core-ambcond, The Core Words, The Core Words
9243: @subsection Implementation Defined Options
9244: @c ---------------------------------------------------------------------
9245: @cindex core words, implementation-defined options
9246: @cindex implementation-defined options, core words
9247:
9248:
9249: @table @i
9250: @item (Cell) aligned addresses:
9251: @cindex cell-aligned addresses
9252: @cindex aligned addresses
9253: processor-dependent. Gforth's alignment words perform natural alignment
9254: (e.g., an address aligned for a datum of size 8 is divisible by
9255: 8). Unaligned accesses usually result in a @code{-23 THROW}.
9256:
9257: @item @code{EMIT} and non-graphic characters:
9258: @cindex @code{EMIT} and non-graphic characters
9259: @cindex non-graphic characters and @code{EMIT}
9260: The character is output using the C library function (actually, macro)
9261: @code{putc}.
9262:
9263: @item character editing of @code{ACCEPT} and @code{EXPECT}:
9264: @cindex character editing of @code{ACCEPT} and @code{EXPECT}
9265: @cindex editing in @code{ACCEPT} and @code{EXPECT}
9266: @cindex @code{ACCEPT}, editing
9267: @cindex @code{EXPECT}, editing
9268: This is modeled on the GNU readline library (@pxref{Readline
9269: Interaction, , Command Line Editing, readline, The GNU Readline
9270: Library}) with Emacs-like key bindings. @kbd{Tab} deviates a little by
9271: producing a full word completion every time you type it (instead of
9272: producing the common prefix of all completions). @xref{Command-line editing}.
9273:
9274: @item character set:
9275: @cindex character set
9276: The character set of your computer and display device. Gforth is
9277: 8-bit-clean (but some other component in your system may make trouble).
9278:
9279: @item Character-aligned address requirements:
9280: @cindex character-aligned address requirements
9281: installation-dependent. Currently a character is represented by a C
9282: @code{unsigned char}; in the future we might switch to @code{wchar_t}
9283: (Comments on that requested).
9284:
9285: @item character-set extensions and matching of names:
9286: @cindex character-set extensions and matching of names
9287: @cindex case-sensitivity for name lookup
9288: @cindex name lookup, case-sensitivity
9289: @cindex locale and case-sensitivity
9290: Any character except the ASCII NUL character can be used in a
9291: name. Matching is case-insensitive (except in @code{TABLE}s). The
9292: matching is performed using the C function @code{strncasecmp}, whose
9293: function is probably influenced by the locale. E.g., the @code{C} locale
9294: does not know about accents and umlauts, so they are matched
9295: case-sensitively in that locale. For portability reasons it is best to
9296: write programs such that they work in the @code{C} locale. Then one can
9297: use libraries written by a Polish programmer (who might use words
9298: containing ISO Latin-2 encoded characters) and by a French programmer
9299: (ISO Latin-1) in the same program (of course, @code{WORDS} will produce
9300: funny results for some of the words (which ones, depends on the font you
9301: are using)). Also, the locale you prefer may not be available in other
9302: operating systems. Hopefully, Unicode will solve these problems one day.
9303:
9304: @item conditions under which control characters match a space delimiter:
9305: @cindex space delimiters
9306: @cindex control characters as delimiters
9307: If @code{WORD} is called with the space character as a delimiter, all
9308: white-space characters (as identified by the C macro @code{isspace()})
9309: are delimiters. @code{PARSE}, on the other hand, treats space like other
9310: delimiters. @code{SWORD} treats space like @code{WORD}, but behaves
9311: like @code{PARSE} otherwise. @code{(NAME)}, which is used by the outer
9312: interpreter (aka text interpreter) by default, treats all white-space
9313: characters as delimiters.
9314:
9315: @item format of the control-flow stack:
9316: @cindex control-flow stack, format
9317: The data stack is used as control-flow stack. The size of a control-flow
9318: stack item in cells is given by the constant @code{cs-item-size}. At the
9319: time of this writing, an item consists of a (pointer to a) locals list
9320: (third), an address in the code (second), and a tag for identifying the
9321: item (TOS). The following tags are used: @code{defstart},
9322: @code{live-orig}, @code{dead-orig}, @code{dest}, @code{do-dest},
9323: @code{scopestart}.
9324:
9325: @item conversion of digits > 35
9326: @cindex digits > 35
9327: The characters @code{[\]^_'} are the digits with the decimal value
9328: 36@minus{}41. There is no way to input many of the larger digits.
9329:
9330: @item display after input terminates in @code{ACCEPT} and @code{EXPECT}:
9331: @cindex @code{EXPECT}, display after end of input
9332: @cindex @code{ACCEPT}, display after end of input
9333: The cursor is moved to the end of the entered string. If the input is
9334: terminated using the @kbd{Return} key, a space is typed.
9335:
9336: @item exception abort sequence of @code{ABORT"}:
9337: @cindex exception abort sequence of @code{ABORT"}
9338: @cindex @code{ABORT"}, exception abort sequence
9339: The error string is stored into the variable @code{"error} and a
9340: @code{-2 throw} is performed.
9341:
9342: @item input line terminator:
9343: @cindex input line terminator
9344: @cindex line terminator on input
9345: @cindex newline character on input
9346: For interactive input, @kbd{C-m} (CR) and @kbd{C-j} (LF) terminate
9347: lines. One of these characters is typically produced when you type the
9348: @kbd{Enter} or @kbd{Return} key.
9349:
9350: @item maximum size of a counted string:
9351: @cindex maximum size of a counted string
9352: @cindex counted string, maximum size
9353: @code{s" /counted-string" environment? drop .}. Currently 255 characters
9354: on all ports, but this may change.
9355:
9356: @item maximum size of a parsed string:
9357: @cindex maximum size of a parsed string
9358: @cindex parsed string, maximum size
9359: Given by the constant @code{/line}. Currently 255 characters.
9360:
9361: @item maximum size of a definition name, in characters:
9362: @cindex maximum size of a definition name, in characters
9363: @cindex name, maximum length
9364: 31
9365:
9366: @item maximum string length for @code{ENVIRONMENT?}, in characters:
9367: @cindex maximum string length for @code{ENVIRONMENT?}, in characters
9368: @cindex @code{ENVIRONMENT?} string length, maximum
9369: 31
9370:
9371: @item method of selecting the user input device:
9372: @cindex user input device, method of selecting
9373: The user input device is the standard input. There is currently no way to
9374: change it from within Gforth. However, the input can typically be
9375: redirected in the command line that starts Gforth.
9376:
9377: @item method of selecting the user output device:
9378: @cindex user output device, method of selecting
9379: @code{EMIT} and @code{TYPE} output to the file-id stored in the value
9380: @code{outfile-id} (@code{stdout} by default). Gforth uses unbuffered
9381: output when the user output device is a terminal, otherwise the output
9382: is buffered.
9383:
9384: @item methods of dictionary compilation:
9385: What are we expected to document here?
9386:
9387: @item number of bits in one address unit:
9388: @cindex number of bits in one address unit
9389: @cindex address unit, size in bits
9390: @code{s" address-units-bits" environment? drop .}. 8 in all current
9391: ports.
9392:
9393: @item number representation and arithmetic:
9394: @cindex number representation and arithmetic
9395: Processor-dependent. Binary two's complement on all current ports.
9396:
9397: @item ranges for integer types:
9398: @cindex ranges for integer types
9399: @cindex integer types, ranges
9400: Installation-dependent. Make environmental queries for @code{MAX-N},
9401: @code{MAX-U}, @code{MAX-D} and @code{MAX-UD}. The lower bounds for
9402: unsigned (and positive) types is 0. The lower bound for signed types on
9403: two's complement and one's complement machines machines can be computed
9404: by adding 1 to the upper bound.
9405:
9406: @item read-only data space regions:
9407: @cindex read-only data space regions
9408: @cindex data-space, read-only regions
9409: The whole Forth data space is writable.
9410:
9411: @item size of buffer at @code{WORD}:
9412: @cindex size of buffer at @code{WORD}
9413: @cindex @code{WORD} buffer size
9414: @code{PAD HERE - .}. 104 characters on 32-bit machines. The buffer is
9415: shared with the pictured numeric output string. If overwriting
9416: @code{PAD} is acceptable, it is as large as the remaining dictionary
9417: space, although only as much can be sensibly used as fits in a counted
9418: string.
9419:
9420: @item size of one cell in address units:
9421: @cindex cell size
9422: @code{1 cells .}.
9423:
9424: @item size of one character in address units:
9425: @cindex char size
9426: @code{1 chars .}. 1 on all current ports.
9427:
9428: @item size of the keyboard terminal buffer:
9429: @cindex size of the keyboard terminal buffer
9430: @cindex terminal buffer, size
9431: Varies. You can determine the size at a specific time using @code{lp@@
9432: tib - .}. It is shared with the locals stack and TIBs of files that
9433: include the current file. You can change the amount of space for TIBs
9434: and locals stack at Gforth startup with the command line option
9435: @code{-l}.
9436:
9437: @item size of the pictured numeric output buffer:
9438: @cindex size of the pictured numeric output buffer
9439: @cindex pictured numeric output buffer, size
9440: @code{PAD HERE - .}. 104 characters on 32-bit machines. The buffer is
9441: shared with @code{WORD}.
9442:
9443: @item size of the scratch area returned by @code{PAD}:
9444: @cindex size of the scratch area returned by @code{PAD}
9445: @cindex @code{PAD} size
9446: The remainder of dictionary space. @code{unused pad here - - .}.
9447:
9448: @item system case-sensitivity characteristics:
9449: @cindex case-sensitivity characteristics
9450: Dictionary searches are case-insensitive (except in
9451: @code{TABLE}s). However, as explained above under @i{character-set
9452: extensions}, the matching for non-ASCII characters is determined by the
9453: locale you are using. In the default @code{C} locale all non-ASCII
9454: characters are matched case-sensitively.
9455:
9456: @item system prompt:
9457: @cindex system prompt
9458: @cindex prompt
9459: @code{ ok} in interpret state, @code{ compiled} in compile state.
9460:
9461: @item division rounding:
9462: @cindex division rounding
9463: installation dependent. @code{s" floored" environment? drop .}. We leave
9464: the choice to @code{gcc} (what to use for @code{/}) and to you (whether
9465: to use @code{fm/mod}, @code{sm/rem} or simply @code{/}).
9466:
9467: @item values of @code{STATE} when true:
9468: @cindex @code{STATE} values
9469: -1.
9470:
9471: @item values returned after arithmetic overflow:
9472: On two's complement machines, arithmetic is performed modulo
9473: 2**bits-per-cell for single arithmetic and 4**bits-per-cell for double
9474: arithmetic (with appropriate mapping for signed types). Division by zero
9475: typically results in a @code{-55 throw} (Floating-point unidentified
9476: fault), although a @code{-10 throw} (divide by zero) would be more
9477: appropriate.
9478:
9479: @item whether the current definition can be found after @t{DOES>}:
9480: @cindex @t{DOES>}, visibility of current definition
9481: No.
9482:
9483: @end table
9484:
9485: @c ---------------------------------------------------------------------
9486: @node core-ambcond, core-other, core-idef, The Core Words
9487: @subsection Ambiguous conditions
9488: @c ---------------------------------------------------------------------
9489: @cindex core words, ambiguous conditions
9490: @cindex ambiguous conditions, core words
9491:
9492: @table @i
9493:
9494: @item a name is neither a word nor a number:
9495: @cindex name not found
9496: @cindex undefined word
9497: @code{-13 throw} (Undefined word). Actually, @code{-13 bounce}, which
9498: preserves the data and FP stack, so you don't lose more work than
9499: necessary.
9500:
9501: @item a definition name exceeds the maximum length allowed:
9502: @cindex word name too long
9503: @code{-19 throw} (Word name too long)
9504:
9505: @item addressing a region not inside the various data spaces of the forth system:
9506: @cindex Invalid memory address
9507: The stacks, code space and header space are accessible. Machine code space is
9508: typically readable. Accessing other addresses gives results dependent on
9509: the operating system. On decent systems: @code{-9 throw} (Invalid memory
9510: address).
9511:
9512: @item argument type incompatible with parameter:
9513: @cindex argument type mismatch
9514: This is usually not caught. Some words perform checks, e.g., the control
9515: flow words, and issue a @code{ABORT"} or @code{-12 THROW} (Argument type
9516: mismatch).
9517:
9518: @item attempting to obtain the execution token of a word with undefined execution semantics:
9519: @cindex Interpreting a compile-only word, for @code{'} etc.
9520: @cindex execution token of words with undefined execution semantics
9521: @code{-14 throw} (Interpreting a compile-only word). In some cases, you
9522: get an execution token for @code{compile-only-error} (which performs a
9523: @code{-14 throw} when executed).
9524:
9525: @item dividing by zero:
9526: @cindex dividing by zero
9527: @cindex floating point unidentified fault, integer division
9528: On better platforms, this produces a @code{-10 throw} (Division by
9529: zero); on other systems, this typically results in a @code{-55 throw}
9530: (Floating-point unidentified fault).
9531:
9532: @item insufficient data stack or return stack space:
9533: @cindex insufficient data stack or return stack space
9534: @cindex stack overflow
9535: @cindex address alignment exception, stack overflow
9536: @cindex Invalid memory address, stack overflow
9537: Depending on the operating system, the installation, and the invocation
9538: of Gforth, this is either checked by the memory management hardware, or
9539: it is not checked. If it is checked, you typically get a @code{-3 throw}
9540: (Stack overflow), @code{-5 throw} (Return stack overflow), or @code{-9
9541: throw} (Invalid memory address) (depending on the platform and how you
9542: achieved the overflow) as soon as the overflow happens. If it is not
9543: checked, overflows typically result in mysterious illegal memory
9544: accesses, producing @code{-9 throw} (Invalid memory address) or
9545: @code{-23 throw} (Address alignment exception); they might also destroy
9546: the internal data structure of @code{ALLOCATE} and friends, resulting in
9547: various errors in these words.
9548:
9549: @item insufficient space for loop control parameters:
9550: @cindex insufficient space for loop control parameters
9551: like other return stack overflows.
9552:
9553: @item insufficient space in the dictionary:
9554: @cindex insufficient space in the dictionary
9555: @cindex dictionary overflow
9556: If you try to allot (either directly with @code{allot}, or indirectly
9557: with @code{,}, @code{create} etc.) more memory than available in the
9558: dictionary, you get a @code{-8 throw} (Dictionary overflow). If you try
9559: to access memory beyond the end of the dictionary, the results are
9560: similar to stack overflows.
9561:
9562: @item interpreting a word with undefined interpretation semantics:
9563: @cindex interpreting a word with undefined interpretation semantics
9564: @cindex Interpreting a compile-only word
9565: For some words, we have defined interpretation semantics. For the
9566: others: @code{-14 throw} (Interpreting a compile-only word).
9567:
9568: @item modifying the contents of the input buffer or a string literal:
9569: @cindex modifying the contents of the input buffer or a string literal
9570: These are located in writable memory and can be modified.
9571:
9572: @item overflow of the pictured numeric output string:
9573: @cindex overflow of the pictured numeric output string
9574: @cindex pictured numeric output string, overflow
9575: @code{-17 throw} (Pictured numeric ouput string overflow).
9576:
9577: @item parsed string overflow:
9578: @cindex parsed string overflow
9579: @code{PARSE} cannot overflow. @code{WORD} does not check for overflow.
9580:
9581: @item producing a result out of range:
9582: @cindex result out of range
9583: On two's complement machines, arithmetic is performed modulo
9584: 2**bits-per-cell for single arithmetic and 4**bits-per-cell for double
9585: arithmetic (with appropriate mapping for signed types). Division by zero
9586: typically results in a @code{-10 throw} (divide by zero) or @code{-55
9587: throw} (floating point unidentified fault). @code{convert} and
9588: @code{>number} currently overflow silently.
9589:
9590: @item reading from an empty data or return stack:
9591: @cindex stack empty
9592: @cindex stack underflow
9593: @cindex return stack underflow
9594: The data stack is checked by the outer (aka text) interpreter after
9595: every word executed. If it has underflowed, a @code{-4 throw} (Stack
9596: underflow) is performed. Apart from that, stacks may be checked or not,
9597: depending on operating system, installation, and invocation. If they are
9598: caught by a check, they typically result in @code{-4 throw} (Stack
9599: underflow), @code{-6 throw} (Return stack underflow) or @code{-9 throw}
9600: (Invalid memory address), depending on the platform and which stack
9601: underflows and by how much. Note that even if the system uses checking
9602: (through the MMU), your program may have to underflow by a significant
9603: number of stack items to trigger the reaction (the reason for this is
9604: that the MMU, and therefore the checking, works with a page-size
9605: granularity). If there is no checking, the symptoms resulting from an
9606: underflow are similar to those from an overflow. Unbalanced return
9607: stack errors result in a variaty of symptoms, including @code{-9 throw}
9608: (Invalid memory address) and Illegal Instruction (typically @code{-260
9609: throw}).
9610:
9611: @item unexpected end of the input buffer, resulting in an attempt to use a zero-length string as a name:
9612: @cindex unexpected end of the input buffer
9613: @cindex zero-length string as a name
9614: @cindex Attempt to use zero-length string as a name
9615: @code{Create} and its descendants perform a @code{-16 throw} (Attempt to
9616: use zero-length string as a name). Words like @code{'} probably will not
9617: find what they search. Note that it is possible to create zero-length
9618: names with @code{nextname} (should it not?).
9619:
9620: @item @code{>IN} greater than input buffer:
9621: @cindex @code{>IN} greater than input buffer
9622: The next invocation of a parsing word returns a string with length 0.
9623:
9624: @item @code{RECURSE} appears after @code{DOES>}:
9625: @cindex @code{RECURSE} appears after @code{DOES>}
9626: Compiles a recursive call to the defining word, not to the defined word.
9627:
9628: @item argument input source different than current input source for @code{RESTORE-INPUT}:
9629: @cindex argument input source different than current input source for @code{RESTORE-INPUT}
9630: @cindex argument type mismatch, @code{RESTORE-INPUT}
9631: @cindex @code{RESTORE-INPUT}, Argument type mismatch
9632: @code{-12 THROW}. Note that, once an input file is closed (e.g., because
9633: the end of the file was reached), its source-id may be
9634: reused. Therefore, restoring an input source specification referencing a
9635: closed file may lead to unpredictable results instead of a @code{-12
9636: THROW}.
9637:
9638: In the future, Gforth may be able to restore input source specifications
9639: from other than the current input source.
9640:
9641: @item data space containing definitions gets de-allocated:
9642: @cindex data space containing definitions gets de-allocated
9643: Deallocation with @code{allot} is not checked. This typically results in
9644: memory access faults or execution of illegal instructions.
9645:
9646: @item data space read/write with incorrect alignment:
9647: @cindex data space read/write with incorrect alignment
9648: @cindex alignment faults
9649: @cindex address alignment exception
9650: Processor-dependent. Typically results in a @code{-23 throw} (Address
9651: alignment exception). Under Linux-Intel on a 486 or later processor with
9652: alignment turned on, incorrect alignment results in a @code{-9 throw}
9653: (Invalid memory address). There are reportedly some processors with
9654: alignment restrictions that do not report violations.
9655:
9656: @item data space pointer not properly aligned, @code{,}, @code{C,}:
9657: @cindex data space pointer not properly aligned, @code{,}, @code{C,}
9658: Like other alignment errors.
9659:
9660: @item less than u+2 stack items (@code{PICK} and @code{ROLL}):
9661: Like other stack underflows.
9662:
9663: @item loop control parameters not available:
9664: @cindex loop control parameters not available
9665: Not checked. The counted loop words simply assume that the top of return
9666: stack items are loop control parameters and behave accordingly.
9667:
9668: @item most recent definition does not have a name (@code{IMMEDIATE}):
9669: @cindex most recent definition does not have a name (@code{IMMEDIATE})
9670: @cindex last word was headerless
9671: @code{abort" last word was headerless"}.
9672:
9673: @item name not defined by @code{VALUE} used by @code{TO}:
9674: @cindex name not defined by @code{VALUE} used by @code{TO}
9675: @cindex @code{TO} on non-@code{VALUE}s
9676: @cindex Invalid name argument, @code{TO}
9677: @code{-32 throw} (Invalid name argument) (unless name is a local or was
9678: defined by @code{CONSTANT}; in the latter case it just changes the constant).
9679:
9680: @item name not found (@code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]}):
9681: @cindex name not found (@code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]})
9682: @cindex undefined word, @code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]}
9683: @code{-13 throw} (Undefined word)
9684:
9685: @item parameters are not of the same type (@code{DO}, @code{?DO}, @code{WITHIN}):
9686: @cindex parameters are not of the same type (@code{DO}, @code{?DO}, @code{WITHIN})
9687: Gforth behaves as if they were of the same type. I.e., you can predict
9688: the behaviour by interpreting all parameters as, e.g., signed.
9689:
9690: @item @code{POSTPONE} or @code{[COMPILE]} applied to @code{TO}:
9691: @cindex @code{POSTPONE} or @code{[COMPILE]} applied to @code{TO}
9692: Assume @code{: X POSTPONE TO ; IMMEDIATE}. @code{X} performs the
9693: compilation semantics of @code{TO}.
9694:
9695: @item String longer than a counted string returned by @code{WORD}:
9696: @cindex string longer than a counted string returned by @code{WORD}
9697: @cindex @code{WORD}, string overflow
9698: Not checked. The string will be ok, but the count will, of course,
9699: contain only the least significant bits of the length.
9700:
9701: @item u greater than or equal to the number of bits in a cell (@code{LSHIFT}, @code{RSHIFT}):
9702: @cindex @code{LSHIFT}, large shift counts
9703: @cindex @code{RSHIFT}, large shift counts
9704: Processor-dependent. Typical behaviours are returning 0 and using only
9705: the low bits of the shift count.
9706:
9707: @item word not defined via @code{CREATE}:
9708: @cindex @code{>BODY} of non-@code{CREATE}d words
9709: @code{>BODY} produces the PFA of the word no matter how it was defined.
9710:
9711: @cindex @code{DOES>} of non-@code{CREATE}d words
9712: @code{DOES>} changes the execution semantics of the last defined word no
9713: matter how it was defined. E.g., @code{CONSTANT DOES>} is equivalent to
9714: @code{CREATE , DOES>}.
9715:
9716: @item words improperly used outside @code{<#} and @code{#>}:
9717: Not checked. As usual, you can expect memory faults.
9718:
9719: @end table
9720:
9721:
9722: @c ---------------------------------------------------------------------
9723: @node core-other, , core-ambcond, The Core Words
9724: @subsection Other system documentation
9725: @c ---------------------------------------------------------------------
9726: @cindex other system documentation, core words
9727: @cindex core words, other system documentation
9728:
9729: @table @i
9730: @item nonstandard words using @code{PAD}:
9731: @cindex @code{PAD} use by nonstandard words
9732: None.
9733:
9734: @item operator's terminal facilities available:
9735: @cindex operator's terminal facilities available
9736: After processing the command line, Gforth goes into interactive mode,
9737: and you can give commands to Gforth interactively. The actual facilities
9738: available depend on how you invoke Gforth.
9739:
9740: @item program data space available:
9741: @cindex program data space available
9742: @cindex data space available
9743: @code{UNUSED .} gives the remaining dictionary space. The total
9744: dictionary space can be specified with the @code{-m} switch
9745: (@pxref{Invoking Gforth}) when Gforth starts up.
9746:
9747: @item return stack space available:
9748: @cindex return stack space available
9749: You can compute the total return stack space in cells with
9750: @code{s" RETURN-STACK-CELLS" environment? drop .}. You can specify it at
9751: startup time with the @code{-r} switch (@pxref{Invoking Gforth}).
9752:
9753: @item stack space available:
9754: @cindex stack space available
9755: You can compute the total data stack space in cells with
9756: @code{s" STACK-CELLS" environment? drop .}. You can specify it at
9757: startup time with the @code{-d} switch (@pxref{Invoking Gforth}).
9758:
9759: @item system dictionary space required, in address units:
9760: @cindex system dictionary space required, in address units
9761: Type @code{here forthstart - .} after startup. At the time of this
9762: writing, this gives 80080 (bytes) on a 32-bit system.
9763: @end table
9764:
9765:
9766: @c =====================================================================
9767: @node The optional Block word set, The optional Double Number word set, The Core Words, ANS conformance
9768: @section The optional Block word set
9769: @c =====================================================================
9770: @cindex system documentation, block words
9771: @cindex block words, system documentation
9772:
9773: @menu
9774: * block-idef:: Implementation Defined Options
9775: * block-ambcond:: Ambiguous Conditions
9776: * block-other:: Other System Documentation
9777: @end menu
9778:
9779:
9780: @c ---------------------------------------------------------------------
9781: @node block-idef, block-ambcond, The optional Block word set, The optional Block word set
9782: @subsection Implementation Defined Options
9783: @c ---------------------------------------------------------------------
9784: @cindex implementation-defined options, block words
9785: @cindex block words, implementation-defined options
9786:
9787: @table @i
9788: @item the format for display by @code{LIST}:
9789: @cindex @code{LIST} display format
9790: First the screen number is displayed, then 16 lines of 64 characters,
9791: each line preceded by the line number.
9792:
9793: @item the length of a line affected by @code{\}:
9794: @cindex length of a line affected by @code{\}
9795: @cindex @code{\}, line length in blocks
9796: 64 characters.
9797: @end table
9798:
9799:
9800: @c ---------------------------------------------------------------------
9801: @node block-ambcond, block-other, block-idef, The optional Block word set
9802: @subsection Ambiguous conditions
9803: @c ---------------------------------------------------------------------
9804: @cindex block words, ambiguous conditions
9805: @cindex ambiguous conditions, block words
9806:
9807: @table @i
9808: @item correct block read was not possible:
9809: @cindex block read not possible
9810: Typically results in a @code{throw} of some OS-derived value (between
9811: -512 and -2048). If the blocks file was just not long enough, blanks are
9812: supplied for the missing portion.
9813:
9814: @item I/O exception in block transfer:
9815: @cindex I/O exception in block transfer
9816: @cindex block transfer, I/O exception
9817: Typically results in a @code{throw} of some OS-derived value (between
9818: -512 and -2048).
9819:
9820: @item invalid block number:
9821: @cindex invalid block number
9822: @cindex block number invalid
9823: @code{-35 throw} (Invalid block number)
9824:
9825: @item a program directly alters the contents of @code{BLK}:
9826: @cindex @code{BLK}, altering @code{BLK}
9827: The input stream is switched to that other block, at the same
9828: position. If the storing to @code{BLK} happens when interpreting
9829: non-block input, the system will get quite confused when the block ends.
9830:
9831: @item no current block buffer for @code{UPDATE}:
9832: @cindex @code{UPDATE}, no current block buffer
9833: @code{UPDATE} has no effect.
9834:
9835: @end table
9836:
9837: @c ---------------------------------------------------------------------
9838: @node block-other, , block-ambcond, The optional Block word set
9839: @subsection Other system documentation
9840: @c ---------------------------------------------------------------------
9841: @cindex other system documentation, block words
9842: @cindex block words, other system documentation
9843:
9844: @table @i
9845: @item any restrictions a multiprogramming system places on the use of buffer addresses:
9846: No restrictions (yet).
9847:
9848: @item the number of blocks available for source and data:
9849: depends on your disk space.
9850:
9851: @end table
9852:
9853:
9854: @c =====================================================================
9855: @node The optional Double Number word set, The optional Exception word set, The optional Block word set, ANS conformance
9856: @section The optional Double Number word set
9857: @c =====================================================================
9858: @cindex system documentation, double words
9859: @cindex double words, system documentation
9860:
9861: @menu
9862: * double-ambcond:: Ambiguous Conditions
9863: @end menu
9864:
9865:
9866: @c ---------------------------------------------------------------------
9867: @node double-ambcond, , The optional Double Number word set, The optional Double Number word set
9868: @subsection Ambiguous conditions
9869: @c ---------------------------------------------------------------------
9870: @cindex double words, ambiguous conditions
9871: @cindex ambiguous conditions, double words
9872:
9873: @table @i
9874: @item @i{d} outside of range of @i{n} in @code{D>S}:
9875: @cindex @code{D>S}, @i{d} out of range of @i{n}
9876: The least significant cell of @i{d} is produced.
9877:
9878: @end table
9879:
9880:
9881: @c =====================================================================
9882: @node The optional Exception word set, The optional Facility word set, The optional Double Number word set, ANS conformance
9883: @section The optional Exception word set
9884: @c =====================================================================
9885: @cindex system documentation, exception words
9886: @cindex exception words, system documentation
9887:
9888: @menu
9889: * exception-idef:: Implementation Defined Options
9890: @end menu
9891:
9892:
9893: @c ---------------------------------------------------------------------
9894: @node exception-idef, , The optional Exception word set, The optional Exception word set
9895: @subsection Implementation Defined Options
9896: @c ---------------------------------------------------------------------
9897: @cindex implementation-defined options, exception words
9898: @cindex exception words, implementation-defined options
9899:
9900: @table @i
9901: @item @code{THROW}-codes used in the system:
9902: @cindex @code{THROW}-codes used in the system
9903: The codes -256@minus{}-511 are used for reporting signals. The mapping
9904: from OS signal numbers to throw codes is -256@minus{}@i{signal}. The
9905: codes -512@minus{}-2047 are used for OS errors (for file and memory
9906: allocation operations). The mapping from OS error numbers to throw codes
9907: is -512@minus{}@code{errno}. One side effect of this mapping is that
9908: undefined OS errors produce a message with a strange number; e.g.,
9909: @code{-1000 THROW} results in @code{Unknown error 488} on my system.
9910: @end table
9911:
9912: @c =====================================================================
9913: @node The optional Facility word set, The optional File-Access word set, The optional Exception word set, ANS conformance
9914: @section The optional Facility word set
9915: @c =====================================================================
9916: @cindex system documentation, facility words
9917: @cindex facility words, system documentation
9918:
9919: @menu
9920: * facility-idef:: Implementation Defined Options
9921: * facility-ambcond:: Ambiguous Conditions
9922: @end menu
9923:
9924:
9925: @c ---------------------------------------------------------------------
9926: @node facility-idef, facility-ambcond, The optional Facility word set, The optional Facility word set
9927: @subsection Implementation Defined Options
9928: @c ---------------------------------------------------------------------
9929: @cindex implementation-defined options, facility words
9930: @cindex facility words, implementation-defined options
9931:
9932: @table @i
9933: @item encoding of keyboard events (@code{EKEY}):
9934: @cindex keyboard events, encoding in @code{EKEY}
9935: @cindex @code{EKEY}, encoding of keyboard events
9936: Keys corresponding to ASCII characters are encoded as ASCII characters.
9937: Other keys are encoded with the constants @code{k-left}, @code{k-right},
9938: @code{k-up}, @code{k-down}, @code{k-home}, @code{k-end}, @code{k1},
9939: @code{k2}, @code{k3}, @code{k4}, @code{k5}, @code{k6}, @code{k7},
9940: @code{k8}, @code{k9}, @code{k10}, @code{k11}, @code{k12}.
9941:
9942:
9943: @item duration of a system clock tick:
9944: @cindex duration of a system clock tick
9945: @cindex clock tick duration
9946: System dependent. With respect to @code{MS}, the time is specified in
9947: microseconds. How well the OS and the hardware implement this, is
9948: another question.
9949:
9950: @item repeatability to be expected from the execution of @code{MS}:
9951: @cindex repeatability to be expected from the execution of @code{MS}
9952: @cindex @code{MS}, repeatability to be expected
9953: System dependent. On Unix, a lot depends on load. If the system is
9954: lightly loaded, and the delay is short enough that Gforth does not get
9955: swapped out, the performance should be acceptable. Under MS-DOS and
9956: other single-tasking systems, it should be good.
9957:
9958: @end table
9959:
9960:
9961: @c ---------------------------------------------------------------------
9962: @node facility-ambcond, , facility-idef, The optional Facility word set
9963: @subsection Ambiguous conditions
9964: @c ---------------------------------------------------------------------
9965: @cindex facility words, ambiguous conditions
9966: @cindex ambiguous conditions, facility words
9967:
9968: @table @i
9969: @item @code{AT-XY} can't be performed on user output device:
9970: @cindex @code{AT-XY} can't be performed on user output device
9971: Largely terminal dependent. No range checks are done on the arguments.
9972: No errors are reported. You may see some garbage appearing, you may see
9973: simply nothing happen.
9974:
9975: @end table
9976:
9977:
9978: @c =====================================================================
9979: @node The optional File-Access word set, The optional Floating-Point word set, The optional Facility word set, ANS conformance
9980: @section The optional File-Access word set
9981: @c =====================================================================
9982: @cindex system documentation, file words
9983: @cindex file words, system documentation
9984:
9985: @menu
9986: * file-idef:: Implementation Defined Options
9987: * file-ambcond:: Ambiguous Conditions
9988: @end menu
9989:
9990: @c ---------------------------------------------------------------------
9991: @node file-idef, file-ambcond, The optional File-Access word set, The optional File-Access word set
9992: @subsection Implementation Defined Options
9993: @c ---------------------------------------------------------------------
9994: @cindex implementation-defined options, file words
9995: @cindex file words, implementation-defined options
9996:
9997: @table @i
9998: @item file access methods used:
9999: @cindex file access methods used
10000: @code{R/O}, @code{R/W} and @code{BIN} work as you would
10001: expect. @code{W/O} translates into the C file opening mode @code{w} (or
10002: @code{wb}): The file is cleared, if it exists, and created, if it does
10003: not (with both @code{open-file} and @code{create-file}). Under Unix
10004: @code{create-file} creates a file with 666 permissions modified by your
10005: umask.
10006:
10007: @item file exceptions:
10008: @cindex file exceptions
10009: The file words do not raise exceptions (except, perhaps, memory access
10010: faults when you pass illegal addresses or file-ids).
10011:
10012: @item file line terminator:
10013: @cindex file line terminator
10014: System-dependent. Gforth uses C's newline character as line
10015: terminator. What the actual character code(s) of this are is
10016: system-dependent.
10017:
10018: @item file name format:
10019: @cindex file name format
10020: System dependent. Gforth just uses the file name format of your OS.
10021:
10022: @item information returned by @code{FILE-STATUS}:
10023: @cindex @code{FILE-STATUS}, returned information
10024: @code{FILE-STATUS} returns the most powerful file access mode allowed
10025: for the file: Either @code{R/O}, @code{W/O} or @code{R/W}. If the file
10026: cannot be accessed, @code{R/O BIN} is returned. @code{BIN} is applicable
10027: along with the returned mode.
10028:
10029: @item input file state after an exception when including source:
10030: @cindex exception when including source
10031: All files that are left via the exception are closed.
10032:
10033: @item @i{ior} values and meaning:
10034: @cindex @i{ior} values and meaning
10035: The @i{ior}s returned by the file and memory allocation words are
10036: intended as throw codes. They typically are in the range
10037: -512@minus{}-2047 of OS errors. The mapping from OS error numbers to
10038: @i{ior}s is -512@minus{}@i{errno}.
10039:
10040: @item maximum depth of file input nesting:
10041: @cindex maximum depth of file input nesting
10042: @cindex file input nesting, maximum depth
10043: limited by the amount of return stack, locals/TIB stack, and the number
10044: of open files available. This should not give you troubles.
10045:
10046: @item maximum size of input line:
10047: @cindex maximum size of input line
10048: @cindex input line size, maximum
10049: @code{/line}. Currently 255.
10050:
10051: @item methods of mapping block ranges to files:
10052: @cindex mapping block ranges to files
10053: @cindex files containing blocks
10054: @cindex blocks in files
10055: By default, blocks are accessed in the file @file{blocks.fb} in the
10056: current working directory. The file can be switched with @code{USE}.
10057:
10058: @item number of string buffers provided by @code{S"}:
10059: @cindex @code{S"}, number of string buffers
10060: 1
10061:
10062: @item size of string buffer used by @code{S"}:
10063: @cindex @code{S"}, size of string buffer
10064: @code{/line}. currently 255.
10065:
10066: @end table
10067:
10068: @c ---------------------------------------------------------------------
10069: @node file-ambcond, , file-idef, The optional File-Access word set
10070: @subsection Ambiguous conditions
10071: @c ---------------------------------------------------------------------
10072: @cindex file words, ambiguous conditions
10073: @cindex ambiguous conditions, file words
10074:
10075: @table @i
10076: @item attempting to position a file outside its boundaries:
10077: @cindex @code{REPOSITION-FILE}, outside the file's boundaries
10078: @code{REPOSITION-FILE} is performed as usual: Afterwards,
10079: @code{FILE-POSITION} returns the value given to @code{REPOSITION-FILE}.
10080:
10081: @item attempting to read from file positions not yet written:
10082: @cindex reading from file positions not yet written
10083: End-of-file, i.e., zero characters are read and no error is reported.
10084:
10085: @item @i{file-id} is invalid (@code{INCLUDE-FILE}):
10086: @cindex @code{INCLUDE-FILE}, @i{file-id} is invalid
10087: An appropriate exception may be thrown, but a memory fault or other
10088: problem is more probable.
10089:
10090: @item I/O exception reading or closing @i{file-id} (@code{INCLUDE-FILE}, @code{INCLUDED}):
10091: @cindex @code{INCLUDE-FILE}, I/O exception reading or closing @i{file-id}
10092: @cindex @code{INCLUDED}, I/O exception reading or closing @i{file-id}
10093: The @i{ior} produced by the operation, that discovered the problem, is
10094: thrown.
10095:
10096: @item named file cannot be opened (@code{INCLUDED}):
10097: @cindex @code{INCLUDED}, named file cannot be opened
10098: The @i{ior} produced by @code{open-file} is thrown.
10099:
10100: @item requesting an unmapped block number:
10101: @cindex unmapped block numbers
10102: There are no unmapped legal block numbers. On some operating systems,
10103: writing a block with a large number may overflow the file system and
10104: have an error message as consequence.
10105:
10106: @item using @code{source-id} when @code{blk} is non-zero:
10107: @cindex @code{SOURCE-ID}, behaviour when @code{BLK} is non-zero
10108: @code{source-id} performs its function. Typically it will give the id of
10109: the source which loaded the block. (Better ideas?)
10110:
10111: @end table
10112:
10113:
10114: @c =====================================================================
10115: @node The optional Floating-Point word set, The optional Locals word set, The optional File-Access word set, ANS conformance
10116: @section The optional Floating-Point word set
10117: @c =====================================================================
10118: @cindex system documentation, floating-point words
10119: @cindex floating-point words, system documentation
10120:
10121: @menu
10122: * floating-idef:: Implementation Defined Options
10123: * floating-ambcond:: Ambiguous Conditions
10124: @end menu
10125:
10126:
10127: @c ---------------------------------------------------------------------
10128: @node floating-idef, floating-ambcond, The optional Floating-Point word set, The optional Floating-Point word set
10129: @subsection Implementation Defined Options
10130: @c ---------------------------------------------------------------------
10131: @cindex implementation-defined options, floating-point words
10132: @cindex floating-point words, implementation-defined options
10133:
10134: @table @i
10135: @item format and range of floating point numbers:
10136: @cindex format and range of floating point numbers
10137: @cindex floating point numbers, format and range
10138: System-dependent; the @code{double} type of C.
10139:
10140: @item results of @code{REPRESENT} when @i{float} is out of range:
10141: @cindex @code{REPRESENT}, results when @i{float} is out of range
10142: System dependent; @code{REPRESENT} is implemented using the C library
10143: function @code{ecvt()} and inherits its behaviour in this respect.
10144:
10145: @item rounding or truncation of floating-point numbers:
10146: @cindex rounding of floating-point numbers
10147: @cindex truncation of floating-point numbers
10148: @cindex floating-point numbers, rounding or truncation
10149: System dependent; the rounding behaviour is inherited from the hosting C
10150: compiler. IEEE-FP-based (i.e., most) systems by default round to
10151: nearest, and break ties by rounding to even (i.e., such that the last
10152: bit of the mantissa is 0).
10153:
10154: @item size of floating-point stack:
10155: @cindex floating-point stack size
10156: @code{s" FLOATING-STACK" environment? drop .} gives the total size of
10157: the floating-point stack (in floats). You can specify this on startup
10158: with the command-line option @code{-f} (@pxref{Invoking Gforth}).
10159:
10160: @item width of floating-point stack:
10161: @cindex floating-point stack width
10162: @code{1 floats}.
10163:
10164: @end table
10165:
10166:
10167: @c ---------------------------------------------------------------------
10168: @node floating-ambcond, , floating-idef, The optional Floating-Point word set
10169: @subsection Ambiguous conditions
10170: @c ---------------------------------------------------------------------
10171: @cindex floating-point words, ambiguous conditions
10172: @cindex ambiguous conditions, floating-point words
10173:
10174: @table @i
10175: @item @code{df@@} or @code{df!} used with an address that is not double-float aligned:
10176: @cindex @code{df@@} or @code{df!} used with an address that is not double-float aligned
10177: System-dependent. Typically results in a @code{-23 THROW} like other
10178: alignment violations.
10179:
10180: @item @code{f@@} or @code{f!} used with an address that is not float aligned:
10181: @cindex @code{f@@} used with an address that is not float aligned
10182: @cindex @code{f!} used with an address that is not float aligned
10183: System-dependent. Typically results in a @code{-23 THROW} like other
10184: alignment violations.
10185:
10186: @item floating-point result out of range:
10187: @cindex floating-point result out of range
10188: System-dependent. Can result in a @code{-55 THROW} (Floating-point
10189: unidentified fault), or can produce a special value representing, e.g.,
10190: Infinity.
10191:
10192: @item @code{sf@@} or @code{sf!} used with an address that is not single-float aligned:
10193: @cindex @code{sf@@} or @code{sf!} used with an address that is not single-float aligned
10194: System-dependent. Typically results in an alignment fault like other
10195: alignment violations.
10196:
10197: @item @code{base} is not decimal (@code{REPRESENT}, @code{F.}, @code{FE.}, @code{FS.}):
10198: @cindex @code{base} is not decimal (@code{REPRESENT}, @code{F.}, @code{FE.}, @code{FS.})
10199: The floating-point number is converted into decimal nonetheless.
10200:
10201: @item Both arguments are equal to zero (@code{FATAN2}):
10202: @cindex @code{FATAN2}, both arguments are equal to zero
10203: System-dependent. @code{FATAN2} is implemented using the C library
10204: function @code{atan2()}.
10205:
10206: @item Using @code{FTAN} on an argument @i{r1} where cos(@i{r1}) is zero:
10207: @cindex @code{FTAN} on an argument @i{r1} where cos(@i{r1}) is zero
10208: System-dependent. Anyway, typically the cos of @i{r1} will not be zero
10209: because of small errors and the tan will be a very large (or very small)
10210: but finite number.
10211:
10212: @item @i{d} cannot be presented precisely as a float in @code{D>F}:
10213: @cindex @code{D>F}, @i{d} cannot be presented precisely as a float
10214: The result is rounded to the nearest float.
10215:
10216: @item dividing by zero:
10217: @cindex dividing by zero, floating-point
10218: @cindex floating-point dividing by zero
10219: @cindex floating-point unidentified fault, FP divide-by-zero
10220: @code{-55 throw} (Floating-point unidentified fault)
10221:
10222: @item exponent too big for conversion (@code{DF!}, @code{DF@@}, @code{SF!}, @code{SF@@}):
10223: @cindex exponent too big for conversion (@code{DF!}, @code{DF@@}, @code{SF!}, @code{SF@@})
10224: System dependent. On IEEE-FP based systems the number is converted into
10225: an infinity.
10226:
10227: @item @i{float}<1 (@code{FACOSH}):
10228: @cindex @code{FACOSH}, @i{float}<1
10229: @cindex floating-point unidentified fault, @code{FACOSH}
10230: @code{-55 throw} (Floating-point unidentified fault)
10231:
10232: @item @i{float}=<-1 (@code{FLNP1}):
10233: @cindex @code{FLNP1}, @i{float}=<-1
10234: @cindex floating-point unidentified fault, @code{FLNP1}
10235: @code{-55 throw} (Floating-point unidentified fault). On IEEE-FP systems
10236: negative infinity is typically produced for @i{float}=-1.
10237:
10238: @item @i{float}=<0 (@code{FLN}, @code{FLOG}):
10239: @cindex @code{FLN}, @i{float}=<0
10240: @cindex @code{FLOG}, @i{float}=<0
10241: @cindex floating-point unidentified fault, @code{FLN} or @code{FLOG}
10242: @code{-55 throw} (Floating-point unidentified fault). On IEEE-FP systems
10243: negative infinity is typically produced for @i{float}=0.
10244:
10245: @item @i{float}<0 (@code{FASINH}, @code{FSQRT}):
10246: @cindex @code{FASINH}, @i{float}<0
10247: @cindex @code{FSQRT}, @i{float}<0
10248: @cindex floating-point unidentified fault, @code{FASINH} or @code{FSQRT}
10249: @code{-55 throw} (Floating-point unidentified fault). @code{fasinh}
10250: produces values for these inputs on my Linux box (Bug in the C library?)
10251:
10252: @item |@i{float}|>1 (@code{FACOS}, @code{FASIN}, @code{FATANH}):
10253: @cindex @code{FACOS}, |@i{float}|>1
10254: @cindex @code{FASIN}, |@i{float}|>1
10255: @cindex @code{FATANH}, |@i{float}|>1
10256: @cindex floating-point unidentified fault, @code{FACOS}, @code{FASIN} or @code{FATANH}
10257: @code{-55 throw} (Floating-point unidentified fault).
10258:
10259: @item integer part of float cannot be represented by @i{d} in @code{F>D}:
10260: @cindex @code{F>D}, integer part of float cannot be represented by @i{d}
10261: @cindex floating-point unidentified fault, @code{F>D}
10262: @code{-55 throw} (Floating-point unidentified fault).
10263:
10264: @item string larger than pictured numeric output area (@code{f.}, @code{fe.}, @code{fs.}):
10265: @cindex string larger than pictured numeric output area (@code{f.}, @code{fe.}, @code{fs.})
10266: This does not happen.
10267: @end table
10268:
10269: @c =====================================================================
10270: @node The optional Locals word set, The optional Memory-Allocation word set, The optional Floating-Point word set, ANS conformance
10271: @section The optional Locals word set
10272: @c =====================================================================
10273: @cindex system documentation, locals words
10274: @cindex locals words, system documentation
10275:
10276: @menu
10277: * locals-idef:: Implementation Defined Options
10278: * locals-ambcond:: Ambiguous Conditions
10279: @end menu
10280:
10281:
10282: @c ---------------------------------------------------------------------
10283: @node locals-idef, locals-ambcond, The optional Locals word set, The optional Locals word set
10284: @subsection Implementation Defined Options
10285: @c ---------------------------------------------------------------------
10286: @cindex implementation-defined options, locals words
10287: @cindex locals words, implementation-defined options
10288:
10289: @table @i
10290: @item maximum number of locals in a definition:
10291: @cindex maximum number of locals in a definition
10292: @cindex locals, maximum number in a definition
10293: @code{s" #locals" environment? drop .}. Currently 15. This is a lower
10294: bound, e.g., on a 32-bit machine there can be 41 locals of up to 8
10295: characters. The number of locals in a definition is bounded by the size
10296: of locals-buffer, which contains the names of the locals.
10297:
10298: @end table
10299:
10300:
10301: @c ---------------------------------------------------------------------
10302: @node locals-ambcond, , locals-idef, The optional Locals word set
10303: @subsection Ambiguous conditions
10304: @c ---------------------------------------------------------------------
10305: @cindex locals words, ambiguous conditions
10306: @cindex ambiguous conditions, locals words
10307:
10308: @table @i
10309: @item executing a named local in interpretation state:
10310: @cindex local in interpretation state
10311: @cindex Interpreting a compile-only word, for a local
10312: Locals have no interpretation semantics. If you try to perform the
10313: interpretation semantics, you will get a @code{-14 throw} somewhere
10314: (Interpreting a compile-only word). If you perform the compilation
10315: semantics, the locals access will be compiled (irrespective of state).
10316:
10317: @item @i{name} not defined by @code{VALUE} or @code{(LOCAL)} (@code{TO}):
10318: @cindex name not defined by @code{VALUE} or @code{(LOCAL)} used by @code{TO}
10319: @cindex @code{TO} on non-@code{VALUE}s and non-locals
10320: @cindex Invalid name argument, @code{TO}
10321: @code{-32 throw} (Invalid name argument)
10322:
10323: @end table
10324:
10325:
10326: @c =====================================================================
10327: @node The optional Memory-Allocation word set, The optional Programming-Tools word set, The optional Locals word set, ANS conformance
10328: @section The optional Memory-Allocation word set
10329: @c =====================================================================
10330: @cindex system documentation, memory-allocation words
10331: @cindex memory-allocation words, system documentation
10332:
10333: @menu
10334: * memory-idef:: Implementation Defined Options
10335: @end menu
10336:
10337:
10338: @c ---------------------------------------------------------------------
10339: @node memory-idef, , The optional Memory-Allocation word set, The optional Memory-Allocation word set
10340: @subsection Implementation Defined Options
10341: @c ---------------------------------------------------------------------
10342: @cindex implementation-defined options, memory-allocation words
10343: @cindex memory-allocation words, implementation-defined options
10344:
10345: @table @i
10346: @item values and meaning of @i{ior}:
10347: @cindex @i{ior} values and meaning
10348: The @i{ior}s returned by the file and memory allocation words are
10349: intended as throw codes. They typically are in the range
10350: -512@minus{}-2047 of OS errors. The mapping from OS error numbers to
10351: @i{ior}s is -512@minus{}@i{errno}.
10352:
10353: @end table
10354:
10355: @c =====================================================================
10356: @node The optional Programming-Tools word set, The optional Search-Order word set, The optional Memory-Allocation word set, ANS conformance
10357: @section The optional Programming-Tools word set
10358: @c =====================================================================
10359: @cindex system documentation, programming-tools words
10360: @cindex programming-tools words, system documentation
10361:
10362: @menu
10363: * programming-idef:: Implementation Defined Options
10364: * programming-ambcond:: Ambiguous Conditions
10365: @end menu
10366:
10367:
10368: @c ---------------------------------------------------------------------
10369: @node programming-idef, programming-ambcond, The optional Programming-Tools word set, The optional Programming-Tools word set
10370: @subsection Implementation Defined Options
10371: @c ---------------------------------------------------------------------
10372: @cindex implementation-defined options, programming-tools words
10373: @cindex programming-tools words, implementation-defined options
10374:
10375: @table @i
10376: @item ending sequence for input following @code{;CODE} and @code{CODE}:
10377: @cindex @code{;CODE} ending sequence
10378: @cindex @code{CODE} ending sequence
10379: @code{END-CODE}
10380:
10381: @item manner of processing input following @code{;CODE} and @code{CODE}:
10382: @cindex @code{;CODE}, processing input
10383: @cindex @code{CODE}, processing input
10384: The @code{ASSEMBLER} vocabulary is pushed on the search order stack, and
10385: the input is processed by the text interpreter, (starting) in interpret
10386: state.
10387:
10388: @item search order capability for @code{EDITOR} and @code{ASSEMBLER}:
10389: @cindex @code{ASSEMBLER}, search order capability
10390: The ANS Forth search order word set.
10391:
10392: @item source and format of display by @code{SEE}:
10393: @cindex @code{SEE}, source and format of output
10394: The source for @code{see} is the intermediate code used by the inner
10395: interpreter. The current @code{see} tries to output Forth source code
10396: as well as possible.
10397:
10398: @end table
10399:
10400: @c ---------------------------------------------------------------------
10401: @node programming-ambcond, , programming-idef, The optional Programming-Tools word set
10402: @subsection Ambiguous conditions
10403: @c ---------------------------------------------------------------------
10404: @cindex programming-tools words, ambiguous conditions
10405: @cindex ambiguous conditions, programming-tools words
10406:
10407: @table @i
10408:
10409: @item deleting the compilation word list (@code{FORGET}):
10410: @cindex @code{FORGET}, deleting the compilation word list
10411: Not implemented (yet).
10412:
10413: @item fewer than @i{u}+1 items on the control-flow stack (@code{CS-PICK}, @code{CS-ROLL}):
10414: @cindex @code{CS-PICK}, fewer than @i{u}+1 items on the control flow-stack
10415: @cindex @code{CS-ROLL}, fewer than @i{u}+1 items on the control flow-stack
10416: @cindex control-flow stack underflow
10417: This typically results in an @code{abort"} with a descriptive error
10418: message (may change into a @code{-22 throw} (Control structure mismatch)
10419: in the future). You may also get a memory access error. If you are
10420: unlucky, this ambiguous condition is not caught.
10421:
10422: @item @i{name} can't be found (@code{FORGET}):
10423: @cindex @code{FORGET}, @i{name} can't be found
10424: Not implemented (yet).
10425:
10426: @item @i{name} not defined via @code{CREATE}:
10427: @cindex @code{;CODE}, @i{name} not defined via @code{CREATE}
10428: @code{;CODE} behaves like @code{DOES>} in this respect, i.e., it changes
10429: the execution semantics of the last defined word no matter how it was
10430: defined.
10431:
10432: @item @code{POSTPONE} applied to @code{[IF]}:
10433: @cindex @code{POSTPONE} applied to @code{[IF]}
10434: @cindex @code{[IF]} and @code{POSTPONE}
10435: After defining @code{: X POSTPONE [IF] ; IMMEDIATE}. @code{X} is
10436: equivalent to @code{[IF]}.
10437:
10438: @item reaching the end of the input source before matching @code{[ELSE]} or @code{[THEN]}:
10439: @cindex @code{[IF]}, end of the input source before matching @code{[ELSE]} or @code{[THEN]}
10440: Continue in the same state of conditional compilation in the next outer
10441: input source. Currently there is no warning to the user about this.
10442:
10443: @item removing a needed definition (@code{FORGET}):
10444: @cindex @code{FORGET}, removing a needed definition
10445: Not implemented (yet).
10446:
10447: @end table
10448:
10449:
10450: @c =====================================================================
10451: @node The optional Search-Order word set, , The optional Programming-Tools word set, ANS conformance
10452: @section The optional Search-Order word set
10453: @c =====================================================================
10454: @cindex system documentation, search-order words
10455: @cindex search-order words, system documentation
10456:
10457: @menu
10458: * search-idef:: Implementation Defined Options
10459: * search-ambcond:: Ambiguous Conditions
10460: @end menu
10461:
10462:
10463: @c ---------------------------------------------------------------------
10464: @node search-idef, search-ambcond, The optional Search-Order word set, The optional Search-Order word set
10465: @subsection Implementation Defined Options
10466: @c ---------------------------------------------------------------------
10467: @cindex implementation-defined options, search-order words
10468: @cindex search-order words, implementation-defined options
10469:
10470: @table @i
10471: @item maximum number of word lists in search order:
10472: @cindex maximum number of word lists in search order
10473: @cindex search order, maximum depth
10474: @code{s" wordlists" environment? drop .}. Currently 16.
10475:
10476: @item minimum search order:
10477: @cindex minimum search order
10478: @cindex search order, minimum
10479: @code{root root}.
10480:
10481: @end table
10482:
10483: @c ---------------------------------------------------------------------
10484: @node search-ambcond, , search-idef, The optional Search-Order word set
10485: @subsection Ambiguous conditions
10486: @c ---------------------------------------------------------------------
10487: @cindex search-order words, ambiguous conditions
10488: @cindex ambiguous conditions, search-order words
10489:
10490: @table @i
10491: @item changing the compilation word list (during compilation):
10492: @cindex changing the compilation word list (during compilation)
10493: @cindex compilation word list, change before definition ends
10494: The word is entered into the word list that was the compilation word list
10495: at the start of the definition. Any changes to the name field (e.g.,
10496: @code{immediate}) or the code field (e.g., when executing @code{DOES>})
10497: are applied to the latest defined word (as reported by @code{last} or
10498: @code{lastxt}), if possible, irrespective of the compilation word list.
10499:
10500: @item search order empty (@code{previous}):
10501: @cindex @code{previous}, search order empty
10502: @cindex vocstack empty, @code{previous}
10503: @code{abort" Vocstack empty"}.
10504:
10505: @item too many word lists in search order (@code{also}):
10506: @cindex @code{also}, too many word lists in search order
10507: @cindex vocstack full, @code{also}
10508: @code{abort" Vocstack full"}.
10509:
10510: @end table
10511:
10512: @c ***************************************************************
10513: @node Model, Integrating Gforth, ANS conformance, Top
10514: @chapter Model
10515:
10516: This chapter has yet to be written. It will contain information, on
10517: which internal structures you can rely.
10518:
10519: @c ***************************************************************
10520: @node Integrating Gforth, Emacs and Gforth, Model, Top
10521: @chapter Integrating Gforth into C programs
10522:
10523: This is not yet implemented.
10524:
10525: Several people like to use Forth as scripting language for applications
10526: that are otherwise written in C, C++, or some other language.
10527:
10528: The Forth system ATLAST provides facilities for embedding it into
10529: applications; unfortunately it has several disadvantages: most
10530: importantly, it is not based on ANS Forth, and it is apparently dead
10531: (i.e., not developed further and not supported). The facilities
10532: provided by Gforth in this area are inspired by ATLAST's facilities, so
10533: making the switch should not be hard.
10534:
10535: We also tried to design the interface such that it can easily be
10536: implemented by other Forth systems, so that we may one day arrive at a
10537: standardized interface. Such a standard interface would allow you to
10538: replace the Forth system without having to rewrite C code.
10539:
10540: You embed the Gforth interpreter by linking with the library
10541: @code{libgforth.a} (give the compiler the option @code{-lgforth}). All
10542: global symbols in this library that belong to the interface, have the
10543: prefix @code{forth_}. (Global symbols that are used internally have the
10544: prefix @code{gforth_}).
10545:
10546: You can include the declarations of Forth types and the functions and
10547: variables of the interface with @code{#include <forth.h>}.
10548:
10549: Types.
10550:
10551: Variables.
10552:
10553: Data and FP Stack pointer. Area sizes.
10554:
10555: functions.
10556:
10557: forth_init(imagefile)
10558: forth_evaluate(string) exceptions?
10559: forth_goto(address) (or forth_execute(xt)?)
10560: forth_continue() (a corountining mechanism)
10561:
10562: Adding primitives.
10563:
10564: No checking.
10565:
10566: Signals?
10567:
10568: Accessing the Stacks
10569:
10570: @c ******************************************************************
10571: @node Emacs and Gforth, Image Files, Integrating Gforth, Top
10572: @chapter Emacs and Gforth
10573: @cindex Emacs and Gforth
10574:
10575: @cindex @file{gforth.el}
10576: @cindex @file{forth.el}
10577: @cindex Rydqvist, Goran
10578: @cindex comment editing commands
10579: @cindex @code{\}, editing with Emacs
10580: @cindex debug tracer editing commands
10581: @cindex @code{~~}, removal with Emacs
10582: @cindex Forth mode in Emacs
10583: Gforth comes with @file{gforth.el}, an improved version of
10584: @file{forth.el} by Goran Rydqvist (included in the TILE package). The
10585: improvements are:
10586:
10587: @itemize @bullet
10588: @item
10589: A better (but still not perfect) handling of indentation.
10590: @item
10591: Comment paragraph filling (@kbd{M-q})
10592: @item
10593: Commenting (@kbd{C-x \}) and uncommenting (@kbd{C-u C-x \}) of regions
10594: @item
10595: Removal of debugging tracers (@kbd{C-x ~}, @pxref{Debugging}).
10596: @item
10597: Support of the @code{info-lookup} feature for looking up the
10598: documentation of a word.
10599: @end itemize
10600:
10601: I left the stuff I do not use alone, even though some of it only makes
10602: sense for TILE. To get a description of these features, enter Forth mode
10603: and type @kbd{C-h m}.
10604:
10605: @cindex source location of error or debugging output in Emacs
10606: @cindex error output, finding the source location in Emacs
10607: @cindex debugging output, finding the source location in Emacs
10608: In addition, Gforth supports Emacs quite well: The source code locations
10609: given in error messages, debugging output (from @code{~~}) and failed
10610: assertion messages are in the right format for Emacs' compilation mode
10611: (@pxref{Compilation, , Running Compilations under Emacs, emacs, Emacs
10612: Manual}) so the source location corresponding to an error or other
10613: message is only a few keystrokes away (@kbd{C-x `} for the next error,
10614: @kbd{C-c C-c} for the error under the cursor).
10615:
10616: @cindex @file{TAGS} file
10617: @cindex @file{etags.fs}
10618: @cindex viewing the source of a word in Emacs
10619: @cindex @code{require}, placement in files
10620: @cindex @code{include}, placement in files
10621: Also, if you @code{require} @file{etags.fs}, a new @file{TAGS} file will
10622: be produced (@pxref{Tags, , Tags Tables, emacs, Emacs Manual}) that
10623: contains the definitions of all words defined afterwards. You can then
10624: find the source for a word using @kbd{M-.}. Note that emacs can use
10625: several tags files at the same time (e.g., one for the Gforth sources
10626: and one for your program, @pxref{Select Tags Table,,Selecting a Tags
10627: Table,emacs, Emacs Manual}). The TAGS file for the preloaded words is
10628: @file{$(datadir)/gforth/$(VERSION)/TAGS} (e.g.,
10629: @file{/usr/local/share/gforth/0.2.0/TAGS}). To get the best behaviour
10630: with @file{etags.fs}, you should avoid putting definitions both before
10631: and after @code{require} etc., otherwise you will see the same file
10632: visited several times by commands like @code{tags-search}.
10633:
10634: @cindex viewing the documentation of a word in Emacs
10635: @cindex context-sensitive help
10636: Moreover, for words documented in this manual, you can look up the
10637: glossary entry quickly by using @kbd{C-h TAB}
10638: (@code{info-lookup-symbol}, see @pxref{Documentation, ,Documentation
10639: Commands, emacs, Emacs Manual}). This feature requires Emacs 20.3 or
10640: later and does not work for words containing @code{:}.
10641:
10642:
10643: @cindex @file{.emacs}
10644: To get all these benefits, add the following lines to your @file{.emacs}
10645: file:
10646:
10647: @example
10648: (autoload 'forth-mode "gforth.el")
10649: (setq auto-mode-alist (cons '("\\.fs\\'" . forth-mode) auto-mode-alist))
10650: @end example
10651:
10652: @c ******************************************************************
10653: @node Image Files, Engine, Emacs and Gforth, Top
10654: @chapter Image Files
10655: @cindex image file
10656: @cindex @file{.fi} files
10657: @cindex precompiled Forth code
10658: @cindex dictionary in persistent form
10659: @cindex persistent form of dictionary
10660:
10661: An image file is a file containing an image of the Forth dictionary,
10662: i.e., compiled Forth code and data residing in the dictionary. By
10663: convention, we use the extension @code{.fi} for image files.
10664:
10665: @menu
10666: * Image Licensing Issues:: Distribution terms for images.
10667: * Image File Background:: Why have image files?
10668: * Non-Relocatable Image Files:: don't always work.
10669: * Data-Relocatable Image Files:: are better.
10670: * Fully Relocatable Image Files:: better yet.
10671: * Stack and Dictionary Sizes:: Setting the default sizes for an image.
10672: * Running Image Files:: @code{gforth -i @i{file}} or @i{file}.
10673: * Modifying the Startup Sequence:: and turnkey applications.
10674: @end menu
10675:
10676: @node Image Licensing Issues, Image File Background, Image Files, Image Files
10677: @section Image Licensing Issues
10678: @cindex license for images
10679: @cindex image license
10680:
10681: An image created with @code{gforthmi} (@pxref{gforthmi}) or
10682: @code{savesystem} (@pxref{Non-Relocatable Image Files}) includes the
10683: original image; i.e., according to copyright law it is a derived work of
10684: the original image.
10685:
10686: Since Gforth is distributed under the GNU GPL, the newly created image
10687: falls under the GNU GPL, too. In particular, this means that if you
10688: distribute the image, you have to make all of the sources for the image
10689: available, including those you wrote. For details see @ref{License, ,
10690: GNU General Public License (Section 3)}.
10691:
10692: If you create an image with @code{cross} (@pxref{cross.fs}), the image
10693: contains only code compiled from the sources you gave it; if none of
10694: these sources is under the GPL, the terms discussed above do not apply
10695: to the image. However, if your image needs an engine (a gforth binary)
10696: that is under the GPL, you should make sure that you distribute both in
10697: a way that is at most a @emph{mere aggregation}, if you don't want the
10698: terms of the GPL to apply to the image.
10699:
10700: @node Image File Background, Non-Relocatable Image Files, Image Licensing Issues, Image Files
10701: @section Image File Background
10702: @cindex image file background
10703:
10704: Our Forth system consists not only of primitives, but also of
10705: definitions written in Forth. Since the Forth compiler itself belongs to
10706: those definitions, it is not possible to start the system with the
10707: primitives and the Forth source alone. Therefore we provide the Forth
10708: code as an image file in nearly executable form. When Gforth starts up,
10709: a C routine loads the image file into memory, optionally relocates the
10710: addresses, then sets up the memory (stacks etc.) according to
10711: information in the image file, and (finally) starts executing Forth
10712: code.
10713:
10714: The image file variants represent different compromises between the
10715: goals of making it easy to generate image files and making them
10716: portable.
10717:
10718: @cindex relocation at run-time
10719: Win32Forth 3.4 and Mitch Bradley's @code{cforth} use relocation at
10720: run-time. This avoids many of the complications discussed below (image
10721: files are data relocatable without further ado), but costs performance
10722: (one addition per memory access).
10723:
10724: @cindex relocation at load-time
10725: By contrast, the Gforth loader performs relocation at image load time. The
10726: loader also has to replace tokens that represent primitive calls with the
10727: appropriate code-field addresses (or code addresses in the case of
10728: direct threading).
10729:
10730: There are three kinds of image files, with different degrees of
10731: relocatability: non-relocatable, data-relocatable, and fully relocatable
10732: image files.
10733:
10734: @cindex image file loader
10735: @cindex relocating loader
10736: @cindex loader for image files
10737: These image file variants have several restrictions in common; they are
10738: caused by the design of the image file loader:
10739:
10740: @itemize @bullet
10741: @item
10742: There is only one segment; in particular, this means, that an image file
10743: cannot represent @code{ALLOCATE}d memory chunks (and pointers to
10744: them). The contents of the stacks are not represented, either.
10745:
10746: @item
10747: The only kinds of relocation supported are: adding the same offset to
10748: all cells that represent data addresses; and replacing special tokens
10749: with code addresses or with pieces of machine code.
10750:
10751: If any complex computations involving addresses are performed, the
10752: results cannot be represented in the image file. Several applications that
10753: use such computations come to mind:
10754: @itemize @minus
10755: @item
10756: Hashing addresses (or data structures which contain addresses) for table
10757: lookup. If you use Gforth's @code{table}s or @code{wordlist}s for this
10758: purpose, you will have no problem, because the hash tables are
10759: recomputed automatically when the system is started. If you use your own
10760: hash tables, you will have to do something similar.
10761:
10762: @item
10763: There's a cute implementation of doubly-linked lists that uses
10764: @code{XOR}ed addresses. You could represent such lists as singly-linked
10765: in the image file, and restore the doubly-linked representation on
10766: startup.@footnote{In my opinion, though, you should think thrice before
10767: using a doubly-linked list (whatever implementation).}
10768:
10769: @item
10770: The code addresses of run-time routines like @code{docol:} cannot be
10771: represented in the image file (because their tokens would be replaced by
10772: machine code in direct threaded implementations). As a workaround,
10773: compute these addresses at run-time with @code{>code-address} from the
10774: executions tokens of appropriate words (see the definitions of
10775: @code{docol:} and friends in @file{kernel.fs}).
10776:
10777: @item
10778: On many architectures addresses are represented in machine code in some
10779: shifted or mangled form. You cannot put @code{CODE} words that contain
10780: absolute addresses in this form in a relocatable image file. Workarounds
10781: are representing the address in some relative form (e.g., relative to
10782: the CFA, which is present in some register), or loading the address from
10783: a place where it is stored in a non-mangled form.
10784: @end itemize
10785: @end itemize
10786:
10787: @node Non-Relocatable Image Files, Data-Relocatable Image Files, Image File Background, Image Files
10788: @section Non-Relocatable Image Files
10789: @cindex non-relocatable image files
10790: @cindex image file, non-relocatable
10791:
10792: These files are simple memory dumps of the dictionary. They are specific
10793: to the executable (i.e., @file{gforth} file) they were created
10794: with. What's worse, they are specific to the place on which the
10795: dictionary resided when the image was created. Now, there is no
10796: guarantee that the dictionary will reside at the same place the next
10797: time you start Gforth, so there's no guarantee that a non-relocatable
10798: image will work the next time (Gforth will complain instead of crashing,
10799: though).
10800:
10801: You can create a non-relocatable image file with
10802:
10803:
10804: doc-savesystem
10805:
10806:
10807: @node Data-Relocatable Image Files, Fully Relocatable Image Files, Non-Relocatable Image Files, Image Files
10808: @section Data-Relocatable Image Files
10809: @cindex data-relocatable image files
10810: @cindex image file, data-relocatable
10811:
10812: These files contain relocatable data addresses, but fixed code addresses
10813: (instead of tokens). They are specific to the executable (i.e.,
10814: @file{gforth} file) they were created with. For direct threading on some
10815: architectures (e.g., the i386), data-relocatable images do not work. You
10816: get a data-relocatable image, if you use @file{gforthmi} with a
10817: Gforth binary that is not doubly indirect threaded (@pxref{Fully
10818: Relocatable Image Files}).
10819:
10820: @node Fully Relocatable Image Files, Stack and Dictionary Sizes, Data-Relocatable Image Files, Image Files
10821: @section Fully Relocatable Image Files
10822: @cindex fully relocatable image files
10823: @cindex image file, fully relocatable
10824:
10825: @cindex @file{kern*.fi}, relocatability
10826: @cindex @file{gforth.fi}, relocatability
10827: These image files have relocatable data addresses, and tokens for code
10828: addresses. They can be used with different binaries (e.g., with and
10829: without debugging) on the same machine, and even across machines with
10830: the same data formats (byte order, cell size, floating point
10831: format). However, they are usually specific to the version of Gforth
10832: they were created with. The files @file{gforth.fi} and @file{kernl*.fi}
10833: are fully relocatable.
10834:
10835: There are two ways to create a fully relocatable image file:
10836:
10837: @menu
10838: * gforthmi:: The normal way
10839: * cross.fs:: The hard way
10840: @end menu
10841:
10842: @node gforthmi, cross.fs, Fully Relocatable Image Files, Fully Relocatable Image Files
10843: @subsection @file{gforthmi}
10844: @cindex @file{comp-i.fs}
10845: @cindex @file{gforthmi}
10846:
10847: You will usually use @file{gforthmi}. If you want to create an
10848: image @i{file} that contains everything you would load by invoking
10849: Gforth with @code{gforth @i{options}}, you simply say:
10850: @example
10851: gforthmi @i{file} @i{options}
10852: @end example
10853:
10854: E.g., if you want to create an image @file{asm.fi} that has the file
10855: @file{asm.fs} loaded in addition to the usual stuff, you could do it
10856: like this:
10857:
10858: @example
10859: gforthmi asm.fi asm.fs
10860: @end example
10861:
10862: @file{gforthmi} is implemented as a sh script and works like this: It
10863: produces two non-relocatable images for different addresses and then
10864: compares them. Its output reflects this: first you see the output (if
10865: any) of the two Gforth invocations that produce the nonrelocatable image
10866: files, then you see the output of the comparing program: It displays the
10867: offset used for data addresses and the offset used for code addresses;
10868: moreover, for each cell that cannot be represented correctly in the
10869: image files, it displays a line like this:
10870:
10871: @example
10872: 78DC BFFFFA50 BFFFFA40
10873: @end example
10874:
10875: This means that at offset $78dc from @code{forthstart}, one input image
10876: contains $bffffa50, and the other contains $bffffa40. Since these cells
10877: cannot be represented correctly in the output image, you should examine
10878: these places in the dictionary and verify that these cells are dead
10879: (i.e., not read before they are written).
10880:
10881: @cindex --application, @code{gforthmi} option
10882: If you insert the option @code{--application} in front of the image file
10883: name, you will get an image that uses the @code{--appl-image} option
10884: instead of the @code{--image-file} option (@pxref{Invoking
10885: Gforth}). When you execute such an image on Unix (by typing the image
10886: name as command), the Gforth engine will pass all options to the image
10887: instead of trying to interpret them as engine options.
10888:
10889: If you type @file{gforthmi} with no arguments, it prints some usage
10890: instructions.
10891:
10892: @cindex @code{savesystem} during @file{gforthmi}
10893: @cindex @code{bye} during @file{gforthmi}
10894: @cindex doubly indirect threaded code
10895: @cindex environment variables
10896: @cindex @code{GFORTHD} -- environment variable
10897: @cindex @code{GFORTH} -- environment variable
10898: @cindex @code{gforth-ditc}
10899: There are a few wrinkles: After processing the passed @i{options}, the
10900: words @code{savesystem} and @code{bye} must be visible. A special doubly
10901: indirect threaded version of the @file{gforth} executable is used for
10902: creating the nonrelocatable images; you can pass the exact filename of
10903: this executable through the environment variable @code{GFORTHD}
10904: (default: @file{gforth-ditc}); if you pass a version that is not doubly
10905: indirect threaded, you will not get a fully relocatable image, but a
10906: data-relocatable image (because there is no code address offset). The
10907: normal @file{gforth} executable is used for creating the relocatable
10908: image; you can pass the exact filename of this executable through the
10909: environment variable @code{GFORTH}.
10910:
10911: @node cross.fs, , gforthmi, Fully Relocatable Image Files
10912: @subsection @file{cross.fs}
10913: @cindex @file{cross.fs}
10914: @cindex cross-compiler
10915: @cindex metacompiler
10916:
10917: You can also use @code{cross}, a batch compiler that accepts a Forth-like
10918: programming language. This @code{cross} language has to be documented
10919: yet.
10920:
10921: @cindex target compiler
10922: @code{cross} also allows you to create image files for machines with
10923: different data sizes and data formats than the one used for generating
10924: the image file. You can also use it to create an application image that
10925: does not contain a Forth compiler. These features are bought with
10926: restrictions and inconveniences in programming. E.g., addresses have to
10927: be stored in memory with special words (@code{A!}, @code{A,}, etc.) in
10928: order to make the code relocatable.
10929:
10930:
10931: @node Stack and Dictionary Sizes, Running Image Files, Fully Relocatable Image Files, Image Files
10932: @section Stack and Dictionary Sizes
10933: @cindex image file, stack and dictionary sizes
10934: @cindex dictionary size default
10935: @cindex stack size default
10936:
10937: If you invoke Gforth with a command line flag for the size
10938: (@pxref{Invoking Gforth}), the size you specify is stored in the
10939: dictionary. If you save the dictionary with @code{savesystem} or create
10940: an image with @file{gforthmi}, this size will become the default
10941: for the resulting image file. E.g., the following will create a
10942: fully relocatable version of @file{gforth.fi} with a 1MB dictionary:
10943:
10944: @example
10945: gforthmi gforth.fi -m 1M
10946: @end example
10947:
10948: In other words, if you want to set the default size for the dictionary
10949: and the stacks of an image, just invoke @file{gforthmi} with the
10950: appropriate options when creating the image.
10951:
10952: @cindex stack size, cache-friendly
10953: Note: For cache-friendly behaviour (i.e., good performance), you should
10954: make the sizes of the stacks modulo, say, 2K, somewhat different. E.g.,
10955: the default stack sizes are: data: 16k (mod 2k=0); fp: 15.5k (mod
10956: 2k=1.5k); return: 15k(mod 2k=1k); locals: 14.5k (mod 2k=0.5k).
10957:
10958: @node Running Image Files, Modifying the Startup Sequence, Stack and Dictionary Sizes, Image Files
10959: @section Running Image Files
10960: @cindex running image files
10961: @cindex invoking image files
10962: @cindex image file invocation
10963:
10964: @cindex -i, invoke image file
10965: @cindex --image file, invoke image file
10966: You can invoke Gforth with an image file @i{image} instead of the
10967: default @file{gforth.fi} with the @code{-i} flag (@pxref{Invoking Gforth}):
10968: @example
10969: gforth -i @i{image}
10970: @end example
10971:
10972: @cindex executable image file
10973: @cindex image file, executable
10974: If your operating system supports starting scripts with a line of the
10975: form @code{#! ...}, you just have to type the image file name to start
10976: Gforth with this image file (note that the file extension @code{.fi} is
10977: just a convention). I.e., to run Gforth with the image file @i{image},
10978: you can just type @i{image} instead of @code{gforth -i @i{image}}.
10979: This works because every @code{.fi} file starts with a line of this
10980: format:
10981:
10982: @example
10983: #! /usr/local/bin/gforth-0.4.0 -i
10984: @end example
10985:
10986: The file and pathname for the Gforth engine specified on this line is
10987: the specific Gforth executable that it was built against; i.e. the value
10988: of the environment variable @code{GFORTH} at the time that
10989: @file{gforthmi} was executed.
10990:
10991: You can make use of the same shell capability to make a Forth source
10992: file into an executable. For example, if you place this text in a file:
10993:
10994: @example
10995: #! /usr/local/bin/gforth
10996:
10997: ." Hello, world" CR
10998: bye
10999: @end example
11000:
11001: @noindent
11002: and then make the file executable (chmod +x in Unix), you can run it
11003: directly from the command line. The sequence @code{#!} is used in two
11004: ways; firstly, it is recognised as a ``magic sequence'' by the operating
11005: system@footnote{The Unix kernel actually recognises two types of files:
11006: executable files and files of data, where the data is processed by an
11007: interpreter that is specified on the ``interpreter line'' -- the first
11008: line of the file, starting with the sequence #!. There may be a small
11009: limit (e.g., 32) on the number of characters that may be specified on
11010: the interpreter line.} secondly it is treated as a comment character by
11011: Gforth. Because of the second usage, a space is required between
11012: @code{#!} and the path to the executable.
11013:
11014: The disadvantage of this latter technique, compared with using
11015: @file{gforthmi}, is that it is slower; the Forth source code is compiled
11016: on-the-fly, each time the program is invoked.
11017:
11018:
11019: doc-#!
11020:
11021:
11022: @node Modifying the Startup Sequence, , Running Image Files, Image Files
11023: @section Modifying the Startup Sequence
11024: @cindex startup sequence for image file
11025: @cindex image file initialization sequence
11026: @cindex initialization sequence of image file
11027:
11028: You can add your own initialization to the startup sequence through the
11029: deferred word @code{'cold}. @code{'cold} is invoked just before the
11030: image-specific command line processing (by default, loading files and
11031: evaluating (@code{-e}) strings) starts.
11032:
11033: A sequence for adding your initialization usually looks like this:
11034:
11035: @example
11036: :noname
11037: Defers 'cold \ do other initialization stuff (e.g., rehashing wordlists)
11038: ... \ your stuff
11039: ; IS 'cold
11040: @end example
11041:
11042: @cindex turnkey image files
11043: @cindex image file, turnkey applications
11044: You can make a turnkey image by letting @code{'cold} execute a word
11045: (your turnkey application) that never returns; instead, it exits Gforth
11046: via @code{bye} or @code{throw}.
11047:
11048: @cindex command-line arguments, access
11049: @cindex arguments on the command line, access
11050: You can access the (image-specific) command-line arguments through the
11051: variables @code{argc} and @code{argv}. @code{arg} provides convenient
11052: access to @code{argv}.
11053:
11054: If @code{'cold} exits normally, Gforth processes the command-line
11055: arguments as files to be loaded and strings to be evaluated. Therefore,
11056: @code{'cold} should remove the arguments it has used in this case.
11057:
11058:
11059:
11060: doc-'cold
11061: doc-argc
11062: doc-argv
11063: doc-arg
11064:
11065:
11066:
11067: @c ******************************************************************
11068: @node Engine, Binding to System Library, Image Files, Top
11069: @chapter Engine
11070: @cindex engine
11071: @cindex virtual machine
11072:
11073: Reading this chapter is not necessary for programming with Gforth. It
11074: may be helpful for finding your way in the Gforth sources.
11075:
11076: The ideas in this section have also been published in the papers
11077: @cite{ANS fig/GNU/??? Forth} (in German) by Bernd Paysan, presented at
11078: the Forth-Tagung '93 and @cite{A Portable Forth Engine} by M. Anton
11079: Ertl, presented at EuroForth '93; the latter is available at
11080: @*@url{http://www.complang.tuwien.ac.at/papers/ertl93.ps.Z}.
11081:
11082: @menu
11083: * Portability::
11084: * Threading::
11085: * Primitives::
11086: * Performance::
11087: @end menu
11088:
11089: @node Portability, Threading, Engine, Engine
11090: @section Portability
11091: @cindex engine portability
11092:
11093: An important goal of the Gforth Project is availability across a wide
11094: range of personal machines. fig-Forth, and, to a lesser extent, F83,
11095: achieved this goal by manually coding the engine in assembly language
11096: for several then-popular processors. This approach is very
11097: labor-intensive and the results are short-lived due to progress in
11098: computer architecture.
11099:
11100: @cindex C, using C for the engine
11101: Others have avoided this problem by coding in C, e.g., Mitch Bradley
11102: (cforth), Mikael Patel (TILE) and Dirk Zoller (pfe). This approach is
11103: particularly popular for UNIX-based Forths due to the large variety of
11104: architectures of UNIX machines. Unfortunately an implementation in C
11105: does not mix well with the goals of efficiency and with using
11106: traditional techniques: Indirect or direct threading cannot be expressed
11107: in C, and switch threading, the fastest technique available in C, is
11108: significantly slower. Another problem with C is that it is very
11109: cumbersome to express double integer arithmetic.
11110:
11111: @cindex GNU C for the engine
11112: @cindex long long
11113: Fortunately, there is a portable language that does not have these
11114: limitations: GNU C, the version of C processed by the GNU C compiler
11115: (@pxref{C Extensions, , Extensions to the C Language Family, gcc.info,
11116: GNU C Manual}). Its labels as values feature (@pxref{Labels as Values, ,
11117: Labels as Values, gcc.info, GNU C Manual}) makes direct and indirect
11118: threading possible, its @code{long long} type (@pxref{Long Long, ,
11119: Double-Word Integers, gcc.info, GNU C Manual}) corresponds to Forth's
11120: double numbers@footnote{Unfortunately, long longs are not implemented
11121: properly on all machines (e.g., on alpha-osf1, long longs are only 64
11122: bits, the same size as longs (and pointers), but they should be twice as
11123: long according to @pxref{Long Long, , Double-Word Integers, gcc.info, GNU
11124: C Manual}). So, we had to implement doubles in C after all. Still, on
11125: most machines we can use long longs and achieve better performance than
11126: with the emulation package.}. GNU C is available for free on all
11127: important (and many unimportant) UNIX machines, VMS, 80386s running
11128: MS-DOS, the Amiga, and the Atari ST, so a Forth written in GNU C can run
11129: on all these machines.
11130:
11131: Writing in a portable language has the reputation of producing code that
11132: is slower than assembly. For our Forth engine we repeatedly looked at
11133: the code produced by the compiler and eliminated most compiler-induced
11134: inefficiencies by appropriate changes in the source code.
11135:
11136: @cindex explicit register declarations
11137: @cindex --enable-force-reg, configuration flag
11138: @cindex -DFORCE_REG
11139: However, register allocation cannot be portably influenced by the
11140: programmer, leading to some inefficiencies on register-starved
11141: machines. We use explicit register declarations (@pxref{Explicit Reg
11142: Vars, , Variables in Specified Registers, gcc.info, GNU C Manual}) to
11143: improve the speed on some machines. They are turned on by using the
11144: configuration flag @code{--enable-force-reg} (@code{gcc} switch
11145: @code{-DFORCE_REG}). Unfortunately, this feature not only depends on the
11146: machine, but also on the compiler version: On some machines some
11147: compiler versions produce incorrect code when certain explicit register
11148: declarations are used. So by default @code{-DFORCE_REG} is not used.
11149:
11150: @node Threading, Primitives, Portability, Engine
11151: @section Threading
11152: @cindex inner interpreter implementation
11153: @cindex threaded code implementation
11154:
11155: @cindex labels as values
11156: GNU C's labels as values extension (available since @code{gcc-2.0},
11157: @pxref{Labels as Values, , Labels as Values, gcc.info, GNU C Manual})
11158: makes it possible to take the address of @i{label} by writing
11159: @code{&&@i{label}}. This address can then be used in a statement like
11160: @code{goto *@i{address}}. I.e., @code{goto *&&x} is the same as
11161: @code{goto x}.
11162:
11163: @cindex @code{NEXT}, indirect threaded
11164: @cindex indirect threaded inner interpreter
11165: @cindex inner interpreter, indirect threaded
11166: With this feature an indirect threaded @code{NEXT} looks like:
11167: @example
11168: cfa = *ip++;
11169: ca = *cfa;
11170: goto *ca;
11171: @end example
11172: @cindex instruction pointer
11173: For those unfamiliar with the names: @code{ip} is the Forth instruction
11174: pointer; the @code{cfa} (code-field address) corresponds to ANS Forths
11175: execution token and points to the code field of the next word to be
11176: executed; The @code{ca} (code address) fetched from there points to some
11177: executable code, e.g., a primitive or the colon definition handler
11178: @code{docol}.
11179:
11180: @cindex @code{NEXT}, direct threaded
11181: @cindex direct threaded inner interpreter
11182: @cindex inner interpreter, direct threaded
11183: Direct threading is even simpler:
11184: @example
11185: ca = *ip++;
11186: goto *ca;
11187: @end example
11188:
11189: Of course we have packaged the whole thing neatly in macros called
11190: @code{NEXT} and @code{NEXT1} (the part of @code{NEXT} after fetching the cfa).
11191:
11192: @menu
11193: * Scheduling::
11194: * Direct or Indirect Threaded?::
11195: * DOES>::
11196: @end menu
11197:
11198: @node Scheduling, Direct or Indirect Threaded?, Threading, Threading
11199: @subsection Scheduling
11200: @cindex inner interpreter optimization
11201:
11202: There is a little complication: Pipelined and superscalar processors,
11203: i.e., RISC and some modern CISC machines can process independent
11204: instructions while waiting for the results of an instruction. The
11205: compiler usually reorders (schedules) the instructions in a way that
11206: achieves good usage of these delay slots. However, on our first tries
11207: the compiler did not do well on scheduling primitives. E.g., for
11208: @code{+} implemented as
11209: @example
11210: n=sp[0]+sp[1];
11211: sp++;
11212: sp[0]=n;
11213: NEXT;
11214: @end example
11215: the @code{NEXT} comes strictly after the other code, i.e., there is nearly no
11216: scheduling. After a little thought the problem becomes clear: The
11217: compiler cannot know that @code{sp} and @code{ip} point to different
11218: addresses (and the version of @code{gcc} we used would not know it even
11219: if it was possible), so it could not move the load of the cfa above the
11220: store to the TOS. Indeed the pointers could be the same, if code on or
11221: very near the top of stack were executed. In the interest of speed we
11222: chose to forbid this probably unused ``feature'' and helped the compiler
11223: in scheduling: @code{NEXT} is divided into the loading part (@code{NEXT_P1})
11224: and the goto part (@code{NEXT_P2}). @code{+} now looks like:
11225: @example
11226: n=sp[0]+sp[1];
11227: sp++;
11228: NEXT_P1;
11229: sp[0]=n;
11230: NEXT_P2;
11231: @end example
11232: This can be scheduled optimally by the compiler.
11233:
11234: This division can be turned off with the switch @code{-DCISC_NEXT}. This
11235: switch is on by default on machines that do not profit from scheduling
11236: (e.g., the 80386), in order to preserve registers.
11237:
11238: @node Direct or Indirect Threaded?, DOES>, Scheduling, Threading
11239: @subsection Direct or Indirect Threaded?
11240: @cindex threading, direct or indirect?
11241:
11242: @cindex -DDIRECT_THREADED
11243: Both! After packaging the nasty details in macro definitions we
11244: realized that we could switch between direct and indirect threading by
11245: simply setting a compilation flag (@code{-DDIRECT_THREADED}) and
11246: defining a few machine-specific macros for the direct-threading case.
11247: On the Forth level we also offer access words that hide the
11248: differences between the threading methods (@pxref{Threading Words}).
11249:
11250: Indirect threading is implemented completely machine-independently.
11251: Direct threading needs routines for creating jumps to the executable
11252: code (e.g. to @code{docol} or @code{dodoes}). These routines are inherently
11253: machine-dependent, but they do not amount to many source lines. Therefore,
11254: even porting direct threading to a new machine requires little effort.
11255:
11256: @cindex --enable-indirect-threaded, configuration flag
11257: @cindex --enable-direct-threaded, configuration flag
11258: The default threading method is machine-dependent. You can enforce a
11259: specific threading method when building Gforth with the configuration
11260: flag @code{--enable-direct-threaded} or
11261: @code{--enable-indirect-threaded}. Note that direct threading is not
11262: supported on all machines.
11263:
11264: @node DOES>, , Direct or Indirect Threaded?, Threading
11265: @subsection DOES>
11266: @cindex @code{DOES>} implementation
11267:
11268: @cindex @code{dodoes} routine
11269: @cindex @code{DOES>}-code
11270: One of the most complex parts of a Forth engine is @code{dodoes}, i.e.,
11271: the chunk of code executed by every word defined by a
11272: @code{CREATE}...@code{DOES>} pair. The main problem here is: How to find
11273: the Forth code to be executed, i.e. the code after the
11274: @code{DOES>} (the @code{DOES>}-code)? There are two solutions:
11275:
11276: In fig-Forth the code field points directly to the @code{dodoes} and the
11277: @code{DOES>}code address is stored in the cell after the code address (i.e. at
11278: @code{@i{CFA} cell+}). It may seem that this solution is illegal in
11279: the Forth-79 and all later standards, because in fig-Forth this address
11280: lies in the body (which is illegal in these standards). However, by
11281: making the code field larger for all words this solution becomes legal
11282: again. We use this approach for the indirect threaded version and for
11283: direct threading on some machines. Leaving a cell unused in most words
11284: is a bit wasteful, but on the machines we are targeting this is hardly a
11285: problem. The other reason for having a code field size of two cells is
11286: to avoid having different image files for direct and indirect threaded
11287: systems (direct threaded systems require two-cell code fields on many
11288: machines).
11289:
11290: @cindex @code{DOES>}-handler
11291: The other approach is that the code field points or jumps to the cell
11292: after @code{DOES>}. In this variant there is a jump to @code{dodoes} at
11293: this address (the @code{DOES>}-handler). @code{dodoes} can then get the
11294: @code{DOES>}-code address by computing the code address, i.e., the address of
11295: the jump to dodoes, and add the length of that jump field. A variant of
11296: this is to have a call to @code{dodoes} after the @code{DOES>}; then the
11297: return address (which can be found in the return register on RISCs) is
11298: the @code{DOES>}-code address. Since the two cells available in the code field
11299: are used up by the jump to the code address in direct threading on many
11300: architectures, we use this approach for direct threading on these
11301: architectures. We did not want to add another cell to the code field.
11302:
11303: @node Primitives, Performance, Threading, Engine
11304: @section Primitives
11305: @cindex primitives, implementation
11306: @cindex virtual machine instructions, implementation
11307:
11308: @menu
11309: * Automatic Generation::
11310: * TOS Optimization::
11311: * Produced code::
11312: @end menu
11313:
11314: @node Automatic Generation, TOS Optimization, Primitives, Primitives
11315: @subsection Automatic Generation
11316: @cindex primitives, automatic generation
11317:
11318: @cindex @file{prims2x.fs}
11319: Since the primitives are implemented in a portable language, there is no
11320: longer any need to minimize the number of primitives. On the contrary,
11321: having many primitives has an advantage: speed. In order to reduce the
11322: number of errors in primitives and to make programming them easier, we
11323: provide a tool, the primitive generator (@file{prims2x.fs}), that
11324: automatically generates most (and sometimes all) of the C code for a
11325: primitive from the stack effect notation. The source for a primitive
11326: has the following form:
11327:
11328: @cindex primitive source format
11329: @format
11330: @i{Forth-name} @i{stack-effect} @i{category} [@i{pronounc.}]
11331: [@code{""}@i{glossary entry}@code{""}]
11332: @i{C code}
11333: [@code{:}
11334: @i{Forth code}]
11335: @end format
11336:
11337: The items in brackets are optional. The category and glossary fields
11338: are there for generating the documentation, the Forth code is there
11339: for manual implementations on machines without GNU C. E.g., the source
11340: for the primitive @code{+} is:
11341: @example
11342: + n1 n2 -- n core plus
11343: n = n1+n2;
11344: @end example
11345:
11346: This looks like a specification, but in fact @code{n = n1+n2} is C
11347: code. Our primitive generation tool extracts a lot of information from
11348: the stack effect notations@footnote{We use a one-stack notation, even
11349: though we have separate data and floating-point stacks; The separate
11350: notation can be generated easily from the unified notation.}: The number
11351: of items popped from and pushed on the stack, their type, and by what
11352: name they are referred to in the C code. It then generates a C code
11353: prelude and postlude for each primitive. The final C code for @code{+}
11354: looks like this:
11355:
11356: @example
11357: I_plus: /* + ( n1 n2 -- n ) */ /* label, stack effect */
11358: /* */ /* documentation */
11359: @{
11360: DEF_CA /* definition of variable ca (indirect threading) */
11361: Cell n1; /* definitions of variables */
11362: Cell n2;
11363: Cell n;
11364: n1 = (Cell) sp[1]; /* input */
11365: n2 = (Cell) TOS;
11366: sp += 1; /* stack adjustment */
11367: NAME("+") /* debugging output (with -DDEBUG) */
11368: @{
11369: n = n1+n2; /* C code taken from the source */
11370: @}
11371: NEXT_P1; /* NEXT part 1 */
11372: TOS = (Cell)n; /* output */
11373: NEXT_P2; /* NEXT part 2 */
11374: @}
11375: @end example
11376:
11377: This looks long and inefficient, but the GNU C compiler optimizes quite
11378: well and produces optimal code for @code{+} on, e.g., the R3000 and the
11379: HP RISC machines: Defining the @code{n}s does not produce any code, and
11380: using them as intermediate storage also adds no cost.
11381:
11382: There are also other optimizations that are not illustrated by this
11383: example: assignments between simple variables are usually for free (copy
11384: propagation). If one of the stack items is not used by the primitive
11385: (e.g. in @code{drop}), the compiler eliminates the load from the stack
11386: (dead code elimination). On the other hand, there are some things that
11387: the compiler does not do, therefore they are performed by
11388: @file{prims2x.fs}: The compiler does not optimize code away that stores
11389: a stack item to the place where it just came from (e.g., @code{over}).
11390:
11391: While programming a primitive is usually easy, there are a few cases
11392: where the programmer has to take the actions of the generator into
11393: account, most notably @code{?dup}, but also words that do not (always)
11394: fall through to @code{NEXT}.
11395:
11396: @node TOS Optimization, Produced code, Automatic Generation, Primitives
11397: @subsection TOS Optimization
11398: @cindex TOS optimization for primitives
11399: @cindex primitives, keeping the TOS in a register
11400:
11401: An important optimization for stack machine emulators, e.g., Forth
11402: engines, is keeping one or more of the top stack items in
11403: registers. If a word has the stack effect @i{in1}...@i{inx} @code{--}
11404: @i{out1}...@i{outy}, keeping the top @i{n} items in registers
11405: @itemize @bullet
11406: @item
11407: is better than keeping @i{n-1} items, if @i{x>=n} and @i{y>=n},
11408: due to fewer loads from and stores to the stack.
11409: @item is slower than keeping @i{n-1} items, if @i{x<>y} and @i{x<n} and
11410: @i{y<n}, due to additional moves between registers.
11411: @end itemize
11412:
11413: @cindex -DUSE_TOS
11414: @cindex -DUSE_NO_TOS
11415: In particular, keeping one item in a register is never a disadvantage,
11416: if there are enough registers. Keeping two items in registers is a
11417: disadvantage for frequent words like @code{?branch}, constants,
11418: variables, literals and @code{i}. Therefore our generator only produces
11419: code that keeps zero or one items in registers. The generated C code
11420: covers both cases; the selection between these alternatives is made at
11421: C-compile time using the switch @code{-DUSE_TOS}. @code{TOS} in the C
11422: code for @code{+} is just a simple variable name in the one-item case,
11423: otherwise it is a macro that expands into @code{sp[0]}. Note that the
11424: GNU C compiler tries to keep simple variables like @code{TOS} in
11425: registers, and it usually succeeds, if there are enough registers.
11426:
11427: @cindex -DUSE_FTOS
11428: @cindex -DUSE_NO_FTOS
11429: The primitive generator performs the TOS optimization for the
11430: floating-point stack, too (@code{-DUSE_FTOS}). For floating-point
11431: operations the benefit of this optimization is even larger:
11432: floating-point operations take quite long on most processors, but can be
11433: performed in parallel with other operations as long as their results are
11434: not used. If the FP-TOS is kept in a register, this works. If
11435: it is kept on the stack, i.e., in memory, the store into memory has to
11436: wait for the result of the floating-point operation, lengthening the
11437: execution time of the primitive considerably.
11438:
11439: The TOS optimization makes the automatic generation of primitives a
11440: bit more complicated. Just replacing all occurrences of @code{sp[0]} by
11441: @code{TOS} is not sufficient. There are some special cases to
11442: consider:
11443: @itemize @bullet
11444: @item In the case of @code{dup ( w -- w w )} the generator must not
11445: eliminate the store to the original location of the item on the stack,
11446: if the TOS optimization is turned on.
11447: @item Primitives with stack effects of the form @code{--}
11448: @i{out1}...@i{outy} must store the TOS to the stack at the start.
11449: Likewise, primitives with the stack effect @i{in1}...@i{inx} @code{--}
11450: must load the TOS from the stack at the end. But for the null stack
11451: effect @code{--} no stores or loads should be generated.
11452: @end itemize
11453:
11454: @node Produced code, , TOS Optimization, Primitives
11455: @subsection Produced code
11456: @cindex primitives, assembly code listing
11457:
11458: @cindex @file{engine.s}
11459: To see what assembly code is produced for the primitives on your machine
11460: with your compiler and your flag settings, type @code{make engine.s} and
11461: look at the resulting file @file{engine.s}.
11462:
11463: @node Performance, , Primitives, Engine
11464: @section Performance
11465: @cindex performance of some Forth interpreters
11466: @cindex engine performance
11467: @cindex benchmarking Forth systems
11468: @cindex Gforth performance
11469:
11470: On RISCs the Gforth engine is very close to optimal; i.e., it is usually
11471: impossible to write a significantly faster engine.
11472:
11473: On register-starved machines like the 386 architecture processors
11474: improvements are possible, because @code{gcc} does not utilize the
11475: registers as well as a human, even with explicit register declarations;
11476: e.g., Bernd Beuster wrote a Forth system fragment in assembly language
11477: and hand-tuned it for the 486; this system is 1.19 times faster on the
11478: Sieve benchmark on a 486DX2/66 than Gforth compiled with
11479: @code{gcc-2.6.3} with @code{-DFORCE_REG}. The situation has improved
11480: with gcc-2.95 and gforth-0.4.9; now the most important virtual machine
11481: registers fit in real registers (and we can even afford to use the TOS
11482: optimization), resulting in a speedup of 1.14 on the sieve over the
11483: earlier results.
11484:
11485: @cindex Win32Forth performance
11486: @cindex NT Forth performance
11487: @cindex eforth performance
11488: @cindex ThisForth performance
11489: @cindex PFE performance
11490: @cindex TILE performance
11491: The potential advantage of assembly language implementations
11492: is not necessarily realized in complete Forth systems: We compared
11493: Gforth-0.4.9 (direct threaded, compiled with @code{gcc-2.95.1} and
11494: @code{-DFORCE_REG}) with Win32Forth 1.2093, LMI's NT Forth (Beta, May
11495: 1994) and Eforth (with and without peephole (aka pinhole) optimization
11496: of the threaded code); all these systems were written in assembly
11497: language. We also compared Gforth with three systems written in C:
11498: PFE-0.9.14 (compiled with @code{gcc-2.6.3} with the default
11499: configuration for Linux: @code{-O2 -fomit-frame-pointer -DUSE_REGS
11500: -DUNROLL_NEXT}), ThisForth Beta (compiled with @code{gcc-2.6.3 -O3
11501: -fomit-frame-pointer}; ThisForth employs peephole optimization of the
11502: threaded code) and TILE (compiled with @code{make opt}). We benchmarked
11503: Gforth, PFE, ThisForth and TILE on a 486DX2/66 under Linux. Kenneth
11504: O'Heskin kindly provided the results for Win32Forth and NT Forth on a
11505: 486DX2/66 with similar memory performance under Windows NT. Marcel
11506: Hendrix ported Eforth to Linux, then extended it to run the benchmarks,
11507: added the peephole optimizer, ran the benchmarks and reported the
11508: results.
11509:
11510: We used four small benchmarks: the ubiquitous Sieve; bubble-sorting and
11511: matrix multiplication come from the Stanford integer benchmarks and have
11512: been translated into Forth by Martin Fraeman; we used the versions
11513: included in the TILE Forth package, but with bigger data set sizes; and
11514: a recursive Fibonacci number computation for benchmarking calling
11515: performance. The following table shows the time taken for the benchmarks
11516: scaled by the time taken by Gforth (in other words, it shows the speedup
11517: factor that Gforth achieved over the other systems).
11518:
11519: @example
11520: relative Win32- NT eforth This-
11521: time Gforth Forth Forth eforth +opt PFE Forth TILE
11522: sieve 1.00 1.58 1.30 1.58 0.97 1.80 3.63 9.79
11523: bubble 1.00 1.55 1.67 1.75 1.04 1.78 4.59
11524: matmul 1.00 1.67 1.53 1.66 0.84 1.79 4.63
11525: fib 1.00 1.75 1.53 1.40 0.99 1.99 3.43 4.93
11526: @end example
11527:
11528: You may be quite surprised by the good performance of Gforth when
11529: compared with systems written in assembly language. One important reason
11530: for the disappointing performance of these other systems is probably
11531: that they are not written optimally for the 486 (e.g., they use the
11532: @code{lods} instruction). In addition, Win32Forth uses a comfortable,
11533: but costly method for relocating the Forth image: like @code{cforth}, it
11534: computes the actual addresses at run time, resulting in two address
11535: computations per @code{NEXT} (@pxref{Image File Background}).
11536:
11537: Only Eforth with the peephole optimizer performs comparable to
11538: Gforth. The speedups achieved with peephole optimization of threaded
11539: code are quite remarkable. Adding a peephole optimizer to Gforth should
11540: cause similar speedups.
11541:
11542: The speedup of Gforth over PFE, ThisForth and TILE can be easily
11543: explained with the self-imposed restriction of the latter systems to
11544: standard C, which makes efficient threading impossible (however, the
11545: measured implementation of PFE uses a GNU C extension: @pxref{Global Reg
11546: Vars, , Defining Global Register Variables, gcc.info, GNU C Manual}).
11547: Moreover, current C compilers have a hard time optimizing other aspects
11548: of the ThisForth and the TILE source.
11549:
11550: The performance of Gforth on 386 architecture processors varies widely
11551: with the version of @code{gcc} used. E.g., @code{gcc-2.5.8} failed to
11552: allocate any of the virtual machine registers into real machine
11553: registers by itself and would not work correctly with explicit register
11554: declarations, giving a 1.5 times slower engine (on a 486DX2/66 running
11555: the Sieve) than the one measured above.
11556:
11557: Note that there have been several releases of Win32Forth since the
11558: release presented here, so the results presented above may have little
11559: predictive value for the performance of Win32Forth today (results for
11560: the current release on an i486DX2/66 are welcome).
11561:
11562: @cindex @file{Benchres}
11563: In @cite{Translating Forth to Efficient C} by M. Anton Ertl and Martin
11564: Maierhofer (presented at EuroForth '95), an indirect threaded version of
11565: Gforth is compared with Win32Forth, NT Forth, PFE, and ThisForth; that
11566: version of Gforth is slower on a 486 than the direct threaded version
11567: used here. The paper available at
11568: @*@url{http://www.complang.tuwien.ac.at/papers/ertl&maierhofer95.ps.gz};
11569: it also contains numbers for some native code systems. You can find a
11570: newer version of these measurements at
11571: @url{http://www.complang.tuwien.ac.at/forth/performance.html}. You can
11572: find numbers for Gforth on various machines in @file{Benchres}.
11573:
11574: @c ******************************************************************
11575: @node Binding to System Library, Cross Compiler, Engine, Top
11576: @chapter Binding to System Library
11577:
11578: @node Cross Compiler, Bugs, Binding to System Library, Top
11579: @chapter Cross Compiler
11580:
11581: Cross Compiler
11582:
11583: @menu
11584: * Using the Cross Compiler::
11585: * How the Cross Compiler Works::
11586: @end menu
11587:
11588: @node Using the Cross Compiler, How the Cross Compiler Works, Cross Compiler, Cross Compiler
11589: @section Using the Cross Compiler
11590:
11591: @node How the Cross Compiler Works, , Using the Cross Compiler, Cross Compiler
11592: @section How the Cross Compiler Works
11593:
11594: @node Bugs, Origin, Cross Compiler, Top
11595: @appendix Bugs
11596: @cindex bug reporting
11597:
11598: Known bugs are described in the file @file{BUGS} in the Gforth distribution.
11599:
11600: If you find a bug, please send a bug report to
11601: @email{bug-gforth@@gnu.org}. A bug report should include this
11602: information:
11603:
11604: @itemize @bullet
11605: @item
11606: The Gforth version used (it is announced at the start of an
11607: interactive Gforth session).
11608: @item
11609: The machine and operating system (on Unix
11610: systems @code{uname -a} will report this information).
11611: @item
11612: The installation options (send the file @file{config.status}).
11613: @item
11614: A complete list of changes (if any) you (or your installer) have made to the
11615: Gforth sources.
11616: @item
11617: A program (or a sequence of keyboard commands) that reproduces the bug.
11618: @item
11619: A description of what you think constitutes the buggy behaviour.
11620: @end itemize
11621:
11622: For a thorough guide on reporting bugs read @ref{Bug Reporting, , How
11623: to Report Bugs, gcc.info, GNU C Manual}.
11624:
11625:
11626: @node Origin, Forth-related information, Bugs, Top
11627: @appendix Authors and Ancestors of Gforth
11628:
11629: @section Authors and Contributors
11630: @cindex authors of Gforth
11631: @cindex contributors to Gforth
11632:
11633: The Gforth project was started in mid-1992 by Bernd Paysan and Anton
11634: Ertl. The third major author was Jens Wilke. Lennart Benschop (who was
11635: one of Gforth's first users, in mid-1993) and Stuart Ramsden inspired us
11636: with their continuous feedback. Lennart Benshop contributed
11637: @file{glosgen.fs}, while Stuart Ramsden has been working on automatic
11638: support for calling C libraries. Helpful comments also came from Paul
11639: Kleinrubatscher, Christian Pirker, Dirk Zoller, Marcel Hendrix, John
11640: Wavrik, Barrie Stott, Marc de Groot, and Jorge Acerada. Since the
11641: release of Gforth-0.2.1 there were also helpful comments from many
11642: others; thank you all, sorry for not listing you here (but digging
11643: through my mailbox to extract your names is on my to-do list). Since the
11644: release of Gforth-0.4.0 Neal Crook worked on the manual.
11645:
11646: Gforth also owes a lot to the authors of the tools we used (GCC, CVS,
11647: and autoconf, among others), and to the creators of the Internet: Gforth
11648: was developed across the Internet, and its authors did not meet
11649: physically for the first 4 years of development.
11650:
11651: @section Pedigree
11652: @cindex pedigree of Gforth
11653:
11654: Gforth descends from bigFORTH (1993) and fig-Forth. Gforth and PFE (by
11655: Dirk Zoller) will cross-fertilize each other. Of course, a significant
11656: part of the design of Gforth was prescribed by ANS Forth.
11657:
11658: Bernd Paysan wrote bigFORTH, a descendent from TurboForth, an unreleased
11659: 32 bit native code version of VolksForth for the Atari ST, written
11660: mostly by Dietrich Weineck.
11661:
11662: VolksForth descends from F83. It was written by Klaus Schleisiek, Bernd
11663: Pennemann, Georg Rehfeld and Dietrich Weineck for the C64 (called
11664: UltraForth there) in the mid-80s and ported to the Atari ST in 1986.
11665:
11666: Henry Laxen and Mike Perry wrote F83 as a model implementation of the
11667: Forth-83 standard. !! Pedigree? When?
11668:
11669: A team led by Bill Ragsdale implemented fig-Forth on many processors in
11670: 1979. Robert Selzer and Bill Ragsdale developed the original
11671: implementation of fig-Forth for the 6502 based on microForth.
11672:
11673: The principal architect of microForth was Dean Sanderson. microForth was
11674: FORTH, Inc.'s first off-the-shelf product. It was developed in 1976 for
11675: the 1802, and subsequently implemented on the 8080, the 6800 and the
11676: Z80.
11677:
11678: All earlier Forth systems were custom-made, usually by Charles Moore,
11679: who discovered (as he puts it) Forth during the late 60s. The first full
11680: Forth existed in 1971.
11681:
11682: A part of the information in this section comes from @cite{The Evolution
11683: of Forth} by Elizabeth D. Rather, Donald R. Colburn and Charles
11684: H. Moore, presented at the HOPL-II conference and preprinted in SIGPLAN
11685: Notices 28(3), 1993. You can find more historical and genealogical
11686: information about Forth there.
11687:
11688: @node Forth-related information, Word Index, Origin, Top
11689: @appendix Other Forth-related information
11690: @cindex Forth-related information
11691:
11692: @menu
11693: * Internet resources::
11694: * Books::
11695: * The Forth Interest Group::
11696: * Conferences::
11697: @end menu
11698:
11699:
11700: @node Internet resources, Books, Forth-related information, Forth-related information
11701: @section Internet resources
11702: @cindex internet resources
11703:
11704: @cindex comp.lang.forth
11705: @cindex frequently asked questions
11706: There is an active newsgroup (comp.lang.forth) discussing Forth and
11707: Forth-related issues. A frequently-asked-questions (FAQ) list
11708: is posted to the newsgroup regulary, and archived at these sites:
11709:
11710: @itemize @bullet
11711: @item
11712: @url{ftp://rtfm.mit.edu/pub/usenet-by-group/comp.lang.forth/}
11713: @item
11714: @url{ftp://ftp.forth.org/pub/Forth/FAQ/}
11715: @end itemize
11716:
11717: The FAQ list should be considered mandatory reading before posting to
11718: the newsgroup.
11719:
11720: Here are some other web sites holding Forth-related material:
11721:
11722: @itemize @bullet
11723: @item
11724: @url{http://www.taygeta.com/forth.html} -- Skip Carter's Forth pages.
11725: @item
11726: @url{http://www.jwdt.com/~paysan/gforth.html} -- the Gforth home page.
11727: @item
11728: @url{http://www.minerva.com/uathena.htm} -- home of ANS Forth Standard.
11729: @item
11730: @url{http://dec.bournemouth.ac.uk/forth/index.html} -- the Forth
11731: Research page, including links to the Journal of Forth Application and
11732: Research (JFAR) and a searchable Forth bibliography.
11733: @end itemize
11734:
11735:
11736: @node Books, The Forth Interest Group, Internet resources, Forth-related information
11737: @section Books
11738: @cindex books on Forth
11739:
11740: As the Standard is relatively new, there are not many books out yet. It
11741: is not recommended to learn Forth by using Gforth and a book that is not
11742: written for ANS Forth, as you will not know your mistakes from the
11743: deviations of the book. However, books based on the Forth-83 standard
11744: should be ok, because ANS Forth is primarily an extension of Forth-83.
11745: Refer to the Forth FAQ for details of Forth-related books.
11746:
11747: @cindex standard document for ANS Forth
11748: @cindex ANS Forth document
11749: The definite reference if you want to write ANS Forth programs is, of
11750: course, the ANS Forth document. It is available in printed form from the
11751: National Standards Institute Sales Department (Tel.: USA (212) 642-4900;
11752: Fax.: USA (212) 302-1286) as document @cite{X3.215-1994} for about
11753: $200. You can also get it from Global Engineering Documents (Tel.: USA
11754: (800) 854-7179; Fax.: (303) 843-9880) for about $300.
11755:
11756: @cite{dpANS6}, the last draft of the standard, which was then submitted
11757: to ANSI for publication is available electronically and for free in some
11758: MS Word format, and it has been converted to HTML
11759: (@url{http://www.taygeta.com/forth/dpans.html}; this HTML version also
11760: includes the answers to Requests for Interpretation (RFIs). Some
11761: pointers to these versions can be found through
11762: @*@url{http://www.complang.tuwien.ac.at/projects/forth.html}.
11763:
11764:
11765: @node The Forth Interest Group, Conferences, Books, Forth-related information
11766: @section The Forth Interest Group
11767: @cindex Forth interest group (FIG)
11768:
11769: The Forth Interest Group (FIG) is a world-wide, non-profit,
11770: member-supported organisation. It publishes a regular magazine,
11771: @var{FORTH Dimensions}, and offers other benefits of membership. You can
11772: contact the FIG through their office email address:
11773: @email{office@@forth.org} or by visiting their web site at
11774: @url{http://www.forth.org/}. This web site also includes links to FIG
11775: chapters in other countries and American cities
11776: (@url{http://www.forth.org/chapters.html}).
11777:
11778: @node Conferences, , The Forth Interest Group, Forth-related information
11779: @section Conferences
11780: @cindex Conferences
11781:
11782: There are several regular conferences related to Forth. They are all
11783: well-publicised in @var{FORTH Dimensions} and on the comp.lang.forth
11784: newsgroup:
11785:
11786: @itemize @bullet
11787: @item
11788: FORML -- the Forth modification laboratory convenes every year near
11789: Monterey, California.
11790: @item
11791: The Rochester Forth Conference -- an annual conference traditionally
11792: held in Rochester, New York.
11793: @item
11794: EuroForth -- this European conference takes place annually.
11795: @end itemize
11796:
11797:
11798: @node Word Index, Name Index, Forth-related information, Top
11799: @unnumbered Word Index
11800:
11801: This index is a list of Forth words that have ``glossary'' entries
11802: within this manual. Each word is listed with its stack effect and
11803: wordset.
11804:
11805: @printindex fn
11806:
11807: @node Name Index, Concept Index, Word Index, Top
11808: @unnumbered Name Index
11809:
11810: This index is a list of Forth words that have ``glossary'' entries
11811: within this manual.
11812:
11813: @printindex ky
11814:
11815: @node Concept Index, , Name Index, Top
11816: @unnumbered Concept and Word Index
11817:
11818: Not all entries listed in this index are present verbatim in the
11819: text. This index also duplicates, in abbreviated form, all of the words
11820: listed in the Word Index (only the names are listed for the words here).
11821:
11822: @printindex cp
11823:
11824: @contents
11825: @bye
11826:
FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>