File:
[gforth] /
gforth /
doc /
gforth.ds
Revision
1.47:
download - view:
text,
annotated -
select for diffs
Fri Mar 17 21:35:32 2000 UTC (24 years, 1 month ago) by
crook
Branches:
MAIN
CVS tags:
HEAD
Re-ordered a couple of sections. Added new section on time. Fixed url
references so they would be real hypertext links in HTML output. Other
minor fixes here and there.
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 15-Mar-2000
99:
100: @comment The following two commands start the copyright page.
101: @page
102: @vskip 0pt plus 1filll
103: Copyright @copyright{} 1995--1999 Free Software Foundation, Inc.
104:
105: @comment !! Published by ... or You can get a copy of this manual ...
106:
107: Permission is granted to make and distribute verbatim copies of
108: this manual provided the copyright notice and this permission notice
109: are preserved on all copies.
110:
111: Permission is granted to copy and distribute modified versions of this
112: manual under the conditions for verbatim copying, provided also that the
113: sections entitled "Distribution" and "General Public License" are
114: included exactly as in the original, and provided that the entire
115: resulting derived work is distributed under the terms of a permission
116: notice identical to this one.
117:
118: Permission is granted to copy and distribute translations of this manual
119: into another language, under the above conditions for modified versions,
120: except that the sections entitled "Distribution" and "General Public
121: License" may be included in a translation approved by the author instead
122: of in the original English.
123: @end titlepage
124:
125: @node Top, License, (dir), (dir)
126: @ifinfo
127: Gforth is a free implementation of ANS Forth available on many
128: personal machines. This manual corresponds to version @value{VERSION}.
129: @end ifinfo
130:
131: @menu
132: * License:: The GPL
133: * Goals:: About the Gforth Project
134: * Gforth Environment:: Starting (and exiting) Gforth
135: * Introduction:: An introduction to ANS Forth
136: * Words:: Forth words available in Gforth
137: * Error messages:: How to interpret them
138: * Tools:: Programming tools
139: * ANS conformance:: Implementation-defined options etc.
140: * Model:: The abstract machine of Gforth
141: * Integrating Gforth:: Forth as scripting language for applications
142: * Emacs and Gforth:: The Gforth Mode
143: * Image Files:: @code{.fi} files contain compiled code
144: * Engine:: The inner interpreter and the primitives
145: * Binding to System Library::
146: * Cross Compiler:: The Cross Compiler
147: * Bugs:: How to report them
148: * Origin:: Authors and ancestors of Gforth
149: * Forth-related information:: Books and places to look on the WWW
150: * Word Index:: An item for each Forth word
151: * Name Index:: Forth words, only names listed
152: * Concept Index:: A menu covering many topics
153:
154: @detailmenu
155: --- The Detailed Node Listing ---
156:
157: Goals of Gforth
158:
159: * Gforth Extensions Sinful?::
160:
161: Gforth Environment
162:
163: * Invoking Gforth:: Getting in
164: * Leaving Gforth:: Getting out
165: * Command-line editing::
166: * Upper and lower case::
167: * Environment variables:: ..that affect how Gforth starts up
168: * Gforth Files:: What gets installed and where
169:
170: An Introduction to ANS Forth
171:
172: * Introducing the Text Interpreter::
173: * Stacks and Postfix notation::
174: * Your first definition::
175: * How does that work?::
176: * Forth is written in Forth::
177: * Review - elements of a Forth system::
178: * Where to go next::
179: * Exercises::
180:
181: Forth Words
182:
183: * Notation::
184: * Comments::
185: * Boolean Flags::
186: * Arithmetic::
187: * Stack Manipulation::
188: * Memory::
189: * Control Structures::
190: * Defining Words::
191: * Interpretation and Compilation Semantics::
192: * Tokens for Words::
193: * The Text Interpreter::
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: * Keeping track of Time::
207: * Miscellaneous Words::
208:
209: Arithmetic
210:
211: * Single precision::
212: * Bitwise operations::
213: * Double precision:: Double-cell integer arithmetic
214: * Numeric comparison::
215: * Mixed precision:: Operations with single and double-cell integers
216: * Floating Point::
217:
218: Stack Manipulation
219:
220: * Data stack::
221: * Floating point stack::
222: * Return stack::
223: * Locals stack::
224: * Stack pointer manipulation::
225:
226: Memory
227:
228: * Memory model::
229: * Dictionary allocation::
230: * Heap Allocation::
231: * Memory Access::
232: * Address arithmetic::
233: * Memory Blocks::
234:
235: Control Structures
236:
237: * Selection:: IF ... ELSE ... ENDIF
238: * Simple Loops:: BEGIN ...
239: * Counted Loops:: DO
240: * Arbitrary control structures::
241: * Calls and returns::
242: * Exception Handling::
243:
244: Defining Words
245:
246: * CREATE::
247: * Variables:: Variables and user variables
248: * Constants::
249: * Values:: Initialised variables
250: * Colon Definitions::
251: * Anonymous Definitions:: Definitions without names
252: * User-defined Defining Words::
253: * Deferred words:: Allow forward references
254: * Aliases::
255: * Supplying names::
256:
257: Interpretation and Compilation Semantics
258:
259: * Combined words::
260:
261: The Text Interpreter
262:
263: * Input Sources::
264: * Number Conversion::
265: * Interpret/Compile states::
266: * Literals::
267: * Interpreter Directives::
268:
269: Word Lists
270:
271: * Why use word lists?::
272: * Word list examples::
273:
274: Files
275:
276: * Forth source files::
277: * General files::
278: * Search Paths::
279: * Forth Search Paths::
280: * General Search Paths::
281:
282: Other I/O
283:
284: * Simple numeric output:: Predefined formats
285: * Formatted numeric output:: Formatted (pictured) output
286: * String Formats:: How Forth stores strings in memory
287: * Displaying characters and strings:: Other stuff
288: * Input:: Input
289:
290: Programming Tools
291:
292: * Debugging:: Simple and quick.
293: * Assertions:: Making your programs self-checking.
294: * Singlestep Debugger:: Executing your program word by word.
295:
296: Locals
297:
298: * Gforth locals::
299: * ANS Forth locals::
300:
301: Gforth locals
302:
303: * Where are locals visible by name?::
304: * How long do locals live?::
305: * Programming Style::
306: * Implementation::
307:
308: Structures
309:
310: * Why explicit structure support?::
311: * Structure Usage::
312: * Structure Naming Convention::
313: * Structure Implementation::
314: * Structure Glossary::
315:
316: Object-oriented Forth
317:
318: * Why object-oriented programming?::
319: * Object-Oriented Terminology::
320: * Objects::
321: * OOF::
322: * Mini-OOF::
323: * Comparison with other object models::
324:
325: The @file{objects.fs} model
326:
327: * Properties of the Objects model::
328: * Basic Objects Usage::
329: * The Objects base class::
330: * Creating objects::
331: * Object-Oriented Programming Style::
332: * Class Binding::
333: * Method conveniences::
334: * Classes and Scoping::
335: * Dividing classes::
336: * Object Interfaces::
337: * Objects Implementation::
338: * Objects Glossary::
339:
340: The @file{oof.fs} model
341:
342: * Properties of the OOF model::
343: * Basic OOF Usage::
344: * The OOF base class::
345: * Class Declaration::
346: * Class Implementation::
347:
348: The @file{mini-oof.fs} model
349:
350: * Basic Mini-OOF Usage::
351: * Mini-OOF Example::
352: * Mini-OOF Implementation::
353:
354: Tools
355:
356: * ANS Report:: Report the words used, sorted by wordset.
357:
358: ANS conformance
359:
360: * The Core Words::
361: * The optional Block word set::
362: * The optional Double Number word set::
363: * The optional Exception word set::
364: * The optional Facility word set::
365: * The optional File-Access word set::
366: * The optional Floating-Point word set::
367: * The optional Locals word set::
368: * The optional Memory-Allocation word set::
369: * The optional Programming-Tools word set::
370: * The optional Search-Order word set::
371:
372: The Core Words
373:
374: * core-idef:: Implementation Defined Options
375: * core-ambcond:: Ambiguous Conditions
376: * core-other:: Other System Documentation
377:
378: The optional Block word set
379:
380: * block-idef:: Implementation Defined Options
381: * block-ambcond:: Ambiguous Conditions
382: * block-other:: Other System Documentation
383:
384: The optional Double Number word set
385:
386: * double-ambcond:: Ambiguous Conditions
387:
388: The optional Exception word set
389:
390: * exception-idef:: Implementation Defined Options
391:
392: The optional Facility word set
393:
394: * facility-idef:: Implementation Defined Options
395: * facility-ambcond:: Ambiguous Conditions
396:
397: The optional File-Access word set
398:
399: * file-idef:: Implementation Defined Options
400: * file-ambcond:: Ambiguous Conditions
401:
402: The optional Floating-Point word set
403:
404: * floating-idef:: Implementation Defined Options
405: * floating-ambcond:: Ambiguous Conditions
406:
407: The optional Locals word set
408:
409: * locals-idef:: Implementation Defined Options
410: * locals-ambcond:: Ambiguous Conditions
411:
412: The optional Memory-Allocation word set
413:
414: * memory-idef:: Implementation Defined Options
415:
416: The optional Programming-Tools word set
417:
418: * programming-idef:: Implementation Defined Options
419: * programming-ambcond:: Ambiguous Conditions
420:
421: The optional Search-Order word set
422:
423: * search-idef:: Implementation Defined Options
424: * search-ambcond:: Ambiguous Conditions
425:
426: Image Files
427:
428: * Image Licensing Issues:: Distribution terms for images.
429: * Image File Background:: Why have image files?
430: * Non-Relocatable Image Files:: don't always work.
431: * Data-Relocatable Image Files:: are better.
432: * Fully Relocatable Image Files:: better yet.
433: * Stack and Dictionary Sizes:: Setting the default sizes for an image.
434: * Running Image Files:: @code{gforth -i @i{file}} or @i{file}.
435: * Modifying the Startup Sequence:: and turnkey applications.
436:
437: Fully Relocatable Image Files
438:
439: * gforthmi:: The normal way
440: * cross.fs:: The hard way
441:
442: Engine
443:
444: * Portability::
445: * Threading::
446: * Primitives::
447: * Performance::
448:
449: Threading
450:
451: * Scheduling::
452: * Direct or Indirect Threaded?::
453: * DOES>::
454:
455: Primitives
456:
457: * Automatic Generation::
458: * TOS Optimization::
459: * Produced code::
460:
461: Cross Compiler
462:
463: * Using the Cross Compiler::
464: * How the Cross Compiler Works::
465:
466: Other Forth-related information
467:
468: * Internet resources::
469: * Books::
470: * The Forth Interest Group::
471: * Conferences::
472:
473: @end detailmenu
474: @end menu
475:
476: @node License, Goals, Top, Top
477: @unnumbered GNU GENERAL PUBLIC LICENSE
478: @center Version 2, June 1991
479:
480: @display
481: Copyright @copyright{} 1989, 1991 Free Software Foundation, Inc.
482: 675 Mass Ave, Cambridge, MA 02139, USA
483:
484: Everyone is permitted to copy and distribute verbatim copies
485: of this license document, but changing it is not allowed.
486: @end display
487:
488: @unnumberedsec Preamble
489:
490: The licenses for most software are designed to take away your
491: freedom to share and change it. By contrast, the GNU General Public
492: License is intended to guarantee your freedom to share and change free
493: software---to make sure the software is free for all its users. This
494: General Public License applies to most of the Free Software
495: Foundation's software and to any other program whose authors commit to
496: using it. (Some other Free Software Foundation software is covered by
497: the GNU Library General Public License instead.) You can apply it to
498: your programs, too.
499:
500: When we speak of free software, we are referring to freedom, not
501: price. Our General Public Licenses are designed to make sure that you
502: have the freedom to distribute copies of free software (and charge for
503: this service if you wish), that you receive source code or can get it
504: if you want it, that you can change the software or use pieces of it
505: in new free programs; and that you know you can do these things.
506:
507: To protect your rights, we need to make restrictions that forbid
508: anyone to deny you these rights or to ask you to surrender the rights.
509: These restrictions translate to certain responsibilities for you if you
510: distribute copies of the software, or if you modify it.
511:
512: For example, if you distribute copies of such a program, whether
513: gratis or for a fee, you must give the recipients all the rights that
514: you have. You must make sure that they, too, receive or can get the
515: source code. And you must show them these terms so they know their
516: rights.
517:
518: We protect your rights with two steps: (1) copyright the software, and
519: (2) offer you this license which gives you legal permission to copy,
520: distribute and/or modify the software.
521:
522: Also, for each author's protection and ours, we want to make certain
523: that everyone understands that there is no warranty for this free
524: software. If the software is modified by someone else and passed on, we
525: want its recipients to know that what they have is not the original, so
526: that any problems introduced by others will not reflect on the original
527: authors' reputations.
528:
529: Finally, any free program is threatened constantly by software
530: patents. We wish to avoid the danger that redistributors of a free
531: program will individually obtain patent licenses, in effect making the
532: program proprietary. To prevent this, we have made it clear that any
533: patent must be licensed for everyone's free use or not licensed at all.
534:
535: The precise terms and conditions for copying, distribution and
536: modification follow.
537:
538: @iftex
539: @unnumberedsec TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION
540: @end iftex
541: @ifinfo
542: @center TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION
543: @end ifinfo
544:
545: @enumerate 0
546: @item
547: This License applies to any program or other work which contains
548: a notice placed by the copyright holder saying it may be distributed
549: under the terms of this General Public License. The ``Program'', below,
550: refers to any such program or work, and a ``work based on the Program''
551: means either the Program or any derivative work under copyright law:
552: that is to say, a work containing the Program or a portion of it,
553: either verbatim or with modifications and/or translated into another
554: language. (Hereinafter, translation is included without limitation in
555: the term ``modification''.) Each licensee is addressed as ``you''.
556:
557: Activities other than copying, distribution and modification are not
558: covered by this License; they are outside its scope. The act of
559: running the Program is not restricted, and the output from the Program
560: is covered only if its contents constitute a work based on the
561: Program (independent of having been made by running the Program).
562: Whether that is true depends on what the Program does.
563:
564: @item
565: You may copy and distribute verbatim copies of the Program's
566: source code as you receive it, in any medium, provided that you
567: conspicuously and appropriately publish on each copy an appropriate
568: copyright notice and disclaimer of warranty; keep intact all the
569: notices that refer to this License and to the absence of any warranty;
570: and give any other recipients of the Program a copy of this License
571: along with the Program.
572:
573: You may charge a fee for the physical act of transferring a copy, and
574: you may at your option offer warranty protection in exchange for a fee.
575:
576: @item
577: You may modify your copy or copies of the Program or any portion
578: of it, thus forming a work based on the Program, and copy and
579: distribute such modifications or work under the terms of Section 1
580: above, provided that you also meet all of these conditions:
581:
582: @enumerate a
583: @item
584: You must cause the modified files to carry prominent notices
585: stating that you changed the files and the date of any change.
586:
587: @item
588: You must cause any work that you distribute or publish, that in
589: whole or in part contains or is derived from the Program or any
590: part thereof, to be licensed as a whole at no charge to all third
591: parties under the terms of this License.
592:
593: @item
594: If the modified program normally reads commands interactively
595: when run, you must cause it, when started running for such
596: interactive use in the most ordinary way, to print or display an
597: announcement including an appropriate copyright notice and a
598: notice that there is no warranty (or else, saying that you provide
599: a warranty) and that users may redistribute the program under
600: these conditions, and telling the user how to view a copy of this
601: License. (Exception: if the Program itself is interactive but
602: does not normally print such an announcement, your work based on
603: the Program is not required to print an announcement.)
604: @end enumerate
605:
606: These requirements apply to the modified work as a whole. If
607: identifiable sections of that work are not derived from the Program,
608: and can be reasonably considered independent and separate works in
609: themselves, then this License, and its terms, do not apply to those
610: sections when you distribute them as separate works. But when you
611: distribute the same sections as part of a whole which is a work based
612: on the Program, the distribution of the whole must be on the terms of
613: this License, whose permissions for other licensees extend to the
614: entire whole, and thus to each and every part regardless of who wrote it.
615:
616: Thus, it is not the intent of this section to claim rights or contest
617: your rights to work written entirely by you; rather, the intent is to
618: exercise the right to control the distribution of derivative or
619: collective works based on the Program.
620:
621: In addition, mere aggregation of another work not based on the Program
622: with the Program (or with a work based on the Program) on a volume of
623: a storage or distribution medium does not bring the other work under
624: the scope of this License.
625:
626: @item
627: You may copy and distribute the Program (or a work based on it,
628: under Section 2) in object code or executable form under the terms of
629: Sections 1 and 2 above provided that you also do one of the following:
630:
631: @enumerate a
632: @item
633: Accompany it with the complete corresponding machine-readable
634: source code, which must be distributed under the terms of Sections
635: 1 and 2 above on a medium customarily used for software interchange; or,
636:
637: @item
638: Accompany it with a written offer, valid for at least three
639: years, to give any third party, for a charge no more than your
640: cost of physically performing source distribution, a complete
641: machine-readable copy of the corresponding source code, to be
642: distributed under the terms of Sections 1 and 2 above on a medium
643: customarily used for software interchange; or,
644:
645: @item
646: Accompany it with the information you received as to the offer
647: to distribute corresponding source code. (This alternative is
648: allowed only for noncommercial distribution and only if you
649: received the program in object code or executable form with such
650: an offer, in accord with Subsection b above.)
651: @end enumerate
652:
653: The source code for a work means the preferred form of the work for
654: making modifications to it. For an executable work, complete source
655: code means all the source code for all modules it contains, plus any
656: associated interface definition files, plus the scripts used to
657: control compilation and installation of the executable. However, as a
658: special exception, the source code distributed need not include
659: anything that is normally distributed (in either source or binary
660: form) with the major components (compiler, kernel, and so on) of the
661: operating system on which the executable runs, unless that component
662: itself accompanies the executable.
663:
664: If distribution of executable or object code is made by offering
665: access to copy from a designated place, then offering equivalent
666: access to copy the source code from the same place counts as
667: distribution of the source code, even though third parties are not
668: compelled to copy the source along with the object code.
669:
670: @item
671: You may not copy, modify, sublicense, or distribute the Program
672: except as expressly provided under this License. Any attempt
673: otherwise to copy, modify, sublicense or distribute the Program is
674: void, and will automatically terminate your rights under this License.
675: However, parties who have received copies, or rights, from you under
676: this License will not have their licenses terminated so long as such
677: parties remain in full compliance.
678:
679: @item
680: You are not required to accept this License, since you have not
681: signed it. However, nothing else grants you permission to modify or
682: distribute the Program or its derivative works. These actions are
683: prohibited by law if you do not accept this License. Therefore, by
684: modifying or distributing the Program (or any work based on the
685: Program), you indicate your acceptance of this License to do so, and
686: all its terms and conditions for copying, distributing or modifying
687: the Program or works based on it.
688:
689: @item
690: Each time you redistribute the Program (or any work based on the
691: Program), the recipient automatically receives a license from the
692: original licensor to copy, distribute or modify the Program subject to
693: these terms and conditions. You may not impose any further
694: restrictions on the recipients' exercise of the rights granted herein.
695: You are not responsible for enforcing compliance by third parties to
696: this License.
697:
698: @item
699: If, as a consequence of a court judgment or allegation of patent
700: infringement or for any other reason (not limited to patent issues),
701: conditions are imposed on you (whether by court order, agreement or
702: otherwise) that contradict the conditions of this License, they do not
703: excuse you from the conditions of this License. If you cannot
704: distribute so as to satisfy simultaneously your obligations under this
705: License and any other pertinent obligations, then as a consequence you
706: may not distribute the Program at all. For example, if a patent
707: license would not permit royalty-free redistribution of the Program by
708: all those who receive copies directly or indirectly through you, then
709: the only way you could satisfy both it and this License would be to
710: refrain entirely from distribution of the Program.
711:
712: If any portion of this section is held invalid or unenforceable under
713: any particular circumstance, the balance of the section is intended to
714: apply and the section as a whole is intended to apply in other
715: circumstances.
716:
717: It is not the purpose of this section to induce you to infringe any
718: patents or other property right claims or to contest validity of any
719: such claims; this section has the sole purpose of protecting the
720: integrity of the free software distribution system, which is
721: implemented by public license practices. Many people have made
722: generous contributions to the wide range of software distributed
723: through that system in reliance on consistent application of that
724: system; it is up to the author/donor to decide if he or she is willing
725: to distribute software through any other system and a licensee cannot
726: impose that choice.
727:
728: This section is intended to make thoroughly clear what is believed to
729: be a consequence of the rest of this License.
730:
731: @item
732: If the distribution and/or use of the Program is restricted in
733: certain countries either by patents or by copyrighted interfaces, the
734: original copyright holder who places the Program under this License
735: may add an explicit geographical distribution limitation excluding
736: those countries, so that distribution is permitted only in or among
737: countries not thus excluded. In such case, this License incorporates
738: the limitation as if written in the body of this License.
739:
740: @item
741: The Free Software Foundation may publish revised and/or new versions
742: of the General Public License from time to time. Such new versions will
743: be similar in spirit to the present version, but may differ in detail to
744: address new problems or concerns.
745:
746: Each version is given a distinguishing version number. If the Program
747: specifies a version number of this License which applies to it and ``any
748: later version'', you have the option of following the terms and conditions
749: either of that version or of any later version published by the Free
750: Software Foundation. If the Program does not specify a version number of
751: this License, you may choose any version ever published by the Free Software
752: Foundation.
753:
754: @item
755: If you wish to incorporate parts of the Program into other free
756: programs whose distribution conditions are different, write to the author
757: to ask for permission. For software which is copyrighted by the Free
758: Software Foundation, write to the Free Software Foundation; we sometimes
759: make exceptions for this. Our decision will be guided by the two goals
760: of preserving the free status of all derivatives of our free software and
761: of promoting the sharing and reuse of software generally.
762:
763: @iftex
764: @heading NO WARRANTY
765: @end iftex
766: @ifinfo
767: @center NO WARRANTY
768: @end ifinfo
769:
770: @item
771: BECAUSE THE PROGRAM IS LICENSED FREE OF CHARGE, THERE IS NO WARRANTY
772: FOR THE PROGRAM, TO THE EXTENT PERMITTED BY APPLICABLE LAW. EXCEPT WHEN
773: OTHERWISE STATED IN WRITING THE COPYRIGHT HOLDERS AND/OR OTHER PARTIES
774: PROVIDE THE PROGRAM ``AS IS'' WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESSED
775: OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
776: MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. THE ENTIRE RISK AS
777: TO THE QUALITY AND PERFORMANCE OF THE PROGRAM IS WITH YOU. SHOULD THE
778: PROGRAM PROVE DEFECTIVE, YOU ASSUME THE COST OF ALL NECESSARY SERVICING,
779: REPAIR OR CORRECTION.
780:
781: @item
782: IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING
783: WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MAY MODIFY AND/OR
784: REDISTRIBUTE THE PROGRAM AS PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES,
785: INCLUDING ANY GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING
786: OUT OF THE USE OR INABILITY TO USE THE PROGRAM (INCLUDING BUT NOT LIMITED
787: TO LOSS OF DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY
788: YOU OR THIRD PARTIES OR A FAILURE OF THE PROGRAM TO OPERATE WITH ANY OTHER
789: PROGRAMS), EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE
790: POSSIBILITY OF SUCH DAMAGES.
791: @end enumerate
792:
793: @iftex
794: @heading END OF TERMS AND CONDITIONS
795: @end iftex
796: @ifinfo
797: @center END OF TERMS AND CONDITIONS
798: @end ifinfo
799:
800: @page
801: @unnumberedsec How to Apply These Terms to Your New Programs
802:
803: If you develop a new program, and you want it to be of the greatest
804: possible use to the public, the best way to achieve this is to make it
805: free software which everyone can redistribute and change under these terms.
806:
807: To do so, attach the following notices to the program. It is safest
808: to attach them to the start of each source file to most effectively
809: convey the exclusion of warranty; and each file should have at least
810: the ``copyright'' line and a pointer to where the full notice is found.
811:
812: @smallexample
813: @var{one line to give the program's name and a brief idea of what it does.}
814: Copyright (C) 19@var{yy} @var{name of author}
815:
816: This program is free software; you can redistribute it and/or modify
817: it under the terms of the GNU General Public License as published by
818: the Free Software Foundation; either version 2 of the License, or
819: (at your option) any later version.
820:
821: This program is distributed in the hope that it will be useful,
822: but WITHOUT ANY WARRANTY; without even the implied warranty of
823: MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
824: GNU General Public License for more details.
825:
826: You should have received a copy of the GNU General Public License
827: along with this program; if not, write to the Free Software
828: Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
829: @end smallexample
830:
831: Also add information on how to contact you by electronic and paper mail.
832:
833: If the program is interactive, make it output a short notice like this
834: when it starts in an interactive mode:
835:
836: @smallexample
837: Gnomovision version 69, Copyright (C) 19@var{yy} @var{name of author}
838: Gnomovision comes with ABSOLUTELY NO WARRANTY; for details
839: type `show w'.
840: This is free software, and you are welcome to redistribute it
841: under certain conditions; type `show c' for details.
842: @end smallexample
843:
844: The hypothetical commands @samp{show w} and @samp{show c} should show
845: the appropriate parts of the General Public License. Of course, the
846: commands you use may be called something other than @samp{show w} and
847: @samp{show c}; they could even be mouse-clicks or menu items---whatever
848: suits your program.
849:
850: You should also get your employer (if you work as a programmer) or your
851: school, if any, to sign a ``copyright disclaimer'' for the program, if
852: necessary. Here is a sample; alter the names:
853:
854: @smallexample
855: Yoyodyne, Inc., hereby disclaims all copyright interest in the program
856: `Gnomovision' (which makes passes at compilers) written by James Hacker.
857:
858: @var{signature of Ty Coon}, 1 April 1989
859: Ty Coon, President of Vice
860: @end smallexample
861:
862: This General Public License does not permit incorporating your program into
863: proprietary programs. If your program is a subroutine library, you may
864: consider it more useful to permit linking proprietary applications with the
865: library. If this is what you want to do, use the GNU Library General
866: Public License instead of this License.
867:
868: @iftex
869: @unnumbered Preface
870: @cindex Preface
871: This manual documents Gforth. Some introductory material is provided for
872: readers who are unfamiliar with Forth or who are migrating to Gforth
873: from other Forth compilers. However, this manual is primarily a
874: reference manual.
875: @end iftex
876:
877: @comment TODO much more blurb here.
878:
879: @c ******************************************************************
880: @node Goals, Gforth Environment, License, Top
881: @comment node-name, next, previous, up
882: @chapter Goals of Gforth
883: @cindex goals of the Gforth project
884: The goal of the Gforth Project is to develop a standard model for
885: ANS Forth. This can be split into several subgoals:
886:
887: @itemize @bullet
888: @item
889: Gforth should conform to the ANS Forth Standard.
890: @item
891: It should be a model, i.e. it should define all the
892: implementation-dependent things.
893: @item
894: It should become standard, i.e. widely accepted and used. This goal
895: is the most difficult one.
896: @end itemize
897:
898: To achieve these goals Gforth should be
899: @itemize @bullet
900: @item
901: Similar to previous models (fig-Forth, F83)
902: @item
903: Powerful. It should provide for all the things that are considered
904: necessary today and even some that are not yet considered necessary.
905: @item
906: Efficient. It should not get the reputation of being exceptionally
907: slow.
908: @item
909: Free.
910: @item
911: Available on many machines/easy to port.
912: @end itemize
913:
914: Have we achieved these goals? Gforth conforms to the ANS Forth
915: standard. It may be considered a model, but we have not yet documented
916: which parts of the model are stable and which parts we are likely to
917: change. It certainly has not yet become a de facto standard, but it
918: appears to be quite popular. It has some similarities to and some
919: differences from previous models. It has some powerful features, but not
920: yet everything that we envisioned. We certainly have achieved our
921: execution speed goals (@pxref{Performance}). It is free and available
922: on many machines.
923:
924: @menu
925: * Gforth Extensions Sinful?::
926: @end menu
927:
928: @node Gforth Extensions Sinful?, , Goals, Goals
929: @comment node-name, next, previous, up
930: @section Is it a Sin to use Gforth Extensions?
931: @cindex Gforth extensions
932:
933: If you've been paying attention, you will have realised that there is an
934: ANS (American National Standard) for Forth. As you read through the rest
935: of this manual, you will see documentation for @i{Standard} words, and
936: documentation for some appealing Gforth @i{extensions}. You might ask
937: yourself the question: @i{``Given that there is a standard, would I be
938: committing a sin if I use (non-Standard) Gforth extensions?''}
939:
940: The answer to that question is somewhat pragmatic and somewhat
941: philosophical. Consider these points:
942:
943: @itemize @bullet
944: @item
945: A number of the Gforth extensions can be implemented in ANS Forth using
946: files provided in the @file{compat/} directory. These are mentioned in
947: the text in passing.
948: @item
949: Forth has a rich historical precedent for programmers taking advantage
950: of implementation-dependent features of their tools (for example,
951: relying on a knowledge of the dictionary structure). Sometimes these
952: techniques are necessary to extract every last bit of performance from
953: the hardware, sometimes they are just a programming shorthand.
954: @item
955: The best way to break the rules is to know what the rules are. To learn
956: the rules, there is no substitute for studying the text of the Standard
957: itself. In particular, Appendix A of the Standard (@var{Rationale})
958: provides a valuable insight into the thought processes of the technical
959: committee.
960: @item
961: The best reason to break a rule is because you have to; because it's
962: more productive to do that, because it makes your code run fast enough
963: or because you can see no Standard way to achieve what you want to
964: achieve.
965: @end itemize
966:
967: The tool @file{ans-report.fs} (@pxref{ANS Report}) makes it easy to
968: analyse your program and determine what non-Standard definitions it
969: relies upon.
970:
971:
972: @c ******************************************************************
973: @node Gforth Environment, Introduction, Goals, Top
974: @chapter Gforth Environment
975: @cindex Gforth environment
976:
977: Note: ultimately, the Gforth man page will be auto-generated from the
978: material in this chapter.
979:
980: @menu
981: * Invoking Gforth:: Getting in
982: * Leaving Gforth:: Getting out
983: * Command-line editing::
984: * Upper and lower case::
985: * Environment variables:: ..that affect how Gforth starts up
986: * Gforth Files:: What gets installed and where
987: @end menu
988:
989: @xref{Image Files} for related information about the creation of images.
990:
991: @comment ----------------------------------------------
992: @node Invoking Gforth, Leaving Gforth, ,Gforth Environment
993: @section Invoking Gforth
994: @cindex invoking Gforth
995: @cindex running Gforth
996: @cindex command-line options
997: @cindex options on the command line
998: @cindex flags on the command line
999:
1000: Gforth is made up of two parts; an executable ``engine'' (named
1001: @file{gforth} or @file{gforth-fast}) and an image file. To start it, you
1002: will usually just say @code{gforth} -- this automatically loads the
1003: default image file @file{gforth.fi}. In many other cases the default
1004: Gforth image will be invoked like this:
1005: @example
1006: gforth [file | -e forth-code] ...
1007: @end example
1008: @noindent
1009: This interprets the contents of the files and the Forth code in the order they
1010: are given.
1011:
1012: In addition to the @file{gforth} engine, there is also an engine called
1013: @file{gforth-fast}, which is faster, but gives less informative error
1014: messages (@pxref{Error messages}).
1015:
1016: In general, the command line looks like this:
1017:
1018: @example
1019: gforth[-fast] [engine options] [image options]
1020: @end example
1021:
1022: The engine options must come before the rest of the command
1023: line. They are:
1024:
1025: @table @code
1026: @cindex -i, command-line option
1027: @cindex --image-file, command-line option
1028: @item --image-file @i{file}
1029: @itemx -i @i{file}
1030: Loads the Forth image @i{file} instead of the default
1031: @file{gforth.fi} (@pxref{Image Files}).
1032:
1033: @cindex --appl-image, command-line option
1034: @item --appl-image @i{file}
1035: Loads the image @i{file} and leaves all further command-line arguments
1036: to the image (instead of processing them as options). This is useful
1037: for building executable application images on Unix, built with
1038: @code{gforthmi --application ...}.
1039:
1040: @cindex --path, command-line option
1041: @cindex -p, command-line option
1042: @item --path @i{path}
1043: @itemx -p @i{path}
1044: Uses @i{path} for searching the image file and Forth source code files
1045: instead of the default in the environment variable @code{GFORTHPATH} or
1046: the path specified at installation time (e.g.,
1047: @file{/usr/local/share/gforth/0.2.0:.}). A path is given as a list of
1048: directories, separated by @samp{:} (on Unix) or @samp{;} (on other OSs).
1049:
1050: @cindex --dictionary-size, command-line option
1051: @cindex -m, command-line option
1052: @cindex @i{size} parameters for command-line options
1053: @cindex size of the dictionary and the stacks
1054: @item --dictionary-size @i{size}
1055: @itemx -m @i{size}
1056: Allocate @i{size} space for the Forth dictionary space instead of
1057: using the default specified in the image (typically 256K). The
1058: @i{size} specification for this and subsequent options consists of
1059: an integer and a unit (e.g.,
1060: @code{4M}). The unit can be one of @code{b} (bytes), @code{e} (element
1061: size, in this case Cells), @code{k} (kilobytes), @code{M} (Megabytes),
1062: @code{G} (Gigabytes), and @code{T} (Terabytes). If no unit is specified,
1063: @code{e} is used.
1064:
1065: @cindex --data-stack-size, command-line option
1066: @cindex -d, command-line option
1067: @item --data-stack-size @i{size}
1068: @itemx -d @i{size}
1069: Allocate @i{size} space for the data stack instead of using the
1070: default specified in the image (typically 16K).
1071:
1072: @cindex --return-stack-size, command-line option
1073: @cindex -r, command-line option
1074: @item --return-stack-size @i{size}
1075: @itemx -r @i{size}
1076: Allocate @i{size} space for the return stack instead of using the
1077: default specified in the image (typically 15K).
1078:
1079: @cindex --fp-stack-size, command-line option
1080: @cindex -f, command-line option
1081: @item --fp-stack-size @i{size}
1082: @itemx -f @i{size}
1083: Allocate @i{size} space for the floating point stack instead of
1084: using the default specified in the image (typically 15.5K). In this case
1085: the unit specifier @code{e} refers to floating point numbers.
1086:
1087: @cindex --locals-stack-size, command-line option
1088: @cindex -l, command-line option
1089: @item --locals-stack-size @i{size}
1090: @itemx -l @i{size}
1091: Allocate @i{size} space for the locals stack instead of using the
1092: default specified in the image (typically 14.5K).
1093:
1094: @cindex -h, command-line option
1095: @cindex --help, command-line option
1096: @item --help
1097: @itemx -h
1098: Print a message about the command-line options
1099:
1100: @cindex -v, command-line option
1101: @cindex --version, command-line option
1102: @item --version
1103: @itemx -v
1104: Print version and exit
1105:
1106: @cindex --debug, command-line option
1107: @item --debug
1108: Print some information useful for debugging on startup.
1109:
1110: @cindex --offset-image, command-line option
1111: @item --offset-image
1112: Start the dictionary at a slightly different position than would be used
1113: otherwise (useful for creating data-relocatable images,
1114: @pxref{Data-Relocatable Image Files}).
1115:
1116: @cindex --no-offset-im, command-line option
1117: @item --no-offset-im
1118: Start the dictionary at the normal position.
1119:
1120: @cindex --clear-dictionary, command-line option
1121: @item --clear-dictionary
1122: Initialize all bytes in the dictionary to 0 before loading the image
1123: (@pxref{Data-Relocatable Image Files}).
1124:
1125: @cindex --die-on-signal, command-line-option
1126: @item --die-on-signal
1127: Normally Gforth handles most signals (e.g., the user interrupt SIGINT,
1128: or the segmentation violation SIGSEGV) by translating it into a Forth
1129: @code{THROW}. With this option, Gforth exits if it receives such a
1130: signal. This option is useful when the engine and/or the image might be
1131: severely broken (such that it causes another signal before recovering
1132: from the first); this option avoids endless loops in such cases.
1133: @end table
1134:
1135: @cindex loading files at startup
1136: @cindex executing code on startup
1137: @cindex batch processing with Gforth
1138: As explained above, the image-specific command-line arguments for the
1139: default image @file{gforth.fi} consist of a sequence of filenames and
1140: @code{-e @var{forth-code}} options that are interpreted in the sequence
1141: in which they are given. The @code{-e @var{forth-code}} or
1142: @code{--evaluate @var{forth-code}} option evaluates the Forth
1143: code. This option takes only one argument; if you want to evaluate more
1144: Forth words, you have to quote them or use @code{-e} several times. To exit
1145: after processing the command line (instead of entering interactive mode)
1146: append @code{-e bye} to the command line.
1147:
1148: @cindex versions, invoking other versions of Gforth
1149: If you have several versions of Gforth installed, @code{gforth} will
1150: invoke the version that was installed last. @code{gforth-@i{version}}
1151: invokes a specific version. If your environment contains the variable
1152: @code{GFORTHPATH}, you may want to override it by using the
1153: @code{--path} option.
1154:
1155: Not yet implemented:
1156: On startup the system first executes the system initialization file
1157: (unless the option @code{--no-init-file} is given; note that the system
1158: resulting from using this option may not be ANS Forth conformant). Then
1159: the user initialization file @file{.gforth.fs} is executed, unless the
1160: option @code{--no-rc} is given; this file is first searched in @file{.},
1161: then in @file{~}, then in the normal path (see above).
1162:
1163:
1164:
1165: @comment ----------------------------------------------
1166: @node Leaving Gforth, Command-line editing, Invoking Gforth, Gforth Environment
1167: @section Leaving Gforth
1168: @cindex Gforth - leaving
1169: @cindex leaving Gforth
1170:
1171: You can leave Gforth by typing @code{bye} or @kbd{Ctrl-d} (at the start
1172: of a line) or (if you invoked Gforth with the @code{--die-on-signal}
1173: option) @kbd{Ctrl-c}. When you leave Gforth, all of your definitions and
1174: data are discarded. @xref{Image Files} for ways of saving the state of
1175: the system before leaving Gforth.
1176:
1177: doc-bye
1178:
1179:
1180: @comment ----------------------------------------------
1181: @node Command-line editing, Upper and lower case,Leaving Gforth,Gforth Environment
1182: @section Command-line editing
1183: @cindex command-line editing
1184:
1185: Gforth maintains a history file that records every line that you type to
1186: the text interpreter. This file is preserved between sessions, and is
1187: used to provide a command-line recall facility; if you type @kbd{Ctrl-P}
1188: repeatedly you can recall successively older commands from this (or
1189: previous) session(s). The full list of command-line editing facilities is:
1190:
1191: @itemize @bullet
1192: @item
1193: @kbd{Ctrl-p} (``previous'') (or up-arrow) to recall successively older
1194: commands from the history buffer.
1195: @item
1196: @kbd{Ctrl-n} (``next'') (or down-arrow) to recall successively newer commands
1197: from the history buffer.
1198: @item
1199: @kbd{Ctrl-f} (or right-arrow) to move the cursor right, non-destructively.
1200: @item
1201: @kbd{Ctrl-b} (or left-arrow) to move the cursor left, non-destructively.
1202: @item
1203: @kbd{Ctrl-h} (backspace) to delete the character to the left of the cursor,
1204: closing up the line.
1205: @item
1206: @kbd{Ctrl-k} to delete (``kill'') from the cursor to the end of the line.
1207: @item
1208: @kbd{Ctrl-a} to move the cursor to the start of the line.
1209: @item
1210: @kbd{Ctrl-e} to move the cursor to the end of the line.
1211: @item
1212: @key{RET} (@kbd{Ctrl-m}) or @key{LFD} (@kbd{Ctrl-j}) to submit the current
1213: line.
1214: @item
1215: @key{TAB} to step through all possible full-word completions of the word
1216: currently being typed.
1217: @item
1218: @kbd{Ctrl-d} at the start of the line to terminate Gforth (gracefully,
1219: using @code{bye}).
1220: @end itemize
1221:
1222: When editing, displayable characters are inserted to the left of the
1223: cursor position; the line is always in ``insert'' (as opposed to
1224: ``overstrike'') mode.
1225:
1226: @cindex history file
1227: @cindex @file{.gforth-history}
1228: On Unix systems, the history file is @file{~/.gforth-history} by
1229: default@footnote{i.e. it is stored in the user's home directory.}. You
1230: can find out the name and location of your history file using:
1231:
1232: @example
1233: history-file type \ Unix-class systems
1234:
1235: history-file type \ Other systems
1236: history-dir type
1237: @end example
1238:
1239: If you enter long definitions by hand, you can use a text editor to
1240: paste them out of the history file into a Forth source file for reuse at
1241: a later time.
1242:
1243: Gforth never trims the size of the history file, so you should do this
1244: periodically, if necessary.
1245:
1246: @comment this is all defined in history.fs
1247: @comment NAC TODO the ctrl-D behaviour can either do a bye or a beep.. how is that option
1248: @comment chosen?
1249:
1250:
1251:
1252: @comment ----------------------------------------------
1253: @node Upper and lower case, Environment variables,Command-line editing,Gforth Environment
1254: @section Upper and lower case
1255: @cindex case-sensitivity
1256: @cindex upper and lower case
1257:
1258: Gforth is case-insensitive; you can enter definitions and invoke
1259: Standard words using upper, lower or mixed case (however,
1260: @pxref{core-idef, Implementation-defined options, Implementation-defined
1261: options}).
1262:
1263: ANS Forth only @i{requires} implementations to recognise Standard words
1264: when they are typed entirely in upper case. Therefore, a Standard
1265: program must use upper case for all Standard words. You can use whatever
1266: case you like for words that you define, but in a standard program you
1267: have to use the words in the same case that you defined them.
1268:
1269: Gforth supports case sensitivity through @code{table}s (case-sensitive
1270: wordlists, @pxref{Word Lists}).
1271:
1272: Two people have asked how to convert Gforth to case sensitivity; while
1273: we think this is a bad idea, you can change all wordlists into tables
1274: like this:
1275:
1276: @example
1277: ' table-find forth-wordlist wordlist-map @ !
1278: @end example
1279:
1280: Note that you now have to type the predefined words in the same case
1281: that we defined them, which are varying. You may want to convert them
1282: to your favourite case before doing this operation (I won't explain how,
1283: because if you are even contemplating to do this, you'd better have
1284: enough knowledge of Forth systems to know this already).
1285:
1286: @comment ----------------------------------------------
1287: @node Environment variables, Gforth Files, Upper and lower case,Gforth Environment
1288: @section Environment variables
1289: @cindex environment variables
1290:
1291: Gforth uses these environment variables:
1292:
1293: @itemize @bullet
1294: @item
1295: @cindex @code{GFORTHHIST} -- environment variable
1296: @code{GFORTHHIST} -- (Unix systems only) specifies the directory in which to
1297: open/create the history file, @file{.gforth-history}. Default:
1298: @code{$HOME}.
1299:
1300: @item
1301: @cindex @code{GFORTHPATH} -- environment variable
1302: @code{GFORTHPATH} -- specifies the path used when searching for the gforth image file and
1303: for Forth source-code files.
1304:
1305: @item
1306: @cindex @code{GFORTH} -- environment variable
1307: @code{GFORTH} -- used by @file{gforthmi} @xref{gforthmi}.
1308:
1309: @item
1310: @cindex @code{GFORTHD} -- environment variable
1311: @code{GFORTHD} -- used by @file{gforthmi} @xref{gforthmi}.
1312:
1313: @item
1314: @cindex @code{TMP}, @code{TEMP} - environment variable
1315: @code{TMP}, @code{TEMP} - (non-Unix systems only) used as a potential
1316: location for the history file.
1317: @end itemize
1318:
1319: @comment also POSIXELY_CORRECT LINES COLUMNS HOME but no interest in
1320: @comment mentioning these.
1321:
1322: All the Gforth environment variables default to sensible values if they
1323: are not set.
1324:
1325:
1326: @comment ----------------------------------------------
1327: @node Gforth Files, ,Environment variables,Gforth Environment
1328: @section Gforth files
1329: @cindex Gforth files
1330:
1331: When you install Gforth on a Unix system, it installs files in these
1332: locations by default:
1333:
1334: @itemize @bullet
1335: @item
1336: @file{/usr/local/bin/gforth}
1337: @item
1338: @file{/usr/local/bin/gforthmi}
1339: @item
1340: @file{/usr/local/man/man1/gforth.1} - man page.
1341: @item
1342: @file{/usr/local/info} - the Info version of this manual.
1343: @item
1344: @file{/usr/local/lib/gforth/<version>/...} - Gforth @file{.fi} files.
1345: @item
1346: @file{/usr/local/share/gforth/<version>/TAGS} - Emacs TAGS file.
1347: @item
1348: @file{/usr/local/share/gforth/<version>/...} - Gforth source files.
1349: @item
1350: @file{.../emacs/site-lisp/gforth.el} - Emacs gforth mode.
1351: @end itemize
1352:
1353: You can select different places for installation by using
1354: @code{configure} options (listed with @code{configure --help}).
1355:
1356: @c ******************************************************************
1357: @node Introduction, Words, Gforth Environment, Top
1358: @comment node-name, next, previous, up
1359: @chapter An Introduction to ANS Forth
1360: @cindex Forth - an introduction
1361:
1362: The primary purpose of this manual is to document Gforth. However, since
1363: Forth is not a widely-known language and there is a lack of up-to-date
1364: teaching material, it seems worthwhile to provide some introductory
1365: material. @xref{Forth-related information} for other sources of Forth-related
1366: information.
1367:
1368: The examples in this section should work on any ANS Forth; the
1369: output shown was produced using Gforth. Each example attempts to
1370: reproduce the exact output that Gforth produces. If you try out the
1371: examples (and you should), what you should type is shown @kbd{like this}
1372: and Gforth's response is shown @code{like this}. The single exception is
1373: that, where the example shows @key{RET} it means that you should
1374: press the ``carriage return'' key. Unfortunately, some output formats for
1375: this manual cannot show the difference between @kbd{this} and
1376: @code{this} which will make trying out the examples harder (but not
1377: impossible).
1378:
1379: Forth is an unusual language. It provides an interactive development
1380: environment which includes both an interpreter and compiler. Forth
1381: programming style encourages you to break a problem down into many
1382: @cindex factoring
1383: small fragments (@dfn{factoring}), and then to develop and test each
1384: fragment interactively. Forth advocates assert that breaking the
1385: edit-compile-test cycle used by conventional programming languages can
1386: lead to great productivity improvements.
1387:
1388: @menu
1389: * Introducing the Text Interpreter::
1390: * Stacks and Postfix notation::
1391: * Your first definition::
1392: * How does that work?::
1393: * Forth is written in Forth::
1394: * Review - elements of a Forth system::
1395: * Where to go next::
1396: * Exercises::
1397: @end menu
1398:
1399: @comment ----------------------------------------------
1400: @node Introducing the Text Interpreter, Stacks and Postfix notation, Introduction, Introduction
1401: @section Introducing the Text Interpreter
1402: @cindex text interpreter
1403: @cindex outer interpreter
1404:
1405: @c IMO this is too detailed and the pace is too slow for
1406: @c an introduction. If you know German, take a look at
1407: @c http://www.complang.tuwien.ac.at/anton/lvas/skriptum-stack.html
1408: @c to see how I do it - anton
1409:
1410: @c nac-> Where I have accepted your comments 100% and modified the text
1411: @c accordingly, I have deleted your comments. Elsewhere I have added a
1412: @c response like this to attempt to rationalise what I have done. Of
1413: @c course, this is a very clumsy mechanism for something that would be
1414: @c done far more efficiently over a beer. Please delete any dialogue
1415: @c you consider closed.
1416:
1417: When you invoke the Forth image, you will see a startup banner printed
1418: and nothing else (if you have Gforth installed on your system, try
1419: invoking it now, by typing @kbd{gforth@key{RET}}). Forth is now running
1420: its command line interpreter, which is called the @dfn{Text Interpreter}
1421: (also known as the @dfn{Outer Interpreter}). (You will learn a lot
1422: about the text interpreter as you read through this chapter, but
1423: @pxref{The Text Interpreter} for more detail).
1424:
1425: Although it's not obvious, Forth is actually waiting for your
1426: input. Type a number and press the @key{RET} key:
1427:
1428: @example
1429: @kbd{45@key{RET}} ok
1430: @end example
1431:
1432: Rather than give you a prompt to invite you to input something, the text
1433: interpreter prints a status message @i{after} it has processed a line
1434: of input. The status message in this case (``@code{ ok}'' followed by
1435: carriage-return) indicates that the text interpreter was able to process
1436: all of your input successfully. Now type something illegal:
1437:
1438: @example
1439: @kbd{qwer341@key{RET}}
1440: :1: Undefined word
1441: qwer341
1442: ^^^^^^^
1443: $400D2BA8 Bounce
1444: $400DBDA8 no.extensions
1445: @end example
1446:
1447: The exact text, other than the ``Undefined word'' may differ slightly on
1448: your system, but the effect is the same; when the text interpreter
1449: detects an error, it discards any remaining text on a line, resets
1450: certain internal state and prints an error message. @xref{Error
1451: messages} for a detailed description of error messages.
1452:
1453: The text interpreter waits for you to press carriage-return, and then
1454: processes your input line. Starting at the beginning of the line, it
1455: breaks the line into groups of characters separated by spaces. For each
1456: group of characters in turn, it makes two attempts to do something:
1457:
1458: @itemize @bullet
1459: @item
1460: @cindex name dictionary
1461: It tries to treat it as a command. It does this by searching a @dfn{name
1462: dictionary}. If the group of characters matches an entry in the name
1463: dictionary, the name dictionary provides the text interpreter with
1464: information that allows the text interpreter perform some actions. In
1465: Forth jargon, we say that the group
1466: @cindex word
1467: @cindex definition
1468: @cindex execution token
1469: @cindex xt
1470: of characters names a @dfn{word}, that the dictionary search returns an
1471: @dfn{execution token (xt)} corresponding to the @dfn{definition} of the
1472: word, and that the text interpreter executes the xt. Often, the terms
1473: @dfn{word} and @dfn{definition} are used interchangeably.
1474: @item
1475: If the text interpreter fails to find a match in the name dictionary, it
1476: tries to treat the group of characters as a number in the current number
1477: base (when you start up Forth, the current number base is base 10). If
1478: the group of characters legitimately represents a number, the text
1479: interpreter pushes the number onto a stack (we'll learn more about that
1480: in the next section).
1481: @end itemize
1482:
1483: If the text interpreter is unable to do either of these things with any
1484: group of characters, it discards the group of characters and the rest of
1485: the line, then prints an error message. If the text interpreter reaches
1486: the end of the line without error, it prints the status message ``@code{ ok}''
1487: followed by carriage-return.
1488:
1489: This is the simplest command we can give to the text interpreter:
1490:
1491: @example
1492: @key{RET} ok
1493: @end example
1494:
1495: The text interpreter did everything we asked it to do (nothing) without
1496: an error, so it said that everything is ``@code{ ok}''. Try a slightly longer
1497: command:
1498:
1499: @example
1500: @kbd{12 dup fred dup@key{RET}}
1501: :1: Undefined word
1502: 12 dup fred dup
1503: ^^^^
1504: $400D2BA8 Bounce
1505: $400DBDA8 no.extensions
1506: @end example
1507:
1508: When you press the carriage-return key, the text interpreter starts to
1509: work its way along the line:
1510:
1511: @itemize @bullet
1512: @item
1513: When it gets to the space after the @code{2}, it takes the group of
1514: characters @code{12} and looks them up in the name
1515: dictionary@footnote{We can't tell if it found them or not, but assume
1516: for now that it did not}. There is no match for this group of characters
1517: in the name dictionary, so it tries to treat them as a number. It is
1518: able to do this successfully, so it puts the number, 12, ``on the stack''
1519: (whatever that means).
1520: @item
1521: The text interpreter resumes scanning the line and gets the next group
1522: of characters, @code{dup}. It looks it up in the name dictionary and
1523: (you'll have to take my word for this) finds it, and executes the word
1524: @code{dup} (whatever that means).
1525: @item
1526: Once again, the text interpreter resumes scanning the line and gets the
1527: group of characters @code{fred}. It looks them up in the name
1528: dictionary, but can't find them. It tries to treat them as a number, but
1529: they don't represent any legal number.
1530: @end itemize
1531:
1532: At this point, the text interpreter gives up and prints an error
1533: message. The error message shows exactly how far the text interpreter
1534: got in processing the line. In particular, it shows that the text
1535: interpreter made no attempt to do anything with the final character
1536: group, @code{dup}, even though we have good reason to believe that the
1537: text interpreter would have no problem looking that word up and
1538: executing it a second time.
1539:
1540:
1541: @comment ----------------------------------------------
1542: @node Stacks and Postfix notation, Your first definition, Introducing the Text Interpreter, Introduction
1543: @section Stacks, postfix notation and parameter passing
1544: @cindex text interpreter
1545: @cindex outer interpreter
1546:
1547: In procedural programming languages (like C and Pascal), the
1548: building-block of programs is the @dfn{function} or @dfn{procedure}. These
1549: functions or procedures are called with @dfn{explicit parameters}. For
1550: example, in C we might write:
1551:
1552: @example
1553: total = total + new_volume(length,height,depth);
1554: @end example
1555:
1556: @noindent
1557: where new_volume is a function-call to another piece of code, and total,
1558: length, height and depth are all variables. length, height and depth are
1559: parameters to the function-call.
1560:
1561: In Forth, the equivalent of the function or procedure is the
1562: @dfn{definition} and parameters are implicitly passed between
1563: definitions using a shared stack that is visible to the
1564: programmer. Although Forth does support variables, the existence of the
1565: stack means that they are used far less often than in most other
1566: programming languages. When the text interpreter encounters a number, it
1567: will place (@dfn{push}) it on the stack. There are several stacks (the
1568: actual number is implementation-dependent ...) and the particular stack
1569: used for any operation is implied unambiguously by the operation being
1570: performed. The stack used for all integer operations is called the @dfn{data
1571: stack} and, since this is the stack used most commonly, references to
1572: ``the data stack'' are often abbreviated to ``the stack''.
1573:
1574: The stacks have a last-in, first-out (LIFO) organisation. If you type:
1575:
1576: @example
1577: @kbd{1 2 3@key{RET}} ok
1578: @end example
1579:
1580: Then this instructs the text interpreter to placed three numbers on the
1581: (data) stack. An analogy for the behaviour of the stack is to take a
1582: pack of playing cards and deal out the ace (1), 2 and 3 into a pile on
1583: the table. The 3 was the last card onto the pile (``last-in'') and if
1584: you take a card off the pile then, unless you're prepared to fiddle a
1585: bit, the card that you take off will be the 3 (``first-out''). The
1586: number that will be first-out of the stack is called the @dfn{top of
1587: stack}, which
1588: @cindex TOS definition
1589: is often abbreviated to @dfn{TOS}.
1590:
1591: To understand how parameters are passed in Forth, consider the
1592: behaviour of the definition @code{+} (pronounced ``plus''). You will not
1593: be surprised to learn that this definition performs addition. More
1594: precisely, it adds two number together and produces a result. Where does
1595: it get the two numbers from? It takes the top two numbers off the
1596: stack. Where does it place the result? On the stack. You can act-out the
1597: behaviour of @code{+} with your playing cards like this:
1598:
1599: @itemize @bullet
1600: @item
1601: Pick up two cards from the stack on the table
1602: @item
1603: Stare at them intently and ask yourself ``what @i{is} the sum of these two
1604: numbers''
1605: @item
1606: Decide that the answer is 5
1607: @item
1608: Shuffle the two cards back into the pack and find a 5
1609: @item
1610: Put a 5 on the remaining ace that's on the table.
1611: @end itemize
1612:
1613: If you don't have a pack of cards handy but you do have Forth running,
1614: you can use the definition @code{.s} to show the current state of the stack,
1615: without affecting the stack. Type:
1616:
1617: @example
1618: @kbd{clearstack 1 2 3@key{RET}} ok
1619: @kbd{.s@key{RET}} <3> 1 2 3 ok
1620: @end example
1621:
1622: The text interpreter looks up the word @code{clearstack} and executes
1623: it; it tidies up the stack and removes any entries that may have been
1624: left on it by earlier examples. The text interpreter pushes each of the
1625: three numbers in turn onto the stack. Finally, the text interpreter
1626: looks up the word @code{.s} and executes it. The effect of executing
1627: @code{.s} is to print the ``<3>'' (the total number of items on the stack)
1628: followed by a list of all the items on the stack; the item on the far
1629: right-hand side is the TOS.
1630:
1631: You can now type:
1632:
1633: @example
1634: @kbd{+ .s@key{RET}} <2> 1 5 ok
1635: @end example
1636:
1637: @noindent
1638: which is correct; there are now 2 items on the stack and the result of
1639: the addition is 5.
1640:
1641: If you're playing with cards, try doing a second addition: pick up the
1642: two cards, work out that their sum is 6, shuffle them into the pack,
1643: look for a 6 and place that on the table. You now have just one item on
1644: the stack. What happens if you try to do a third addition? Pick up the
1645: first card, pick up the second card -- ah! There is no second card. This
1646: is called a @dfn{stack underflow} and consitutes an error. If you try to
1647: do the same thing with Forth it will report an error (probably a Stack
1648: Underflow or an Invalid Memory Address error).
1649:
1650: The opposite situation to a stack underflow is a @dfn{stack overflow},
1651: which simply accepts that there is a finite amount of storage space
1652: reserved for the stack. To stretch the playing card analogy, if you had
1653: enough packs of cards and you piled the cards up on the table, you would
1654: eventually be unable to add another card; you'd hit the ceiling. Gforth
1655: allows you to set the maximum size of the stacks. In general, the only
1656: time that you will get a stack overflow is because a definition has a
1657: bug in it and is generating data on the stack uncontrollably.
1658:
1659: There's one final use for the playing card analogy. If you model your
1660: stack using a pack of playing cards, the maximum number of items on
1661: your stack will be 52 (I assume you didn't use the Joker). The maximum
1662: @i{value} of any item on the stack is 13 (the King). In fact, the only
1663: possible numbers are positive integer numbers 1 through 13; you can't
1664: have (for example) 0 or 27 or 3.52 or -2. If you change the way you
1665: think about some of the cards, you can accommodate different
1666: numbers. For example, you could think of the Jack as representing 0,
1667: the Queen as representing -1 and the King as representing -2. Your
1668: @i{range} remains unchanged (you can still only represent a total of 13
1669: numbers) but the numbers that you can represent are -2 through 10.
1670:
1671: In that analogy, the limit was the amount of information that a single
1672: stack entry could hold, and Forth has a similar limit. In Forth, the
1673: size of a stack entry is called a @dfn{cell}. The actual size of a cell is
1674: implementation dependent and affects the maximum value that a stack
1675: entry can hold. A Standard Forth provides a cell size of at least
1676: 16-bits, and most desktop systems use a cell size of 32-bits.
1677:
1678: Forth does not do any type checking for you, so you are free to
1679: manipulate and combine stack items in any way you wish. A convenient way
1680: of treating stack items is as 2's complement signed integers, and that
1681: is what Standard words like @code{+} do. Therefore you can type:
1682:
1683: @example
1684: @kbd{-5 12 + .s@key{RET}} <1> 7 ok
1685: @end example
1686:
1687: If you use numbers and definitions like @code{+} in order to turn Forth
1688: into a great big pocket calculator, you will realise that it's rather
1689: different from a normal calculator. Rather than typing 2 + 3 = you had
1690: to type 2 3 + (ignore the fact that you had to use @code{.s} to see the
1691: result). The terminology used to describe this difference is to say that
1692: your calculator uses @dfn{Infix Notation} (parameters and operators are
1693: mixed) whilst Forth uses @dfn{Postfix Notation} (parameters and
1694: operators are separate), also called @dfn{Reverse Polish Notation}.
1695:
1696: Whilst postfix notation might look confusing to begin with, it has
1697: several important advantages:
1698:
1699: @itemize @bullet
1700: @item
1701: it is unambiguous
1702: @item
1703: it is more concise
1704: @item
1705: it fits naturally with a stack-based system
1706: @end itemize
1707:
1708: To examine these claims in more detail, consider these sums:
1709:
1710: @example
1711: 6 + 5 * 4 =
1712: 4 * 5 + 6 =
1713: @end example
1714:
1715: If you're just learning maths or your maths is very rusty, you will
1716: probably come up with the answer 44 for the first and 26 for the
1717: second. If you are a bit of a whizz at maths you will remember the
1718: @i{convention} that multiplication takes precendence over addition, and
1719: you'd come up with the answer 26 both times. To explain the answer 26
1720: to someone who got the answer 44, you'd probably rewrite the first sum
1721: like this:
1722:
1723: @example
1724: 6 + (5 * 4) =
1725: @end example
1726:
1727: If what you really wanted was to perform the addition before the
1728: multiplication, you would have to use parentheses to force it.
1729:
1730: If you did the first two sums on a pocket calculator you would probably
1731: get the right answers, unless you were very cautious and entered them using
1732: these keystroke sequences:
1733:
1734: 6 + 5 = * 4 =
1735: 4 * 5 = + 6 =
1736:
1737: Postfix notation is unambiguous because the order that the operators
1738: are applied is always explicit; that also means that parentheses are
1739: never required. The operators are @i{active} (the act of quoting the
1740: operator makes the operation occur) which removes the need for ``=''.
1741:
1742: The sum 6 + 5 * 4 can be written (in postfix notation) in two
1743: equivalent ways:
1744:
1745: @example
1746: 6 5 4 * + or:
1747: 5 4 * 6 +
1748: @end example
1749:
1750: An important thing that you should notice about this notation is that
1751: the @i{order} of the numbers does not change; if you want to subtract
1752: 2 from 10 you type @code{10 2 -}.
1753:
1754: The reason that Forth uses postfix notation is very simple to explain: it
1755: makes the implementation extremely simple, and it follows naturally from
1756: using the stack as a mechanism for passing parameters. Another way of
1757: thinking about this is to realise that all Forth definitions are
1758: @i{active}; they execute as they are encountered by the text
1759: interpreter. The result of this is that the syntax of Forth is trivially
1760: simple.
1761:
1762:
1763:
1764: @comment ----------------------------------------------
1765: @node Your first definition, How does that work?, Stacks and Postfix notation, Introduction
1766: @section Your first Forth definition
1767: @cindex first definition
1768:
1769: Until now, the examples we've seen have been trivial; we've just been
1770: using Forth as a bigger-than-pocket calculator. Also, each calculation
1771: we've shown has been a ``one-off'' -- to repeat it we'd need to type it in
1772: again@footnote{That's not quite true. If you press the up-arrow key on
1773: your keyboard you should be able to scroll back to any earlier command,
1774: edit it and re-enter it.} In this section we'll see how to add new
1775: words to Forth's vocabulary.
1776:
1777: The easiest way to create a new word is to use a @dfn{colon
1778: definition}. We'll define a few and try them out before worrying too
1779: much about how they work. Try typing in these examples; be careful to
1780: copy the spaces accurately:
1781:
1782: @example
1783: : add-two 2 + . ;
1784: : greet ." Hello and welcome" ;
1785: : demo 5 add-two ;
1786: @end example
1787:
1788: @noindent
1789: Now try them out:
1790:
1791: @example
1792: @kbd{greet@key{RET}} Hello and welcome ok
1793: @kbd{greet greet@key{RET}} Hello and welcomeHello and welcome ok
1794: @kbd{4 add-two@key{RET}} 6 ok
1795: @kbd{demo@key{RET}} 7 ok
1796: @kbd{9 greet demo add-two@key{RET}} Hello and welcome7 11 ok
1797: @end example
1798:
1799: The first new thing that we've introduced here is the pair of words
1800: @code{:} and @code{;}. These are used to start and terminate a new
1801: definition, respectively. The first word after the @code{:} is the name
1802: for the new definition.
1803:
1804: As you can see from the examples, a definition is built up of words that
1805: have already been defined; Forth makes no distinction between
1806: definitions that existed when you started the system up, and those that
1807: you define yourself.
1808:
1809: The examples also introduce the words @code{.} (dot), @code{."}
1810: (dot-quote) and @code{dup} (dewp). Dot takes the value from the top of
1811: the stack and displays it. It's like @code{.s} except that it only
1812: displays the top item of the stack and it is destructive; after it has
1813: executed, the number is no longer on the stack. There is always one
1814: space printed after the number, and no spaces before it. Dot-quote
1815: defines a string (a sequence of characters) that will be printed when
1816: the word is executed. The string can contain any printable characters
1817: except @code{"}. A @code{"} has a special function; it is not a Forth
1818: word but it acts as a delimiter (the way that delimiters work is
1819: described in the next section). Finally, @code{dup} duplicates the value
1820: at the top of the stack. Try typing @code{5 dup .s} to see what it does.
1821:
1822: We already know that the text interpreter searches through the
1823: dictionary to locate names. If you've followed the examples earlier, you
1824: will already have a definition called @code{add-two}. Lets try modifying
1825: it by typing in a new definition:
1826:
1827: @example
1828: @kbd{: add-two dup . ." + 2 =" 2 + . ;@key{RET}} redefined add-two ok
1829: @end example
1830:
1831: Forth recognised that we were defining a word that already exists, and
1832: printed a message to warn us of that fact. Let's try out the new
1833: definition:
1834:
1835: @example
1836: @kbd{9 add-two@key{RET}} 9 + 2 =11 ok
1837: @end example
1838:
1839: @noindent
1840: All that we've actually done here, though, is to create a new
1841: definition, with a particular name. The fact that there was already a
1842: definition with the same name did not make any difference to the way
1843: that the new definition was created (except that Forth printed a warning
1844: message). The old definition of add-two still exists (try @code{demo}
1845: again to see that this is true). Any new definition will use the new
1846: definition of @code{add-two}, but old definitions continue to use the
1847: version that already existed at the time that they were @code{compiled}.
1848:
1849: Before you go on to the next section, try defining and redefining some
1850: words of your own.
1851:
1852: @comment ----------------------------------------------
1853: @node How does that work?, Forth is written in Forth, Your first definition, Introduction
1854: @section How does that work?
1855: @cindex parsing words
1856:
1857: @c That's pretty deep (IMO way too deep) for an introduction. - anton
1858:
1859: @c Is it a good idea to talk about the interpretation semantics of a
1860: @c number? We don't have an xt to go along with it. - anton
1861:
1862: @c Now that I have eliminated execution semantics, I wonder if it would not
1863: @c be better to keep them (or add run-time semantics), to make it easier to
1864: @c explain what compilation semantics usually does. - anton
1865:
1866: @c nac-> I removed the term ``default compilation sematics'' from the
1867: @c introductory chapter. Removing ``execution semantics'' was making
1868: @c everything simpler to explain, then I think the use of this term made
1869: @c everything more complex again. I replaced it with ``default
1870: @c semantics'' (which is used elsewhere in the manual) by which I mean
1871: @c ``a definition that has neither the immediate nor the compile-only
1872: @c flag set''. I reworded big chunks of the ``how does that work''
1873: @c section (and, unusually for me, I think I even made it shorter!). See
1874: @c what you think -- I know I have not addressed your primary concern
1875: @c that it is too heavy-going for an introduction. From what I understood
1876: @c of your course notes it looks as though they might be a good framework.
1877: @c Things that I've tried to capture here are some things that came as a
1878: @c great revelation here when I first understood them. Also, I like the
1879: @c fact that a very simple code example shows up almost all of the issues
1880: @c that you need to understand to see how Forth works. That's unique and
1881: @c worthwhile to emphasise.
1882:
1883: Now we're going to take another look at the definition of @code{add-two}
1884: from the previous section. From our knowledge of the way that the text
1885: interpreter works, we would have expected this result when we tried to
1886: define @code{add-two}:
1887:
1888: @example
1889: @kbd{: add-two 2 + . ;@key{RET}}
1890: ^^^^^^^
1891: Error: Undefined word
1892: @end example
1893:
1894: The reason that this didn't happen is bound up in the way that @code{:}
1895: works. The word @code{:} does two special things. The first special
1896: thing that it does prevents the text interpreter from ever seeing the
1897: characters @code{add-two}. The text interpreter uses a variable called
1898: @cindex modifying >IN
1899: @code{>IN} (pronounced ``to-in'') to keep track of where it is in the
1900: input line. When it encounters the word @code{:} it behaves in exactly
1901: the same way as it does for any other word; it looks it up in the name
1902: dictionary, finds its xt and executes it. When @code{:} executes, it
1903: looks at the input buffer, finds the word @code{add-two} and advances the
1904: value of @code{>IN} to point past it. It then does some other stuff
1905: associated with creating the new definition (including creating an entry
1906: for @code{add-two} in the name dictionary). When the execution of @code{:}
1907: completes, control returns to the text interpreter, which is oblivious
1908: to the fact that it has been tricked into ignoring part of the input
1909: line.
1910:
1911: @cindex parsing words
1912: Words like @code{:} -- words that advance the value of @code{>IN} and so
1913: prevent the text interpreter from acting on the whole of the input line
1914: -- are called @dfn{parsing words}.
1915:
1916: @cindex @code{state} - effect on the text interpreter
1917: @cindex text interpreter - effect of state
1918: The second special thing that @code{:} does is change the value of a
1919: variable called @code{state}, which affects the way that the text
1920: interpreter behaves. When Gforth starts up, @code{state} has the value
1921: 0, and the text interpreter is said to be @dfn{interpreting}. During a
1922: colon definition (started with @code{:}), @code{state} is set to -1 and
1923: the text interpreter is said to be @dfn{compiling}.
1924:
1925: In this example, the text interpreter is compiling when it processes the
1926: string ``@code{2 + . ;}''. It still breaks the string down into
1927: character sequences in the same way. However, instead of pushing the
1928: number @code{2} onto the stack, it lays down (@dfn{compiles}) some magic
1929: into the definition of @code{add-two} that will make the number @code{2} get
1930: pushed onto the stack when @code{add-two} is @dfn{executed}. Similarly,
1931: the behaviours of @code{+} and @code{.} are also compiled into the
1932: definition.
1933:
1934: One category of words don't get compiled. These so-called @dfn{immediate
1935: words} get executed (performed @i{now}) regardless of whether the text
1936: interpreter is interpreting or compiling. The word @code{;} is an
1937: immediate word. Rather than being compiled into the definition, it
1938: executes. Its effect is to terminate the current definition, which
1939: includes changing the value of @code{state} back to 0.
1940:
1941: When you execute @code{add-two}, it has a @dfn{run-time effect} that is
1942: exactly the same as if you had typed @code{2 + . @key{RET}} outside of a
1943: definition.
1944:
1945: In Forth, every word or number can be described in terms of two
1946: properties:
1947:
1948: @itemize @bullet
1949: @item
1950: @cindex interpretation semantics
1951: Its @dfn{interpretation semantics} describe how it will behave when the
1952: text interpreter encounters it in @dfn{interpret} state. The
1953: interpretation semantics of a word are represented by an @dfn{execution
1954: token}.
1955: @item
1956: @cindex compilation semantics
1957: Its @dfn{compilation semantics} describe how it will behave when the
1958: text interpreter encounters it in @dfn{compile} state. The compilation
1959: semantics of a word are represented in an implementation-dependent way;
1960: Gforth uses a @dfn{compilation token}.
1961: @end itemize
1962:
1963: @noindent
1964: Numbers are always treated in a fixed way:
1965:
1966: @itemize @bullet
1967: @item
1968: When the number is @dfn{interpreted}, its behaviour is to push the
1969: number onto the stack.
1970: @item
1971: When the number is @dfn{compiled}, a piece of code is appended to the
1972: current definition that pushes the number when it runs. (In other words,
1973: the compilation semantics of a number are to postpone its interpretation
1974: semantics until the run-time of the definition that it is being compiled
1975: into.)
1976: @end itemize
1977:
1978: Words don't behave in such a regular way, but most have @i{default
1979: semantics} which means that they behave like this:
1980:
1981: @itemize @bullet
1982: @item
1983: The @dfn{interpretation semantics} of the word are to do something useful.
1984: @item
1985: The @dfn{compilation semantics} of the word are to append its
1986: @dfn{interpretation semantics} to the current definition (so that its
1987: run-time behaviour is to do something useful).
1988: @end itemize
1989:
1990: @cindex immediate words
1991: The actual behaviour of any particular word can be controlled by using
1992: the words @code{immediate} and @code{compile-only} when the word is
1993: defined. These words set flags in the name dictionary entry of the most
1994: recently defined word, and these flags are retrieved by the text
1995: interpreter when it finds the word in the name dictionary.
1996:
1997: A word that is marked as @dfn{immediate} has compilation semantics that
1998: are identical to its interpretation semantics. In other words, it
1999: behaves like this:
2000:
2001: @itemize @bullet
2002: @item
2003: The @dfn{interpretation semantics} of the word are to do something useful.
2004: @item
2005: The @dfn{compilation semantics} of the word are to do something useful
2006: (and actually the same thing); i.e., it is executed during compilation.
2007: @end itemize
2008:
2009: Marking a word as @dfn{compile-only} prohibits the text interpreter from
2010: performing the interpretation semantics of the word directly; an attempt
2011: to do so will generate an error. It is never necessary to use
2012: @code{compile-only} (and it is not even part of ANS Forth, though it is
2013: provided by many implementations) but it is good etiquette to apply it
2014: to a word that will not behave correctly (and might have unexpected
2015: side-effects) in interpret state. For example, it is only legal to use
2016: the conditional word @code{IF} within a definition. If you forget this
2017: and try to use it elsewhere, the fact that (in Gforth) it is marked as
2018: @code{compile-only} allows the text interpreter to generate a helpful
2019: error message rather than subjecting you to the consequences of your
2020: folly.
2021:
2022: This example shows the difference between an immediate and a
2023: non-immediate word:
2024:
2025: @example
2026: : show-state state @@ . ;
2027: : show-state-now show-state ; immediate
2028: : word1 show-state ;
2029: : word2 show-state-now ;
2030: @end example
2031:
2032: The word @code{immediate} after the definition of @code{show-state-now}
2033: makes that word an immediate word. These definitions introduce a new
2034: word: @code{@@} (pronounced ``fetch''). This word fetches the value of a
2035: variable, and leaves it on the stack. Therefore, the behaviour of
2036: @code{show-state} is to print a number that represents the current value
2037: of @code{state}.
2038:
2039: When you execute @code{word1}, it prints the number 0, indicating that
2040: the system is interpreting. When the text interpreter compiled the
2041: definition of @code{word1}, it encountered @code{show-state} whose
2042: compilation semantics are to append its interpretation semantics to the
2043: current definition. When you execute @code{word1}, it performs the
2044: interpretation semantics of @code{show-state}. At the time that @code{word1}
2045: (and therefore @code{show-state}) are executed, the system is
2046: interpreting.
2047:
2048: When you pressed @key{RET} after entering the definition of @code{word2},
2049: you should have seen the number -1 printed, followed by ``@code{
2050: ok}''. When the text interpreter compiled the definition of
2051: @code{word2}, it encountered @code{show-state-now}, an immediate word,
2052: whose compilation semantics are therefore to perform its interpretation
2053: semantics. It is executed straight away (even before the text
2054: interpreter has moved on to process another group of characters; the
2055: @code{;} in this example). The effect of executing it are to display the
2056: value of @code{state} @i{at the time that the definition of}
2057: @code{word2} @i{is being defined}. Printing -1 demonstrates that the
2058: system is compiling at this time. If you execute @code{word2} it does
2059: nothing at all.
2060:
2061: @cindex @code{."}, how it works
2062: Before leaving the subject of immediate words, consider the behaviour of
2063: @code{."} in the definition of @code{greet}, in the previous
2064: section. This word is both a parsing word and an immediate word. Notice
2065: that there is a space between @code{."} and the start of the text
2066: @code{Hello and welcome}, but that there is no space between the last
2067: letter of @code{welcome} and the @code{"} character. The reason for this
2068: is that @code{."} is a Forth word; it must have a space after it so that
2069: the text interpreter can identify it. The @code{"} is not a Forth word;
2070: it is a @dfn{delimiter}. The examples earlier show that, when the string
2071: is displayed, there is neither a space before the @code{H} nor after the
2072: @code{e}. Since @code{."} is an immediate word, it executes at the time
2073: that @code{greet} is defined. When it executes, its behaviour is to
2074: search forward in the input line looking for the delimiter. When it
2075: finds the delimiter, it updates @code{>IN} to point past the
2076: delimiter. It also compiles some magic code into the definition of
2077: @code{greet}; the xt of a run-time routine that prints a text string. It
2078: compiles the string @code{Hello and welcome} into memory so that it is
2079: available to be printed later. When the text interpreter gains control,
2080: the next word it finds in the input stream is @code{;} and so it
2081: terminates the definition of @code{greet}.
2082:
2083:
2084: @comment ----------------------------------------------
2085: @node Forth is written in Forth, Review - elements of a Forth system, How does that work?, Introduction
2086: @section Forth is written in Forth
2087: @cindex structure of Forth programs
2088:
2089: When you start up a Forth compiler, a large number of definitions
2090: already exist. In Forth, you develop a new application using bottom-up
2091: programming techniques to create new definitions that are defined in
2092: terms of existing definitions. As you create each definition you can
2093: test and debug it interactively.
2094:
2095: If you have tried out the examples in this section, you will probably
2096: have typed them in by hand; when you leave Gforth, your definitions will
2097: be lost. You can avoid this by using a text editor to enter Forth source
2098: code into a file, and then loading code from the file using
2099: @code{include} (@xref{Forth source files}). A Forth source file is
2100: processed by the text interpreter, just as though you had typed it in by
2101: hand@footnote{Actually, there are some subtle differences -- see
2102: @ref{The Text Interpreter}.}.
2103:
2104: Gforth also supports the traditional Forth alternative to using text
2105: files for program entry (@xref{Blocks}).
2106:
2107: In common with many, if not most, Forth compilers, most of Gforth is
2108: actually written in Forth. All of the @file{.fs} files in the
2109: installation directory@footnote{For example,
2110: @file{/usr/local/share/gforth...}} are Forth source files, which you can
2111: study to see examples of Forth programming.
2112:
2113: Gforth maintains a history file that records every line that you type to
2114: the text interpreter. This file is preserved between sessions, and is
2115: used to provide a command-line recall facility. If you enter long
2116: definitions by hand, you can use a text editor to paste them out of the
2117: history file into a Forth source file for reuse at a later time
2118: (@pxref{Command-line editing} for more information).
2119:
2120:
2121: @comment ----------------------------------------------
2122: @node Review - elements of a Forth system, Where to go next, Forth is written in Forth, Introduction
2123: @section Review - elements of a Forth system
2124: @cindex elements of a Forth system
2125:
2126: To summarise this chapter:
2127:
2128: @itemize @bullet
2129: @item
2130: Forth programs use @dfn{factoring} to break a problem down into small
2131: fragments called @dfn{words} or @dfn{definitions}.
2132: @item
2133: Forth program development is an interactive process.
2134: @item
2135: The main command loop that accepts input, and controls both
2136: interpretation and compilation, is called the @dfn{text interpreter}
2137: (also known as the @dfn{outer interpreter}).
2138: @item
2139: Forth has a very simple syntax, consisting of words and numbers
2140: separated by spaces or carriage-return characters. Any additional syntax
2141: is imposed by @dfn{parsing words}.
2142: @item
2143: Forth uses a stack to pass parameters between words. As a result, it
2144: uses postfix notation.
2145: @item
2146: To use a word that has previously been defined, the text interpreter
2147: searches for the word in the @dfn{name dictionary}.
2148: @item
2149: Words have @dfn{interpretation semantics} and @dfn{compilation semantics}.
2150: @item
2151: The text interpreter uses the value of @code{state} to select between
2152: the use of the @dfn{interpretation semantics} and the @dfn{compilation
2153: semantics} of a word that it encounters.
2154: @item
2155: The relationship between the @dfn{interpretation semantics} and
2156: @dfn{compilation semantics} for a word
2157: depend upon the way in which the word was defined (for example, whether
2158: it is an @dfn{immediate} word).
2159: @item
2160: Forth definitions can be implemented in Forth (called @dfn{high-level
2161: definitions}) or in some other way (usually a lower-level language and
2162: as a result often called @dfn{low-level definitions}, @dfn{code
2163: definitions} or @dfn{primitives}).
2164: @item
2165: Many Forth systems are implemented mainly in Forth.
2166: @end itemize
2167:
2168:
2169: @comment ----------------------------------------------
2170: @node Where to go next,Exercises,Review - elements of a Forth system, Introduction
2171: @section Where To Go Next
2172: @cindex where to go next
2173:
2174: Amazing as it may seem, if you have read (and understood) this far, you
2175: know almost all the fundamentals about the inner workings of a Forth
2176: system. You certainly know enough to be able to read and understand the
2177: rest of this manual and the ANS Forth document, to learn more about the
2178: facilities that Forth in general and Gforth in particular provide. Even
2179: scarier, you know almost enough to implement your own Forth system.
2180: However, that's not a good idea just yet... better to try writing some
2181: programs in Gforth.
2182:
2183: Forth has such a rich vocabulary that it can be hard to know where to
2184: start in learning it. This section suggests a few sets of words that are
2185: enough to write small but useful programs. Use the word index in this
2186: document to learn more about each word, then try it out and try to write
2187: small definitions using it. Start by experimenting with these words:
2188:
2189: @itemize @bullet
2190: @item
2191: Arithmetic: @code{+ - * / /MOD */ ABS INVERT}
2192: @item
2193: Comparison: @code{MIN MAX =}
2194: @item
2195: Logic: @code{AND OR XOR NOT}
2196: @item
2197: Stack manipulation: @code{DUP DROP SWAP OVER}
2198: @item
2199: Loops and decisions: @code{IF ELSE ENDIF ?DO I LOOP}
2200: @item
2201: Input/Output: @code{. ." EMIT CR KEY}
2202: @item
2203: Defining words: @code{: ; CREATE}
2204: @item
2205: Memory allocation words: @code{ALLOT ,}
2206: @item
2207: Tools: @code{SEE WORDS .S MARKER}
2208: @end itemize
2209:
2210: When you have mastered those, go on to:
2211:
2212: @itemize @bullet
2213: @item
2214: More defining words: @code{VARIABLE CONSTANT VALUE TO CREATE DOES>}
2215: @item
2216: Memory access: @code{@@ !}
2217: @end itemize
2218:
2219: When you have mastered these, there's nothing for it but to read through
2220: the whole of this manual and find out what you've missed.
2221:
2222: @comment ----------------------------------------------
2223: @node Exercises, ,Where to go next, Introduction
2224: @section Exercises
2225: @cindex exercises
2226:
2227: TODO: provide a set of programming excercises linked into the stuff done
2228: already and into other sections of the manual. Provide solutions to all
2229: the exercises in a .fs file in the distribution.
2230:
2231: @c Get some inspiration from Starting Forth and Kelly&Spies.
2232:
2233: @c excercises:
2234: @c 1. take inches and convert to feet and inches.
2235: @c 2. take temperature and convert from fahrenheight to celcius;
2236: @c may need to care about symmetric vs floored??
2237: @c 3. take input line and do character substitution
2238: @c to encipher or decipher
2239: @c 4. as above but work on a file for in and out
2240: @c 5. take input line and convert to pig-latin
2241: @c
2242: @c thing of sets of things to exercise then come up with
2243: @c problems that need those things.
2244:
2245:
2246: @c ******************************************************************
2247: @node Words, Error messages, Introduction, Top
2248: @chapter Forth Words
2249: @cindex words
2250:
2251: @menu
2252: * Notation::
2253: * Comments::
2254: * Boolean Flags::
2255: * Arithmetic::
2256: * Stack Manipulation::
2257: * Memory::
2258: * Control Structures::
2259: * Defining Words::
2260: * Interpretation and Compilation Semantics::
2261: * Tokens for Words::
2262: * The Text Interpreter::
2263: * Word Lists::
2264: * Environmental Queries::
2265: * Files::
2266: * Blocks::
2267: * Other I/O::
2268: * Programming Tools::
2269: * Assembler and Code Words::
2270: * Threading Words::
2271: * Locals::
2272: * Structures::
2273: * Object-oriented Forth::
2274: * Passing Commands to the OS::
2275: * Keeping track of Time::
2276: * Miscellaneous Words::
2277: @end menu
2278:
2279: @node Notation, Comments, Words, Words
2280: @section Notation
2281: @cindex notation of glossary entries
2282: @cindex format of glossary entries
2283: @cindex glossary notation format
2284: @cindex word glossary entry format
2285:
2286: The Forth words are described in this section in the glossary notation
2287: that has become a de-facto standard for Forth texts, i.e.,
2288:
2289: @format
2290: @i{word} @i{Stack effect} @i{wordset} @i{pronunciation}
2291: @end format
2292: @i{Description}
2293:
2294: @table @var
2295: @item word
2296: The name of the word.
2297:
2298: @item Stack effect
2299: @cindex stack effect
2300: The stack effect is written in the notation @code{@i{before} --
2301: @i{after}}, where @i{before} and @i{after} describe the top of
2302: stack entries before and after the execution of the word. The rest of
2303: the stack is not touched by the word. The top of stack is rightmost,
2304: i.e., a stack sequence is written as it is typed in. Note that Gforth
2305: uses a separate floating point stack, but a unified stack
2306: notation. Also, return stack effects are not shown in @i{stack
2307: effect}, but in @i{Description}. The name of a stack item describes
2308: the type and/or the function of the item. See below for a discussion of
2309: the types.
2310:
2311: All words have two stack effects: A compile-time stack effect and a
2312: run-time stack effect. The compile-time stack-effect of most words is
2313: @i{ -- }. If the compile-time stack-effect of a word deviates from
2314: this standard behaviour, or the word does other unusual things at
2315: compile time, both stack effects are shown; otherwise only the run-time
2316: stack effect is shown.
2317:
2318: @cindex pronounciation of words
2319: @item pronunciation
2320: How the word is pronounced.
2321:
2322: @cindex wordset
2323: @item wordset
2324: The ANS Forth standard is divided into several word sets. A standard
2325: system need not support all of them. Therefore, in theory, the fewer
2326: word sets your program uses the more portable it will be. However, we
2327: suspect that most ANS Forth systems on personal machines will feature
2328: all word sets. Words that are not defined in ANS Forth have
2329: @code{gforth} or @code{gforth-internal} as word set. @code{gforth}
2330: describes words that will work in future releases of Gforth;
2331: @code{gforth-internal} words are more volatile. Environmental query
2332: strings are also displayed like words; you can recognize them by the
2333: @code{environment} in the word set field.
2334:
2335: @item Description
2336: A description of the behaviour of the word.
2337: @end table
2338:
2339: @cindex types of stack items
2340: @cindex stack item types
2341: The type of a stack item is specified by the character(s) the name
2342: starts with:
2343:
2344: @table @code
2345: @item f
2346: @cindex @code{f}, stack item type
2347: Boolean flags, i.e. @code{false} or @code{true}.
2348: @item c
2349: @cindex @code{c}, stack item type
2350: Char
2351: @item w
2352: @cindex @code{w}, stack item type
2353: Cell, can contain an integer or an address
2354: @item n
2355: @cindex @code{n}, stack item type
2356: signed integer
2357: @item u
2358: @cindex @code{u}, stack item type
2359: unsigned integer
2360: @item d
2361: @cindex @code{d}, stack item type
2362: double sized signed integer
2363: @item ud
2364: @cindex @code{ud}, stack item type
2365: double sized unsigned integer
2366: @item r
2367: @cindex @code{r}, stack item type
2368: Float (on the FP stack)
2369: @item a-
2370: @cindex @code{a_}, stack item type
2371: Cell-aligned address
2372: @item c-
2373: @cindex @code{c_}, stack item type
2374: Char-aligned address (note that a Char may have two bytes in Windows NT)
2375: @item f-
2376: @cindex @code{f_}, stack item type
2377: Float-aligned address
2378: @item df-
2379: @cindex @code{df_}, stack item type
2380: Address aligned for IEEE double precision float
2381: @item sf-
2382: @cindex @code{sf_}, stack item type
2383: Address aligned for IEEE single precision float
2384: @item xt
2385: @cindex @code{xt}, stack item type
2386: Execution token, same size as Cell
2387: @item wid
2388: @cindex @code{wid}, stack item type
2389: Word list ID, same size as Cell
2390: @item f83name
2391: @cindex @code{f83name}, stack item type
2392: Pointer to a name structure
2393: @item "
2394: @cindex @code{"}, stack item type
2395: string in the input stream (not on the stack). The terminating character
2396: is a blank by default. If it is not a blank, it is shown in @code{<>}
2397: quotes.
2398: @end table
2399:
2400: @node Comments, Boolean Flags, Notation, Words
2401: @section Comments
2402: @cindex comments
2403:
2404: Forth supports two styles of comment; the traditional @i{in-line} comment,
2405: @code{(} and its modern cousin, the @i{comment to end of line}; @code{\}.
2406:
2407:
2408: doc-(
2409: doc-\
2410: doc-\G
2411:
2412:
2413: @node Boolean Flags, Arithmetic, Comments, Words
2414: @section Boolean Flags
2415: @cindex Boolean flags
2416:
2417: A Boolean flag is cell-sized. A cell with all bits clear represents the
2418: flag @code{false} and a flag with all bits set represents the flag
2419: @code{true}. Words that check a flag (for example, @code{IF}) will treat
2420: a cell that has @i{any} bit set as @code{true}.
2421:
2422:
2423: doc-true
2424: doc-false
2425: doc-on
2426: doc-off
2427:
2428:
2429: @node Arithmetic, Stack Manipulation, Boolean Flags, Words
2430: @section Arithmetic
2431: @cindex arithmetic words
2432:
2433: @cindex division with potentially negative operands
2434: Forth arithmetic is not checked, i.e., you will not hear about integer
2435: overflow on addition or multiplication, you may hear about division by
2436: zero if you are lucky. The operator is written after the operands, but
2437: the operands are still in the original order. I.e., the infix @code{2-1}
2438: corresponds to @code{2 1 -}. Forth offers a variety of division
2439: operators. If you perform division with potentially negative operands,
2440: you do not want to use @code{/} or @code{/mod} with its undefined
2441: behaviour, but rather @code{fm/mod} or @code{sm/mod} (probably the
2442: former, @pxref{Mixed precision}).
2443: @comment TODO discuss the different division forms and the std approach
2444:
2445: @menu
2446: * Single precision::
2447: * Bitwise operations::
2448: * Double precision:: Double-cell integer arithmetic
2449: * Numeric comparison::
2450: * Mixed precision:: Operations with single and double-cell integers
2451: * Floating Point::
2452: @end menu
2453:
2454: @node Single precision, Bitwise operations, Arithmetic, Arithmetic
2455: @subsection Single precision
2456: @cindex single precision arithmetic words
2457:
2458: By default, numbers in Forth are single-precision integers that are 1
2459: cell in size. They can be signed or unsigned, depending upon how you
2460: treat them. @xref{Number Conversion} for the rules used by the text
2461: interpreter for recognising single-precision integers.
2462:
2463:
2464: doc-+
2465: doc-1+
2466: doc--
2467: doc-1-
2468: doc-*
2469: doc-/
2470: doc-mod
2471: doc-/mod
2472: doc-negate
2473: doc-abs
2474: doc-min
2475: doc-max
2476: doc-d>s
2477: doc-floored
2478:
2479:
2480: @node Bitwise operations, Double precision, Single precision, Arithmetic
2481: @subsection Bitwise operations
2482: @cindex bitwise operation words
2483:
2484:
2485: doc-and
2486: doc-or
2487: doc-xor
2488: doc-invert
2489: doc-lshift
2490: doc-rshift
2491: doc-2*
2492: doc-d2*
2493: doc-2/
2494: doc-d2/
2495:
2496:
2497: @node Double precision, Numeric comparison, Bitwise operations, Arithmetic
2498: @subsection Double precision
2499: @cindex double precision arithmetic words
2500:
2501: @xref{Number Conversion} for the rules used by the text interpreter for
2502: recognising double-precision integers.
2503:
2504: A double precision number is represented by a cell pair, with the most
2505: significant cell at the TOS. It is trivial to convert an unsigned
2506: single to an (unsigned) double; simply push a @code{0} onto the
2507: TOS. Since numbers are represented by Gforth using 2's complement
2508: arithmetic, converting a signed single to a (signed) double requires
2509: sign-extension across the most significant cell. This can be achieved
2510: using @code{s>d}. The moral of the story is that you cannot convert a
2511: number without knowing whether it represents an unsigned or a
2512: signed number.
2513:
2514:
2515: doc-s>d
2516: doc-d+
2517: doc-d-
2518: doc-dnegate
2519: doc-dabs
2520: doc-dmin
2521: doc-dmax
2522:
2523:
2524: @node Numeric comparison, Mixed precision, Double precision, Arithmetic
2525: @subsection Numeric comparison
2526: @cindex numeric comparison words
2527:
2528:
2529: doc-<
2530: doc-<=
2531: doc-<>
2532: doc-=
2533: doc->
2534: doc->=
2535:
2536: doc-0<
2537: doc-0<=
2538: doc-0<>
2539: doc-0=
2540: doc-0>
2541: doc-0>=
2542:
2543: doc-u<
2544: doc-u<=
2545: @c u<> and u= exist but are the same as <> and =
2546: @c doc-u<>
2547: @c doc-u=
2548: doc-u>
2549: doc-u>=
2550:
2551: doc-within
2552:
2553: doc-d<
2554: doc-d<=
2555: doc-d<>
2556: doc-d=
2557: doc-d>
2558: doc-d>=
2559:
2560: doc-d0<
2561: doc-d0<=
2562: doc-d0<>
2563: doc-d0=
2564: doc-d0>
2565: doc-d0>=
2566:
2567: doc-du<
2568: doc-du<=
2569: @c du<> and du= exist but are the same as d<> and d=
2570: @c doc-du<>
2571: @c doc-du=
2572: doc-du>
2573: doc-du>=
2574:
2575:
2576: @node Mixed precision, Floating Point, Numeric comparison, Arithmetic
2577: @subsection Mixed precision
2578: @cindex mixed precision arithmetic words
2579:
2580:
2581: doc-m+
2582: doc-*/
2583: doc-*/mod
2584: doc-m*
2585: doc-um*
2586: doc-m*/
2587: doc-um/mod
2588: doc-fm/mod
2589: doc-sm/rem
2590:
2591:
2592: @node Floating Point, , Mixed precision, Arithmetic
2593: @subsection Floating Point
2594: @cindex floating point arithmetic words
2595:
2596: @xref{Number Conversion} for the rules used by the text interpreter for
2597: recognising floating-point numbers.
2598:
2599: Gforth has a separate floating point
2600: stack, but the documentation uses the unified notation.
2601:
2602: @cindex floating-point arithmetic, pitfalls
2603: Floating point numbers have a number of unpleasant surprises for the
2604: unwary (e.g., floating point addition is not associative) and even a few
2605: for the wary. You should not use them unless you know what you are doing
2606: or you don't care that the results you get are totally bogus. If you
2607: want to learn about the problems of floating point numbers (and how to
2608: avoid them), you might start with @cite{David Goldberg, What Every
2609: Computer Scientist Should Know About Floating-Point Arithmetic, ACM
2610: Computing Surveys 23(1):5@minus{}48, March 1991}
2611: (@uref{http://www.validgh.com/goldberg/paper.ps}).
2612:
2613:
2614: doc-d>f
2615: doc-f>d
2616: doc-f+
2617: doc-f-
2618: doc-f*
2619: doc-f/
2620: doc-fnegate
2621: doc-fabs
2622: doc-fmax
2623: doc-fmin
2624: doc-floor
2625: doc-fround
2626: doc-f**
2627: doc-fsqrt
2628: doc-fexp
2629: doc-fexpm1
2630: doc-fln
2631: doc-flnp1
2632: doc-flog
2633: doc-falog
2634: doc-f2*
2635: doc-f2/
2636: doc-1/f
2637: doc-precision
2638: doc-set-precision
2639:
2640: @cindex angles in trigonometric operations
2641: @cindex trigonometric operations
2642: Angles in floating point operations are given in radians (a full circle
2643: has 2 pi radians).
2644:
2645: doc-fsin
2646: doc-fcos
2647: doc-fsincos
2648: doc-ftan
2649: doc-fasin
2650: doc-facos
2651: doc-fatan
2652: doc-fatan2
2653: doc-fsinh
2654: doc-fcosh
2655: doc-ftanh
2656: doc-fasinh
2657: doc-facosh
2658: doc-fatanh
2659: doc-pi
2660:
2661: @cindex equality of floats
2662: @cindex floating-point comparisons
2663: One particular problem with floating-point arithmetic is that comparison
2664: for equality often fails when you would expect it to succeed. For this
2665: reason approximate equality is often preferred (but you still have to
2666: know what you are doing). The comparison words are:
2667:
2668: doc-f~rel
2669: doc-f~abs
2670: doc-f=
2671: doc-f~
2672: doc-f<>
2673:
2674: doc-f<
2675: doc-f<=
2676: doc-f>
2677: doc-f>=
2678:
2679: doc-f0<
2680: doc-f0<=
2681: doc-f0<>
2682: doc-f0=
2683: doc-f0>
2684: doc-f0>=
2685:
2686:
2687: @node Stack Manipulation, Memory, Arithmetic, Words
2688: @section Stack Manipulation
2689: @cindex stack manipulation words
2690:
2691: @cindex floating-point stack in the standard
2692: Gforth maintains a number of separate stacks:
2693:
2694: @cindex data stack
2695: @cindex parameter stack
2696: @itemize @bullet
2697: @item
2698: A data stack (also known as the @dfn{parameter stack}) -- for
2699: characters, cells, addresses, and double cells.
2700:
2701: @cindex floating-point stack
2702: @item
2703: A floating point stack -- for holding floating point (FP) numbers.
2704:
2705: @cindex return stack
2706: @item
2707: A return stack -- for holding the return addresses of colon
2708: definitions and other (non-FP) data.
2709:
2710: @cindex locals stack
2711: @item
2712: A locals stack -- for holding local variables.
2713: @end itemize
2714:
2715: @menu
2716: * Data stack::
2717: * Floating point stack::
2718: * Return stack::
2719: * Locals stack::
2720: * Stack pointer manipulation::
2721: @end menu
2722:
2723: @node Data stack, Floating point stack, Stack Manipulation, Stack Manipulation
2724: @subsection Data stack
2725: @cindex data stack manipulation words
2726: @cindex stack manipulations words, data stack
2727:
2728:
2729: doc-drop
2730: doc-nip
2731: doc-dup
2732: doc-over
2733: doc-tuck
2734: doc-swap
2735: doc-pick
2736: doc-rot
2737: doc--rot
2738: doc-?dup
2739: doc-roll
2740: doc-2drop
2741: doc-2nip
2742: doc-2dup
2743: doc-2over
2744: doc-2tuck
2745: doc-2swap
2746: doc-2rot
2747:
2748:
2749: @node Floating point stack, Return stack, Data stack, Stack Manipulation
2750: @subsection Floating point stack
2751: @cindex floating-point stack manipulation words
2752: @cindex stack manipulation words, floating-point stack
2753:
2754: Whilst every sane Forth has a separate floating-point stack, it is not
2755: strictly required; an ANS Forth system could theoretically keep
2756: floating-point numbers on the data stack. As an additional difficulty,
2757: you don't know how many cells a floating-point number takes. It is
2758: reportedly possible to write words in a way that they work also for a
2759: unified stack model, but we do not recommend trying it. Instead, just
2760: say that your program has an environmental dependency on a separate
2761: floating-point stack.
2762:
2763: doc-floating-stack
2764:
2765: doc-fdrop
2766: doc-fnip
2767: doc-fdup
2768: doc-fover
2769: doc-ftuck
2770: doc-fswap
2771: doc-fpick
2772: doc-frot
2773:
2774:
2775: @node Return stack, Locals stack, Floating point stack, Stack Manipulation
2776: @subsection Return stack
2777: @cindex return stack manipulation words
2778: @cindex stack manipulation words, return stack
2779:
2780: @cindex return stack and locals
2781: @cindex locals and return stack
2782: A Forth system is allowed to keep local variables on the
2783: return stack. This is reasonable, as local variables usually eliminate
2784: the need to use the return stack explicitly. So, if you want to produce
2785: a standard compliant program and you are using local variables in a
2786: word, forget about return stack manipulations in that word (refer to the
2787: standard document for the exact rules).
2788:
2789: doc->r
2790: doc-r>
2791: doc-r@
2792: doc-rdrop
2793: doc-2>r
2794: doc-2r>
2795: doc-2r@
2796: doc-2rdrop
2797:
2798:
2799: @node Locals stack, Stack pointer manipulation, Return stack, Stack Manipulation
2800: @subsection Locals stack
2801:
2802: Gforth uses an extra locals stack. It is described, along with the
2803: reasons for its existence, in @ref{Implementation,Implementation of locals}.
2804:
2805: @node Stack pointer manipulation, , Locals stack, Stack Manipulation
2806: @subsection Stack pointer manipulation
2807: @cindex stack pointer manipulation words
2808:
2809: @c removed s0 r0 l0 -- they are obsolete aliases for sp0 rp0 lp0
2810: doc-sp0
2811: doc-sp@
2812: doc-sp!
2813: doc-fp0
2814: doc-fp@
2815: doc-fp!
2816: doc-rp0
2817: doc-rp@
2818: doc-rp!
2819: doc-lp0
2820: doc-lp@
2821: doc-lp!
2822:
2823:
2824: @node Memory, Control Structures, Stack Manipulation, Words
2825: @section Memory
2826: @cindex memory words
2827:
2828: @menu
2829: * Memory model::
2830: * Dictionary allocation::
2831: * Heap Allocation::
2832: * Memory Access::
2833: * Address arithmetic::
2834: * Memory Blocks::
2835: @end menu
2836:
2837: @node Memory model, Dictionary allocation, Memory, Memory
2838: @subsection ANS Forth and Gforth memory models
2839:
2840: @c The ANS Forth description is a mess (e.g., is the heap part of
2841: @c the dictionary?), so let's not stick to closely with it.
2842:
2843: ANS Forth considers a Forth system as consisting of several memories, of
2844: which only @dfn{data space} is managed and accessible with the memory
2845: words. Memory not necessarily in data space includes the stacks, the
2846: code (called code space) and the headers (called name space). In Gforth
2847: everything is in data space, but the code for the primitives is usually
2848: read-only.
2849:
2850: Data space is divided into a number of areas: The (data space portion of
2851: the) dictionary@footnote{Sometimes, the term @dfn{dictionary} is used to
2852: refer to the search data structure embodied in word lists and headers,
2853: because it is used for looking up names, just as you would in a
2854: conventional dictionary.}, the heap, and a number of system-allocated
2855: buffers.
2856:
2857: In ANS Forth data space is also divided into contiguous regions. You
2858: can only use address arithmetic within a contiguous region, not between
2859: them. Usually each allocation gives you one contiguous region, but the
2860: dictionary allocation words have additional rules (@pxref{Dictionary
2861: allocation}).
2862:
2863: Gforth provides one big address space, and address arithmetic can be
2864: performed between any addresses. However, in the dictionary headers or
2865: code are interleaved with data, so almost the only contiguous data space
2866: regions there are those described by ANS Forth as contiguous; but you
2867: can be sure that the dictionary is allocated towards increasing
2868: addresses even between contiguous regions. The memory order of
2869: allocations in the heap is platform-dependent (and possibly different
2870: from one run to the next).
2871:
2872: @subsubsection ANS Forth dictionary details
2873:
2874: This section is just informative, you can skip it if you are in a hurry.
2875:
2876: When you create a colon definition, the text interpreter compiles the
2877: code for the definition into the code space and compiles the name
2878: of the definition into the header space, together with other
2879: information about the definition (such as its execution token).
2880:
2881: When you create a variable, the execution of @code{Variable} will
2882: compile some code, assign one cell in data space, and compile the name
2883: of the variable into the header space.
2884:
2885: @cindex memory regions - relationship between them
2886: ANS Forth does not specify the relationship between the three memory
2887: regions, and specifies that a Standard program must not access code or
2888: data space directly -- it may only access data space directly. In
2889: addition, the Standard defines what relationships you may and may not
2890: rely on when allocating regions in data space. These constraints are
2891: simply a reflection of the many diverse techniques that are used to
2892: implement Forth systems; understanding and following the requirements of
2893: the Standard allows you to write portable programs -- programs that run
2894: in the same way on any of these diverse systems. Another way of looking
2895: at this is to say that ANS Forth was designed to permit compliant Forth
2896: systems to be implemented in many diverse ways.
2897:
2898: @cindex memory regions - how they are assigned
2899: Here are some examples of ways in which name, code and data spaces
2900: might be assigned in different Forth implementations:
2901:
2902: @itemize @bullet
2903: @item
2904: For a Forth system that runs from RAM under a general-purpose operating
2905: system, it can be convenient to interleave name, code and data spaces in
2906: a single contiguous memory region. This organisation can be
2907: memory-efficient (for example, because the relationship between the name
2908: dictionary entry and the associated code space entry can be
2909: implicit, rather than requiring an explicit memory pointer to reference
2910: from the header space and the code space). This is the
2911: organisation used by Gforth, as this example@footnote{The addresses
2912: in the example have been truncated to fit it onto the page, and the
2913: addresses and data shown will not match the output from your system} shows:
2914: @example
2915: hex
2916: variable fred 123456 fred !
2917: variable jim abcd jim !
2918: : foo + / - ;
2919: ' fred 10 - 50 dump
2920: ..80: 5C 46 0E 40 84 66 72 65 - 64 20 20 20 20 20 20 20 \F.@.fred
2921: ..90: D0 9B 04 08 00 00 00 00 - 56 34 12 00 80 46 0E 40 ........V4...F.@
2922: ..A0: 83 6A 69 6D 20 20 20 20 - D0 9B 04 08 00 00 00 00 .jim ........
2923: ..B0: CD AB 00 00 9C 46 0E 40 - 83 66 6F 6F 20 20 20 20 .....F.@.foo
2924: ..C0: 80 9B 04 08 00 00 00 00 - E4 2E 05 08 0C 2F 05 08 ............./..
2925: @end example
2926:
2927: @item
2928: For a high-performance system running on a modern RISC processor with a
2929: modified Harvard architecture (one that has a unified main memory but
2930: separate instruction and data caches), it is desirable to separate
2931: processor instructions from processor data. This encourages a high cache
2932: density and therefore a high cache hit rate. The Forth code space
2933: is not necessarily made up entirely of processor instructions; its
2934: nature is dependent upon the Forth implementation.
2935:
2936: @item
2937: A Forth compiler that runs on a segmented 8086 processor could be
2938: designed to interleave the name, code and data spaces within a single
2939: 64Kbyte segment. A more common implementation choice is to use a
2940: separate 64Kbyte segment for each region, which provides more memory
2941: overall but provides an address map in which only the data space is
2942: accessible.
2943:
2944: @item
2945: Microprocessors exist that run Forth (or many of the primitives required
2946: to implement the Forth virtual machine efficiently) directly. On these
2947: processors, the relationship between name, code and data spaces may be
2948: imposed as a side-effect of the architecture of the processor.
2949:
2950: @item
2951: A Forth compiler that executes from ROM on an embedded system needs its
2952: data space separated from the name and code spaces so that the data
2953: space can be mapped to a RAM area.
2954:
2955: @item
2956: A Forth compiler that runs on an embedded system may have a requirement
2957: for a small memory footprint. On such a system it can be useful to
2958: separate the header space from the data and code spaces; once the
2959: application has been compiled, the header space is no longer
2960: required@footnote{more strictly speaking, most applications can be
2961: designed so that this is the case}. The header space can be deleted
2962: entirely, or could be stored in memory on a remote @i{host} system for
2963: debug and development purposes. In the latter case, the compiler running
2964: on the @i{target} system could implement a protocol across a
2965: communication link that would allow it to interrogate the header space.
2966: @end itemize
2967:
2968: @node Dictionary allocation, Heap Allocation, Memory model, Memory
2969: @subsection Dictionary allocation
2970: @cindex reserving data space
2971: @cindex data space - reserving some
2972:
2973: Dictionary allocation is a stack-oriented allocation scheme, i.e., if
2974: you want to deallocate X, you also deallocate everything
2975: allocated after X.
2976:
2977: The allocations using the words below are contiguous and grow the region
2978: towards increasing addresses. Other words that allocate dictionary
2979: memory of any kind (i.e., defining words including @code{:noname}) end
2980: the contiguous region and start a new one.
2981:
2982: In ANS Forth only @code{create}d words are guaranteed to produce an
2983: address that is the start of the following contiguous region. In
2984: particular, the cell allocated by @code{variable} is not guaranteed to
2985: be contiguous with following @code{allot}ed memory.
2986:
2987: You can deallocate memory by using @code{allot} with a negative argument
2988: (with some restrictions, see @code{allot}). For larger deallocations use
2989: @code{marker}.
2990:
2991:
2992: doc-here
2993: doc-unused
2994: doc-allot
2995: doc-c,
2996: doc-f,
2997: doc-,
2998: doc-2,
2999: @cindex user space
3000: doc-udp
3001: doc-uallot
3002:
3003: Memory accesses have to be aligned (@pxref{Address arithmetic}). So of
3004: course you should allocate memory in an aligned way, too. I.e., before
3005: allocating allocating a cell, @code{here} must be cell-aligned, etc.
3006: The words below align @code{here} if it is not already. Basically it is
3007: only already aligned for a type, if the last allocation was a multiple
3008: of the size of this type and if @code{here} was aligned for this type
3009: before.
3010:
3011: After freshly @code{create}ing a word, @code{here} is @code{align}ed in
3012: ANS Forth (@code{maxalign}ed in Gforth).
3013:
3014: doc-align
3015: doc-falign
3016: doc-sfalign
3017: doc-dfalign
3018: doc-maxalign
3019: doc-cfalign
3020:
3021:
3022: @node Heap Allocation, Memory Access, Dictionary allocation, Memory
3023: @subsection Heap allocation
3024: @cindex heap allocation
3025: @cindex dynamic allocation of memory
3026: @cindex memory-allocation word set
3027:
3028: Heap allocation supports deallocation of allocated memory in any
3029: order. Dictionary allocation is not affected by it (i.e., it does not
3030: end a contiguous region). In Gforth, these words are implemented using
3031: the standard C library calls malloc(), free() and resize().
3032:
3033: doc-allocate
3034: doc-free
3035: doc-resize
3036:
3037:
3038: @node Memory Access, Address arithmetic, Heap Allocation, Memory
3039: @subsection Memory Access
3040: @cindex memory access words
3041:
3042:
3043: doc-@
3044: doc-!
3045: doc-+!
3046: doc-c@
3047: doc-c!
3048: doc-2@
3049: doc-2!
3050: doc-f@
3051: doc-f!
3052: doc-sf@
3053: doc-sf!
3054: doc-df@
3055: doc-df!
3056:
3057: @node Address arithmetic, Memory Blocks, Memory Access, Memory
3058: @subsection Address arithmetic
3059: @cindex address arithmetic words
3060:
3061: Address arithmetic is the foundation on which data structures like
3062: arrays, records (@pxref{Structures}) and objects (@pxref{Object-oriented
3063: Forth}) are built.
3064:
3065: ANS Forth does not specify the sizes of the data types. Instead, it
3066: offers a number of words for computing sizes and doing address
3067: arithmetic. Address arithmetic is performed in terms of address units
3068: (aus); on most systems the address unit is one byte. Note that a
3069: character may have more than one au, so @code{chars} is no noop (on
3070: systems where it is a noop, it compiles to nothing).
3071:
3072: @cindex alignment of addresses for types
3073: ANS Forth also defines words for aligning addresses for specific
3074: types. Many computers require that accesses to specific data types
3075: must only occur at specific addresses; e.g., that cells may only be
3076: accessed at addresses divisible by 4. Even if a machine allows unaligned
3077: accesses, it can usually perform aligned accesses faster.
3078:
3079: For the performance-conscious: alignment operations are usually only
3080: necessary during the definition of a data structure, not during the
3081: (more frequent) accesses to it.
3082:
3083: ANS Forth defines no words for character-aligning addresses. This is not
3084: an oversight, but reflects the fact that addresses that are not
3085: char-aligned have no use in the standard and therefore will not be
3086: created.
3087:
3088: @cindex @code{CREATE} and alignment
3089: ANS Forth guarantees that addresses returned by @code{CREATE}d words
3090: are cell-aligned; in addition, Gforth guarantees that these addresses
3091: are aligned for all purposes.
3092:
3093: Note that the ANS Forth word @code{char} has nothing to do with address
3094: arithmetic.
3095:
3096:
3097: doc-chars
3098: doc-char+
3099: doc-cells
3100: doc-cell+
3101: doc-cell
3102: doc-aligned
3103: doc-floats
3104: doc-float+
3105: doc-float
3106: doc-faligned
3107: doc-sfloats
3108: doc-sfloat+
3109: doc-sfaligned
3110: doc-dfloats
3111: doc-dfloat+
3112: doc-dfaligned
3113: doc-maxaligned
3114: doc-cfaligned
3115: doc-address-unit-bits
3116:
3117:
3118: @node Memory Blocks, , Address arithmetic, Memory
3119: @subsection Memory Blocks
3120: @cindex memory block words
3121: @cindex character strings - moving and copying
3122:
3123: Memory blocks often represent character strings; @xref{String Formats}
3124: for ways of storing character strings in memory. @xref{Displaying
3125: characters and strings} for other string-processing words.
3126:
3127: Some of these words work on address units. Others work on character
3128: units (increments of @code{CHAR}), and expect a @code{CHAR}-aligned
3129: address. Choose the correct operation depending upon your data type.
3130:
3131: When copying characters between overlapping memory regions, choose
3132: carefully between @code{cmove} and @code{cmove>}.
3133:
3134: You can only use any of these words @i{portably} to access data space.
3135:
3136: @comment TODO - think the naming of the arguments is wrong for move
3137: @comment well, really it seems to be the Standard that's wrong; it
3138: @comment describes MOVE as a word that requires a CELL-aligned source
3139: @comment and destination address but a xtranfer count that need not
3140: @comment be a multiple of CELL.
3141:
3142: doc-move
3143: doc-erase
3144: doc-cmove
3145: doc-cmove>
3146: doc-fill
3147: doc-blank
3148: doc-compare
3149: doc-search
3150: doc--trailing
3151: doc-/string
3152:
3153:
3154: @comment TODO examples
3155:
3156:
3157: @node Control Structures, Defining Words, Memory, Words
3158: @section Control Structures
3159: @cindex control structures
3160:
3161: Control structures in Forth cannot be used interpretively, only in a
3162: colon definition@footnote{To be precise, they have no interpretation
3163: semantics (@pxref{Interpretation and Compilation Semantics}).}. We do
3164: not like this limitation, but have not seen a satisfying way around it
3165: yet, although many schemes have been proposed.
3166:
3167: @menu
3168: * Selection:: IF ... ELSE ... ENDIF
3169: * Simple Loops:: BEGIN ...
3170: * Counted Loops:: DO
3171: * Arbitrary control structures::
3172: * Calls and returns::
3173: * Exception Handling::
3174: @end menu
3175:
3176: @node Selection, Simple Loops, Control Structures, Control Structures
3177: @subsection Selection
3178: @cindex selection control structures
3179: @cindex control structures for selection
3180:
3181: @c what's the purpose of all these @i? Maybe we should define a macro
3182: @c so we can produce logical markup. - anton
3183:
3184: @c nac-> When I started working on the manual, a mixture of @i and @var
3185: @c were used inconsistently in code examples and \Glossary entries. These
3186: @c two behave differently in info format so I decided to standardize on @i.
3187: @c Logical markup would be better but texi isn't really upto it, and
3188: @c texi2html just ignores macros.
3189: @c nac02dec1999-> update: the latest texinfo release can spit out html
3190: @c and it handles macros, so we could do some logical markup. Unfortunately
3191: @c texinfo will not split html output, which would be a big pain if you
3192: @c wanted to put the document on the web, which would be nice.
3193:
3194: @cindex @code{IF} control structure
3195: @example
3196: @i{flag}
3197: IF
3198: @i{code}
3199: ENDIF
3200: @end example
3201: @noindent
3202:
3203: If @i{flag} is non-zero (as far as @code{IF} etc. are concerned, a cell
3204: with any bit set represents truth) @i{code} is executed.
3205:
3206: @example
3207: @i{flag}
3208: IF
3209: @i{code1}
3210: ELSE
3211: @i{code2}
3212: ENDIF
3213: @end example
3214:
3215: If @var{flag} is true, @i{code1} is executed, otherwise @i{code2} is
3216: executed.
3217:
3218: You can use @code{THEN} instead of @code{ENDIF}. Indeed, @code{THEN} is
3219: standard, and @code{ENDIF} is not, although it is quite popular. We
3220: recommend using @code{ENDIF}, because it is less confusing for people
3221: who also know other languages (and is not prone to reinforcing negative
3222: prejudices against Forth in these people). Adding @code{ENDIF} to a
3223: system that only supplies @code{THEN} is simple:
3224: @example
3225: : ENDIF POSTPONE THEN ; immediate
3226: @end example
3227:
3228: [According to @cite{Webster's New Encyclopedic Dictionary}, @dfn{then
3229: (adv.)} has the following meanings:
3230: @quotation
3231: ... 2b: following next after in order ... 3d: as a necessary consequence
3232: (if you were there, then you saw them).
3233: @end quotation
3234: Forth's @code{THEN} has the meaning 2b, whereas @code{THEN} in Pascal
3235: and many other programming languages has the meaning 3d.]
3236:
3237: Gforth also provides the words @code{?DUP-IF} and @code{?DUP-0=-IF}, so
3238: you can avoid using @code{?dup}. Using these alternatives is also more
3239: efficient than using @code{?dup}. Definitions in ANS Forth
3240: for @code{ENDIF}, @code{?DUP-IF} and @code{?DUP-0=-IF} are provided in
3241: @file{compat/control.fs}.
3242:
3243: @cindex @code{CASE} control structure
3244: @example
3245: @i{n}
3246: CASE
3247: @i{n1} OF @i{code1} ENDOF
3248: @i{n2} OF @i{code2} ENDOF
3249: @dots{}
3250: ENDCASE
3251: @end example
3252:
3253: Executes the first @i{codei}, where the @i{ni} is equal to
3254: @i{n}. A default case can be added by simply writing the code after
3255: the last @code{ENDOF}. It may use @i{n}, which is on top of the stack,
3256: but must not consume it.
3257:
3258: @node Simple Loops, Counted Loops, Selection, Control Structures
3259: @subsection Simple Loops
3260: @cindex simple loops
3261: @cindex loops without count
3262:
3263: @cindex @code{WHILE} loop
3264: @example
3265: BEGIN
3266: @i{code1}
3267: @i{flag}
3268: WHILE
3269: @i{code2}
3270: REPEAT
3271: @end example
3272:
3273: @i{code1} is executed and @i{flag} is computed. If it is true,
3274: @i{code2} is executed and the loop is restarted; If @i{flag} is
3275: false, execution continues after the @code{REPEAT}.
3276:
3277: @cindex @code{UNTIL} loop
3278: @example
3279: BEGIN
3280: @i{code}
3281: @i{flag}
3282: UNTIL
3283: @end example
3284:
3285: @i{code} is executed. The loop is restarted if @code{flag} is false.
3286:
3287: @cindex endless loop
3288: @cindex loops, endless
3289: @example
3290: BEGIN
3291: @i{code}
3292: AGAIN
3293: @end example
3294:
3295: This is an endless loop.
3296:
3297: @node Counted Loops, Arbitrary control structures, Simple Loops, Control Structures
3298: @subsection Counted Loops
3299: @cindex counted loops
3300: @cindex loops, counted
3301: @cindex @code{DO} loops
3302:
3303: The basic counted loop is:
3304: @example
3305: @i{limit} @i{start}
3306: ?DO
3307: @i{body}
3308: LOOP
3309: @end example
3310:
3311: This performs one iteration for every integer, starting from @i{start}
3312: and up to, but excluding @i{limit}. The counter, or @i{index}, can be
3313: accessed with @code{i}. For example, the loop:
3314: @example
3315: 10 0 ?DO
3316: i .
3317: LOOP
3318: @end example
3319: @noindent
3320: prints @code{0 1 2 3 4 5 6 7 8 9}
3321:
3322: The index of the innermost loop can be accessed with @code{i}, the index
3323: of the next loop with @code{j}, and the index of the third loop with
3324: @code{k}.
3325:
3326:
3327: doc-i
3328: doc-j
3329: doc-k
3330:
3331:
3332: The loop control data are kept on the return stack, so there are some
3333: restrictions on mixing return stack accesses and counted loop words. In
3334: particuler, if you put values on the return stack outside the loop, you
3335: cannot read them inside the loop@footnote{well, not in a way that is
3336: portable.}. If you put values on the return stack within a loop, you
3337: have to remove them before the end of the loop and before accessing the
3338: index of the loop.
3339:
3340: There are several variations on the counted loop:
3341:
3342: @itemize @bullet
3343: @item
3344: @code{LEAVE} leaves the innermost counted loop immediately; execution
3345: continues after the associated @code{LOOP} or @code{NEXT}. For example:
3346:
3347: @example
3348: 10 0 ?DO i DUP . 3 = IF LEAVE THEN LOOP
3349: @end example
3350: prints @code{0 1 2 3}
3351:
3352:
3353: @item
3354: @code{UNLOOP} prepares for an abnormal loop exit, e.g., via
3355: @code{EXIT}. @code{UNLOOP} removes the loop control parameters from the
3356: return stack so @code{EXIT} can get to its return address. For example:
3357:
3358: @example
3359: : demo 10 0 ?DO i DUP . 3 = IF UNLOOP EXIT THEN LOOP ." Done" ;
3360: @end example
3361: prints @code{0 1 2 3}
3362:
3363:
3364: @item
3365: If @i{start} is greater than @i{limit}, a @code{?DO} loop is entered
3366: (and @code{LOOP} iterates until they become equal by wrap-around
3367: arithmetic). This behaviour is usually not what you want. Therefore,
3368: Gforth offers @code{+DO} and @code{U+DO} (as replacements for
3369: @code{?DO}), which do not enter the loop if @i{start} is greater than
3370: @i{limit}; @code{+DO} is for signed loop parameters, @code{U+DO} for
3371: unsigned loop parameters.
3372:
3373: @item
3374: @code{?DO} can be replaced by @code{DO}. @code{DO} always enters
3375: the loop, independent of the loop parameters. Do not use @code{DO}, even
3376: if you know that the loop is entered in any case. Such knowledge tends
3377: to become invalid during maintenance of a program, and then the
3378: @code{DO} will make trouble.
3379:
3380: @item
3381: @code{LOOP} can be replaced with @code{@i{n} +LOOP}; this updates the
3382: index by @i{n} instead of by 1. The loop is terminated when the border
3383: between @i{limit-1} and @i{limit} is crossed. E.g.:
3384:
3385: @example
3386: 4 0 +DO i . 2 +LOOP
3387: @end example
3388: @noindent
3389: prints @code{0 2}
3390:
3391: @example
3392: 4 1 +DO i . 2 +LOOP
3393: @end example
3394: @noindent
3395: prints @code{1 3}
3396:
3397:
3398: @cindex negative increment for counted loops
3399: @cindex counted loops with negative increment
3400: The behaviour of @code{@i{n} +LOOP} is peculiar when @i{n} is negative:
3401:
3402: @example
3403: -1 0 ?DO i . -1 +LOOP
3404: @end example
3405: @noindent
3406: prints @code{0 -1}
3407:
3408: @example
3409: 0 0 ?DO i . -1 +LOOP
3410: @end example
3411: prints nothing.
3412:
3413: Therefore we recommend avoiding @code{@i{n} +LOOP} with negative
3414: @i{n}. One alternative is @code{@i{u} -LOOP}, which reduces the
3415: index by @i{u} each iteration. The loop is terminated when the border
3416: between @i{limit+1} and @i{limit} is crossed. Gforth also provides
3417: @code{-DO} and @code{U-DO} for down-counting loops. E.g.:
3418:
3419: @example
3420: -2 0 -DO i . 1 -LOOP
3421: @end example
3422: @noindent
3423: prints @code{0 -1}
3424:
3425: @example
3426: -1 0 -DO i . 1 -LOOP
3427: @end example
3428: @noindent
3429: prints @code{0}
3430:
3431: @example
3432: 0 0 -DO i . 1 -LOOP
3433: @end example
3434: @noindent
3435: prints nothing.
3436:
3437: @end itemize
3438:
3439: Unfortunately, @code{+DO}, @code{U+DO}, @code{-DO}, @code{U-DO} and
3440: @code{-LOOP} are not defined in ANS Forth. However, an implementation
3441: for these words that uses only standard words is provided in
3442: @file{compat/loops.fs}.
3443:
3444:
3445: @cindex @code{FOR} loops
3446: Another counted loop is:
3447: @example
3448: @i{n}
3449: FOR
3450: @i{body}
3451: NEXT
3452: @end example
3453: This is the preferred loop of native code compiler writers who are too
3454: lazy to optimize @code{?DO} loops properly. This loop structure is not
3455: defined in ANS Forth. In Gforth, this loop iterates @i{n+1} times;
3456: @code{i} produces values starting with @i{n} and ending with 0. Other
3457: Forth systems may behave differently, even if they support @code{FOR}
3458: loops. To avoid problems, don't use @code{FOR} loops.
3459:
3460: @node Arbitrary control structures, Calls and returns, Counted Loops, Control Structures
3461: @subsection Arbitrary control structures
3462: @cindex control structures, user-defined
3463:
3464: @cindex control-flow stack
3465: ANS Forth permits and supports using control structures in a non-nested
3466: way. Information about incomplete control structures is stored on the
3467: control-flow stack. This stack may be implemented on the Forth data
3468: stack, and this is what we have done in Gforth.
3469:
3470: @cindex @code{orig}, control-flow stack item
3471: @cindex @code{dest}, control-flow stack item
3472: An @i{orig} entry represents an unresolved forward branch, a @i{dest}
3473: entry represents a backward branch target. A few words are the basis for
3474: building any control structure possible (except control structures that
3475: need storage, like calls, coroutines, and backtracking).
3476:
3477:
3478: doc-if
3479: doc-ahead
3480: doc-then
3481: doc-begin
3482: doc-until
3483: doc-again
3484: doc-cs-pick
3485: doc-cs-roll
3486:
3487:
3488: The Standard words @code{CS-PICK} and @code{CS-ROLL} allow you to
3489: manipulate the control-flow stack in a portable way. Without them, you
3490: would need to know how many stack items are occupied by a control-flow
3491: entry (many systems use one cell. In Gforth they currently take three,
3492: but this may change in the future).
3493:
3494: Some standard control structure words are built from these words:
3495:
3496:
3497: doc-else
3498: doc-while
3499: doc-repeat
3500:
3501:
3502: @noindent
3503: Gforth adds some more control-structure words:
3504:
3505:
3506: doc-endif
3507: doc-?dup-if
3508: doc-?dup-0=-if
3509:
3510:
3511: @noindent
3512: Counted loop words constitute a separate group of words:
3513:
3514:
3515: doc-?do
3516: doc-+do
3517: doc-u+do
3518: doc--do
3519: doc-u-do
3520: doc-do
3521: doc-for
3522: doc-loop
3523: doc-+loop
3524: doc--loop
3525: doc-next
3526: doc-leave
3527: doc-?leave
3528: doc-unloop
3529: doc-done
3530:
3531:
3532: The standard does not allow using @code{CS-PICK} and @code{CS-ROLL} on
3533: @i{do-sys}. Gforth allows it, but it's your job to ensure that for
3534: every @code{?DO} etc. there is exactly one @code{UNLOOP} on any path
3535: through the definition (@code{LOOP} etc. compile an @code{UNLOOP} on the
3536: fall-through path). Also, you have to ensure that all @code{LEAVE}s are
3537: resolved (by using one of the loop-ending words or @code{DONE}).
3538:
3539: @noindent
3540: Another group of control structure words are:
3541:
3542:
3543: doc-case
3544: doc-endcase
3545: doc-of
3546: doc-endof
3547:
3548:
3549: @i{case-sys} and @i{of-sys} cannot be processed using @code{CS-PICK} and
3550: @code{CS-ROLL}.
3551:
3552: @subsubsection Programming Style
3553: @cindex control structures programming style
3554: @cindex programming style, arbitrary control structures
3555:
3556: In order to ensure readability we recommend that you do not create
3557: arbitrary control structures directly, but define new control structure
3558: words for the control structure you want and use these words in your
3559: program. For example, instead of writing:
3560:
3561: @example
3562: BEGIN
3563: ...
3564: IF [ 1 CS-ROLL ]
3565: ...
3566: AGAIN THEN
3567: @end example
3568:
3569: @noindent
3570: we recommend defining control structure words, e.g.,
3571:
3572: @example
3573: : WHILE ( DEST -- ORIG DEST )
3574: POSTPONE IF
3575: 1 CS-ROLL ; immediate
3576:
3577: : REPEAT ( orig dest -- )
3578: POSTPONE AGAIN
3579: POSTPONE THEN ; immediate
3580: @end example
3581:
3582: @noindent
3583: and then using these to create the control structure:
3584:
3585: @example
3586: BEGIN
3587: ...
3588: WHILE
3589: ...
3590: REPEAT
3591: @end example
3592:
3593: That's much easier to read, isn't it? Of course, @code{REPEAT} and
3594: @code{WHILE} are predefined, so in this example it would not be
3595: necessary to define them.
3596:
3597: @node Calls and returns, Exception Handling, Arbitrary control structures, Control Structures
3598: @subsection Calls and returns
3599: @cindex calling a definition
3600: @cindex returning from a definition
3601:
3602: @cindex recursive definitions
3603: A definition can be called simply be writing the name of the definition
3604: to be called. Normally a definition is invisible during its own
3605: definition. If you want to write a directly recursive definition, you
3606: can use @code{recursive} to make the current definition visible, or
3607: @code{recurse} to call the current definition directly.
3608:
3609:
3610: doc-recursive
3611: doc-recurse
3612:
3613:
3614: @comment TODO add example of the two recursion methods
3615: @quotation
3616: @progstyle
3617: I prefer using @code{recursive} to @code{recurse}, because calling the
3618: definition by name is more descriptive (if the name is well-chosen) than
3619: the somewhat cryptic @code{recurse}. E.g., in a quicksort
3620: implementation, it is much better to read (and think) ``now sort the
3621: partitions'' than to read ``now do a recursive call''.
3622: @end quotation
3623:
3624: For mutual recursion, use @code{Defer}red words, like this:
3625:
3626: @example
3627: Defer foo
3628:
3629: : bar ( ... -- ... )
3630: ... foo ... ;
3631:
3632: :noname ( ... -- ... )
3633: ... bar ... ;
3634: IS foo
3635: @end example
3636:
3637: Deferred words are discussed in more detail in @ref{Deferred words}.
3638:
3639: The current definition returns control to the calling definition when
3640: the end of the definition is reached or @code{EXIT} is encountered.
3641:
3642: doc-exit
3643: doc-;s
3644:
3645:
3646: @node Exception Handling, , Calls and returns, Control Structures
3647: @subsection Exception Handling
3648: @cindex exceptions
3649:
3650: If your program detects a fatal error condition, the simplest action
3651: that it can take is to @code{quit}. This resets the return stack and
3652: restarts the text interpreter, but does not print any error message.
3653:
3654: The next stage in severity is to execute @code{abort}, which has the
3655: same effect as @code{quit}, with the addition that it resets the data
3656: stack.
3657:
3658: A slightly more sophisticated approach is use use @code{abort"}, which
3659: compiles a string to be used as an error message and does a conditional
3660: @code{abort} at run-time. For example:
3661:
3662: @example
3663: @kbd{: checker abort" That flag was true" ." A false flag" ;@key{RET}} ok
3664: @kbd{0 checker@key{RET}} A false flag ok
3665: @kbd{1 checker@key{RET}}
3666: :1: That flag was true
3667: 1 checker
3668: ^^^^^^^
3669: $400D1648 throw
3670: $400E4660
3671: @end example
3672:
3673: These simple techniques allow a program to react to a fatal error
3674: condition, but they are not exactly user-friendly. The ANS Forth
3675: Exception word set provides the pair of words @code{throw} and
3676: @code{catch}, which can be used to provide sophisticated error-handling.
3677:
3678: @code{catch} has a similar behaviour to @code{execute}, in that it takes
3679: an @i{xt} as a parameter and starts execution of the xt. However,
3680: before passing control to the xt, @code{catch} pushes an
3681: @dfn{exception frame} onto the @dfn{exception stack}. This exception
3682: frame is used to restore the system to a known state if a detected error
3683: occurs during the execution of the xt. A typical way to use @code{catch}
3684: would be:
3685:
3686: @example
3687: ... ['] foo catch IF ...
3688: @end example
3689:
3690: @c TOS is undefined. - anton
3691:
3692: @c nac-> TODO -- I need to look at this example again.
3693:
3694: Whilst @code{foo} executes, it can call other words to any level of
3695: nesting, as usual. If @code{foo} (and all the words that it calls)
3696: execute successfully, control will ultimately pass to the word following
3697: the @code{catch}, and there will be a 0 at TOS. However, if any word
3698: detects an error, it can terminate the execution of @code{foo} by
3699: pushing a non-zero error code onto the stack and then performing a
3700: @code{throw}. The execution of @code{throw} will pass control to the
3701: word following the @code{catch}, but this time the TOS will hold the
3702: error code. Therefore, the @code{IF} in the example can be used to
3703: determine whether @code{foo} executed successfully.
3704:
3705: This simple example shows how you can use @code{throw} and @code{catch}
3706: to ``take over'' exception handling from the system:
3707: @example
3708: : my-div ['] / catch if ." DIVIDE ERROR" else ." OK.. " . then ;
3709: @end example
3710:
3711: The next example is more sophisticated and shows a multi-level
3712: @code{throw} and @code{catch}. To understand this example, start at the
3713: definition of @code{top-level} and work backwards:
3714:
3715: @example
3716: : lowest-level ( -- c )
3717: key dup 27 = if
3718: 1 throw \ ESCAPE key pressed
3719: else
3720: ." lowest-level successful" CR
3721: then
3722: ;
3723:
3724: : lower-level ( -- c )
3725: lowest-level
3726: \ at this level consider a CTRL-U to be a fatal error
3727: dup 21 = if \ CTRL-U
3728: 2 throw
3729: else
3730: ." lower-level successful" CR
3731: then
3732: ;
3733:
3734: : low-level ( -- c )
3735: ['] lower-level catch
3736: ?dup if
3737: \ error occurred - do we recognise it?
3738: dup 1 = if
3739: \ ESCAPE key pressed.. pretend it was an E
3740: [char] E
3741: else throw \ propogate the error upwards
3742: then
3743: then
3744: ." low-level successfull" CR
3745: ;
3746:
3747: : top-level ( -- )
3748: CR ['] low-level catch \ CATCH is used like EXECUTE
3749: ?dup if \ error occurred..
3750: ." Error " . ." occurred - contact your supplier"
3751: else
3752: ." The '" emit ." ' key was pressed" CR
3753: then
3754: ;
3755: @end example
3756:
3757: The ANS Forth document assigns @code{throw} codes thus:
3758:
3759: @itemize @bullet
3760: @item
3761: codes in the range -1 -- -255 are reserved to be assigned by the
3762: Standard. Assignments for codes in the range -1 -- -58 are currently
3763: documented in the Standard. In particular, @code{-1 throw} is equivalent
3764: to @code{abort} and @code{-2 throw} is equivalent to @code{abort"}.
3765: @item
3766: codes in the range -256 -- -4095 are reserved to be assigned by the system.
3767: @item
3768: all other codes may be assigned by programs.
3769: @end itemize
3770:
3771: Gforth provides the word @code{exception} as a mechanism for assigning
3772: system throw codes to applications. This allows multiple applications to
3773: co-exist in memory without any clash of @code{throw} codes. A definition
3774: of @code{exception} in ANS Forth is provided in
3775: @file{compat/exception.fs}.
3776:
3777:
3778: doc-quit
3779: doc-abort
3780: doc-abort"
3781:
3782: doc-catch
3783: doc-throw
3784: doc---exception-exception
3785:
3786:
3787:
3788: @c -------------------------------------------------------------
3789: @node Defining Words, Interpretation and Compilation Semantics, Control Structures, Words
3790: @section Defining Words
3791: @cindex defining words
3792:
3793: Defining words are used to extend Forth by creating new entries in the dictionary.
3794:
3795: @menu
3796: * CREATE::
3797: * Variables:: Variables and user variables
3798: * Constants::
3799: * Values:: Initialised variables
3800: * Colon Definitions::
3801: * Anonymous Definitions:: Definitions without names
3802: * User-defined Defining Words::
3803: * Deferred words:: Allow forward references
3804: * Aliases::
3805: * Supplying names::
3806: @end menu
3807:
3808: @node CREATE, Variables, Defining Words, Defining Words
3809: @subsection @code{CREATE}
3810: @cindex simple defining words
3811: @cindex defining words, simple
3812:
3813: Defining words are used to create new entries in the dictionary. The
3814: simplest defining word is @code{CREATE}. @code{CREATE} is used like
3815: this:
3816:
3817: @example
3818: CREATE new-word1
3819: @end example
3820:
3821: @code{CREATE} is a parsing word that generates a dictionary entry for
3822: @code{new-word1}. When @code{new-word1} is executed, all that it does is
3823: leave an address on the stack. The address represents the value of
3824: the data space pointer (@code{HERE}) at the time that @code{new-word1}
3825: was defined. Therefore, @code{CREATE} is a way of associating a name
3826: with the address of a region of memory.
3827:
3828: doc-create
3829:
3830: By extending this example to reserve some memory in data space, we end
3831: up with a @i{variable}. Here are two different ways to do it:
3832:
3833: @example
3834: CREATE new-word2 1 cells allot \ reserve 1 cell - initial value undefined
3835: CREATE new-word3 4 , \ reserve 1 cell and initialise it (to 4)
3836: @end example
3837:
3838: The variable can be examined and modified using @code{@@} (``fetch'') and
3839: @code{!} (``store'') like this:
3840:
3841: @example
3842: new-word2 @@ . \ get address, fetch from it and display
3843: 1234 new-word2 ! \ new value, get address, store to it
3844: @end example
3845:
3846: @cindex arrays
3847: A similar mechanism can be used to create arrays. For example, an
3848: 80-character text input buffer:
3849:
3850: @example
3851: CREATE text-buf 80 chars allot
3852:
3853: text-buf 0 chars c@@ \ the 1st character (offset 0)
3854: text-buf 3 chars c@@ \ the 4th character (offset 3)
3855: @end example
3856:
3857: You can build arbitrarily complex data structures by allocating
3858: appropriate areas of memory. @xref{Structures} for further discussions
3859: of this, and to learn about some Gforth tools that make it easier.
3860:
3861:
3862: @node Variables, Constants, CREATE, Defining Words
3863: @subsection Variables
3864: @cindex variables
3865:
3866: The previous section showed how a sequence of commands could be used to
3867: generate a variable. As a final refinement, the whole code sequence can
3868: be wrapped up in a defining word (pre-empting the subject of the next
3869: section), making it easier to create new variables:
3870:
3871: @example
3872: : myvariableX ( "name" -- a-addr ) CREATE 1 cells allot ;
3873: : myvariable0 ( "name" -- a-addr ) CREATE 0 , ;
3874:
3875: myvariableX foo \ variable foo starts off with an unknown value
3876: myvariable0 joe \ whilst joe is initialised to 0
3877:
3878: 45 3 * foo ! \ set foo to 135
3879: 1234 joe ! \ set joe to 1234
3880: 3 joe +! \ increment joe by 3.. to 1237
3881: @end example
3882:
3883: Not surprisingly, there is no need to define @code{myvariable}, since
3884: Forth already has a definition @code{Variable}. ANS Forth does not
3885: require a @code{Variable} to be initialised when it is created (i.e., it
3886: behaves like @code{myvariableX}). In contrast, Gforth's @code{Variable}
3887: initialises the variable to 0 (i.e., it behaves exactly like
3888: @code{myvariable0}). Forth also provides @code{2Variable} and
3889: @code{fvariable} for double and floating-point variables, respectively
3890: -- both are initialised to 0 in Gforth. If you use a @code{Variable} to
3891: store a boolean, you can use @code{on} and @code{off} to toggle its
3892: state.
3893:
3894: doc-variable
3895: doc-2variable
3896: doc-fvariable
3897:
3898: @cindex user variables
3899: @cindex user space
3900: The defining word @code{User} behaves in the same way as @code{Variable}.
3901: The difference is that it reserves space in @i{user (data) space} rather
3902: than normal data space. In a Forth system that has a multi-tasker, each
3903: task has its own set of user variables.
3904:
3905: doc-user
3906:
3907: @comment TODO is that stuff about user variables strictly correct? Is it
3908: @comment just terminal tasks that have user variables?
3909: @comment should document tasker.fs (with some examples) elsewhere
3910: @comment in this manual, then expand on user space and user variables.
3911:
3912:
3913: @node Constants, Values, Variables, Defining Words
3914: @subsection Constants
3915: @cindex constants
3916:
3917: @code{Constant} allows you to declare a fixed value and refer to it by
3918: name. For example:
3919:
3920: @example
3921: 12 Constant INCHES-PER-FOOT
3922: 3E+08 fconstant SPEED-O-LIGHT
3923: @end example
3924:
3925: A @code{Variable} can be both read and written, so its run-time
3926: behaviour is to supply an address through which its current value can be
3927: manipulated. In contrast, the value of a @code{Constant} cannot be
3928: changed once it has been declared@footnote{Well, often it can be -- but
3929: not in a Standard, portable way. It's safer to use a @code{Value} (read
3930: on).} so it's not necessary to supply the address -- it is more
3931: efficient to return the value of the constant directly. That's exactly
3932: what happens; the run-time effect of a constant is to put its value on
3933: the top of the stack (@ref{User-defined Defining Words} describes one
3934: way of implementing @code{Constant}).
3935:
3936: Gforth also provides @code{2Constant} and @code{fconstant} for defining
3937: double and floating-point constants, respectively.
3938:
3939: doc-constant
3940: doc-2constant
3941: doc-fconstant
3942:
3943: @c that's too deep, and it's not necessarily true for all ANS Forths. - anton
3944: @c nac-> How could that not be true in an ANS Forth? You can't define a
3945: @c constant, use it and then delete the definition of the constant..
3946: @c I agree that it's rather deep, but IMO it is an important difference
3947: @c relative to other programming languages.. often it's annoying: it
3948: @c certainly changes my programming style relative to C.
3949:
3950: Constants in Forth behave differently from their equivalents in other
3951: programming languages. In other languages, a constant (such as an EQU in
3952: assembler or a #define in C) only exists at compile-time; in the
3953: executable program the constant has been translated into an absolute
3954: number and, unless you are using a symbolic debugger, it's impossible to
3955: know what abstract thing that number represents. In Forth a constant has
3956: an entry in the header space and remains there after the code that uses
3957: it has been defined. In fact, it must remain in the dictionary since it
3958: has run-time duties to perform. For example:
3959:
3960: @example
3961: 12 Constant INCHES-PER-FOOT
3962: : FEET-TO-INCHES ( n1 -- n2 ) INCHES-PER-FOOT * ;
3963: @end example
3964:
3965: @cindex in-lining of constants
3966: When @code{FEET-TO-INCHES} is executed, it will in turn execute the xt
3967: associated with the constant @code{INCHES-PER-FOOT}. If you use
3968: @code{see} to decompile the definition of @code{FEET-TO-INCHES}, you can
3969: see that it makes a call to @code{INCHES-PER-FOOT}. Some Forth compilers
3970: attempt to optimise constants by in-lining them where they are used. You
3971: can force Gforth to in-line a constant like this:
3972:
3973: @example
3974: : FEET-TO-INCHES ( n1 -- n2 ) [ INCHES-PER-FOOT ] LITERAL * ;
3975: @end example
3976:
3977: If you use @code{see} to decompile @i{this} version of
3978: @code{FEET-TO-INCHES}, you can see that @code{INCHES-PER-FOOT} is no
3979: longer present. @xref{Interpret/Compile states} and @ref{Literals} on
3980: how this works.
3981:
3982: In-lining constants in this way might improve execution time
3983: fractionally, and can ensure that a constant is now only referenced at
3984: compile-time. However, the definition of the constant still remains in
3985: the dictionary. Some Forth compilers provide a mechanism for controlling
3986: a second dictionary for holding transient words such that this second
3987: dictionary can be deleted later in order to recover memory
3988: space. However, there is no standard way of doing this.
3989:
3990:
3991: @node Values, Colon Definitions, Constants, Defining Words
3992: @subsection Values
3993: @cindex values
3994:
3995: A @code{Value} is like a @code{Variable} but with two important
3996: differences:
3997:
3998: @itemize @bullet
3999: @item
4000: A @code{Value} is initialised when it is declared; like a
4001: @code{Constant} but unlike a @code{Variable}.
4002: @item
4003: A @code{Value} returns its value rather than its address when it is
4004: executed; i.e., it has the same run-time behaviour as @code{Constant}.
4005: @end itemize
4006:
4007: A @code{Value} needs an additional word, @code{TO} to allow its value to
4008: be changed. Here are some examples:
4009:
4010: @example
4011: 12 Value APPLES \ Define APPLES with an initial value of 12
4012: 34 TO APPLES \ Change the value of APPLES. TO is a parsing word
4013: APPLES \ puts 34 on the top of the stack.
4014: @end example
4015:
4016: doc-value
4017: doc-to
4018:
4019:
4020: @node Colon Definitions, Anonymous Definitions, Values, Defining Words
4021: @subsection Colon Definitions
4022: @cindex colon definitions
4023:
4024: @example
4025: : name ( ... -- ... )
4026: word1 word2 word3 ;
4027: @end example
4028:
4029: @noindent
4030: Creates a word called @code{name} that, upon execution, executes
4031: @code{word1 word2 word3}. @code{name} is a @dfn{(colon) definition}.
4032:
4033: The explanation above is somewhat superficial. @xref{Your first
4034: definition} for simple examples of colon definitions, then
4035: @xref{Interpretation and Compilation Semantics} for an in-depth
4036: discussion of some of the issues involved.
4037:
4038: doc-:
4039: doc-;
4040:
4041:
4042: @node Anonymous Definitions, User-defined Defining Words, Colon Definitions, Defining Words
4043: @subsection Anonymous Definitions
4044: @cindex colon definitions
4045: @cindex defining words without name
4046:
4047: Sometimes you want to define an @dfn{anonymous word}; a word without a
4048: name. You can do this with:
4049:
4050: doc-:noname
4051:
4052: This leaves the execution token for the word on the stack after the
4053: closing @code{;}. Here's an example in which a deferred word is
4054: initialised with an @code{xt} from an anonymous colon definition:
4055:
4056: @example
4057: Defer deferred
4058: :noname ( ... -- ... )
4059: ... ;
4060: IS deferred
4061: @end example
4062:
4063: @noindent
4064: Gforth provides an alternative way of doing this, using two separate
4065: words:
4066:
4067: doc-noname
4068: @cindex execution token of last defined word
4069: doc-lastxt
4070:
4071: @noindent
4072: The previous example can be rewritten using @code{noname} and
4073: @code{lastxt}:
4074:
4075: @example
4076: Defer deferred
4077: noname : ( ... -- ... )
4078: ... ;
4079: lastxt IS deferred
4080: @end example
4081:
4082: @noindent
4083: @code{noname} works with any defining word, not just @code{:}.
4084:
4085: @code{lastxt} also works when the last word was not defined as
4086: @code{noname}. It also has the useful property that is is valid as soon
4087: as the header for a definition has been built. Thus:
4088:
4089: @example
4090: lastxt . : foo [ lastxt . ] ; ' foo .
4091: @end example
4092:
4093: @noindent
4094: prints 3 numbers; the last two are the same.
4095:
4096:
4097: @node User-defined Defining Words, Deferred words, Anonymous Definitions, Defining Words
4098: @subsection User-defined Defining Words
4099: @cindex user-defined defining words
4100: @cindex defining words, user-defined
4101:
4102: You can create a new defining word by wrapping defining-time code around
4103: an existing defining word and putting the sequence in a colon
4104: definition. For example, suppose that you have a word @code{stats} that
4105: gathers statistics about colon definitions given the @i{xt} of the
4106: definition, and you want every colon definition in your application to
4107: make a call to @code{stats}. You can define and use a new version of
4108: @code{:} like this:
4109:
4110: @example
4111: : stats ( xt -- ) DUP ." (Gathering statistics for " . ." )"
4112: ... ; \ other code
4113:
4114: : my: : lastxt postpone literal ['] stats compile, ;
4115:
4116: my: foo + - ;
4117: @end example
4118:
4119: When @code{foo} is defined using @code{my:} these steps occur:
4120:
4121: @itemize @bullet
4122: @item
4123: @code{my:} is executed.
4124: @item
4125: The @code{:} within the definition (the one between @code{my:} and
4126: @code{lastxt}) is executed, and does just what it always does; it parses
4127: the input stream for a name, builds a dictionary header for the name
4128: @code{foo} and switches @code{state} from interpret to compile.
4129: @item
4130: The word @code{lastxt} is executed. It puts the @i{xt} for the word that is
4131: being defined -- @code{foo} -- onto the stack.
4132: @item
4133: The code that was produced by @code{postpone literal} is executed; this
4134: causes the value on the stack to be compiled as a literal in the code
4135: area of @code{foo}.
4136: @item
4137: The code @code{['] stats} compiles a literal into the definition of
4138: @code{my:}. When @code{compile,} is executed, that literal -- the
4139: execution token for @code{stats} -- is layed down in the code area of
4140: @code{foo} , following the literal@footnote{Strictly speaking, the
4141: mechanism that @code{compile,} uses to convert an @i{xt} into something
4142: in the code area is implementation-dependent. A threaded implementation
4143: might spit out the execution token directly whilst another
4144: implementation might spit out a native code sequence.}.
4145: @item
4146: At this point, the execution of @code{my:} is complete, and control
4147: returns to the text interpreter. The text interpreter is in compile
4148: state, so subsequent text @code{+ -} is compiled into the definition of
4149: @code{foo} and the @code{;} terminates the definition as always.
4150: @end itemize
4151:
4152: You can use @code{see} to decompile a word that was defined using
4153: @code{my:} and see how it is different from a normal @code{:}
4154: definition. For example:
4155:
4156: @example
4157: : bar + - ; \ like foo but using : rather than my:
4158: see bar
4159: : bar
4160: + - ;
4161: see foo
4162: : foo
4163: 107645672 stats + - ;
4164:
4165: \ use ' stats . to show that 107645672 is the xt for stats
4166: @end example
4167:
4168: You can use techniques like this to make new defining words in terms of
4169: @i{any} existing defining word.
4170:
4171:
4172: @cindex defining defining words
4173: @cindex @code{CREATE} ... @code{DOES>}
4174: If you want the words defined with your defining words to behave
4175: differently from words defined with standard defining words, you can
4176: write your defining word like this:
4177:
4178: @example
4179: : def-word ( "name" -- )
4180: CREATE @i{code1}
4181: DOES> ( ... -- ... )
4182: @i{code2} ;
4183:
4184: def-word name
4185: @end example
4186:
4187: @cindex child words
4188: This fragment defines a @dfn{defining word} @code{def-word} and then
4189: executes it. When @code{def-word} executes, it @code{CREATE}s a new
4190: word, @code{name}, and executes the code @i{code1}. The code @i{code2}
4191: is not executed at this time. The word @code{name} is sometimes called a
4192: @dfn{child} of @code{def-word}.
4193:
4194: When you execute @code{name}, the address of the body of @code{name} is
4195: put on the data stack and @i{code2} is executed (the address of the body
4196: of @code{name} is the address @code{HERE} returns immediately after the
4197: @code{CREATE}).
4198:
4199: @cindex atavism in child words
4200: You can use @code{def-word} to define a set of child words that behave
4201: differently, though atavistically; they all have a common run-time
4202: behaviour determined by @i{code2}. Typically, the @i{code1} sequence
4203: builds a data area in the body of the child word. The structure of the
4204: data is common to all children of @code{def-word}, but the data values
4205: are specific -- and private -- to each child word. When a child word is
4206: executed, the address of its private data area is passed as a parameter
4207: on TOS to be used and manipulated@footnote{It is legitimate both to read
4208: and write to this data area.} by @i{code2}.
4209:
4210: The two fragments of code that make up the defining words act (are
4211: executed) at two completely separate times:
4212:
4213: @itemize @bullet
4214: @item
4215: At @i{define time}, the defining word executes @i{code1} to generate a
4216: child word
4217: @item
4218: At @i{child execution time}, when a child word is invoked, @i{code2}
4219: is executed, using parameters (data) that are private and specific to
4220: the child word.
4221: @end itemize
4222:
4223: Another way of understanding the behaviour of @code{def-word} and
4224: @code{name} is to say that, if you make the following definitions:
4225: @example
4226: : def-word1 ( "name" -- )
4227: CREATE @i{code1} ;
4228:
4229: : action1 ( ... -- ... )
4230: @i{code2} ;
4231:
4232: def-word1 name1
4233: @end example
4234:
4235: @noindent
4236: Then using @code{name1 action1} is equivalent to using @code{name}.
4237:
4238: The classic example is that you can define @code{CONSTANT} in this way:
4239:
4240: @example
4241: : CONSTANT ( w "name" -- )
4242: CREATE ,
4243: DOES> ( -- w )
4244: @@ ;
4245: @end example
4246:
4247: @comment There is a beautiful description of how this works and what
4248: @comment it does in the Forthwrite 100th edition.. as well as an elegant
4249: @comment commentary on the Counting Fruits problem.
4250:
4251: When you create a constant with @code{5 CONSTANT five}, a set of
4252: define-time actions take place; first a new word @code{five} is created,
4253: then the value 5 is laid down in the body of @code{five} with
4254: @code{,}. When @code{five} is executed, the address of the body is put on
4255: the stack, and @code{@@} retrieves the value 5. The word @code{five} has
4256: no code of its own; it simply contains a data field and a pointer to the
4257: code that follows @code{DOES>} in its defining word. That makes words
4258: created in this way very compact.
4259:
4260: The final example in this section is intended to remind you that space
4261: reserved in @code{CREATE}d words is @i{data} space and therefore can be
4262: both read and written by a Standard program@footnote{Exercise: use this
4263: example as a starting point for your own implementation of @code{Value}
4264: and @code{TO} -- if you get stuck, investigate the behaviour of @code{'} and
4265: @code{[']}.}:
4266:
4267: @example
4268: : foo ( "name" -- )
4269: CREATE -1 ,
4270: DOES> ( -- )
4271: @@ . ;
4272:
4273: foo first-word
4274: foo second-word
4275:
4276: 123 ' first-word >BODY !
4277: @end example
4278:
4279: If @code{first-word} had been a @code{CREATE}d word, we could simply
4280: have executed it to get the address of its data field. However, since it
4281: was defined to have @code{DOES>} actions, its execution semantics are to
4282: perform those @code{DOES>} actions. To get the address of its data field
4283: it's necessary to use @code{'} to get its xt, then @code{>BODY} to
4284: translate the xt into the address of the data field. When you execute
4285: @code{first-word}, it will display @code{123}. When you execute
4286: @code{second-word} it will display @code{-1}.
4287:
4288: @cindex stack effect of @code{DOES>}-parts
4289: @cindex @code{DOES>}-parts, stack effect
4290: In the examples above the stack comment after the @code{DOES>} specifies
4291: the stack effect of the defined words, not the stack effect of the
4292: following code (the following code expects the address of the body on
4293: the top of stack, which is not reflected in the stack comment). This is
4294: the convention that I use and recommend (it clashes a bit with using
4295: locals declarations for stack effect specification, though).
4296:
4297: @subsubsection Applications of @code{CREATE..DOES>}
4298: @cindex @code{CREATE} ... @code{DOES>}, applications
4299:
4300: You may wonder how to use this feature. Here are some usage patterns:
4301:
4302: @cindex factoring similar colon definitions
4303: When you see a sequence of code occurring several times, and you can
4304: identify a meaning, you will factor it out as a colon definition. When
4305: you see similar colon definitions, you can factor them using
4306: @code{CREATE..DOES>}. E.g., an assembler usually defines several words
4307: that look very similar:
4308: @example
4309: : ori, ( reg-target reg-source n -- )
4310: 0 asm-reg-reg-imm ;
4311: : andi, ( reg-target reg-source n -- )
4312: 1 asm-reg-reg-imm ;
4313: @end example
4314:
4315: @noindent
4316: This could be factored with:
4317: @example
4318: : reg-reg-imm ( op-code -- )
4319: CREATE ,
4320: DOES> ( reg-target reg-source n -- )
4321: @@ asm-reg-reg-imm ;
4322:
4323: 0 reg-reg-imm ori,
4324: 1 reg-reg-imm andi,
4325: @end example
4326:
4327: @cindex currying
4328: Another view of @code{CREATE..DOES>} is to consider it as a crude way to
4329: supply a part of the parameters for a word (known as @dfn{currying} in
4330: the functional language community). E.g., @code{+} needs two
4331: parameters. Creating versions of @code{+} with one parameter fixed can
4332: be done like this:
4333: @example
4334: : curry+ ( n1 -- )
4335: CREATE ,
4336: DOES> ( n2 -- n1+n2 )
4337: @@ + ;
4338:
4339: 3 curry+ 3+
4340: -2 curry+ 2-
4341: @end example
4342:
4343: @subsubsection The gory details of @code{CREATE..DOES>}
4344: @cindex @code{CREATE} ... @code{DOES>}, details
4345:
4346: doc-does>
4347:
4348: @cindex @code{DOES>} in a separate definition
4349: This means that you need not use @code{CREATE} and @code{DOES>} in the
4350: same definition; you can put the @code{DOES>}-part in a separate
4351: definition. This allows us to, e.g., select among different @code{DOES>}-parts:
4352: @example
4353: : does1
4354: DOES> ( ... -- ... )
4355: ... ;
4356:
4357: : does2
4358: DOES> ( ... -- ... )
4359: ... ;
4360:
4361: : def-word ( ... -- ... )
4362: create ...
4363: IF
4364: does1
4365: ELSE
4366: does2
4367: ENDIF ;
4368: @end example
4369:
4370: In this example, the selection of whether to use @code{does1} or
4371: @code{does2} is made at compile-time; at the time that the child word is
4372: @code{CREATE}d.
4373:
4374: @cindex @code{DOES>} in interpretation state
4375: In a standard program you can apply a @code{DOES>}-part only if the last
4376: word was defined with @code{CREATE}. In Gforth, the @code{DOES>}-part
4377: will override the behaviour of the last word defined in any case. In a
4378: standard program, you can use @code{DOES>} only in a colon
4379: definition. In Gforth, you can also use it in interpretation state, in a
4380: kind of one-shot mode; for example:
4381: @example
4382: CREATE name ( ... -- ... )
4383: @i{initialization}
4384: DOES>
4385: @i{code} ;
4386: @end example
4387:
4388: @noindent
4389: is equivalent to the standard:
4390: @example
4391: :noname
4392: DOES>
4393: @i{code} ;
4394: CREATE name EXECUTE ( ... -- ... )
4395: @i{initialization}
4396: @end example
4397:
4398:
4399: doc->body
4400:
4401:
4402: @node Deferred words, Aliases, User-defined Defining Words, Defining Words
4403: @subsection Deferred words
4404: @cindex deferred words
4405:
4406: The defining word @code{Defer} allows you to define a word by name
4407: without defining its behaviour; the definition of its behaviour is
4408: deferred. Here are two situation where this can be useful:
4409:
4410: @itemize @bullet
4411: @item
4412: Where you want to allow the behaviour of a word to be altered later, and
4413: for all precompiled references to the word to change when its behaviour
4414: is changed.
4415: @item
4416: For mutual recursion; @xref{Calls and returns}.
4417: @end itemize
4418:
4419: In the following example, @code{foo} always invokes the version of
4420: @code{greet} that prints ``@code{Good morning}'' whilst @code{bar}
4421: always invokes the version that prints ``@code{Hello}''. There is no way
4422: of getting @code{foo} to use the later version without re-ordering the
4423: source code and recompiling it.
4424:
4425: @example
4426: : greet ." Good morning" ;
4427: : foo ... greet ... ;
4428: : greet ." Hello" ;
4429: : bar ... greet ... ;
4430: @end example
4431:
4432: This problem can be solved by defining @code{greet} as a @code{Defer}red
4433: word. The behaviour of a @code{Defer}red word can be defined and
4434: redefined at any time by using @code{IS} to associate the xt of a
4435: previously-defined word with it. The previous example becomes:
4436:
4437: @example
4438: Defer greet
4439: : foo ... greet ... ;
4440: : bar ... greet ... ;
4441: : greet1 ." Good morning" ;
4442: : greet2 ." Hello" ;
4443: ' greet2 <IS> greet \ make greet behave like greet2
4444: @end example
4445:
4446: A deferred word can be used to improve the statistics-gathering example
4447: from @ref{User-defined Defining Words}; rather than edit the
4448: application's source code to change every @code{:} to a @code{my:}, do
4449: this:
4450:
4451: @example
4452: : real: : ; \ retain access to the original
4453: defer : \ redefine as a deferred word
4454: ' my: IS : \ use special version of :
4455: \
4456: \ load application here
4457: \
4458: ' real: IS : \ go back to the original
4459: @end example
4460:
4461:
4462: One thing to note is that @code{<IS>} consumes its name when it is
4463: executed. If you want to specify the name at compile time, use
4464: @code{[IS]}:
4465:
4466: @example
4467: : set-greet ( xt -- )
4468: [IS] greet ;
4469:
4470: ' greet1 set-greet
4471: @end example
4472:
4473: A deferred word can only inherit default semantics from the xt (because
4474: that is all that an xt can represent -- @pxref{Tokens for Words} for
4475: more discussion of this). However, the semantics of the deferred word
4476: itself can be modified at the time that it is defined. For example:
4477:
4478: @example
4479: : bar .... ; compile-only
4480: Defer fred immediate
4481: Defer jim
4482:
4483: ' bar <IS> jim \ jim has default semantics
4484: ' bar <IS> fred \ fred is immediate
4485: @end example
4486:
4487: doc-defer
4488: doc-<is>
4489: doc-[is]
4490: doc-is
4491: @comment TODO document these: what's defers [is]
4492: doc-what's
4493: doc-defers
4494:
4495: @c Use @code{words-deferred} to see a list of deferred words.
4496:
4497: Definitions in ANS Forth for @code{defer}, @code{<is>} and @code{[is]}
4498: are provided in @file{compat/defer.fs}.
4499:
4500:
4501: @node Aliases, Supplying names, Deferred words, Defining Words
4502: @subsection Aliases
4503: @cindex aliases
4504:
4505: The defining word @code{Alias} allows you to define a word by name that
4506: has the same behaviour as some other word. Here are two situation where
4507: this can be useful:
4508:
4509: @itemize @bullet
4510: @item
4511: When you want access to a word's definition from a different word list
4512: (for an example of this, see the definition of the @code{Root} word list
4513: in the Gforth source).
4514: @item
4515: When you want to create a synonym; a definition that can be known by
4516: either of two names (for example, @code{THEN} and @code{ENDIF} are
4517: aliases).
4518: @end itemize
4519:
4520: The word whose behaviour the alias is to inherit is represented by an
4521: xt. Therefore, the alias only inherits default semantics from its
4522: ancestor. The semantics of the alias itself can be modified at the time
4523: that it is defined. For example:
4524:
4525: @example
4526: : foo ... ; immediate
4527:
4528: ' foo Alias bar \ bar is not an immediate word
4529: ' foo Alias fooby immediate \ fooby is an immediate word
4530: @end example
4531:
4532: Words that are aliases have the same xt, different headers in the
4533: dictionary, and consequently different name tokens (@pxref{Tokens for
4534: Words}) and possibly different immediate flags. An alias can only have
4535: default or immediate compilation semantics; you can define aliases for
4536: combined words with @code{interpret/compile:} -- see @ref{Combined words}.
4537:
4538: doc-alias
4539:
4540:
4541: @node Supplying names, , Aliases, Defining Words
4542: @subsection Supplying the name of a defined word
4543: @cindex names for defined words
4544: @cindex defining words, name given in a string
4545:
4546: By default, a defining word takes the name for the defined word from the
4547: input stream. Sometimes you want to supply the name from a string. You
4548: can do this with:
4549:
4550: doc-nextname
4551:
4552: For example:
4553:
4554: @example
4555: s" foo" nextname create
4556: @end example
4557:
4558: @noindent
4559: is equivalent to:
4560:
4561: @example
4562: create foo
4563: @end example
4564:
4565: @noindent
4566: @code{nextname} works with any defining word, not just @code{:}.
4567:
4568:
4569: @node Interpretation and Compilation Semantics, Tokens for Words, Defining Words, Words
4570: @section Interpretation and Compilation Semantics
4571: @cindex semantics, interpretation and compilation
4572:
4573: @cindex interpretation semantics
4574: The @dfn{interpretation semantics} of a word are what the text
4575: interpreter does when it encounters the word in interpret state. It also
4576: appears in some other contexts, e.g., the execution token returned by
4577: @code{' @i{word}} identifies the interpretation semantics of
4578: @i{word} (in other words, @code{' @i{word} execute} is equivalent to
4579: interpret-state text interpretation of @code{@i{word}}).
4580:
4581: @cindex compilation semantics
4582: The @dfn{compilation semantics} of a word are what the text interpreter
4583: does when it encounters the word in compile state. It also appears in
4584: other contexts, e.g, @code{POSTPONE @i{word}} compiles@footnote{In
4585: standard terminology, ``appends to the current definition''.} the
4586: compilation semantics of @i{word}.
4587:
4588: @cindex execution semantics
4589: The standard also talks about @dfn{execution semantics}. They are used
4590: only for defining the interpretation and compilation semantics of many
4591: words. By default, the interpretation semantics of a word are to
4592: @code{execute} its execution semantics, and the compilation semantics of
4593: a word are to @code{compile,} its execution semantics.@footnote{In
4594: standard terminology: The default interpretation semantics are its
4595: execution semantics; the default compilation semantics are to append its
4596: execution semantics to the execution semantics of the current
4597: definition.}
4598:
4599: @comment TODO expand, make it co-operate with new sections on text interpreter.
4600:
4601: @cindex immediate words
4602: @cindex compile-only words
4603: You can change the semantics of the most-recently defined word:
4604:
4605:
4606: doc-immediate
4607: doc-compile-only
4608: doc-restrict
4609:
4610:
4611: Note that ticking (@code{'}) a compile-only word gives an error
4612: (``Interpreting a compile-only word'').
4613:
4614: @menu
4615: * Combined words::
4616: @end menu
4617:
4618: @node Combined words, ,Interpretation and Compilation Semantics, Interpretation and Compilation Semantics
4619: @subsection Combined Words
4620: @cindex combined words
4621:
4622: Gforth allows you to define @dfn{combined words} -- words that have an
4623: arbitrary combination of interpretation and compilation semantics.
4624:
4625:
4626: doc-interpret/compile:
4627:
4628:
4629: This feature was introduced for implementing @code{TO} and @code{S"}. I
4630: recommend that you do not define such words, as cute as they may be:
4631: they make it hard to get at both parts of the word in some contexts.
4632: E.g., assume you want to get an execution token for the compilation
4633: part. Instead, define two words, one that embodies the interpretation
4634: part, and one that embodies the compilation part. Once you have done
4635: that, you can define a combined word with @code{interpret/compile:} for
4636: the convenience of your users.
4637:
4638: You might try to use this feature to provide an optimizing
4639: implementation of the default compilation semantics of a word. For
4640: example, by defining:
4641: @example
4642: :noname
4643: foo bar ;
4644: :noname
4645: POSTPONE foo POSTPONE bar ;
4646: interpret/compile: opti-foobar
4647: @end example
4648:
4649: @noindent
4650: as an optimizing version of:
4651:
4652: @example
4653: : foobar
4654: foo bar ;
4655: @end example
4656:
4657: Unfortunately, this does not work correctly with @code{[compile]},
4658: because @code{[compile]} assumes that the compilation semantics of all
4659: @code{interpret/compile:} words are non-default. I.e., @code{[compile]
4660: opti-foobar} would compile compilation semantics, whereas
4661: @code{[compile] foobar} would compile interpretation semantics.
4662:
4663: @cindex state-smart words (are a bad idea)
4664: Some people try to use @dfn{state-smart} words to emulate the feature provided
4665: by @code{interpret/compile:} (words are state-smart if they check
4666: @code{STATE} during execution). E.g., they would try to code
4667: @code{foobar} like this:
4668:
4669: @example
4670: : foobar
4671: STATE @@
4672: IF ( compilation state )
4673: POSTPONE foo POSTPONE bar
4674: ELSE
4675: foo bar
4676: ENDIF ; immediate
4677: @end example
4678:
4679: Although this works if @code{foobar} is only processed by the text
4680: interpreter, it does not work in other contexts (like @code{'} or
4681: @code{POSTPONE}). E.g., @code{' foobar} will produce an execution token
4682: for a state-smart word, not for the interpretation semantics of the
4683: original @code{foobar}; when you execute this execution token (directly
4684: with @code{EXECUTE} or indirectly through @code{COMPILE,}) in compile
4685: state, the result will not be what you expected (i.e., it will not
4686: perform @code{foo bar}). State-smart words are a bad idea. Simply don't
4687: write them@footnote{For a more detailed discussion of this topic, see
4688: @cite{@code{State}-smartness -- Why it is Evil and How to Exorcise it} by Anton
4689: Ertl; presented at EuroForth '98 and available from
4690: @*@uref{http://www.complang.tuwien.ac.at/papers/ertl98.ps.gz}}!
4691:
4692: @cindex defining words with arbitrary semantics combinations
4693: It is also possible to write defining words that define words with
4694: arbitrary combinations of interpretation and compilation semantics. In
4695: general, they look like this:
4696:
4697: @example
4698: : def-word
4699: create-interpret/compile
4700: @i{code1}
4701: interpretation>
4702: @i{code2}
4703: <interpretation
4704: compilation>
4705: @i{code3}
4706: <compilation ;
4707: @end example
4708:
4709: For a @i{word} defined with @code{def-word}, the interpretation
4710: semantics are to push the address of the body of @i{word} and perform
4711: @i{code2}, and the compilation semantics are to push the address of
4712: the body of @i{word} and perform @i{code3}. E.g., @code{constant}
4713: can also be defined like this (except that the defined constants don't
4714: behave correctly when @code{[compile]}d):
4715:
4716: @example
4717: : constant ( n "name" -- )
4718: create-interpret/compile
4719: ,
4720: interpretation> ( -- n )
4721: @@
4722: <interpretation
4723: compilation> ( compilation. -- ; run-time. -- n )
4724: @@ postpone literal
4725: <compilation ;
4726: @end example
4727:
4728:
4729: doc-create-interpret/compile
4730: doc-interpretation>
4731: doc-<interpretation
4732: doc-compilation>
4733: doc-<compilation
4734:
4735:
4736: Words defined with @code{interpret/compile:} and
4737: @code{create-interpret/compile} have an extended header structure that
4738: differs from other words; however, unless you try to access them with
4739: plain address arithmetic, you should not notice this. Words for
4740: accessing the header structure usually know how to deal with this; e.g.,
4741: @code{'} @i{word} @code{>body} also gives you the body of a word created
4742: with @code{create-interpret/compile}.
4743:
4744:
4745: doc-postpone
4746:
4747: @comment TODO -- expand glossary text for POSTPONE
4748:
4749:
4750: @c -------------------------------------------------------------
4751: @node Tokens for Words, The Text Interpreter, Interpretation and Compilation Semantics, Words
4752: @section Tokens for Words
4753: @cindex tokens for words
4754:
4755: This section describes the creation and use of tokens that represent
4756: words.
4757:
4758: Named words have information stored in their header space entries to
4759: indicate any non-default semantics (@pxref{Interpretation and
4760: Compilation Semantics}). The semantics can be modified, using
4761: @code{immediate} and/or @code{compile-only}, at the time that the words
4762: are defined. Unnamed words have (by definition) no header space
4763: entry, and therefore must have default semantics.
4764:
4765: Named words have interpretation and compilation semantics. Unnamed words
4766: just have execution semantics.
4767:
4768: @cindex xt
4769: @cindex execution token
4770: The execution semantics of an unnamed word are represented by an
4771: @dfn{execution token} (@i{xt}). As explained in @ref{Supplying names},
4772: the execution token of the last word defined can be produced with
4773: @code{lastxt}.
4774:
4775: The interpretation semantics of a named word are also represented by an
4776: execution token. You can produce the execution token using @code{'} or
4777: @code{[']}. A simple example shows the difference between the two:
4778:
4779: @example
4780: : greet ( -- ) ." Hello" ;
4781: : foo ( -- xt ) ['] greet execute ; \ ['] parses greet at compile-time
4782: : bar ( -- ) ' execute ; \ ' parses at run-time
4783:
4784: \ the next four lines all do the same thing
4785: foo
4786: bar greet
4787: greet
4788: ' greet EXECUTE
4789: @end example
4790:
4791: An execution token occupies one cell.
4792: @cindex code field address
4793: @cindex CFA
4794: In Gforth, the abstract data type @i{execution token} is implemented
4795: as a code field address (CFA).
4796: @comment TODO note that the standard does not say what it represents..
4797: @comment and you cannot necessarily compile it in all Forths (eg native
4798: @comment compilers?).
4799:
4800: For literals, use @code{'} in interpreted code and @code{[']} in
4801: compiled code. Gforth's @code{'} and @code{[']} behave somewhat
4802: unusually by complaining about compile-only words. To get the execution
4803: token for a compile-only word @i{name}, use @code{COMP' @i{name} DROP}
4804: or @code{[COMP'] @i{name} DROP}.
4805:
4806: @cindex compilation token
4807: The compilation semantics of a named word are represented by a
4808: @dfn{compilation token} consisting of two cells: @i{w xt}. The top cell
4809: @i{xt} is an execution token. The compilation semantics represented by
4810: the compilation token can be performed with @code{execute}, which
4811: consumes the whole compilation token, with an additional stack effect
4812: determined by the represented compilation semantics.
4813:
4814: At present, the @i{w} part of a compilation token is an execution token,
4815: and the @i{xt} part represents either @code{execute} or
4816: @code{compile,}@footnote{Depending upon the compilation semantics of the
4817: word. If the word has default compilation semantics, the @i{xt} will
4818: represent @code{compile,}. Otherwise (e.g., for immediate words), the
4819: @i{xt} will represent @code{execute}.}. However, don't rely on that
4820: knowledge, unless necessary; future versions of Gforth may introduce
4821: unusual compilation tokens (e.g., a compilation token that represents
4822: the compilation semantics of a literal).
4823:
4824: You can compile the compilation semantics with @code{postpone,}. I.e.,
4825: @code{COMP' @i{word} postpone,} is equivalent to @code{postpone
4826: @i{word}}.
4827:
4828: @cindex name token
4829: @cindex name field address
4830: @cindex NFA
4831: Named words are also represented by the @dfn{name token}, (@i{nt}). In
4832: Gforth, the abstract data type @emph{name token} is implemented as a
4833: name field address (NFA).
4834:
4835:
4836: doc-execute
4837: doc-perform
4838: doc-compile,
4839: doc-[']
4840: doc-'
4841: doc-[comp']
4842: doc-comp'
4843: doc-postpone,
4844:
4845: doc-find-name
4846: doc-name>int
4847: doc-name?int
4848: doc-name>comp
4849: doc-name>string
4850:
4851:
4852: @c ----------------------------------------------------------
4853: @node The Text Interpreter, Word Lists, Tokens for Words, Words
4854: @section The Text Interpreter
4855: @cindex interpreter - outer
4856: @cindex text interpreter
4857: @cindex outer interpreter
4858:
4859: @c Should we really describe all these ugly details? IMO the text
4860: @c interpreter should be much cleaner, but that may not be possible within
4861: @c ANS Forth. - anton
4862: @c nac-> I wanted to explain how it works to show how you can exploit
4863: @c it in your own programs. When I was writing a cross-compiler, figuring out
4864: @c some of these gory details was very helpful to me. None of the textbooks
4865: @c I've seen cover it, and the most modern Forth textbook -- Forth Inc's,
4866: @c seems to positively avoid going into too much detail for some of
4867: @c the internals.
4868:
4869: The text interpreter@footnote{This is an expanded version of the
4870: material in @ref{Introducing the Text Interpreter}.} is an endless loop
4871: that processes input from the current input device. It is also called
4872: the outer interpreter, in contrast to the inner interpreter
4873: (@pxref{Engine}) which executes the compiled Forth code on interpretive
4874: implementations.
4875:
4876: @cindex interpret state
4877: @cindex compile state
4878: The text interpreter operates in one of two states: @dfn{interpret
4879: state} and @dfn{compile state}. The current state is defined by the
4880: aptly-named variable, @code{state}.
4881:
4882: This section starts by describing how the text interpreter behaves when
4883: it is in interpret state, processing input from the user input device --
4884: the keyboard. This is the mode that a Forth system is in after it starts
4885: up.
4886:
4887: @cindex input buffer
4888: @cindex terminal input buffer
4889: The text interpreter works from an area of memory called the @dfn{input
4890: buffer}@footnote{When the text interpreter is processing input from the
4891: keyboard, this area of memory is called the @dfn{terminal input buffer}
4892: (TIB) and is addressed by the (obsolescent) words @code{TIB} and
4893: @code{#TIB}.}, which stores your keyboard input when you press the
4894: @key{RET} key. Starting at the beginning of the input buffer, it skips
4895: leading spaces (called @dfn{delimiters}) then parses a string (a
4896: sequence of non-space characters) until it reaches either a space
4897: character or the end of the buffer. Having parsed a string, it makes two
4898: attempts to process it:
4899:
4900: @cindex dictionary
4901: @itemize @bullet
4902: @item
4903: It looks for the string in a @dfn{dictionary} of definitions. If the
4904: string is found, the string names a @dfn{definition} (also known as a
4905: @dfn{word}) and the dictionary search returns information that allows
4906: the text interpreter to perform the word's @dfn{interpretation
4907: semantics}. In most cases, this simply means that the word will be
4908: executed.
4909: @item
4910: If the string is not found in the dictionary, the text interpreter
4911: attempts to treat it as a number, using the rules described in
4912: @ref{Number Conversion}. If the string represents a legal number in the
4913: current radix, the number is pushed onto a parameter stack (the data
4914: stack for integers, the floating-point stack for floating-point
4915: numbers).
4916: @end itemize
4917:
4918: If both attempts fail, or if the word is found in the dictionary but has
4919: no interpretation semantics@footnote{This happens if the word was
4920: defined as @code{COMPILE-ONLY}.} the text interpreter discards the
4921: remainder of the input buffer, issues an error message and waits for
4922: more input. If one of the attempts succeeds, the text interpreter
4923: repeats the parsing process until the whole of the input buffer has been
4924: processed, at which point it prints the status message ``@code{ ok}''
4925: and waits for more input.
4926:
4927: @cindex parse area
4928: The text interpreter keeps track of its position in the input buffer by
4929: updating a variable called @code{>IN} (pronounced ``to-in''). The value
4930: of @code{>IN} starts out as 0, indicating an offset of 0 from the start
4931: of the input buffer. The region from offset @code{>IN @@} to the end of
4932: the input buffer is called the @dfn{parse area}@footnote{In other words,
4933: the text interpreter processes the contents of the input buffer by
4934: parsing strings from the parse area until the parse area is empty.}.
4935: This example shows how @code{>IN} changes as the text interpreter parses
4936: the input buffer:
4937:
4938: @example
4939: : remaining >IN @@ SOURCE 2 PICK - -ROT + SWAP
4940: CR ." ->" TYPE ." <-" ; IMMEDIATE
4941:
4942: 1 2 3 remaining + remaining .
4943:
4944: : foo 1 2 3 remaining SWAP remaining ;
4945: @end example
4946:
4947: @noindent
4948: The result is:
4949:
4950: @example
4951: ->+ remaining .<-
4952: ->.<-5 ok
4953:
4954: ->SWAP remaining ;-<
4955: ->;<- ok
4956: @end example
4957:
4958: @cindex parsing words
4959: The value of @code{>IN} can also be modified by a word in the input
4960: buffer that is executed by the text interpreter. This means that a word
4961: can ``trick'' the text interpreter into either skipping a section of the
4962: input buffer@footnote{This is how parsing words work.} or into parsing a
4963: section twice. For example:
4964:
4965: @example
4966: : lat ." <<lat>>" ;
4967: : flat ." <<flat>>" >IN DUP @@ 3 - SWAP ! ;
4968: @end example
4969:
4970: @noindent
4971: When @code{flat} is executed, this output is produced@footnote{Exercise
4972: for the reader: what would happen if the @code{3} were replaced with
4973: @code{4}?}:
4974:
4975: @example
4976: <<flat>><<lat>>
4977: @end example
4978:
4979: @noindent
4980: Two important notes about the behaviour of the text interpreter:
4981:
4982: @itemize @bullet
4983: @item
4984: It processes each input string to completion before parsing additional
4985: characters from the input buffer.
4986: @item
4987: It treats the input buffer as a read-only region (and so must your code).
4988: @end itemize
4989:
4990: @noindent
4991: When the text interpreter is in compile state, its behaviour changes in
4992: these ways:
4993:
4994: @itemize @bullet
4995: @item
4996: If a parsed string is found in the dictionary, the text interpreter will
4997: perform the word's @dfn{compilation semantics}. In most cases, this
4998: simply means that the execution semantics of the word will be appended
4999: to the current definition.
5000: @item
5001: When a number is encountered, it is compiled into the current definition
5002: (as a literal) rather than being pushed onto a parameter stack.
5003: @item
5004: If an error occurs, @code{state} is modified to put the text interpreter
5005: back into interpret state.
5006: @item
5007: Each time a line is entered from the keyboard, Gforth prints
5008: ``@code{ compiled}'' rather than `` @code{ok}''.
5009: @end itemize
5010:
5011: @cindex text interpreter - input sources
5012: When the text interpreter is using an input device other than the
5013: keyboard, its behaviour changes in these ways:
5014:
5015: @itemize @bullet
5016: @item
5017: When the parse area is empty, the text interpreter attempts to refill
5018: the input buffer from the input source. When the input source is
5019: exhausted, the input source is set back to the user input device.
5020: @item
5021: It doesn't print out ``@code{ ok}'' or ``@code{ compiled}'' messages each
5022: time the parse area is emptied.
5023: @item
5024: If an error occurs, the input source is set back to the user input
5025: device.
5026: @end itemize
5027:
5028: @ref{Input Sources} describes this in more detail.
5029:
5030:
5031: doc->in
5032: doc-source
5033:
5034: doc-tib
5035: doc-#tib
5036:
5037:
5038: @menu
5039: * Input Sources::
5040: * Number Conversion::
5041: * Interpret/Compile states::
5042: * Literals::
5043: * Interpreter Directives::
5044: @end menu
5045:
5046: @node Input Sources, Number Conversion, The Text Interpreter, The Text Interpreter
5047: @subsection Input Sources
5048: @cindex input sources
5049: @cindex text interpreter - input sources
5050:
5051: By default, the text interpreter processes input from the user input
5052: device (the keyboard) when Forth starts up. The text interpreter can
5053: process input from any of these sources:
5054:
5055: @itemize @bullet
5056: @item
5057: The user input device -- the keyboard.
5058: @item
5059: A file, using the words described in @ref{Forth source files}.
5060: @item
5061: A block, using the words described in @ref{Blocks}.
5062: @item
5063: A text string, using @code{evaluate}.
5064: @end itemize
5065:
5066: A program can identify the current input device from the values of
5067: @code{source-id} and @code{blk}.
5068:
5069:
5070: doc-source-id
5071: doc-blk
5072:
5073: doc-save-input
5074: doc-restore-input
5075:
5076: doc-evaluate
5077:
5078:
5079:
5080: @node Number Conversion, Interpret/Compile states, Input Sources, The Text Interpreter
5081: @subsection Number Conversion
5082: @cindex number conversion
5083: @cindex double-cell numbers, input format
5084: @cindex input format for double-cell numbers
5085: @cindex single-cell numbers, input format
5086: @cindex input format for single-cell numbers
5087: @cindex floating-point numbers, input format
5088: @cindex input format for floating-point numbers
5089:
5090: This section describes the rules that the text interpreter uses when it
5091: tries to convert a string into a number.
5092:
5093: Let <digit> represent any character that is a legal digit in the current
5094: number base@footnote{For example, 0-9 when the number base is decimal or
5095: 0-9, A-F when the number base is hexadecimal.}.
5096:
5097: Let <decimal digit> represent any character in the range 0-9.
5098:
5099: Let @{@i{a b}@} represent the @i{optional} presence of any of the characters
5100: in the braces (@i{a} or @i{b} or neither).
5101:
5102: Let * represent any number of instances of the previous character
5103: (including none).
5104:
5105: Let any other character represent itself.
5106:
5107: @noindent
5108: Now, the conversion rules are:
5109:
5110: @itemize @bullet
5111: @item
5112: A string of the form <digit><digit>* is treated as a single-precision
5113: (cell-sized) positive integer. Examples are 0 123 6784532 32343212343456 42
5114: @item
5115: A string of the form -<digit><digit>* is treated as a single-precision
5116: (cell-sized) negative integer, and is represented using 2's-complement
5117: arithmetic. Examples are -45 -5681 -0
5118: @item
5119: A string of the form <digit><digit>*.<digit>* is treated as a double-precision
5120: (double-cell-sized) positive integer. Examples are 3465. 3.465 34.65
5121: (all three of these represent the same number).
5122: @item
5123: A string of the form -<digit><digit>*.<digit>* is treated as a
5124: double-precision (double-cell-sized) negative integer, and is
5125: represented using 2's-complement arithmetic. Examples are -3465. -3.465
5126: -34.65 (all three of these represent the same number).
5127: @item
5128: A string of the form @{+ -@}<decimal digit>@{.@}<decimal digit>*@{e
5129: E@}@{+ -@}<decimal digit><decimal digit>* is treated as a floating-point
5130: number. Examples are 1e 1e0 1.e 1.e0 +1e+0 (which all represent the same
5131: number) +12.E-4
5132: @end itemize
5133:
5134: By default, the number base used for integer number conversion is given
5135: by the contents of the variable @code{base}. Note that a lot of
5136: confusion can result from unexpected values of @code{base}. If you
5137: change @code{base} anywhere, make sure to save the old value and restore
5138: it afterwards. In general I recommend keeping @code{base} decimal, and
5139: using the prefixes described below for the popular non-decimal bases.
5140:
5141: doc-dpl
5142: doc-base
5143: doc-hex
5144: doc-decimal
5145:
5146:
5147: @cindex '-prefix for character strings
5148: @cindex &-prefix for decimal numbers
5149: @cindex %-prefix for binary numbers
5150: @cindex $-prefix for hexadecimal numbers
5151: Gforth allows you to override the value of @code{base} by using a
5152: prefix@footnote{Some Forth implementations provide a similar scheme by
5153: implementing @code{$} etc. as parsing words that process the subsequent
5154: number in the input stream and push it onto the stack. For example, see
5155: @cite{Number Conversion and Literals}, by Wil Baden; Forth Dimensions
5156: 20(3) pages 26--27. In such implementations, unlike in Gforth, a space
5157: is required between the prefix and the number.} before the first digit
5158: of an (integer) number. Four prefixes are supported:
5159:
5160: @itemize @bullet
5161: @item
5162: @code{&} -- decimal
5163: @item
5164: @code{%} -- binary
5165: @item
5166: @code{$} -- hexadecimal
5167: @item
5168: @code{'} -- base @code{max-char+1}
5169: @end itemize
5170:
5171: Here are some examples, with the equivalent decimal number shown after
5172: in braces:
5173:
5174: -$41 (-65), %1001101 (205), %1001.0001 (145 - a double-precision number),
5175: 'AB (16706; ascii A is 65, ascii B is 66, number is 65*256 + 66),
5176: 'ab (24930; ascii a is 97, ascii B is 98, number is 97*256 + 98),
5177: &905 (905), $abc (2478), $ABC (2478).
5178:
5179: @cindex number conversion - traps for the unwary
5180: @noindent
5181: Number conversion has a number of traps for the unwary:
5182:
5183: @itemize @bullet
5184: @item
5185: You cannot determine the current number base using the code sequence
5186: @code{base @@ .} -- the number base is always 10 in the current number
5187: base. Instead, use something like @code{base @@ dec.}
5188: @item
5189: If the number base is set to a value greater than 14 (for example,
5190: hexadecimal), the number 123E4 is ambiguous; the conversion rules allow
5191: it to be intepreted as either a single-precision integer or a
5192: floating-point number (Gforth treats it as an integer). The ambiguity
5193: can be resolved by explicitly stating the sign of the mantissa and/or
5194: exponent: 123E+4 or +123E4 -- if the number base is decimal, no
5195: ambiguity arises; either representation will be treated as a
5196: floating-point number.
5197: @item
5198: There is a word @code{bin} but it does @i{not} set the number base!
5199: It is used to specify file types.
5200: @item
5201: ANS Forth requires the @code{.} of a double-precision number to
5202: be the final character in the string. Allowing the @code{.} to be
5203: anywhere after the first digit is a Gforth extension.
5204: @item
5205: The number conversion process does not check for overflow.
5206: @item
5207: In Gforth, number conversion to floating-point numbers always use base
5208: 10, irrespective of the value of @code{base}. In ANS Forth,
5209: conversion to floating-point numbers whilst the value of
5210: @code{base} is not 10 is an ambiguous condition.
5211: @end itemize
5212:
5213: @ref{Input} describes words that you can use to read numbers into your
5214: programs.
5215:
5216: @node Interpret/Compile states, Literals, Number Conversion, The Text Interpreter
5217: @subsection Interpret/Compile states
5218: @cindex Interpret/Compile states
5219:
5220: A standard program is not permitted to change @code{state}
5221: explicitly. However, it can change @code{state} implicitly, using the
5222: words @code{[} and @code{]}. When @code{[} is executed it switches
5223: @code{state} to interpret state, and therefore the text interpreter
5224: starts interpreting. When @code{]} is executed it switches @code{state}
5225: to compile state and therefore the text interpreter starts
5226: compiling. The most common usage for these words is for switching into
5227: interpret state and back from within a colon definition; this technique
5228: can be used to compile a literal (@pxref{Literals} for an example) or
5229: for conditional compilation (@pxref{Interpreter Directives} for an
5230: example).
5231:
5232:
5233: @c This is a bad example: It's non-standard, and it's not necessary.
5234: @c However, I can't think of a good example for switching into compile
5235: @c state when there is no current word (@code{state}-smart words are not a
5236: @c good reason). So maybe we should use an example for switching into
5237: @c interpret @code{state} in a colon def. - anton
5238: @c nac-> I agree. I started out by putting in the example, then realised
5239: @c that it was non-ANS, so wrote more words around it. I hope this
5240: @c re-written version is acceptable to you. I do want to keep the example
5241: @c as it is helpful for showing what is and what is not portable, particularly
5242: @c where it outlaws a style in common use.
5243:
5244:
5245: @code{[} and @code{]} also give you the ability to switch into compile
5246: state and back, but we cannot think of any useful Standard application
5247: for this ability. Pre-ANS Forth textbooks have examples like this:
5248:
5249: @example
5250: : AA ." this is A" ;
5251: : BB ." this is B" ;
5252: : CC ." this is C" ;
5253:
5254: create table ] aa bb cc [
5255:
5256: : go ( n -- ) \ n is offset into table.. 0 for 1st entry
5257: cells table + @ execute ;
5258: @end example
5259:
5260: This example builds a jump table; @code{0 go} will display ``@code{this
5261: is A}''. Using @code{[} and @code{]} in this example is equivalent to
5262: defining @code{table} like this:
5263:
5264: @example
5265: create table ' aa COMPILE, ' bb COMPILE, ' cc COMPILE,
5266: @end example
5267:
5268: The problem with this code is that the definition of @code{table} is not
5269: portable -- it @i{compile}s execution tokens into code space. Whilst it
5270: @i{may} work on systems where code space and data space co-incide, the
5271: Standard only allows data space to be assigned for a @code{CREATE}d
5272: word. In addition, the Standard only allows @code{@@} to access data
5273: space, whilst this example is using it to access code space. The only
5274: portable, Standard way to build this table is to build it in data space,
5275: like this:
5276:
5277: @example
5278: create table ' aa , ' bb , ' cc ,
5279: @end example
5280:
5281: doc-state
5282: doc-[
5283: doc-]
5284:
5285:
5286: @node Literals, Interpreter Directives, Interpret/Compile states, The Text Interpreter
5287: @subsection Literals
5288: @cindex Literals
5289:
5290: Often, you want to use a number within a colon definition. When you do
5291: this, the text interpreter automatically compiles the number as a
5292: @i{literal}. A literal is a number whose run-time effect is to be pushed
5293: onto the stack. If you had to do some maths to generate the number, you
5294: might write it like this:
5295:
5296: @example
5297: : HOUR-TO-SEC ( n1 -- n2 )
5298: 60 * \ to minutes
5299: 60 * ; \ to seconds
5300: @end example
5301:
5302: It is very clear what this definition is doing, but it's inefficient
5303: since it is performing 2 multiples at run-time. An alternative would be
5304: to write:
5305:
5306: @example
5307: : HOUR-TO-SEC ( n1 -- n2 )
5308: 3600 * ; \ to seconds
5309: @end example
5310:
5311: Which does the same thing, and has the advantage of using a single
5312: multiply. Ideally, we'd like the efficiency of the second with the
5313: readability of the first.
5314:
5315: @code{Literal} allows us to achieve that. It takes a number from the
5316: stack and lays it down in the current definition just as though the
5317: number had been typed directly into the definition. Our first attempt
5318: might look like this:
5319:
5320: @example
5321: 60 \ mins per hour
5322: 60 * \ seconds per minute
5323: : HOUR-TO-SEC ( n1 -- n2 )
5324: Literal * ; \ to seconds
5325: @end example
5326:
5327: But this produces the error message @code{unstructured}. What happened?
5328: The stack notation for @code{:} is (@i{ -- colon-sys}) and the size of
5329: @i{colon-sys} is implementation-defined. In other words, once we start a
5330: colon definition we can't portably access anything that was on the stack
5331: before the definition began@footnote{@cite{Two Problems in ANS Forth},
5332: by Thomas Worthington; Forth Dimensions 20(2) pages 32--34 describes
5333: some situations where you might want to access stack items above
5334: colon-sys, and provides a solution to the problem.}. The correct way of
5335: solving this problem in this instance is to use @code{[ ]} like this:
5336:
5337: @example
5338: : HOUR-TO-SEC ( n1 -- n2 )
5339: [ 60 \ minutes per hour
5340: 60 * ] \ seconds per minute
5341: LITERAL * ; \ to seconds
5342: @end example
5343:
5344:
5345: doc-literal
5346: doc-]L
5347: doc-2literal
5348: doc-fliteral
5349:
5350:
5351: @node Interpreter Directives, , Literals, The Text Interpreter
5352: @subsection Interpreter Directives
5353: @cindex interpreter directives
5354:
5355: These words are usually used in interpret state; typically to control
5356: which parts of a source file are processed by the text
5357: interpreter. There are only a few ANS Forth Standard words, but Gforth
5358: supplements these with a rich set of immediate control structure words
5359: to compensate for the fact that the non-immediate versions can only be
5360: used in compile state (@pxref{Control Structures}). Typical usages:
5361:
5362: @example
5363: FALSE Constant ASSEMBLER
5364: .
5365: .
5366: ASSEMBLER [IF]
5367: : ASSEMBLER-FEATURE
5368: ...
5369: ;
5370: [ENDIF]
5371: .
5372: .
5373: : SEE
5374: ... \ general-purpose SEE code
5375: [ ASSEMBLER [IF] ]
5376: ... \ assembler-specific SEE code
5377: [ [ENDIF] ]
5378: ;
5379: @end example
5380:
5381:
5382: doc-[IF]
5383: doc-[ELSE]
5384: doc-[THEN]
5385: doc-[ENDIF]
5386:
5387: doc-[IFDEF]
5388: doc-[IFUNDEF]
5389:
5390: doc-[?DO]
5391: doc-[DO]
5392: doc-[FOR]
5393: doc-[LOOP]
5394: doc-[+LOOP]
5395: doc-[NEXT]
5396:
5397: doc-[BEGIN]
5398: doc-[UNTIL]
5399: doc-[AGAIN]
5400: doc-[WHILE]
5401: doc-[REPEAT]
5402:
5403:
5404: @c -------------------------------------------------------------
5405: @node Word Lists, Environmental Queries, The Text Interpreter, Words
5406: @section Word Lists
5407: @cindex word lists
5408: @cindex header space
5409:
5410: A wordlist is a list of named words; you can add new words and look up
5411: words by name (and you can remove words in a restricted way with
5412: markers). Every named (and @code{reveal}ed) word is in one wordlist.
5413:
5414: @cindex search order stack
5415: The text interpreter searches the wordlists present in the search order
5416: (a stack of wordlists), from the top to the bottom. Within each
5417: wordlist, the search starts conceptually at the newest word; i.e., if
5418: two words in a wordlist have the same name, the newer word is found.
5419:
5420: @cindex compilation word list
5421: New words are added to the @dfn{compilation wordlist} (aka current
5422: wordlist).
5423:
5424: @cindex wid
5425: A word list is identified by a cell-sized word list identifier (@i{wid})
5426: in much the same way as a file is identified by a file handle. The
5427: numerical value of the wid has no (portable) meaning, and might change
5428: from session to session.
5429:
5430: The ANS Forth ``Search order'' word set is intended to provide a set of
5431: low-level tools that allow various different schemes to be
5432: implemented. Gforth provides @code{vocabulary}, a traditional Forth
5433: word. @file{compat/vocabulary.fs} provides an implementation in ANS
5434: Forth.
5435:
5436: @comment TODO: locals section refers to here, saying that every word list (aka
5437: @comment vocabulary) has its own methods for searching etc. Need to document that.
5438:
5439: @comment TODO: document markers, reveal, tables, mappedwordlist
5440:
5441: @comment the gforthman- prefix is used to pick out the true definition of a
5442: @comment word from the source files, rather than some alias.
5443:
5444: doc-forth-wordlist
5445: doc-definitions
5446: doc-get-current
5447: doc-set-current
5448: doc-get-order
5449: doc---gforthman-set-order
5450: doc-wordlist
5451: doc-table
5452: doc-push-order
5453: doc-previous
5454: doc-also
5455: doc---gforthman-forth
5456: doc-only
5457: doc---gforthman-order
5458:
5459: doc-find
5460: doc-search-wordlist
5461:
5462: doc-words
5463: doc-vlist
5464: @c doc-words-deferred
5465:
5466: doc-mappedwordlist
5467: doc-root
5468: doc-vocabulary
5469: doc-seal
5470: doc-vocs
5471: doc-current
5472: doc-context
5473:
5474:
5475: @menu
5476: * Why use word lists?::
5477: * Word list examples::
5478: @end menu
5479:
5480: @node Why use word lists?, Word list examples, Word Lists, Word Lists
5481: @subsection Why use word lists?
5482: @cindex word lists - why use them?
5483:
5484: Here are some reasons for using multiple word lists:
5485:
5486: @itemize @bullet
5487: @item
5488: To improve compilation speed by reducing the number of header space
5489: entries that must be searched. This is achieved by creating a new
5490: word list that contains all of the definitions that are used in the
5491: definition of a Forth system but which would not usually be used by
5492: programs running on that system. That word list would be on the search
5493: list when the Forth system was compiled but would be removed from the
5494: search list for normal operation. This can be a useful technique for
5495: low-performance systems (for example, 8-bit processors in embedded
5496: systems) but is unlikely to be necessary in high-performance desktop
5497: systems.
5498: @item
5499: To prevent a set of words from being used outside the context in which
5500: they are valid. Two classic examples of this are an integrated editor
5501: (all of the edit commands are defined in a separate word list; the
5502: search order is set to the editor word list when the editor is invoked;
5503: the old search order is restored when the editor is terminated) and an
5504: integrated assembler (the op-codes for the machine are defined in a
5505: separate word list which is used when a @code{CODE} word is defined).
5506: @item
5507: To prevent a name-space clash between multiple definitions with the same
5508: name. For example, when building a cross-compiler you might have a word
5509: @code{IF} that generates conditional code for your target system. By
5510: placing this definition in a different word list you can control whether
5511: the host system's @code{IF} or the target system's @code{IF} get used in
5512: any particular context by controlling the order of the word lists on the
5513: search order stack.
5514: @end itemize
5515:
5516: @node Word list examples, ,Why use word lists?, Word Lists
5517: @subsection Word list examples
5518: @cindex word lists - examples
5519:
5520: Here is an example of creating and using a new wordlist using ANS
5521: Forth Standard words:
5522:
5523: @example
5524: wordlist constant my-new-words-wordlist
5525: : my-new-words get-order nip my-new-words-wordlist swap set-order ;
5526:
5527: \ add it to the search order
5528: also my-new-words
5529:
5530: \ alternatively, add it to the search order and make it
5531: \ the compilation word list
5532: also my-new-words definitions
5533: \ type "order" to see the problem
5534: @end example
5535:
5536: The problem with this example is that @code{order} has no way to
5537: associate the name @code{my-new-words} with the wid of the word list (in
5538: Gforth, @code{order} and @code{vocs} will display @code{???} for a wid
5539: that has no associated name). There is no Standard way of associating a
5540: name with a wid.
5541:
5542: In Gforth, this example can be re-coded using @code{vocabulary}, which
5543: associates a name with a wid:
5544:
5545: @example
5546: vocabulary my-new-words
5547:
5548: \ add it to the search order
5549: also my-new-words
5550:
5551: \ alternatively, add it to the search order and make it
5552: \ the compilation word list
5553: my-new-words definitions
5554: \ type "order" to see that the problem is solved
5555: @end example
5556:
5557: @c -------------------------------------------------------------
5558: @node Environmental Queries, Files, Word Lists, Words
5559: @section Environmental Queries
5560: @cindex environmental queries
5561:
5562: ANS Forth introduced the idea of ``environmental queries'' as a way
5563: for a program running on a system to determine certain characteristics of the system.
5564: The Standard specifies a number of strings that might be recognised by a system.
5565:
5566: The Standard requires that the header space used for environmental queries
5567: be distinct from the header space used for definitions.
5568:
5569: Typically, environmental queries are supported by creating a set of
5570: definitions in a word list that is @i{only} used during environmental
5571: queries; that is what Gforth does. There is no Standard way of adding
5572: definitions to the set of recognised environmental queries, but any
5573: implementation that supports the loading of optional word sets must have
5574: some mechanism for doing this (after loading the word set, the
5575: associated environmental query string must return @code{true}). In
5576: Gforth, the word list used to honour environmental queries can be
5577: manipulated just like any other word list.
5578:
5579:
5580: doc-environment?
5581: doc-environment-wordlist
5582:
5583: doc-gforth
5584: doc-os-class
5585:
5586:
5587: Note that, whilst the documentation for (e.g.) @code{gforth} shows it
5588: returning two items on the stack, querying it using @code{environment?}
5589: will return an additional item; the @code{true} flag that shows that the
5590: string was recognised.
5591:
5592: @comment TODO Document the standard strings or note where they are documented herein
5593:
5594: Here are some examples of using environmental queries:
5595:
5596: @example
5597: s" address-unit-bits" environment? 0=
5598: [IF]
5599: cr .( environmental attribute address-units-bits unknown... ) cr
5600: [THEN]
5601:
5602: s" block" environment? [IF] DROP include block.fs [THEN]
5603:
5604: s" gforth" environment? [IF] 2DROP include compat/vocabulary.fs [THEN]
5605:
5606: s" gforth" environment? [IF] .( Gforth version ) TYPE
5607: [ELSE] .( Not Gforth..) [THEN]
5608: @end example
5609:
5610:
5611: Here is an example of adding a definition to the environment word list:
5612:
5613: @example
5614: get-current environment-wordlist set-current
5615: true constant block
5616: true constant block-ext
5617: set-current
5618: @end example
5619:
5620: You can see what definitions are in the environment word list like this:
5621:
5622: @example
5623: get-order 1+ environment-wordlist swap set-order words previous
5624: @end example
5625:
5626:
5627: @c -------------------------------------------------------------
5628: @node Files, Blocks, Environmental Queries, Words
5629: @section Files
5630: @cindex files
5631: @cindex I/O - file-handling
5632:
5633: Gforth provides facilities for accessing files that are stored in the
5634: host operating system's file-system. Files that are processed by Gforth
5635: can be divided into two categories:
5636:
5637: @itemize @bullet
5638: @item
5639: Files that are processed by the Text Interpreter (@dfn{Forth source files}).
5640: @item
5641: Files that are processed by some other program (@dfn{general files}).
5642: @end itemize
5643:
5644: doc-loadfilename
5645: doc-sourcefilename
5646: doc-sourceline#
5647:
5648: @menu
5649: * Forth source files::
5650: * General files::
5651: * Search Paths::
5652: * Forth Search Paths::
5653: * General Search Paths::
5654: @end menu
5655:
5656:
5657: @c -------------------------------------------------------------
5658: @node Forth source files, General files, Files, Files
5659: @subsection Forth source files
5660: @cindex including files
5661: @cindex Forth source files
5662:
5663: The simplest way to interpret the contents of a file is to use one of
5664: these two formats:
5665:
5666: @example
5667: include mysource.fs
5668: s" mysource.fs" included
5669: @end example
5670:
5671: Sometimes you want to include a file only if it is not included already
5672: (by, say, another source file). In that case, you can use one of these
5673: three formats:
5674:
5675: @example
5676: require mysource.fs
5677: needs mysource.fs
5678: s" mysource.fs" required
5679: @end example
5680:
5681: @cindex stack effect of included files
5682: @cindex including files, stack effect
5683: It is good practice to write your source files such that interpreting them
5684: does not change the stack. Source files designed in this way can be used with
5685: @code{required} and friends without complications. For example:
5686:
5687: @example
5688: 1 require foo.fs drop
5689: @end example
5690:
5691:
5692: doc-include-file
5693: doc-included
5694: doc-included?
5695: doc-include
5696: doc-required
5697: doc-require
5698: doc-needs
5699: doc-init-included-files
5700:
5701:
5702: A definition in ANS Forth for @code{required} is provided in
5703: @file{compat/required.fs}.
5704:
5705: @c -------------------------------------------------------------
5706: @node General files, Search Paths, Forth source files, Files
5707: @subsection General files
5708: @cindex general files
5709: @cindex file-handling
5710:
5711: Files are opened/created by name and type. The following types are
5712: recognised:
5713:
5714:
5715: doc-r/o
5716: doc-r/w
5717: doc-w/o
5718: doc-bin
5719:
5720:
5721: When a file is opened/created, it returns a file identifier,
5722: @i{wfileid} that is used for all other file commands. All file
5723: commands also return a status value, @i{wior}, that is 0 for a
5724: successful operation and an implementation-defined non-zero value in the
5725: case of an error.
5726:
5727:
5728: doc-open-file
5729: doc-create-file
5730:
5731: doc-close-file
5732: doc-delete-file
5733: doc-rename-file
5734: doc-read-file
5735: doc-read-line
5736: doc-write-file
5737: doc-write-line
5738: doc-emit-file
5739: doc-flush-file
5740:
5741: doc-file-status
5742: doc-file-position
5743: doc-reposition-file
5744: doc-file-size
5745: doc-resize-file
5746:
5747:
5748: @c ---------------------------------------------------------
5749: @node Search Paths, Forth Search Paths, General files, Files
5750: @subsection Search Paths
5751: @cindex path for @code{included}
5752: @cindex file search path
5753: @cindex @code{include} search path
5754: @cindex search path for files
5755:
5756: If you specify an absolute filename (i.e., a filename starting with
5757: @file{/} or @file{~}, or with @file{:} in the second position (as in
5758: @samp{C:...})) for @code{included} and friends, that file is included
5759: just as you would expect.
5760:
5761: For relative filenames, Gforth uses a search path similar to Forth's
5762: search order (@pxref{Word Lists}). It tries to find the given filename
5763: in the directories present in the path, and includes the first one it
5764: finds. There are separate search paths for Forth source files and
5765: general files.
5766:
5767: If the search path contains the directory @file{.} (as it should), this
5768: refers to the directory that the present file was @code{included}
5769: from. This allows files to include other files relative to their own
5770: position (irrespective of the current working directory or the absolute
5771: position). This feature is essential for libraries consisting of
5772: several files, where a file may include other files from the library.
5773: It corresponds to @code{#include "..."} in C. If the current input
5774: source is not a file, @file{.} refers to the directory of the innermost
5775: file being included, or, if there is no file being included, to the
5776: current working directory.
5777:
5778: Use @file{~+} to refer to the current working directory (as in the
5779: @code{bash}).
5780:
5781: If the filename starts with @file{./}, the search path is not searched
5782: (just as with absolute filenames), and the @file{.} has the same meaning
5783: as described above.
5784:
5785: @c ---------------------------------------------------------
5786: @node Forth Search Paths, General Search Paths, Search Paths, Files
5787: @subsubsection Forth Search Paths
5788: @cindex search path control - Forth
5789:
5790: The search path is initialized when you start Gforth (@pxref{Invoking
5791: Gforth}). You can display it and change it using these words:
5792:
5793:
5794: doc-.fpath
5795: doc-fpath+
5796: doc-fpath=
5797: doc-open-fpath-file
5798:
5799:
5800: @noindent
5801: Here is an example of using @code{fpath} and @code{require}:
5802:
5803: @example
5804: fpath= /usr/lib/forth/|./
5805: require timer.fs
5806: @end example
5807:
5808: @c ---------------------------------------------------------
5809: @node General Search Paths, , Forth Search Paths, Files
5810: @subsubsection General Search Paths
5811: @cindex search path control - for user applications
5812:
5813: Your application may need to search files in several directories, like
5814: @code{included} does. To facilitate this, Gforth allows you to define
5815: and use your own search paths, by providing generic equivalents of the
5816: Forth search path words:
5817:
5818:
5819: doc-.path
5820: doc-path+
5821: doc-path=
5822: doc-open-path-file
5823:
5824:
5825: Here's an example of creating a search path:
5826:
5827: @example
5828: \ Make a buffer for the path:
5829: create mypath 100 chars , \ maximum length (is checked)
5830: 0 , \ real len
5831: 100 chars allot \ space for path
5832: @end example
5833:
5834: @c -------------------------------------------------------------
5835: @node Blocks, Other I/O, Files, Words
5836: @section Blocks
5837: @cindex I/O - blocks
5838: @cindex blocks
5839:
5840: When you run Gforth on a modern desk-top computer, it runs under the
5841: control of an operating system which provides certain services. One of
5842: these services is @var{file services}, which allows Forth source code
5843: and data to be stored in files and read into Gforth (@pxref{Files}).
5844:
5845: Traditionally, Forth has been an important programming language on
5846: systems where it has interfaced directly to the underlying hardware with
5847: no intervening operating system. Forth provides a mechanism, called
5848: @dfn{blocks}, for accessing mass storage on such systems.
5849:
5850: A block is a 1024-byte data area, which can be used to hold data or
5851: Forth source code. No structure is imposed on the contents of the
5852: block. A block is identified by its number; blocks are numbered
5853: contiguously from 1 to an implementation-defined maximum.
5854:
5855: A typical system that used blocks but no operating system might use a
5856: single floppy-disk drive for mass storage, with the disks formatted to
5857: provide 256-byte sectors. Blocks would be implemented by assigning the
5858: first four sectors of the disk to block 1, the second four sectors to
5859: block 2 and so on, up to the limit of the capacity of the disk. The disk
5860: would not contain any file system information, just the set of blocks.
5861:
5862: @cindex blocks file
5863: On systems that do provide file services, blocks are typically
5864: implemented by storing a sequence of blocks within a single @dfn{blocks
5865: file}. The size of the blocks file will be an exact multiple of 1024
5866: bytes, corresponding to the number of blocks it contains. This is the
5867: mechanism that Gforth uses.
5868:
5869: @cindex @file{blocks.fb}
5870: Only 1 blocks file can be open at a time. If you use block words without
5871: having specified a blocks file, Gforth defaults to the blocks file
5872: @file{blocks.fb}. Gforth uses the Forth search path when attempting to
5873: locate a blocks file (@pxref{Forth Search Paths}).
5874:
5875: @cindex block buffers
5876: When you read and write blocks under program control, Gforth uses a
5877: number of @dfn{block buffers} as intermediate storage. These buffers are
5878: not used when you use @code{load} to interpret the contents of a block.
5879:
5880: The behaviour of the block buffers is directly analagous to that of a
5881: cache. Each block buffer has three states:
5882:
5883: @itemize @bullet
5884: @item
5885: Unassigned
5886: @item
5887: Assigned-clean
5888: @item
5889: Assigned-dirty
5890: @end itemize
5891:
5892: Initially, all block buffers are @i{unassigned}. In order to access a
5893: block, the block (specified by its block number) must be assigned to a
5894: block buffer.
5895:
5896: The assignment of a block to a block buffer is performed by @code{block}
5897: or @code{buffer}. Use @code{block} when you wish to modify the existing
5898: contents of a block. Use @code{buffer} when you don't care about the
5899: existing contents of the block@footnote{The ANS Forth definition of
5900: @code{buffer} is intended not to cause disk I/O; if the data associated
5901: with the particular block is already stored in a block buffer due to an
5902: earlier @code{block} command, @code{buffer} will return that block
5903: buffer and the existing contents of the block will be
5904: available. Otherwise, @code{buffer} will simply assign a new, empty
5905: block buffer for the block.}.
5906:
5907: Once a block has been assigned to a block buffer using @code{block} or
5908: @code{buffer}, that block buffer becomes the @i{current block buffer}
5909: and its state changes to @i{assigned-clean}. Data may only be
5910: manipulated (read or written) within the current block buffer.
5911:
5912: When the contents of the current block buffer has been modified it is
5913: necessary, @i{before calling} @code{block} @i{or} @code{buffer}
5914: @i{again}, to either abandon the changes (by doing nothing) or commit
5915: the changes, using @code{update}. Using @code{update} does not change
5916: the blocks file; it simply changes a block buffer's state to
5917: @i{assigned-dirty}.
5918:
5919: The word @code{flush} causes all @i{assigned-dirty} blocks to be
5920: written back to the blocks file on disk. Leaving Gforth using @code{bye}
5921: also causes a @code{flush} to be performed.
5922:
5923: In Gforth, @code{block} and @code{buffer} use a @i{direct-mapped}
5924: algorithm to assign a block buffer to a block. That means that any
5925: particular block can only be assigned to one specific block buffer,
5926: called (for the particular operation) the @i{victim buffer}. If the
5927: victim buffer is @i{unassigned} or @i{assigned-clean} it is allocated to
5928: the new block immediately. If it is @i{assigned-dirty} its current
5929: contents are written back to the blocks file on disk before it is
5930: allocated to the new block.
5931:
5932: Although no structure is imposed on the contents of a block, it is
5933: traditional to display the contents as 16 lines each of 64 characters. A
5934: block provides a single, continuous stream of input (for example, it
5935: acts as a single parse area) -- there are no end-of-line characters
5936: within a block, and no end-of-file character at the end of a
5937: block. There are two consequences of this:
5938:
5939: @itemize @bullet
5940: @item
5941: The last character of one line wraps straight into the first character
5942: of the following line
5943: @item
5944: The word @code{\} -- comment to end of line -- requires special
5945: treatment; in the context of a block it causes all characters until the
5946: end of the current 64-character ``line'' to be ignored.
5947: @end itemize
5948:
5949: In Gforth, when you use @code{block} with a non-existent block number,
5950: the current blocks file will be extended to the appropriate size and the
5951: block buffer will be initialised with spaces.
5952:
5953: Gforth includes a simple block editor (type @code{use blocked.fb 0 list}
5954: for details) but doesn't encourage the use of blocks; the mechanism is
5955: only provided for backward compatibility -- ANS Forth requires blocks to
5956: be available when files are.
5957:
5958: Common techniques that are used when working with blocks include:
5959:
5960: @itemize @bullet
5961: @item
5962: A screen editor that allows you to edit blocks without leaving the Forth
5963: environment.
5964: @item
5965: Shadow screens; where every code block has an associated block
5966: containing comments (for example: code in odd block numbers, comments in
5967: even block numbers). Typically, the block editor provides a convenient
5968: mechanism to toggle between code and comments.
5969: @item
5970: Load blocks; a single block (typically block 1) contains a number of
5971: @code{thru} commands which @code{load} the whole of the application.
5972: @end itemize
5973:
5974: See Frank Sergeant's Pygmy Forth to see just how well blocks can be
5975: integrated into a Forth programming environment.
5976:
5977: @comment TODO what about errors on open-blocks?
5978:
5979: doc-open-blocks
5980: doc-use
5981: doc-get-block-fid
5982: doc-block-position
5983:
5984: doc-scr
5985: doc-list
5986:
5987: doc---gforthman-block
5988: doc-buffer
5989:
5990: doc-update
5991: doc-updated?
5992: doc-save-buffers
5993: doc-empty-buffers
5994: doc-empty-buffer
5995: doc-flush
5996:
5997: doc-load
5998: doc-thru
5999: doc-+load
6000: doc-+thru
6001: doc---gforthman--->
6002: doc-block-included
6003:
6004:
6005: @c -------------------------------------------------------------
6006: @node Other I/O, Programming Tools, Blocks, Words
6007: @section Other I/O
6008: @cindex I/O - keyboard and display
6009:
6010: @menu
6011: * Simple numeric output:: Predefined formats
6012: * Formatted numeric output:: Formatted (pictured) output
6013: * String Formats:: How Forth stores strings in memory
6014: * Displaying characters and strings:: Other stuff
6015: * Input:: Input
6016: @end menu
6017:
6018: @node Simple numeric output, Formatted numeric output, Other I/O, Other I/O
6019: @subsection Simple numeric output
6020: @cindex numeric output - simple/free-format
6021:
6022: The simplest output functions are those that display numbers from the
6023: data or floating-point stacks. Floating-point output is always displayed
6024: using base 10. Numbers displayed from the data stack use the value stored
6025: in @code{base}.
6026:
6027:
6028: doc-.
6029: doc-dec.
6030: doc-hex.
6031: doc-u.
6032: doc-.r
6033: doc-u.r
6034: doc-d.
6035: doc-ud.
6036: doc-d.r
6037: doc-ud.r
6038: doc-f.
6039: doc-fe.
6040: doc-fs.
6041:
6042:
6043: Examples of printing the number 1234.5678E23 in the different floating-point output
6044: formats are shown below:
6045:
6046: @example
6047: f. 123456779999999000000000000.
6048: fe. 123.456779999999E24
6049: fs. 1.23456779999999E26
6050: @end example
6051:
6052:
6053: @node Formatted numeric output, String Formats, Simple numeric output, Other I/O
6054: @subsection Formatted numeric output
6055: @cindex formatted numeric output
6056: @cindex pictured numeric output
6057: @cindex numeric output - formatted
6058:
6059: Forth traditionally uses a technique called @dfn{pictured numeric
6060: output} for formatted printing of integers. In this technique, digits
6061: are extracted from the number (using the current output radix defined by
6062: @code{base}), converted to ASCII codes and appended to a string that is
6063: built in a scratch-pad area of memory (@pxref{core-idef,
6064: Implementation-defined options, Implementation-defined
6065: options}). Arbitrary characters can be appended to the string during the
6066: extraction process. The completed string is specified by an address
6067: and length and can be manipulated (@code{TYPE}ed, copied, modified)
6068: under program control.
6069:
6070: All of the words described in the previous section for simple numeric
6071: output are implemented in Gforth using pictured numeric output.
6072:
6073: Three important things to remember about pictured numeric output:
6074:
6075: @itemize @bullet
6076: @item
6077: It always operates on double-precision numbers; to display a
6078: single-precision number, convert it first (@pxref{Double precision} for
6079: ways of doing this).
6080: @item
6081: It always treats the double-precision number as though it were
6082: unsigned. The examples below show ways of printing signed numbers.
6083: @item
6084: The string is built up from right to left; least significant digit first.
6085: @end itemize
6086:
6087:
6088: doc-<#
6089: doc-<<#
6090: doc-#
6091: doc-#s
6092: doc-hold
6093: doc-sign
6094: doc-#>
6095: doc-#>>
6096:
6097: doc-represent
6098:
6099:
6100: @noindent
6101: Here are some examples of using pictured numeric output:
6102:
6103: @example
6104: : my-u. ( u -- )
6105: \ Simplest use of pns.. behaves like Standard u.
6106: 0 \ convert to unsigned double
6107: <# \ start conversion
6108: #s \ convert all digits
6109: #> \ complete conversion
6110: TYPE SPACE ; \ display, with trailing space
6111:
6112: : cents-only ( u -- )
6113: 0 \ convert to unsigned double
6114: <# \ start conversion
6115: # # \ convert two least-significant digits
6116: #> \ complete conversion, discard other digits
6117: TYPE SPACE ; \ display, with trailing space
6118:
6119: : dollars-and-cents ( u -- )
6120: 0 \ convert to unsigned double
6121: <# \ start conversion
6122: # # \ convert two least-significant digits
6123: [char] . hold \ insert decimal point
6124: #s \ convert remaining digits
6125: [char] $ hold \ append currency symbol
6126: #> \ complete conversion
6127: TYPE SPACE ; \ display, with trailing space
6128:
6129: : my-. ( n -- )
6130: \ handling negatives.. behaves like Standard .
6131: s>d \ convert to signed double
6132: swap over dabs \ leave sign byte followed by unsigned double
6133: <# \ start conversion
6134: #s \ convert all digits
6135: rot sign \ get at sign byte, append "-" if needed
6136: #> \ complete conversion
6137: TYPE SPACE ; \ display, with trailing space
6138:
6139: : account. ( n -- )
6140: \ accountants don't like minus signs, they use braces
6141: \ for negative numbers
6142: s>d \ convert to signed double
6143: swap over dabs \ leave sign byte followed by unsigned double
6144: <# \ start conversion
6145: 2 pick \ get copy of sign byte
6146: 0< IF [char] ) hold THEN \ right-most character of output
6147: #s \ convert all digits
6148: rot \ get at sign byte
6149: 0< IF [char] ( hold THEN
6150: #> \ complete conversion
6151: TYPE SPACE ; \ display, with trailing space
6152: @end example
6153:
6154: Here are some examples of using these words:
6155:
6156: @example
6157: 1 my-u. 1
6158: hex -1 my-u. decimal FFFFFFFF
6159: 1 cents-only 01
6160: 1234 cents-only 34
6161: 2 dollars-and-cents $0.02
6162: 1234 dollars-and-cents $12.34
6163: 123 my-. 123
6164: -123 my. -123
6165: 123 account. 123
6166: -456 account. (456)
6167: @end example
6168:
6169:
6170: @node String Formats, Displaying characters and strings, Formatted numeric output, Other I/O
6171: @subsection String Formats
6172: @cindex strings - see character strings
6173: @cindex character strings - formats
6174: @cindex I/O - see character strings
6175:
6176: Forth commonly uses two different methods for representing character
6177: strings:
6178:
6179: @itemize @bullet
6180: @item
6181: @cindex address of counted string
6182: @cindex counted string
6183: As a @dfn{counted string}, represented by a @i{c-addr}. The char
6184: addressed by @i{c-addr} contains a character-count, @i{n}, of the
6185: string and the string occupies the subsequent @i{n} char addresses in
6186: memory.
6187: @item
6188: As cell pair on the stack; @i{c-addr u}, where @i{u} is the length
6189: of the string in characters, and @i{c-addr} is the address of the
6190: first byte of the string.
6191: @end itemize
6192:
6193: ANS Forth encourages the use of the second format when representing
6194: strings on the stack, whilst conceeding that the counted string format
6195: remains useful as a way of storing strings in memory.
6196:
6197:
6198: doc-count
6199:
6200:
6201: @xref{Memory Blocks} for words that move, copy and search
6202: for strings. @xref{Displaying characters and strings,} for words that
6203: display characters and strings.
6204:
6205:
6206: @node Displaying characters and strings, Input, String Formats, Other I/O
6207: @subsection Displaying characters and strings
6208: @cindex characters - compiling and displaying
6209: @cindex character strings - compiling and displaying
6210:
6211: This section starts with a glossary of Forth words and ends with a set
6212: of examples.
6213:
6214:
6215: doc-bl
6216: doc-space
6217: doc-spaces
6218: doc-emit
6219: doc-toupper
6220: doc-."
6221: doc-.(
6222: doc-type
6223: doc-typewhite
6224: doc-cr
6225: @cindex cursor control
6226: doc-at-xy
6227: doc-page
6228: doc-s"
6229: doc-c"
6230: doc-char
6231: doc-[char]
6232: doc-sliteral
6233:
6234:
6235: @noindent
6236: As an example, consider the following text, stored in a file @file{test.fs}:
6237:
6238: @example
6239: .( text-1)
6240: : my-word
6241: ." text-2" cr
6242: .( text-3)
6243: ;
6244:
6245: ." text-4"
6246:
6247: : my-char
6248: [char] ALPHABET emit
6249: char emit
6250: ;
6251: @end example
6252:
6253: When you load this code into Gforth, the following output is generated:
6254:
6255: @example
6256: @kbd{include test.fs @key{RET}} text-1text-3text-4 ok
6257: @end example
6258:
6259: @itemize @bullet
6260: @item
6261: Messages @code{text-1} and @code{text-3} are displayed because @code{.(}
6262: is an immediate word; it behaves in the same way whether it is used inside
6263: or outside a colon definition.
6264: @item
6265: Message @code{text-4} is displayed because of Gforth's added interpretation
6266: semantics for @code{."}.
6267: @item
6268: Message @code{text-2} is @i{not} displayed, because the text interpreter
6269: performs the compilation semantics for @code{."} within the definition of
6270: @code{my-word}.
6271: @end itemize
6272:
6273: Here are some examples of executing @code{my-word} and @code{my-char}:
6274:
6275: @example
6276: @kbd{my-word @key{RET}} text-2
6277: ok
6278: @kbd{my-char fred @key{RET}} Af ok
6279: @kbd{my-char jim @key{RET}} Aj ok
6280: @end example
6281:
6282: @itemize @bullet
6283: @item
6284: Message @code{text-2} is displayed because of the run-time behaviour of
6285: @code{."}.
6286: @item
6287: @code{[char]} compiles the ``A'' from ``ALPHABET'' and puts its display code
6288: on the stack at run-time. @code{emit} always displays the character
6289: when @code{my-char} is executed.
6290: @item
6291: @code{char} parses a string at run-time and the second @code{emit} displays
6292: the first character of the string.
6293: @item
6294: If you type @code{see my-char} you can see that @code{[char]} discarded
6295: the text ``LPHABET'' and only compiled the display code for ``A'' into the
6296: definition of @code{my-char}.
6297: @end itemize
6298:
6299:
6300:
6301: @node Input, , Displaying characters and strings, Other I/O
6302: @subsection Input
6303: @cindex input
6304: @cindex I/O - see input
6305: @cindex parsing a string
6306:
6307: @xref{String Formats} for ways of storing character strings in memory.
6308:
6309: @comment TODO examples for >number >float accept key key? pad parse word refill
6310: @comment then index them
6311:
6312:
6313: doc-key
6314: doc-key?
6315: doc-ekey
6316: doc-ekey?
6317: doc-ekey>char
6318: doc->number
6319: doc->float
6320: doc-accept
6321: doc-pad
6322: doc-parse
6323: doc-word
6324: doc-sword
6325: doc-(name)
6326: doc-refill
6327: @comment obsolescent words..
6328: doc-convert
6329: doc-query
6330: doc-expect
6331: doc-span
6332:
6333:
6334:
6335: @c -------------------------------------------------------------
6336: @node Programming Tools, Assembler and Code Words, Other I/O, Words
6337: @section Programming Tools
6338: @cindex programming tools
6339:
6340: @menu
6341: * Debugging:: Simple and quick.
6342: * Assertions:: Making your programs self-checking.
6343: * Singlestep Debugger:: Executing your program word by word.
6344: @end menu
6345:
6346: @node Debugging, Assertions, Programming Tools, Programming Tools
6347: @subsection Debugging
6348: @cindex debugging
6349:
6350: Languages with a slow edit/compile/link/test development loop tend to
6351: require sophisticated tracing/stepping debuggers to facilate
6352: productive debugging.
6353:
6354: A much better (faster) way in fast-compiling languages is to add
6355: printing code at well-selected places, let the program run, look at
6356: the output, see where things went wrong, add more printing code, etc.,
6357: until the bug is found.
6358:
6359: The simple debugging aids provided in @file{debugs.fs}
6360: are meant to support this style of debugging. In addition, there are
6361: words for non-destructively inspecting the stack and memory:
6362:
6363:
6364: doc-.s
6365: doc-f.s
6366:
6367:
6368: There is a word @code{.r} but it does @i{not} display the return
6369: stack! It is used for formatted numeric output.
6370:
6371:
6372: doc-depth
6373: doc-fdepth
6374: doc-clearstack
6375: doc-?
6376: doc-dump
6377:
6378:
6379: The word @code{~~} prints debugging information (by default the source
6380: location and the stack contents). It is easy to insert. If you use Emacs
6381: it is also easy to remove (@kbd{C-x ~} in the Emacs Forth mode to
6382: query-replace them with nothing). The deferred words
6383: @code{printdebugdata} and @code{printdebugline} control the output of
6384: @code{~~}. The default source location output format works well with
6385: Emacs' compilation mode, so you can step through the program at the
6386: source level using @kbd{C-x `} (the advantage over a stepping debugger
6387: is that you can step in any direction and you know where the crash has
6388: happened or where the strange data has occurred).
6389:
6390: The default actions of @code{~~} clobber the contents of the pictured
6391: numeric output string, so you should not use @code{~~}, e.g., between
6392: @code{<#} and @code{#>}.
6393:
6394:
6395: doc-~~
6396: doc-printdebugdata
6397: doc-printdebugline
6398:
6399: doc-see
6400: doc-marker
6401:
6402:
6403: Here's an example of using @code{marker} at the start of a source file
6404: that you are debugging; it ensures that you only ever have one copy of
6405: the file's definitions compiled at any time:
6406:
6407: @example
6408: [IFDEF] my-code
6409: my-code
6410: [ENDIF]
6411:
6412: marker my-code
6413: init-included-files
6414:
6415: \ .. definitions start here
6416: \ .
6417: \ .
6418: \ end
6419: @end example
6420:
6421:
6422:
6423: @node Assertions, Singlestep Debugger, Debugging, Programming Tools
6424: @subsection Assertions
6425: @cindex assertions
6426:
6427: It is a good idea to make your programs self-checking, especially if you
6428: make an assumption that may become invalid during maintenance (for
6429: example, that a certain field of a data structure is never zero). Gforth
6430: supports @dfn{assertions} for this purpose. They are used like this:
6431:
6432: @example
6433: assert( @i{flag} )
6434: @end example
6435:
6436: The code between @code{assert(} and @code{)} should compute a flag, that
6437: should be true if everything is alright and false otherwise. It should
6438: not change anything else on the stack. The overall stack effect of the
6439: assertion is @code{( -- )}. E.g.
6440:
6441: @example
6442: assert( 1 1 + 2 = ) \ what we learn in school
6443: assert( dup 0<> ) \ assert that the top of stack is not zero
6444: assert( false ) \ this code should not be reached
6445: @end example
6446:
6447: The need for assertions is different at different times. During
6448: debugging, we want more checking, in production we sometimes care more
6449: for speed. Therefore, assertions can be turned off, i.e., the assertion
6450: becomes a comment. Depending on the importance of an assertion and the
6451: time it takes to check it, you may want to turn off some assertions and
6452: keep others turned on. Gforth provides several levels of assertions for
6453: this purpose:
6454:
6455:
6456: doc-assert0(
6457: doc-assert1(
6458: doc-assert2(
6459: doc-assert3(
6460: doc-assert(
6461: doc-)
6462:
6463:
6464: The variable @code{assert-level} specifies the highest assertions that
6465: are turned on. I.e., at the default @code{assert-level} of one,
6466: @code{assert0(} and @code{assert1(} assertions perform checking, while
6467: @code{assert2(} and @code{assert3(} assertions are treated as comments.
6468:
6469: The value of @code{assert-level} is evaluated at compile-time, not at
6470: run-time. Therefore you cannot turn assertions on or off at run-time;
6471: you have to set the @code{assert-level} appropriately before compiling a
6472: piece of code. You can compile different pieces of code at different
6473: @code{assert-level}s (e.g., a trusted library at level 1 and
6474: newly-written code at level 3).
6475:
6476:
6477: doc-assert-level
6478:
6479:
6480: If an assertion fails, a message compatible with Emacs' compilation mode
6481: is produced and the execution is aborted (currently with @code{ABORT"}.
6482: If there is interest, we will introduce a special throw code. But if you
6483: intend to @code{catch} a specific condition, using @code{throw} is
6484: probably more appropriate than an assertion).
6485:
6486: Definitions in ANS Forth for these assertion words are provided
6487: in @file{compat/assert.fs}.
6488:
6489:
6490: @node Singlestep Debugger, , Assertions, Programming Tools
6491: @subsection Singlestep Debugger
6492: @cindex singlestep Debugger
6493: @cindex debugging Singlestep
6494:
6495: When you create a new word there's often the need to check whether it
6496: behaves correctly or not. You can do this by typing @code{dbg
6497: badword}. A debug session might look like this:
6498:
6499: @example
6500: : badword 0 DO i . LOOP ; ok
6501: 2 dbg badword
6502: : badword
6503: Scanning code...
6504:
6505: Nesting debugger ready!
6506:
6507: 400D4738 8049BC4 0 -> [ 2 ] 00002 00000
6508: 400D4740 8049F68 DO -> [ 0 ]
6509: 400D4744 804A0C8 i -> [ 1 ] 00000
6510: 400D4748 400C5E60 . -> 0 [ 0 ]
6511: 400D474C 8049D0C LOOP -> [ 0 ]
6512: 400D4744 804A0C8 i -> [ 1 ] 00001
6513: 400D4748 400C5E60 . -> 1 [ 0 ]
6514: 400D474C 8049D0C LOOP -> [ 0 ]
6515: 400D4758 804B384 ; -> ok
6516: @end example
6517:
6518: Each line displayed is one step. You always have to hit return to
6519: execute the next word that is displayed. If you don't want to execute
6520: the next word in a whole, you have to type @kbd{n} for @code{nest}. Here is
6521: an overview what keys are available:
6522:
6523: @table @i
6524:
6525: @item @key{RET}
6526: Next; Execute the next word.
6527:
6528: @item n
6529: Nest; Single step through next word.
6530:
6531: @item u
6532: Unnest; Stop debugging and execute rest of word. If we got to this word
6533: with nest, continue debugging with the calling word.
6534:
6535: @item d
6536: Done; Stop debugging and execute rest.
6537:
6538: @item s
6539: Stop; Abort immediately.
6540:
6541: @end table
6542:
6543: Debugging large application with this mechanism is very difficult, because
6544: you have to nest very deeply into the program before the interesting part
6545: begins. This takes a lot of time.
6546:
6547: To do it more directly put a @code{BREAK:} command into your source code.
6548: When program execution reaches @code{BREAK:} the single step debugger is
6549: invoked and you have all the features described above.
6550:
6551: If you have more than one part to debug it is useful to know where the
6552: program has stopped at the moment. You can do this by the
6553: @code{BREAK" string"} command. This behaves like @code{BREAK:} except that
6554: string is typed out when the ``breakpoint'' is reached.
6555:
6556:
6557: doc-dbg
6558: doc-break:
6559: doc-break"
6560:
6561:
6562:
6563: @c -------------------------------------------------------------
6564: @node Assembler and Code Words, Threading Words, Programming Tools, Words
6565: @section Assembler and Code Words
6566: @cindex assembler
6567: @cindex code words
6568:
6569: Gforth provides some words for defining primitives (words written in
6570: machine code), and for defining the machine-code equivalent of
6571: @code{DOES>}-based defining words. However, the machine-independent
6572: nature of Gforth poses a few problems: First of all, Gforth runs on
6573: several architectures, so it can provide no standard assembler. What's
6574: worse is that the register allocation not only depends on the processor,
6575: but also on the @code{gcc} version and options used.
6576:
6577: The words that Gforth offers encapsulate some system dependences (e.g.,
6578: the header structure), so a system-independent assembler may be used in
6579: Gforth. If you do not have an assembler, you can compile machine code
6580: directly with @code{,} and @code{c,}@footnote{This isn't portable,
6581: because these words emit stuff in @i{data} space; it works because
6582: Gforth has unified code/data spaces. Assembler isn't likely to be
6583: portable anyway.}.
6584:
6585:
6586: doc-assembler
6587: doc-init-asm
6588: doc-code
6589: doc-end-code
6590: doc-;code
6591: doc-flush-icache
6592:
6593:
6594: If @code{flush-icache} does not work correctly, @code{code} words
6595: etc. will not work (reliably), either.
6596:
6597: The typical usage of these @code{code} words can be shown most easily by
6598: analogy to the equivalent high-level defining words:
6599:
6600: @example
6601: : foo code foo
6602: <high-level Forth words> <assembler>
6603: ; end-code
6604:
6605: : bar : bar
6606: <high-level Forth words> <high-level Forth words>
6607: CREATE CREATE
6608: <high-level Forth words> <high-level Forth words>
6609: DOES> ;code
6610: <high-level Forth words> <assembler>
6611: ; end-code
6612: @end example
6613:
6614: @code{flush-icache} is always present. The other words are rarely used
6615: and reside in @code{code.fs}, which is usually not loaded. You can load
6616: it with @code{require code.fs}.
6617:
6618: @cindex registers of the inner interpreter
6619: In the assembly code you will want to refer to the inner interpreter's
6620: registers (e.g., the data stack pointer) and you may want to use other
6621: registers for temporary storage. Unfortunately, the register allocation
6622: is installation-dependent.
6623:
6624: The easiest solution is to use explicit register declarations
6625: (@pxref{Explicit Reg Vars, , Variables in Specified Registers, gcc.info,
6626: GNU C Manual}) for all of the inner interpreter's registers: You have to
6627: compile Gforth with @code{-DFORCE_REG} (configure option
6628: @code{--enable-force-reg}) and the appropriate declarations must be
6629: present in the @code{machine.h} file (see @code{mips.h} for an example;
6630: you can find a full list of all declarable register symbols with
6631: @code{grep register engine.c}). If you give explicit registers to all
6632: variables that are declared at the beginning of @code{engine()}, you
6633: should be able to use the other caller-saved registers for temporary
6634: storage. Alternatively, you can use the @code{gcc} option
6635: @code{-ffixed-REG} (@pxref{Code Gen Options, , Options for Code
6636: Generation Conventions, gcc.info, GNU C Manual}) to reserve a register
6637: (however, this restriction on register allocation may slow Gforth
6638: significantly).
6639:
6640: If this solution is not viable (e.g., because @code{gcc} does not allow
6641: you to explicitly declare all the registers you need), you have to find
6642: out by looking at the code where the inner interpreter's registers
6643: reside and which registers can be used for temporary storage. You can
6644: get an assembly listing of the engine's code with @code{make engine.s}.
6645:
6646: In any case, it is good practice to abstract your assembly code from the
6647: actual register allocation. E.g., if the data stack pointer resides in
6648: register @code{$17}, create an alias for this register called @code{sp},
6649: and use that in your assembly code.
6650:
6651: @cindex code words, portable
6652: Another option for implementing normal and defining words efficiently
6653: is to add the desired functionality to the source of Gforth. For normal
6654: words you just have to edit @file{primitives} (@pxref{Automatic
6655: Generation}). Defining words (equivalent to @code{;CODE} words, for fast
6656: defined words) may require changes in @file{engine.c}, @file{kernel.fs},
6657: @file{prims2x.fs}, and possibly @file{cross.fs}.
6658:
6659:
6660: @c -------------------------------------------------------------
6661: @node Threading Words, Locals, Assembler and Code Words, Words
6662: @section Threading Words
6663: @cindex threading words
6664:
6665: @cindex code address
6666: These words provide access to code addresses and other threading stuff
6667: in Gforth (and, possibly, other interpretive Forths). It more or less
6668: abstracts away the differences between direct and indirect threading
6669: (and, for direct threading, the machine dependences). However, at
6670: present this wordset is still incomplete. It is also pretty low-level;
6671: some day it will hopefully be made unnecessary by an internals wordset
6672: that abstracts implementation details away completely.
6673:
6674:
6675: doc-threading-method
6676: doc->code-address
6677: doc->does-code
6678: doc-code-address!
6679: doc-does-code!
6680: doc-does-handler!
6681: doc-/does-handler
6682:
6683:
6684: The code addresses produced by various defining words are produced by
6685: the following words:
6686:
6687:
6688: doc-docol:
6689: doc-docon:
6690: doc-dovar:
6691: doc-douser:
6692: doc-dodefer:
6693: doc-dofield:
6694:
6695:
6696: You can recognize words defined by a @code{CREATE}...@code{DOES>} word
6697: with @code{>does-code}. If the word was defined in that way, the value
6698: returned is non-zero and identifies the @code{DOES>} used by the
6699: defining word.
6700: @comment TODO should that be ``identifies the xt of the DOES> ??''
6701:
6702: @c -------------------------------------------------------------
6703: @node Locals, Structures, Threading Words, Words
6704: @section Locals
6705: @cindex locals
6706:
6707: Local variables can make Forth programming more enjoyable and Forth
6708: programs easier to read. Unfortunately, the locals of ANS Forth are
6709: laden with restrictions. Therefore, we provide not only the ANS Forth
6710: locals wordset, but also our own, more powerful locals wordset (we
6711: implemented the ANS Forth locals wordset through our locals wordset).
6712:
6713: The ideas in this section have also been published in the paper
6714: @cite{Automatic Scoping of Local Variables} by M. Anton Ertl, presented
6715: at EuroForth '94; it is available at
6716: @*@uref{http://www.complang.tuwien.ac.at/papers/ertl94l.ps.gz}.
6717:
6718: @menu
6719: * Gforth locals::
6720: * ANS Forth locals::
6721: @end menu
6722:
6723: @node Gforth locals, ANS Forth locals, Locals, Locals
6724: @subsection Gforth locals
6725: @cindex Gforth locals
6726: @cindex locals, Gforth style
6727:
6728: Locals can be defined with
6729:
6730: @example
6731: @{ local1 local2 ... -- comment @}
6732: @end example
6733: or
6734: @example
6735: @{ local1 local2 ... @}
6736: @end example
6737:
6738: E.g.,
6739: @example
6740: : max @{ n1 n2 -- n3 @}
6741: n1 n2 > if
6742: n1
6743: else
6744: n2
6745: endif ;
6746: @end example
6747:
6748: The similarity of locals definitions with stack comments is intended. A
6749: locals definition often replaces the stack comment of a word. The order
6750: of the locals corresponds to the order in a stack comment and everything
6751: after the @code{--} is really a comment.
6752:
6753: This similarity has one disadvantage: It is too easy to confuse locals
6754: declarations with stack comments, causing bugs and making them hard to
6755: find. However, this problem can be avoided by appropriate coding
6756: conventions: Do not use both notations in the same program. If you do,
6757: they should be distinguished using additional means, e.g. by position.
6758:
6759: @cindex types of locals
6760: @cindex locals types
6761: The name of the local may be preceded by a type specifier, e.g.,
6762: @code{F:} for a floating point value:
6763:
6764: @example
6765: : CX* @{ F: Ar F: Ai F: Br F: Bi -- Cr Ci @}
6766: \ complex multiplication
6767: Ar Br f* Ai Bi f* f-
6768: Ar Bi f* Ai Br f* f+ ;
6769: @end example
6770:
6771: @cindex flavours of locals
6772: @cindex locals flavours
6773: @cindex value-flavoured locals
6774: @cindex variable-flavoured locals
6775: Gforth currently supports cells (@code{W:}, @code{W^}), doubles
6776: (@code{D:}, @code{D^}), floats (@code{F:}, @code{F^}) and characters
6777: (@code{C:}, @code{C^}) in two flavours: a value-flavoured local (defined
6778: with @code{W:}, @code{D:} etc.) produces its value and can be changed
6779: with @code{TO}. A variable-flavoured local (defined with @code{W^} etc.)
6780: produces its address (which becomes invalid when the variable's scope is
6781: left). E.g., the standard word @code{emit} can be defined in terms of
6782: @code{type} like this:
6783:
6784: @example
6785: : emit @{ C^ char* -- @}
6786: char* 1 type ;
6787: @end example
6788:
6789: @cindex default type of locals
6790: @cindex locals, default type
6791: A local without type specifier is a @code{W:} local. Both flavours of
6792: locals are initialized with values from the data or FP stack.
6793:
6794: Currently there is no way to define locals with user-defined data
6795: structures, but we are working on it.
6796:
6797: Gforth allows defining locals everywhere in a colon definition. This
6798: poses the following questions:
6799:
6800: @menu
6801: * Where are locals visible by name?::
6802: * How long do locals live?::
6803: * Programming Style::
6804: * Implementation::
6805: @end menu
6806:
6807: @node Where are locals visible by name?, How long do locals live?, Gforth locals, Gforth locals
6808: @subsubsection Where are locals visible by name?
6809: @cindex locals visibility
6810: @cindex visibility of locals
6811: @cindex scope of locals
6812:
6813: Basically, the answer is that locals are visible where you would expect
6814: it in block-structured languages, and sometimes a little longer. If you
6815: want to restrict the scope of a local, enclose its definition in
6816: @code{SCOPE}...@code{ENDSCOPE}.
6817:
6818:
6819: doc-scope
6820: doc-endscope
6821:
6822:
6823: These words behave like control structure words, so you can use them
6824: with @code{CS-PICK} and @code{CS-ROLL} to restrict the scope in
6825: arbitrary ways.
6826:
6827: If you want a more exact answer to the visibility question, here's the
6828: basic principle: A local is visible in all places that can only be
6829: reached through the definition of the local@footnote{In compiler
6830: construction terminology, all places dominated by the definition of the
6831: local.}. In other words, it is not visible in places that can be reached
6832: without going through the definition of the local. E.g., locals defined
6833: in @code{IF}...@code{ENDIF} are visible until the @code{ENDIF}, locals
6834: defined in @code{BEGIN}...@code{UNTIL} are visible after the
6835: @code{UNTIL} (until, e.g., a subsequent @code{ENDSCOPE}).
6836:
6837: The reasoning behind this solution is: We want to have the locals
6838: visible as long as it is meaningful. The user can always make the
6839: visibility shorter by using explicit scoping. In a place that can
6840: only be reached through the definition of a local, the meaning of a
6841: local name is clear. In other places it is not: How is the local
6842: initialized at the control flow path that does not contain the
6843: definition? Which local is meant, if the same name is defined twice in
6844: two independent control flow paths?
6845:
6846: This should be enough detail for nearly all users, so you can skip the
6847: rest of this section. If you really must know all the gory details and
6848: options, read on.
6849:
6850: In order to implement this rule, the compiler has to know which places
6851: are unreachable. It knows this automatically after @code{AHEAD},
6852: @code{AGAIN}, @code{EXIT} and @code{LEAVE}; in other cases (e.g., after
6853: most @code{THROW}s), you can use the word @code{UNREACHABLE} to tell the
6854: compiler that the control flow never reaches that place. If
6855: @code{UNREACHABLE} is not used where it could, the only consequence is
6856: that the visibility of some locals is more limited than the rule above
6857: says. If @code{UNREACHABLE} is used where it should not (i.e., if you
6858: lie to the compiler), buggy code will be produced.
6859:
6860:
6861: doc-unreachable
6862:
6863:
6864: Another problem with this rule is that at @code{BEGIN}, the compiler
6865: does not know which locals will be visible on the incoming
6866: back-edge. All problems discussed in the following are due to this
6867: ignorance of the compiler (we discuss the problems using @code{BEGIN}
6868: loops as examples; the discussion also applies to @code{?DO} and other
6869: loops). Perhaps the most insidious example is:
6870: @example
6871: AHEAD
6872: BEGIN
6873: x
6874: [ 1 CS-ROLL ] THEN
6875: @{ x @}
6876: ...
6877: UNTIL
6878: @end example
6879:
6880: This should be legal according to the visibility rule. The use of
6881: @code{x} can only be reached through the definition; but that appears
6882: textually below the use.
6883:
6884: From this example it is clear that the visibility rules cannot be fully
6885: implemented without major headaches. Our implementation treats common
6886: cases as advertised and the exceptions are treated in a safe way: The
6887: compiler makes a reasonable guess about the locals visible after a
6888: @code{BEGIN}; if it is too pessimistic, the
6889: user will get a spurious error about the local not being defined; if the
6890: compiler is too optimistic, it will notice this later and issue a
6891: warning. In the case above the compiler would complain about @code{x}
6892: being undefined at its use. You can see from the obscure examples in
6893: this section that it takes quite unusual control structures to get the
6894: compiler into trouble, and even then it will often do fine.
6895:
6896: If the @code{BEGIN} is reachable from above, the most optimistic guess
6897: is that all locals visible before the @code{BEGIN} will also be
6898: visible after the @code{BEGIN}. This guess is valid for all loops that
6899: are entered only through the @code{BEGIN}, in particular, for normal
6900: @code{BEGIN}...@code{WHILE}...@code{REPEAT} and
6901: @code{BEGIN}...@code{UNTIL} loops and it is implemented in our
6902: compiler. When the branch to the @code{BEGIN} is finally generated by
6903: @code{AGAIN} or @code{UNTIL}, the compiler checks the guess and
6904: warns the user if it was too optimistic:
6905: @example
6906: IF
6907: @{ x @}
6908: BEGIN
6909: \ x ?
6910: [ 1 cs-roll ] THEN
6911: ...
6912: UNTIL
6913: @end example
6914:
6915: Here, @code{x} lives only until the @code{BEGIN}, but the compiler
6916: optimistically assumes that it lives until the @code{THEN}. It notices
6917: this difference when it compiles the @code{UNTIL} and issues a
6918: warning. The user can avoid the warning, and make sure that @code{x}
6919: is not used in the wrong area by using explicit scoping:
6920: @example
6921: IF
6922: SCOPE
6923: @{ x @}
6924: ENDSCOPE
6925: BEGIN
6926: [ 1 cs-roll ] THEN
6927: ...
6928: UNTIL
6929: @end example
6930:
6931: Since the guess is optimistic, there will be no spurious error messages
6932: about undefined locals.
6933:
6934: If the @code{BEGIN} is not reachable from above (e.g., after
6935: @code{AHEAD} or @code{EXIT}), the compiler cannot even make an
6936: optimistic guess, as the locals visible after the @code{BEGIN} may be
6937: defined later. Therefore, the compiler assumes that no locals are
6938: visible after the @code{BEGIN}. However, the user can use
6939: @code{ASSUME-LIVE} to make the compiler assume that the same locals are
6940: visible at the BEGIN as at the point where the top control-flow stack
6941: item was created.
6942:
6943:
6944: doc-assume-live
6945:
6946:
6947: @noindent
6948: E.g.,
6949: @example
6950: @{ x @}
6951: AHEAD
6952: ASSUME-LIVE
6953: BEGIN
6954: x
6955: [ 1 CS-ROLL ] THEN
6956: ...
6957: UNTIL
6958: @end example
6959:
6960: Other cases where the locals are defined before the @code{BEGIN} can be
6961: handled by inserting an appropriate @code{CS-ROLL} before the
6962: @code{ASSUME-LIVE} (and changing the control-flow stack manipulation
6963: behind the @code{ASSUME-LIVE}).
6964:
6965: Cases where locals are defined after the @code{BEGIN} (but should be
6966: visible immediately after the @code{BEGIN}) can only be handled by
6967: rearranging the loop. E.g., the ``most insidious'' example above can be
6968: arranged into:
6969: @example
6970: BEGIN
6971: @{ x @}
6972: ... 0=
6973: WHILE
6974: x
6975: REPEAT
6976: @end example
6977:
6978: @node How long do locals live?, Programming Style, Where are locals visible by name?, Gforth locals
6979: @subsubsection How long do locals live?
6980: @cindex locals lifetime
6981: @cindex lifetime of locals
6982:
6983: The right answer for the lifetime question would be: A local lives at
6984: least as long as it can be accessed. For a value-flavoured local this
6985: means: until the end of its visibility. However, a variable-flavoured
6986: local could be accessed through its address far beyond its visibility
6987: scope. Ultimately, this would mean that such locals would have to be
6988: garbage collected. Since this entails un-Forth-like implementation
6989: complexities, I adopted the same cowardly solution as some other
6990: languages (e.g., C): The local lives only as long as it is visible;
6991: afterwards its address is invalid (and programs that access it
6992: afterwards are erroneous).
6993:
6994: @node Programming Style, Implementation, How long do locals live?, Gforth locals
6995: @subsubsection Programming Style
6996: @cindex locals programming style
6997: @cindex programming style, locals
6998:
6999: The freedom to define locals anywhere has the potential to change
7000: programming styles dramatically. In particular, the need to use the
7001: return stack for intermediate storage vanishes. Moreover, all stack
7002: manipulations (except @code{PICK}s and @code{ROLL}s with run-time
7003: determined arguments) can be eliminated: If the stack items are in the
7004: wrong order, just write a locals definition for all of them; then
7005: write the items in the order you want.
7006:
7007: This seems a little far-fetched and eliminating stack manipulations is
7008: unlikely to become a conscious programming objective. Still, the number
7009: of stack manipulations will be reduced dramatically if local variables
7010: are used liberally (e.g., compare @code{max} in @ref{Gforth locals} with
7011: a traditional implementation of @code{max}).
7012:
7013: This shows one potential benefit of locals: making Forth programs more
7014: readable. Of course, this benefit will only be realized if the
7015: programmers continue to honour the principle of factoring instead of
7016: using the added latitude to make the words longer.
7017:
7018: @cindex single-assignment style for locals
7019: Using @code{TO} can and should be avoided. Without @code{TO},
7020: every value-flavoured local has only a single assignment and many
7021: advantages of functional languages apply to Forth. I.e., programs are
7022: easier to analyse, to optimize and to read: It is clear from the
7023: definition what the local stands for, it does not turn into something
7024: different later.
7025:
7026: E.g., a definition using @code{TO} might look like this:
7027: @example
7028: : strcmp @{ addr1 u1 addr2 u2 -- n @}
7029: u1 u2 min 0
7030: ?do
7031: addr1 c@@ addr2 c@@ -
7032: ?dup-if
7033: unloop exit
7034: then
7035: addr1 char+ TO addr1
7036: addr2 char+ TO addr2
7037: loop
7038: u1 u2 - ;
7039: @end example
7040: Here, @code{TO} is used to update @code{addr1} and @code{addr2} at
7041: every loop iteration. @code{strcmp} is a typical example of the
7042: readability problems of using @code{TO}. When you start reading
7043: @code{strcmp}, you think that @code{addr1} refers to the start of the
7044: string. Only near the end of the loop you realize that it is something
7045: else.
7046:
7047: This can be avoided by defining two locals at the start of the loop that
7048: are initialized with the right value for the current iteration.
7049: @example
7050: : strcmp @{ addr1 u1 addr2 u2 -- n @}
7051: addr1 addr2
7052: u1 u2 min 0
7053: ?do @{ s1 s2 @}
7054: s1 c@@ s2 c@@ -
7055: ?dup-if
7056: unloop exit
7057: then
7058: s1 char+ s2 char+
7059: loop
7060: 2drop
7061: u1 u2 - ;
7062: @end example
7063: Here it is clear from the start that @code{s1} has a different value
7064: in every loop iteration.
7065:
7066: @node Implementation, , Programming Style, Gforth locals
7067: @subsubsection Implementation
7068: @cindex locals implementation
7069: @cindex implementation of locals
7070:
7071: @cindex locals stack
7072: Gforth uses an extra locals stack. The most compelling reason for
7073: this is that the return stack is not float-aligned; using an extra stack
7074: also eliminates the problems and restrictions of using the return stack
7075: as locals stack. Like the other stacks, the locals stack grows toward
7076: lower addresses. A few primitives allow an efficient implementation:
7077:
7078:
7079: doc-@local#
7080: doc-f@local#
7081: doc-laddr#
7082: doc-lp+!#
7083: doc-lp!
7084: doc->l
7085: doc-f>l
7086:
7087:
7088: In addition to these primitives, some specializations of these
7089: primitives for commonly occurring inline arguments are provided for
7090: efficiency reasons, e.g., @code{@@local0} as specialization of
7091: @code{@@local#} for the inline argument 0. The following compiling words
7092: compile the right specialized version, or the general version, as
7093: appropriate:
7094:
7095:
7096: doc-compile-@local
7097: doc-compile-f@local
7098: doc-compile-lp+!
7099:
7100:
7101: Combinations of conditional branches and @code{lp+!#} like
7102: @code{?branch-lp+!#} (the locals pointer is only changed if the branch
7103: is taken) are provided for efficiency and correctness in loops.
7104:
7105: A special area in the dictionary space is reserved for keeping the
7106: local variable names. @code{@{} switches the dictionary pointer to this
7107: area and @code{@}} switches it back and generates the locals
7108: initializing code. @code{W:} etc.@ are normal defining words. This
7109: special area is cleared at the start of every colon definition.
7110:
7111: @cindex word list for defining locals
7112: A special feature of Gforth's dictionary is used to implement the
7113: definition of locals without type specifiers: every word list (aka
7114: vocabulary) has its own methods for searching
7115: etc. (@pxref{Word Lists}). For the present purpose we defined a word list
7116: with a special search method: When it is searched for a word, it
7117: actually creates that word using @code{W:}. @code{@{} changes the search
7118: order to first search the word list containing @code{@}}, @code{W:} etc.,
7119: and then the word list for defining locals without type specifiers.
7120:
7121: The lifetime rules support a stack discipline within a colon
7122: definition: The lifetime of a local is either nested with other locals
7123: lifetimes or it does not overlap them.
7124:
7125: At @code{BEGIN}, @code{IF}, and @code{AHEAD} no code for locals stack
7126: pointer manipulation is generated. Between control structure words
7127: locals definitions can push locals onto the locals stack. @code{AGAIN}
7128: is the simplest of the other three control flow words. It has to
7129: restore the locals stack depth of the corresponding @code{BEGIN}
7130: before branching. The code looks like this:
7131: @format
7132: @code{lp+!#} current-locals-size @minus{} dest-locals-size
7133: @code{branch} <begin>
7134: @end format
7135:
7136: @code{UNTIL} is a little more complicated: If it branches back, it
7137: must adjust the stack just like @code{AGAIN}. But if it falls through,
7138: the locals stack must not be changed. The compiler generates the
7139: following code:
7140: @format
7141: @code{?branch-lp+!#} <begin> current-locals-size @minus{} dest-locals-size
7142: @end format
7143: The locals stack pointer is only adjusted if the branch is taken.
7144:
7145: @code{THEN} can produce somewhat inefficient code:
7146: @format
7147: @code{lp+!#} current-locals-size @minus{} orig-locals-size
7148: <orig target>:
7149: @code{lp+!#} orig-locals-size @minus{} new-locals-size
7150: @end format
7151: The second @code{lp+!#} adjusts the locals stack pointer from the
7152: level at the @i{orig} point to the level after the @code{THEN}. The
7153: first @code{lp+!#} adjusts the locals stack pointer from the current
7154: level to the level at the orig point, so the complete effect is an
7155: adjustment from the current level to the right level after the
7156: @code{THEN}.
7157:
7158: @cindex locals information on the control-flow stack
7159: @cindex control-flow stack items, locals information
7160: In a conventional Forth implementation a dest control-flow stack entry
7161: is just the target address and an orig entry is just the address to be
7162: patched. Our locals implementation adds a word list to every orig or dest
7163: item. It is the list of locals visible (or assumed visible) at the point
7164: described by the entry. Our implementation also adds a tag to identify
7165: the kind of entry, in particular to differentiate between live and dead
7166: (reachable and unreachable) orig entries.
7167:
7168: A few unusual operations have to be performed on locals word lists:
7169:
7170:
7171: doc-common-list
7172: doc-sub-list?
7173: doc-list-size
7174:
7175:
7176: Several features of our locals word list implementation make these
7177: operations easy to implement: The locals word lists are organised as
7178: linked lists; the tails of these lists are shared, if the lists
7179: contain some of the same locals; and the address of a name is greater
7180: than the address of the names behind it in the list.
7181:
7182: Another important implementation detail is the variable
7183: @code{dead-code}. It is used by @code{BEGIN} and @code{THEN} to
7184: determine if they can be reached directly or only through the branch
7185: that they resolve. @code{dead-code} is set by @code{UNREACHABLE},
7186: @code{AHEAD}, @code{EXIT} etc., and cleared at the start of a colon
7187: definition, by @code{BEGIN} and usually by @code{THEN}.
7188:
7189: Counted loops are similar to other loops in most respects, but
7190: @code{LEAVE} requires special attention: It performs basically the same
7191: service as @code{AHEAD}, but it does not create a control-flow stack
7192: entry. Therefore the information has to be stored elsewhere;
7193: traditionally, the information was stored in the target fields of the
7194: branches created by the @code{LEAVE}s, by organizing these fields into a
7195: linked list. Unfortunately, this clever trick does not provide enough
7196: space for storing our extended control flow information. Therefore, we
7197: introduce another stack, the leave stack. It contains the control-flow
7198: stack entries for all unresolved @code{LEAVE}s.
7199:
7200: Local names are kept until the end of the colon definition, even if
7201: they are no longer visible in any control-flow path. In a few cases
7202: this may lead to increased space needs for the locals name area, but
7203: usually less than reclaiming this space would cost in code size.
7204:
7205:
7206: @node ANS Forth locals, , Gforth locals, Locals
7207: @subsection ANS Forth locals
7208: @cindex locals, ANS Forth style
7209:
7210: The ANS Forth locals wordset does not define a syntax for locals, but
7211: words that make it possible to define various syntaxes. One of the
7212: possible syntaxes is a subset of the syntax we used in the Gforth locals
7213: wordset, i.e.:
7214:
7215: @example
7216: @{ local1 local2 ... -- comment @}
7217: @end example
7218: @noindent
7219: or
7220: @example
7221: @{ local1 local2 ... @}
7222: @end example
7223:
7224: The order of the locals corresponds to the order in a stack comment. The
7225: restrictions are:
7226:
7227: @itemize @bullet
7228: @item
7229: Locals can only be cell-sized values (no type specifiers are allowed).
7230: @item
7231: Locals can be defined only outside control structures.
7232: @item
7233: Locals can interfere with explicit usage of the return stack. For the
7234: exact (and long) rules, see the standard. If you don't use return stack
7235: accessing words in a definition using locals, you will be all right. The
7236: purpose of this rule is to make locals implementation on the return
7237: stack easier.
7238: @item
7239: The whole definition must be in one line.
7240: @end itemize
7241:
7242: Locals defined in this way behave like @code{VALUE}s
7243: (@xref{Values}). I.e., they are initialized from the stack. Using their
7244: name produces their value. Their value can be changed using @code{TO}.
7245:
7246: Since this syntax is supported by Gforth directly, you need not do
7247: anything to use it. If you want to port a program using this syntax to
7248: another ANS Forth system, use @file{compat/anslocal.fs} to implement the
7249: syntax on the other system.
7250:
7251: Note that a syntax shown in the standard, section A.13 looks
7252: similar, but is quite different in having the order of locals
7253: reversed. Beware!
7254:
7255: The ANS Forth locals wordset itself consists of a word:
7256:
7257:
7258: doc-(local)
7259:
7260:
7261: The ANS Forth locals extension wordset defines a syntax using @code{locals|}, but it is so
7262: awful that we strongly recommend not to use it. We have implemented this
7263: syntax to make porting to Gforth easy, but do not document it here. The
7264: problem with this syntax is that the locals are defined in an order
7265: reversed with respect to the standard stack comment notation, making
7266: programs harder to read, and easier to misread and miswrite. The only
7267: merit of this syntax is that it is easy to implement using the ANS Forth
7268: locals wordset.
7269:
7270:
7271: @c ----------------------------------------------------------
7272: @node Structures, Object-oriented Forth, Locals, Words
7273: @section Structures
7274: @cindex structures
7275: @cindex records
7276:
7277: This section presents the structure package that comes with Gforth. A
7278: version of the package implemented in ANS Forth is available in
7279: @file{compat/struct.fs}. This package was inspired by a posting on
7280: comp.lang.forth in 1989 (unfortunately I don't remember, by whom;
7281: possibly John Hayes). A version of this section has been published in
7282: ???. Marcel Hendrix provided helpful comments.
7283:
7284: @menu
7285: * Why explicit structure support?::
7286: * Structure Usage::
7287: * Structure Naming Convention::
7288: * Structure Implementation::
7289: * Structure Glossary::
7290: @end menu
7291:
7292: @node Why explicit structure support?, Structure Usage, Structures, Structures
7293: @subsection Why explicit structure support?
7294:
7295: @cindex address arithmetic for structures
7296: @cindex structures using address arithmetic
7297: If we want to use a structure containing several fields, we could simply
7298: reserve memory for it, and access the fields using address arithmetic
7299: (@pxref{Address arithmetic}). As an example, consider a structure with
7300: the following fields
7301:
7302: @table @code
7303: @item a
7304: is a float
7305: @item b
7306: is a cell
7307: @item c
7308: is a float
7309: @end table
7310:
7311: Given the (float-aligned) base address of the structure we get the
7312: address of the field
7313:
7314: @table @code
7315: @item a
7316: without doing anything further.
7317: @item b
7318: with @code{float+}
7319: @item c
7320: with @code{float+ cell+ faligned}
7321: @end table
7322:
7323: It is easy to see that this can become quite tiring.
7324:
7325: Moreover, it is not very readable, because seeing a
7326: @code{cell+} tells us neither which kind of structure is
7327: accessed nor what field is accessed; we have to somehow infer the kind
7328: of structure, and then look up in the documentation, which field of
7329: that structure corresponds to that offset.
7330:
7331: Finally, this kind of address arithmetic also causes maintenance
7332: troubles: If you add or delete a field somewhere in the middle of the
7333: structure, you have to find and change all computations for the fields
7334: afterwards.
7335:
7336: So, instead of using @code{cell+} and friends directly, how
7337: about storing the offsets in constants:
7338:
7339: @example
7340: 0 constant a-offset
7341: 0 float+ constant b-offset
7342: 0 float+ cell+ faligned c-offset
7343: @end example
7344:
7345: Now we can get the address of field @code{x} with @code{x-offset
7346: +}. This is much better in all respects. Of course, you still
7347: have to change all later offset definitions if you add a field. You can
7348: fix this by declaring the offsets in the following way:
7349:
7350: @example
7351: 0 constant a-offset
7352: a-offset float+ constant b-offset
7353: b-offset cell+ faligned constant c-offset
7354: @end example
7355:
7356: Since we always use the offsets with @code{+}, we could use a defining
7357: word @code{cfield} that includes the @code{+} in the action of the
7358: defined word:
7359:
7360: @example
7361: : cfield ( n "name" -- )
7362: create ,
7363: does> ( name execution: addr1 -- addr2 )
7364: @@ + ;
7365:
7366: 0 cfield a
7367: 0 a float+ cfield b
7368: 0 b cell+ faligned cfield c
7369: @end example
7370:
7371: Instead of @code{x-offset +}, we now simply write @code{x}.
7372:
7373: The structure field words now can be used quite nicely. However,
7374: their definition is still a bit cumbersome: We have to repeat the
7375: name, the information about size and alignment is distributed before
7376: and after the field definitions etc. The structure package presented
7377: here addresses these problems.
7378:
7379: @node Structure Usage, Structure Naming Convention, Why explicit structure support?, Structures
7380: @subsection Structure Usage
7381: @cindex structure usage
7382:
7383: @cindex @code{field} usage
7384: @cindex @code{struct} usage
7385: @cindex @code{end-struct} usage
7386: You can define a structure for a (data-less) linked list with:
7387: @example
7388: struct
7389: cell% field list-next
7390: end-struct list%
7391: @end example
7392:
7393: With the address of the list node on the stack, you can compute the
7394: address of the field that contains the address of the next node with
7395: @code{list-next}. E.g., you can determine the length of a list
7396: with:
7397:
7398: @example
7399: : list-length ( list -- n )
7400: \ "list" is a pointer to the first element of a linked list
7401: \ "n" is the length of the list
7402: 0 BEGIN ( list1 n1 )
7403: over
7404: WHILE ( list1 n1 )
7405: 1+ swap list-next @@ swap
7406: REPEAT
7407: nip ;
7408: @end example
7409:
7410: You can reserve memory for a list node in the dictionary with
7411: @code{list% %allot}, which leaves the address of the list node on the
7412: stack. For the equivalent allocation on the heap you can use @code{list%
7413: %alloc} (or, for an @code{allocate}-like stack effect (i.e., with ior),
7414: use @code{list% %allocate}). You can get the the size of a list
7415: node with @code{list% %size} and its alignment with @code{list%
7416: %alignment}.
7417:
7418: Note that in ANS Forth the body of a @code{create}d word is
7419: @code{aligned} but not necessarily @code{faligned};
7420: therefore, if you do a:
7421: @example
7422: create @emph{name} foo% %allot
7423: @end example
7424:
7425: @noindent
7426: then the memory alloted for @code{foo%} is
7427: guaranteed to start at the body of @code{@emph{name}} only if
7428: @code{foo%} contains only character, cell and double fields.
7429:
7430: @cindex structures containing structures
7431: You can include a structure @code{foo%} as a field of
7432: another structure, like this:
7433: @example
7434: struct
7435: ...
7436: foo% field ...
7437: ...
7438: end-struct ...
7439: @end example
7440:
7441: @cindex structure extension
7442: @cindex extended records
7443: Instead of starting with an empty structure, you can extend an
7444: existing structure. E.g., a plain linked list without data, as defined
7445: above, is hardly useful; You can extend it to a linked list of integers,
7446: like this:@footnote{This feature is also known as @emph{extended
7447: records}. It is the main innovation in the Oberon language; in other
7448: words, adding this feature to Modula-2 led Wirth to create a new
7449: language, write a new compiler etc. Adding this feature to Forth just
7450: required a few lines of code.}
7451:
7452: @example
7453: list%
7454: cell% field intlist-int
7455: end-struct intlist%
7456: @end example
7457:
7458: @code{intlist%} is a structure with two fields:
7459: @code{list-next} and @code{intlist-int}.
7460:
7461: @cindex structures containing arrays
7462: You can specify an array type containing @emph{n} elements of
7463: type @code{foo%} like this:
7464:
7465: @example
7466: foo% @emph{n} *
7467: @end example
7468:
7469: You can use this array type in any place where you can use a normal
7470: type, e.g., when defining a @code{field}, or with
7471: @code{%allot}.
7472:
7473: @cindex first field optimization
7474: The first field is at the base address of a structure and the word
7475: for this field (e.g., @code{list-next}) actually does not change
7476: the address on the stack. You may be tempted to leave it away in the
7477: interest of run-time and space efficiency. This is not necessary,
7478: because the structure package optimizes this case and compiling such
7479: words does not generate any code. So, in the interest of readability
7480: and maintainability you should include the word for the field when
7481: accessing the field.
7482:
7483: @node Structure Naming Convention, Structure Implementation, Structure Usage, Structures
7484: @subsection Structure Naming Convention
7485: @cindex structure naming convention
7486:
7487: The field names that come to (my) mind are often quite generic, and,
7488: if used, would cause frequent name clashes. E.g., many structures
7489: probably contain a @code{counter} field. The structure names
7490: that come to (my) mind are often also the logical choice for the names
7491: of words that create such a structure.
7492:
7493: Therefore, I have adopted the following naming conventions:
7494:
7495: @itemize @bullet
7496: @cindex field naming convention
7497: @item
7498: The names of fields are of the form
7499: @code{@emph{struct}-@emph{field}}, where
7500: @code{@emph{struct}} is the basic name of the structure, and
7501: @code{@emph{field}} is the basic name of the field. You can
7502: think of field words as converting the (address of the)
7503: structure into the (address of the) field.
7504:
7505: @cindex structure naming convention
7506: @item
7507: The names of structures are of the form
7508: @code{@emph{struct}%}, where
7509: @code{@emph{struct}} is the basic name of the structure.
7510: @end itemize
7511:
7512: This naming convention does not work that well for fields of extended
7513: structures; e.g., the integer list structure has a field
7514: @code{intlist-int}, but has @code{list-next}, not
7515: @code{intlist-next}.
7516:
7517: @node Structure Implementation, Structure Glossary, Structure Naming Convention, Structures
7518: @subsection Structure Implementation
7519: @cindex structure implementation
7520: @cindex implementation of structures
7521:
7522: The central idea in the implementation is to pass the data about the
7523: structure being built on the stack, not in some global
7524: variable. Everything else falls into place naturally once this design
7525: decision is made.
7526:
7527: The type description on the stack is of the form @emph{align
7528: size}. Keeping the size on the top-of-stack makes dealing with arrays
7529: very simple.
7530:
7531: @code{field} is a defining word that uses @code{Create}
7532: and @code{DOES>}. The body of the field contains the offset
7533: of the field, and the normal @code{DOES>} action is simply:
7534:
7535: @example
7536: @ +
7537: @end example
7538:
7539: @noindent
7540: i.e., add the offset to the address, giving the stack effect
7541: @i{addr1 -- addr2} for a field.
7542:
7543: @cindex first field optimization, implementation
7544: This simple structure is slightly complicated by the optimization
7545: for fields with offset 0, which requires a different
7546: @code{DOES>}-part (because we cannot rely on there being
7547: something on the stack if such a field is invoked during
7548: compilation). Therefore, we put the different @code{DOES>}-parts
7549: in separate words, and decide which one to invoke based on the
7550: offset. For a zero offset, the field is basically a noop; it is
7551: immediate, and therefore no code is generated when it is compiled.
7552:
7553: @node Structure Glossary, , Structure Implementation, Structures
7554: @subsection Structure Glossary
7555: @cindex structure glossary
7556:
7557:
7558: doc-%align
7559: doc-%alignment
7560: doc-%alloc
7561: doc-%allocate
7562: doc-%allot
7563: doc-cell%
7564: doc-char%
7565: doc-dfloat%
7566: doc-double%
7567: doc-end-struct
7568: doc-field
7569: doc-float%
7570: doc-naligned
7571: doc-sfloat%
7572: doc-%size
7573: doc-struct
7574:
7575:
7576: @c -------------------------------------------------------------
7577: @node Object-oriented Forth, Passing Commands to the OS, Structures, Words
7578: @section Object-oriented Forth
7579:
7580: Gforth comes with three packages for object-oriented programming:
7581: @file{objects.fs}, @file{oof.fs}, and @file{mini-oof.fs}; none of them
7582: is preloaded, so you have to @code{include} them before use. The most
7583: important differences between these packages (and others) are discussed
7584: in @ref{Comparison with other object models}. All packages are written
7585: in ANS Forth and can be used with any other ANS Forth.
7586:
7587: @menu
7588: * Why object-oriented programming?::
7589: * Object-Oriented Terminology::
7590: * Objects::
7591: * OOF::
7592: * Mini-OOF::
7593: * Comparison with other object models::
7594: @end menu
7595:
7596:
7597: @node Why object-oriented programming?, Object-Oriented Terminology, , Object-oriented Forth
7598: @subsubsection Why object-oriented programming?
7599: @cindex object-oriented programming motivation
7600: @cindex motivation for object-oriented programming
7601:
7602: Often we have to deal with several data structures (@emph{objects}),
7603: that have to be treated similarly in some respects, but differently in
7604: others. Graphical objects are the textbook example: circles, triangles,
7605: dinosaurs, icons, and others, and we may want to add more during program
7606: development. We want to apply some operations to any graphical object,
7607: e.g., @code{draw} for displaying it on the screen. However, @code{draw}
7608: has to do something different for every kind of object.
7609: @comment TODO add some other operations eg perimeter, area
7610: @comment and tie in to concrete examples later..
7611:
7612: We could implement @code{draw} as a big @code{CASE}
7613: control structure that executes the appropriate code depending on the
7614: kind of object to be drawn. This would be not be very elegant, and,
7615: moreover, we would have to change @code{draw} every time we add
7616: a new kind of graphical object (say, a spaceship).
7617:
7618: What we would rather do is: When defining spaceships, we would tell
7619: the system: ``Here's how you @code{draw} a spaceship; you figure
7620: out the rest''.
7621:
7622: This is the problem that all systems solve that (rightfully) call
7623: themselves object-oriented; the object-oriented packages presented here
7624: solve this problem (and not much else).
7625: @comment TODO ?list properties of oo systems.. oo vs o-based?
7626:
7627: @node Object-Oriented Terminology, Objects, Why object-oriented programming?, Object-oriented Forth
7628: @subsubsection Object-Oriented Terminology
7629: @cindex object-oriented terminology
7630: @cindex terminology for object-oriented programming
7631:
7632: This section is mainly for reference, so you don't have to understand
7633: all of it right away. The terminology is mainly Smalltalk-inspired. In
7634: short:
7635:
7636: @table @emph
7637: @cindex class
7638: @item class
7639: a data structure definition with some extras.
7640:
7641: @cindex object
7642: @item object
7643: an instance of the data structure described by the class definition.
7644:
7645: @cindex instance variables
7646: @item instance variables
7647: fields of the data structure.
7648:
7649: @cindex selector
7650: @cindex method selector
7651: @cindex virtual function
7652: @item selector
7653: (or @emph{method selector}) a word (e.g.,
7654: @code{draw}) that performs an operation on a variety of data
7655: structures (classes). A selector describes @emph{what} operation to
7656: perform. In C++ terminology: a (pure) virtual function.
7657:
7658: @cindex method
7659: @item method
7660: the concrete definition that performs the operation
7661: described by the selector for a specific class. A method specifies
7662: @emph{how} the operation is performed for a specific class.
7663:
7664: @cindex selector invocation
7665: @cindex message send
7666: @cindex invoking a selector
7667: @item selector invocation
7668: a call of a selector. One argument of the call (the TOS (top-of-stack))
7669: is used for determining which method is used. In Smalltalk terminology:
7670: a message (consisting of the selector and the other arguments) is sent
7671: to the object.
7672:
7673: @cindex receiving object
7674: @item receiving object
7675: the object used for determining the method executed by a selector
7676: invocation. In the @file{objects.fs} model, it is the object that is on
7677: the TOS when the selector is invoked. (@emph{Receiving} comes from
7678: the Smalltalk @emph{message} terminology.)
7679:
7680: @cindex child class
7681: @cindex parent class
7682: @cindex inheritance
7683: @item child class
7684: a class that has (@emph{inherits}) all properties (instance variables,
7685: selectors, methods) from a @emph{parent class}. In Smalltalk
7686: terminology: The subclass inherits from the superclass. In C++
7687: terminology: The derived class inherits from the base class.
7688:
7689: @end table
7690:
7691: @c If you wonder about the message sending terminology, it comes from
7692: @c a time when each object had it's own task and objects communicated via
7693: @c message passing; eventually the Smalltalk developers realized that
7694: @c they can do most things through simple (indirect) calls. They kept the
7695: @c terminology.
7696:
7697:
7698: @node Objects, OOF, Object-Oriented Terminology, Object-oriented Forth
7699: @subsection The @file{objects.fs} model
7700: @cindex objects
7701: @cindex object-oriented programming
7702:
7703: @cindex @file{objects.fs}
7704: @cindex @file{oof.fs}
7705:
7706: This section describes the @file{objects.fs} package. This material also
7707: has been published in @cite{Yet Another Forth Objects Package} by Anton
7708: Ertl and appeared in Forth Dimensions 19(2), pages 37--43
7709: (@uref{http://www.complang.tuwien.ac.at/forth/objects/objects.html}).
7710: @c McKewan's and Zsoter's packages
7711:
7712: This section assumes that you have read @ref{Structures}.
7713:
7714: The techniques on which this model is based have been used to implement
7715: the parser generator, Gray, and have also been used in Gforth for
7716: implementing the various flavours of word lists (hashed or not,
7717: case-sensitive or not, special-purpose word lists for locals etc.).
7718:
7719:
7720: @menu
7721: * Properties of the Objects model::
7722: * Basic Objects Usage::
7723: * The Objects base class::
7724: * Creating objects::
7725: * Object-Oriented Programming Style::
7726: * Class Binding::
7727: * Method conveniences::
7728: * Classes and Scoping::
7729: * Dividing classes::
7730: * Object Interfaces::
7731: * Objects Implementation::
7732: * Objects Glossary::
7733: @end menu
7734:
7735: Marcel Hendrix provided helpful comments on this section. Andras Zsoter
7736: and Bernd Paysan helped me with the related works section.
7737:
7738: @node Properties of the Objects model, Basic Objects Usage, Objects, Objects
7739: @subsubsection Properties of the @file{objects.fs} model
7740: @cindex @file{objects.fs} properties
7741:
7742: @itemize @bullet
7743: @item
7744: It is straightforward to pass objects on the stack. Passing
7745: selectors on the stack is a little less convenient, but possible.
7746:
7747: @item
7748: Objects are just data structures in memory, and are referenced by their
7749: address. You can create words for objects with normal defining words
7750: like @code{constant}. Likewise, there is no difference between instance
7751: variables that contain objects and those that contain other data.
7752:
7753: @item
7754: Late binding is efficient and easy to use.
7755:
7756: @item
7757: It avoids parsing, and thus avoids problems with state-smartness
7758: and reduced extensibility; for convenience there are a few parsing
7759: words, but they have non-parsing counterparts. There are also a few
7760: defining words that parse. This is hard to avoid, because all standard
7761: defining words parse (except @code{:noname}); however, such
7762: words are not as bad as many other parsing words, because they are not
7763: state-smart.
7764:
7765: @item
7766: It does not try to incorporate everything. It does a few things and does
7767: them well (IMO). In particular, this model was not designed to support
7768: information hiding (although it has features that may help); you can use
7769: a separate package for achieving this.
7770:
7771: @item
7772: It is layered; you don't have to learn and use all features to use this
7773: model. Only a few features are necessary (@xref{Basic Objects Usage},
7774: @xref{The Objects base class}, @xref{Creating objects}.), the others
7775: are optional and independent of each other.
7776:
7777: @item
7778: An implementation in ANS Forth is available.
7779:
7780: @end itemize
7781:
7782:
7783: @node Basic Objects Usage, The Objects base class, Properties of the Objects model, Objects
7784: @subsubsection Basic @file{objects.fs} Usage
7785: @cindex basic objects usage
7786: @cindex objects, basic usage
7787:
7788: You can define a class for graphical objects like this:
7789:
7790: @cindex @code{class} usage
7791: @cindex @code{end-class} usage
7792: @cindex @code{selector} usage
7793: @example
7794: object class \ "object" is the parent class
7795: selector draw ( x y graphical -- )
7796: end-class graphical
7797: @end example
7798:
7799: This code defines a class @code{graphical} with an
7800: operation @code{draw}. We can perform the operation
7801: @code{draw} on any @code{graphical} object, e.g.:
7802:
7803: @example
7804: 100 100 t-rex draw
7805: @end example
7806:
7807: @noindent
7808: where @code{t-rex} is a word (say, a constant) that produces a
7809: graphical object.
7810:
7811: @comment TODO add a 2nd operation eg perimeter.. and use for
7812: @comment a concrete example
7813:
7814: @cindex abstract class
7815: How do we create a graphical object? With the present definitions,
7816: we cannot create a useful graphical object. The class
7817: @code{graphical} describes graphical objects in general, but not
7818: any concrete graphical object type (C++ users would call it an
7819: @emph{abstract class}); e.g., there is no method for the selector
7820: @code{draw} in the class @code{graphical}.
7821:
7822: For concrete graphical objects, we define child classes of the
7823: class @code{graphical}, e.g.:
7824:
7825: @cindex @code{overrides} usage
7826: @cindex @code{field} usage in class definition
7827: @example
7828: graphical class \ "graphical" is the parent class
7829: cell% field circle-radius
7830:
7831: :noname ( x y circle -- )
7832: circle-radius @@ draw-circle ;
7833: overrides draw
7834:
7835: :noname ( n-radius circle -- )
7836: circle-radius ! ;
7837: overrides construct
7838:
7839: end-class circle
7840: @end example
7841:
7842: Here we define a class @code{circle} as a child of @code{graphical},
7843: with field @code{circle-radius} (which behaves just like a field
7844: (@pxref{Structures}); it defines (using @code{overrides}) new methods
7845: for the selectors @code{draw} and @code{construct} (@code{construct} is
7846: defined in @code{object}, the parent class of @code{graphical}).
7847:
7848: Now we can create a circle on the heap (i.e.,
7849: @code{allocate}d memory) with:
7850:
7851: @cindex @code{heap-new} usage
7852: @example
7853: 50 circle heap-new constant my-circle
7854: @end example
7855:
7856: @noindent
7857: @code{heap-new} invokes @code{construct}, thus
7858: initializing the field @code{circle-radius} with 50. We can draw
7859: this new circle at (100,100) with:
7860:
7861: @example
7862: 100 100 my-circle draw
7863: @end example
7864:
7865: @cindex selector invocation, restrictions
7866: @cindex class definition, restrictions
7867: Note: You can only invoke a selector if the object on the TOS
7868: (the receiving object) belongs to the class where the selector was
7869: defined or one of its descendents; e.g., you can invoke
7870: @code{draw} only for objects belonging to @code{graphical}
7871: or its descendents (e.g., @code{circle}). Immediately before
7872: @code{end-class}, the search order has to be the same as
7873: immediately after @code{class}.
7874:
7875: @node The Objects base class, Creating objects, Basic Objects Usage, Objects
7876: @subsubsection The @file{object.fs} base class
7877: @cindex @code{object} class
7878:
7879: When you define a class, you have to specify a parent class. So how do
7880: you start defining classes? There is one class available from the start:
7881: @code{object}. It is ancestor for all classes and so is the
7882: only class that has no parent. It has two selectors: @code{construct}
7883: and @code{print}.
7884:
7885: @node Creating objects, Object-Oriented Programming Style, The Objects base class, Objects
7886: @subsubsection Creating objects
7887: @cindex creating objects
7888: @cindex object creation
7889: @cindex object allocation options
7890:
7891: @cindex @code{heap-new} discussion
7892: @cindex @code{dict-new} discussion
7893: @cindex @code{construct} discussion
7894: You can create and initialize an object of a class on the heap with
7895: @code{heap-new} ( ... class -- object ) and in the dictionary
7896: (allocation with @code{allot}) with @code{dict-new} (
7897: ... class -- object ). Both words invoke @code{construct}, which
7898: consumes the stack items indicated by "..." above.
7899:
7900: @cindex @code{init-object} discussion
7901: @cindex @code{class-inst-size} discussion
7902: If you want to allocate memory for an object yourself, you can get its
7903: alignment and size with @code{class-inst-size 2@@} ( class --
7904: align size ). Once you have memory for an object, you can initialize
7905: it with @code{init-object} ( ... class object -- );
7906: @code{construct} does only a part of the necessary work.
7907:
7908: @node Object-Oriented Programming Style, Class Binding, Creating objects, Objects
7909: @subsubsection Object-Oriented Programming Style
7910: @cindex object-oriented programming style
7911: @cindex programming style, object-oriented
7912:
7913: This section is not exhaustive.
7914:
7915: @cindex stack effects of selectors
7916: @cindex selectors and stack effects
7917: In general, it is a good idea to ensure that all methods for the
7918: same selector have the same stack effect: when you invoke a selector,
7919: you often have no idea which method will be invoked, so, unless all
7920: methods have the same stack effect, you will not know the stack effect
7921: of the selector invocation.
7922:
7923: One exception to this rule is methods for the selector
7924: @code{construct}. We know which method is invoked, because we
7925: specify the class to be constructed at the same place. Actually, I
7926: defined @code{construct} as a selector only to give the users a
7927: convenient way to specify initialization. The way it is used, a
7928: mechanism different from selector invocation would be more natural
7929: (but probably would take more code and more space to explain).
7930:
7931: @node Class Binding, Method conveniences, Object-Oriented Programming Style, Objects
7932: @subsubsection Class Binding
7933: @cindex class binding
7934: @cindex early binding
7935:
7936: @cindex late binding
7937: Normal selector invocations determine the method at run-time depending
7938: on the class of the receiving object. This run-time selection is called
7939: @i{late binding}.
7940:
7941: Sometimes it's preferable to invoke a different method. For example,
7942: you might want to use the simple method for @code{print}ing
7943: @code{object}s instead of the possibly long-winded @code{print} method
7944: of the receiver class. You can achieve this by replacing the invocation
7945: of @code{print} with:
7946:
7947: @cindex @code{[bind]} usage
7948: @example
7949: [bind] object print
7950: @end example
7951:
7952: @noindent
7953: in compiled code or:
7954:
7955: @cindex @code{bind} usage
7956: @example
7957: bind object print
7958: @end example
7959:
7960: @cindex class binding, alternative to
7961: @noindent
7962: in interpreted code. Alternatively, you can define the method with a
7963: name (e.g., @code{print-object}), and then invoke it through the
7964: name. Class binding is just a (often more convenient) way to achieve
7965: the same effect; it avoids name clutter and allows you to invoke
7966: methods directly without naming them first.
7967:
7968: @cindex superclass binding
7969: @cindex parent class binding
7970: A frequent use of class binding is this: When we define a method
7971: for a selector, we often want the method to do what the selector does
7972: in the parent class, and a little more. There is a special word for
7973: this purpose: @code{[parent]}; @code{[parent]
7974: @emph{selector}} is equivalent to @code{[bind] @emph{parent
7975: selector}}, where @code{@emph{parent}} is the parent
7976: class of the current class. E.g., a method definition might look like:
7977:
7978: @cindex @code{[parent]} usage
7979: @example
7980: :noname
7981: dup [parent] foo \ do parent's foo on the receiving object
7982: ... \ do some more
7983: ; overrides foo
7984: @end example
7985:
7986: @cindex class binding as optimization
7987: In @cite{Object-oriented programming in ANS Forth} (Forth Dimensions,
7988: March 1997), Andrew McKewan presents class binding as an optimization
7989: technique. I recommend not using it for this purpose unless you are in
7990: an emergency. Late binding is pretty fast with this model anyway, so the
7991: benefit of using class binding is small; the cost of using class binding
7992: where it is not appropriate is reduced maintainability.
7993:
7994: While we are at programming style questions: You should bind
7995: selectors only to ancestor classes of the receiving object. E.g., say,
7996: you know that the receiving object is of class @code{foo} or its
7997: descendents; then you should bind only to @code{foo} and its
7998: ancestors.
7999:
8000: @node Method conveniences, Classes and Scoping, Class Binding, Objects
8001: @subsubsection Method conveniences
8002: @cindex method conveniences
8003:
8004: In a method you usually access the receiving object pretty often. If
8005: you define the method as a plain colon definition (e.g., with
8006: @code{:noname}), you may have to do a lot of stack
8007: gymnastics. To avoid this, you can define the method with @code{m:
8008: ... ;m}. E.g., you could define the method for
8009: @code{draw}ing a @code{circle} with
8010:
8011: @cindex @code{this} usage
8012: @cindex @code{m:} usage
8013: @cindex @code{;m} usage
8014: @example
8015: m: ( x y circle -- )
8016: ( x y ) this circle-radius @@ draw-circle ;m
8017: @end example
8018:
8019: @cindex @code{exit} in @code{m: ... ;m}
8020: @cindex @code{exitm} discussion
8021: @cindex @code{catch} in @code{m: ... ;m}
8022: When this method is executed, the receiver object is removed from the
8023: stack; you can access it with @code{this} (admittedly, in this
8024: example the use of @code{m: ... ;m} offers no advantage). Note
8025: that I specify the stack effect for the whole method (i.e. including
8026: the receiver object), not just for the code between @code{m:}
8027: and @code{;m}. You cannot use @code{exit} in
8028: @code{m:...;m}; instead, use
8029: @code{exitm}.@footnote{Moreover, for any word that calls
8030: @code{catch} and was defined before loading
8031: @code{objects.fs}, you have to redefine it like I redefined
8032: @code{catch}: @code{: catch this >r catch r> to-this ;}}
8033:
8034: @cindex @code{inst-var} usage
8035: You will frequently use sequences of the form @code{this
8036: @emph{field}} (in the example above: @code{this
8037: circle-radius}). If you use the field only in this way, you can
8038: define it with @code{inst-var} and eliminate the
8039: @code{this} before the field name. E.g., the @code{circle}
8040: class above could also be defined with:
8041:
8042: @example
8043: graphical class
8044: cell% inst-var radius
8045:
8046: m: ( x y circle -- )
8047: radius @@ draw-circle ;m
8048: overrides draw
8049:
8050: m: ( n-radius circle -- )
8051: radius ! ;m
8052: overrides construct
8053:
8054: end-class circle
8055: @end example
8056:
8057: @code{radius} can only be used in @code{circle} and its
8058: descendent classes and inside @code{m:...;m}.
8059:
8060: @cindex @code{inst-value} usage
8061: You can also define fields with @code{inst-value}, which is
8062: to @code{inst-var} what @code{value} is to
8063: @code{variable}. You can change the value of such a field with
8064: @code{[to-inst]}. E.g., we could also define the class
8065: @code{circle} like this:
8066:
8067: @example
8068: graphical class
8069: inst-value radius
8070:
8071: m: ( x y circle -- )
8072: radius draw-circle ;m
8073: overrides draw
8074:
8075: m: ( n-radius circle -- )
8076: [to-inst] radius ;m
8077: overrides construct
8078:
8079: end-class circle
8080: @end example
8081:
8082: Finally, you can define named methods with @code{:m}. One use of this
8083: feature is the definition of words that occur only in one class and are
8084: not intended to be overridden, but which still need method context
8085: (e.g., for accessing @code{inst-var}s). Another use is for methods that
8086: would be bound frequently, if defined anonymously.
8087:
8088:
8089: @node Classes and Scoping, Dividing classes, Method conveniences, Objects
8090: @subsubsection Classes and Scoping
8091: @cindex classes and scoping
8092: @cindex scoping and classes
8093:
8094: Inheritance is frequent, unlike structure extension. This exacerbates
8095: the problem with the field name convention (@pxref{Structure Naming
8096: Convention}): One always has to remember in which class the field was
8097: originally defined; changing a part of the class structure would require
8098: changes for renaming in otherwise unaffected code.
8099:
8100: @cindex @code{inst-var} visibility
8101: @cindex @code{inst-value} visibility
8102: To solve this problem, I added a scoping mechanism (which was not in my
8103: original charter): A field defined with @code{inst-var} (or
8104: @code{inst-value}) is visible only in the class where it is defined and in
8105: the descendent classes of this class. Using such fields only makes
8106: sense in @code{m:}-defined methods in these classes anyway.
8107:
8108: This scoping mechanism allows us to use the unadorned field name,
8109: because name clashes with unrelated words become much less likely.
8110:
8111: @cindex @code{protected} discussion
8112: @cindex @code{private} discussion
8113: Once we have this mechanism, we can also use it for controlling the
8114: visibility of other words: All words defined after
8115: @code{protected} are visible only in the current class and its
8116: descendents. @code{public} restores the compilation
8117: (i.e. @code{current}) word list that was in effect before. If you
8118: have several @code{protected}s without an intervening
8119: @code{public} or @code{set-current}, @code{public}
8120: will restore the compilation word list in effect before the first of
8121: these @code{protected}s.
8122:
8123: @node Dividing classes, Object Interfaces, Classes and Scoping, Objects
8124: @subsubsection Dividing classes
8125: @cindex Dividing classes
8126: @cindex @code{methods}...@code{end-methods}
8127:
8128: You may want to do the definition of methods separate from the
8129: definition of the class, its selectors, fields, and instance variables,
8130: i.e., separate the implementation from the definition. You can do this
8131: in the following way:
8132:
8133: @example
8134: graphical class
8135: inst-value radius
8136: end-class circle
8137:
8138: ... \ do some other stuff
8139:
8140: circle methods \ now we are ready
8141:
8142: m: ( x y circle -- )
8143: radius draw-circle ;m
8144: overrides draw
8145:
8146: m: ( n-radius circle -- )
8147: [to-inst] radius ;m
8148: overrides construct
8149:
8150: end-methods
8151: @end example
8152:
8153: You can use several @code{methods}...@code{end-methods} sections. The
8154: only things you can do to the class in these sections are: defining
8155: methods, and overriding the class's selectors. You must not define new
8156: selectors or fields.
8157:
8158: Note that you often have to override a selector before using it. In
8159: particular, you usually have to override @code{construct} with a new
8160: method before you can invoke @code{heap-new} and friends. E.g., you
8161: must not create a circle before the @code{overrides construct} sequence
8162: in the example above.
8163:
8164: @node Object Interfaces, Objects Implementation, Dividing classes, Objects
8165: @subsubsection Object Interfaces
8166: @cindex object interfaces
8167: @cindex interfaces for objects
8168:
8169: In this model you can only call selectors defined in the class of the
8170: receiving objects or in one of its ancestors. If you call a selector
8171: with a receiving object that is not in one of these classes, the
8172: result is undefined; if you are lucky, the program crashes
8173: immediately.
8174:
8175: @cindex selectors common to hardly-related classes
8176: Now consider the case when you want to have a selector (or several)
8177: available in two classes: You would have to add the selector to a
8178: common ancestor class, in the worst case to @code{object}. You
8179: may not want to do this, e.g., because someone else is responsible for
8180: this ancestor class.
8181:
8182: The solution for this problem is interfaces. An interface is a
8183: collection of selectors. If a class implements an interface, the
8184: selectors become available to the class and its descendents. A class
8185: can implement an unlimited number of interfaces. For the problem
8186: discussed above, we would define an interface for the selector(s), and
8187: both classes would implement the interface.
8188:
8189: As an example, consider an interface @code{storage} for
8190: writing objects to disk and getting them back, and a class
8191: @code{foo} that implements it. The code would look like this:
8192:
8193: @cindex @code{interface} usage
8194: @cindex @code{end-interface} usage
8195: @cindex @code{implementation} usage
8196: @example
8197: interface
8198: selector write ( file object -- )
8199: selector read1 ( file object -- )
8200: end-interface storage
8201:
8202: bar class
8203: storage implementation
8204:
8205: ... overrides write
8206: ... overrides read1
8207: ...
8208: end-class foo
8209: @end example
8210:
8211: @noindent
8212: (I would add a word @code{read} @i{( file -- object )} that uses
8213: @code{read1} internally, but that's beyond the point illustrated
8214: here.)
8215:
8216: Note that you cannot use @code{protected} in an interface; and
8217: of course you cannot define fields.
8218:
8219: In the Neon model, all selectors are available for all classes;
8220: therefore it does not need interfaces. The price you pay in this model
8221: is slower late binding, and therefore, added complexity to avoid late
8222: binding.
8223:
8224: @node Objects Implementation, Objects Glossary, Object Interfaces, Objects
8225: @subsubsection @file{objects.fs} Implementation
8226: @cindex @file{objects.fs} implementation
8227:
8228: @cindex @code{object-map} discussion
8229: An object is a piece of memory, like one of the data structures
8230: described with @code{struct...end-struct}. It has a field
8231: @code{object-map} that points to the method map for the object's
8232: class.
8233:
8234: @cindex method map
8235: @cindex virtual function table
8236: The @emph{method map}@footnote{This is Self terminology; in C++
8237: terminology: virtual function table.} is an array that contains the
8238: execution tokens (@i{xt}s) of the methods for the object's class. Each
8239: selector contains an offset into a method map.
8240:
8241: @cindex @code{selector} implementation, class
8242: @code{selector} is a defining word that uses
8243: @code{CREATE} and @code{DOES>}. The body of the
8244: selector contains the offset; the @code{DOES>} action for a
8245: class selector is, basically:
8246:
8247: @example
8248: ( object addr ) @@ over object-map @@ + @@ execute
8249: @end example
8250:
8251: Since @code{object-map} is the first field of the object, it
8252: does not generate any code. As you can see, calling a selector has a
8253: small, constant cost.
8254:
8255: @cindex @code{current-interface} discussion
8256: @cindex class implementation and representation
8257: A class is basically a @code{struct} combined with a method
8258: map. During the class definition the alignment and size of the class
8259: are passed on the stack, just as with @code{struct}s, so
8260: @code{field} can also be used for defining class
8261: fields. However, passing more items on the stack would be
8262: inconvenient, so @code{class} builds a data structure in memory,
8263: which is accessed through the variable
8264: @code{current-interface}. After its definition is complete, the
8265: class is represented on the stack by a pointer (e.g., as parameter for
8266: a child class definition).
8267:
8268: A new class starts off with the alignment and size of its parent,
8269: and a copy of the parent's method map. Defining new fields extends the
8270: size and alignment; likewise, defining new selectors extends the
8271: method map. @code{overrides} just stores a new @i{xt} in the method
8272: map at the offset given by the selector.
8273:
8274: @cindex class binding, implementation
8275: Class binding just gets the @i{xt} at the offset given by the selector
8276: from the class's method map and @code{compile,}s (in the case of
8277: @code{[bind]}) it.
8278:
8279: @cindex @code{this} implementation
8280: @cindex @code{catch} and @code{this}
8281: @cindex @code{this} and @code{catch}
8282: I implemented @code{this} as a @code{value}. At the
8283: start of an @code{m:...;m} method the old @code{this} is
8284: stored to the return stack and restored at the end; and the object on
8285: the TOS is stored @code{TO this}. This technique has one
8286: disadvantage: If the user does not leave the method via
8287: @code{;m}, but via @code{throw} or @code{exit},
8288: @code{this} is not restored (and @code{exit} may
8289: crash). To deal with the @code{throw} problem, I have redefined
8290: @code{catch} to save and restore @code{this}; the same
8291: should be done with any word that can catch an exception. As for
8292: @code{exit}, I simply forbid it (as a replacement, there is
8293: @code{exitm}).
8294:
8295: @cindex @code{inst-var} implementation
8296: @code{inst-var} is just the same as @code{field}, with
8297: a different @code{DOES>} action:
8298: @example
8299: @@ this +
8300: @end example
8301: Similar for @code{inst-value}.
8302:
8303: @cindex class scoping implementation
8304: Each class also has a word list that contains the words defined with
8305: @code{inst-var} and @code{inst-value}, and its protected
8306: words. It also has a pointer to its parent. @code{class} pushes
8307: the word lists of the class and all its ancestors onto the search order stack,
8308: and @code{end-class} drops them.
8309:
8310: @cindex interface implementation
8311: An interface is like a class without fields, parent and protected
8312: words; i.e., it just has a method map. If a class implements an
8313: interface, its method map contains a pointer to the method map of the
8314: interface. The positive offsets in the map are reserved for class
8315: methods, therefore interface map pointers have negative
8316: offsets. Interfaces have offsets that are unique throughout the
8317: system, unlike class selectors, whose offsets are only unique for the
8318: classes where the selector is available (invokable).
8319:
8320: This structure means that interface selectors have to perform one
8321: indirection more than class selectors to find their method. Their body
8322: contains the interface map pointer offset in the class method map, and
8323: the method offset in the interface method map. The
8324: @code{does>} action for an interface selector is, basically:
8325:
8326: @example
8327: ( object selector-body )
8328: 2dup selector-interface @@ ( object selector-body object interface-offset )
8329: swap object-map @@ + @@ ( object selector-body map )
8330: swap selector-offset @@ + @@ execute
8331: @end example
8332:
8333: where @code{object-map} and @code{selector-offset} are
8334: first fields and generate no code.
8335:
8336: As a concrete example, consider the following code:
8337:
8338: @example
8339: interface
8340: selector if1sel1
8341: selector if1sel2
8342: end-interface if1
8343:
8344: object class
8345: if1 implementation
8346: selector cl1sel1
8347: cell% inst-var cl1iv1
8348:
8349: ' m1 overrides construct
8350: ' m2 overrides if1sel1
8351: ' m3 overrides if1sel2
8352: ' m4 overrides cl1sel2
8353: end-class cl1
8354:
8355: create obj1 object dict-new drop
8356: create obj2 cl1 dict-new drop
8357: @end example
8358:
8359: The data structure created by this code (including the data structure
8360: for @code{object}) is shown in the <a
8361: href="objects-implementation.eps">figure</a>, assuming a cell size of 4.
8362: @comment TODO add this diagram..
8363:
8364: @node Objects Glossary, , Objects Implementation, Objects
8365: @subsubsection @file{objects.fs} Glossary
8366: @cindex @file{objects.fs} Glossary
8367:
8368:
8369: doc---objects-bind
8370: doc---objects-<bind>
8371: doc---objects-bind'
8372: doc---objects-[bind]
8373: doc---objects-class
8374: doc---objects-class->map
8375: doc---objects-class-inst-size
8376: doc---objects-class-override!
8377: doc---objects-construct
8378: doc---objects-current'
8379: doc---objects-[current]
8380: doc---objects-current-interface
8381: doc---objects-dict-new
8382: doc---objects-drop-order
8383: doc---objects-end-class
8384: doc---objects-end-class-noname
8385: doc---objects-end-interface
8386: doc---objects-end-interface-noname
8387: doc---objects-end-methods
8388: doc---objects-exitm
8389: doc---objects-heap-new
8390: doc---objects-implementation
8391: doc---objects-init-object
8392: doc---objects-inst-value
8393: doc---objects-inst-var
8394: doc---objects-interface
8395: doc---objects-m:
8396: doc---objects-:m
8397: doc---objects-;m
8398: doc---objects-method
8399: doc---objects-methods
8400: doc---objects-object
8401: doc---objects-overrides
8402: doc---objects-[parent]
8403: doc---objects-print
8404: doc---objects-protected
8405: doc---objects-public
8406: doc---objects-push-order
8407: doc---objects-selector
8408: doc---objects-this
8409: doc---objects-<to-inst>
8410: doc---objects-[to-inst]
8411: doc---objects-to-this
8412: doc---objects-xt-new
8413:
8414:
8415: @c -------------------------------------------------------------
8416: @node OOF, Mini-OOF, Objects, Object-oriented Forth
8417: @subsection The @file{oof.fs} model
8418: @cindex oof
8419: @cindex object-oriented programming
8420:
8421: @cindex @file{objects.fs}
8422: @cindex @file{oof.fs}
8423:
8424: This section describes the @file{oof.fs} package.
8425:
8426: The package described in this section has been used in bigFORTH since 1991, and
8427: used for two large applications: a chromatographic system used to
8428: create new medicaments, and a graphic user interface library (MINOS).
8429:
8430: You can find a description (in German) of @file{oof.fs} in @cite{Object
8431: oriented bigFORTH} by Bernd Paysan, published in @cite{Vierte Dimension}
8432: 10(2), 1994.
8433:
8434: @menu
8435: * Properties of the OOF model::
8436: * Basic OOF Usage::
8437: * The OOF base class::
8438: * Class Declaration::
8439: * Class Implementation::
8440: @end menu
8441:
8442: @node Properties of the OOF model, Basic OOF Usage, OOF, OOF
8443: @subsubsection Properties of the @file{oof.fs} model
8444: @cindex @file{oof.fs} properties
8445:
8446: @itemize @bullet
8447: @item
8448: This model combines object oriented programming with information
8449: hiding. It helps you writing large application, where scoping is
8450: necessary, because it provides class-oriented scoping.
8451:
8452: @item
8453: Named objects, object pointers, and object arrays can be created,
8454: selector invocation uses the ``object selector'' syntax. Selector invocation
8455: to objects and/or selectors on the stack is a bit less convenient, but
8456: possible.
8457:
8458: @item
8459: Selector invocation and instance variable usage of the active object is
8460: straightforward, since both make use of the active object.
8461:
8462: @item
8463: Late binding is efficient and easy to use.
8464:
8465: @item
8466: State-smart objects parse selectors. However, extensibility is provided
8467: using a (parsing) selector @code{postpone} and a selector @code{'}.
8468:
8469: @item
8470: An implementation in ANS Forth is available.
8471:
8472: @end itemize
8473:
8474:
8475: @node Basic OOF Usage, The OOF base class, Properties of the OOF model, OOF
8476: @subsubsection Basic @file{oof.fs} Usage
8477: @cindex @file{oof.fs} usage
8478:
8479: This section uses the same example as for @code{objects} (@pxref{Basic Objects Usage}).
8480:
8481: You can define a class for graphical objects like this:
8482:
8483: @cindex @code{class} usage
8484: @cindex @code{class;} usage
8485: @cindex @code{method} usage
8486: @example
8487: object class graphical \ "object" is the parent class
8488: method draw ( x y graphical -- )
8489: class;
8490: @end example
8491:
8492: This code defines a class @code{graphical} with an
8493: operation @code{draw}. We can perform the operation
8494: @code{draw} on any @code{graphical} object, e.g.:
8495:
8496: @example
8497: 100 100 t-rex draw
8498: @end example
8499:
8500: @noindent
8501: where @code{t-rex} is an object or object pointer, created with e.g.
8502: @code{graphical : t-rex}.
8503:
8504: @cindex abstract class
8505: How do we create a graphical object? With the present definitions,
8506: we cannot create a useful graphical object. The class
8507: @code{graphical} describes graphical objects in general, but not
8508: any concrete graphical object type (C++ users would call it an
8509: @emph{abstract class}); e.g., there is no method for the selector
8510: @code{draw} in the class @code{graphical}.
8511:
8512: For concrete graphical objects, we define child classes of the
8513: class @code{graphical}, e.g.:
8514:
8515: @example
8516: graphical class circle \ "graphical" is the parent class
8517: cell var circle-radius
8518: how:
8519: : draw ( x y -- )
8520: circle-radius @@ draw-circle ;
8521:
8522: : init ( n-radius -- (
8523: circle-radius ! ;
8524: class;
8525: @end example
8526:
8527: Here we define a class @code{circle} as a child of @code{graphical},
8528: with a field @code{circle-radius}; it defines new methods for the
8529: selectors @code{draw} and @code{init} (@code{init} is defined in
8530: @code{object}, the parent class of @code{graphical}).
8531:
8532: Now we can create a circle in the dictionary with:
8533:
8534: @example
8535: 50 circle : my-circle
8536: @end example
8537:
8538: @noindent
8539: @code{:} invokes @code{init}, thus initializing the field
8540: @code{circle-radius} with 50. We can draw this new circle at (100,100)
8541: with:
8542:
8543: @example
8544: 100 100 my-circle draw
8545: @end example
8546:
8547: @cindex selector invocation, restrictions
8548: @cindex class definition, restrictions
8549: Note: You can only invoke a selector if the receiving object belongs to
8550: the class where the selector was defined or one of its descendents;
8551: e.g., you can invoke @code{draw} only for objects belonging to
8552: @code{graphical} or its descendents (e.g., @code{circle}). The scoping
8553: mechanism will check if you try to invoke a selector that is not
8554: defined in this class hierarchy, so you'll get an error at compilation
8555: time.
8556:
8557:
8558: @node The OOF base class, Class Declaration, Basic OOF Usage, OOF
8559: @subsubsection The @file{oof.fs} base class
8560: @cindex @file{oof.fs} base class
8561:
8562: When you define a class, you have to specify a parent class. So how do
8563: you start defining classes? There is one class available from the start:
8564: @code{object}. You have to use it as ancestor for all classes. It is the
8565: only class that has no parent. Classes are also objects, except that
8566: they don't have instance variables; class manipulation such as
8567: inheritance or changing definitions of a class is handled through
8568: selectors of the class @code{object}.
8569:
8570: @code{object} provides a number of selectors:
8571:
8572: @itemize @bullet
8573: @item
8574: @code{class} for subclassing, @code{definitions} to add definitions
8575: later on, and @code{class?} to get type informations (is the class a
8576: subclass of the class passed on the stack?).
8577:
8578: doc---object-class
8579: doc---object-definitions
8580: doc---object-class?
8581:
8582:
8583: @item
8584: @code{init} and @code{dispose} as constructor and destructor of the
8585: object. @code{init} is invocated after the object's memory is allocated,
8586: while @code{dispose} also handles deallocation. Thus if you redefine
8587: @code{dispose}, you have to call the parent's dispose with @code{super
8588: dispose}, too.
8589:
8590: doc---object-init
8591: doc---object-dispose
8592:
8593:
8594: @item
8595: @code{new}, @code{new[]}, @code{:}, @code{ptr}, @code{asptr}, and
8596: @code{[]} to create named and unnamed objects and object arrays or
8597: object pointers.
8598:
8599: doc---object-new
8600: doc---object-new[]
8601: doc---object-:
8602: doc---object-ptr
8603: doc---object-asptr
8604: doc---object-[]
8605:
8606:
8607: @item
8608: @code{::} and @code{super} for explicit scoping. You should use explicit
8609: scoping only for super classes or classes with the same set of instance
8610: variables. Explicitly-scoped selectors use early binding.
8611:
8612: doc---object-::
8613: doc---object-super
8614:
8615:
8616: @item
8617: @code{self} to get the address of the object
8618:
8619: doc---object-self
8620:
8621:
8622: @item
8623: @code{bind}, @code{bound}, @code{link}, and @code{is} to assign object
8624: pointers and instance defers.
8625:
8626: doc---object-bind
8627: doc---object-bound
8628: doc---object-link
8629: doc---object-is
8630:
8631:
8632: @item
8633: @code{'} to obtain selector tokens, @code{send} to invocate selectors
8634: form the stack, and @code{postpone} to generate selector invocation code.
8635:
8636: doc---object-'
8637: doc---object-postpone
8638:
8639:
8640: @item
8641: @code{with} and @code{endwith} to select the active object from the
8642: stack, and enable its scope. Using @code{with} and @code{endwith}
8643: also allows you to create code using selector @code{postpone} without being
8644: trapped by the state-smart objects.
8645:
8646: doc---object-with
8647: doc---object-endwith
8648:
8649:
8650: @end itemize
8651:
8652: @node Class Declaration, Class Implementation, The OOF base class, OOF
8653: @subsubsection Class Declaration
8654: @cindex class declaration
8655:
8656: @itemize @bullet
8657: @item
8658: Instance variables
8659:
8660: doc---oof-var
8661:
8662:
8663: @item
8664: Object pointers
8665:
8666: doc---oof-ptr
8667: doc---oof-asptr
8668:
8669:
8670: @item
8671: Instance defers
8672:
8673: doc---oof-defer
8674:
8675:
8676: @item
8677: Method selectors
8678:
8679: doc---oof-early
8680: doc---oof-method
8681:
8682:
8683: @item
8684: Class-wide variables
8685:
8686: doc---oof-static
8687:
8688:
8689: @item
8690: End declaration
8691:
8692: doc---oof-how:
8693: doc---oof-class;
8694:
8695:
8696: @end itemize
8697:
8698: @c -------------------------------------------------------------
8699: @node Class Implementation, , Class Declaration, OOF
8700: @subsubsection Class Implementation
8701: @cindex class implementation
8702:
8703: @c -------------------------------------------------------------
8704: @node Mini-OOF, Comparison with other object models, OOF, Object-oriented Forth
8705: @subsection The @file{mini-oof.fs} model
8706: @cindex mini-oof
8707:
8708: Gforth's third object oriented Forth package is a 12-liner. It uses a
8709: mixture of the @file{object.fs} and the @file{oof.fs} syntax,
8710: and reduces to the bare minimum of features. This is based on a posting
8711: of Bernd Paysan in comp.arch.
8712:
8713: @menu
8714: * Basic Mini-OOF Usage::
8715: * Mini-OOF Example::
8716: * Mini-OOF Implementation::
8717: @end menu
8718:
8719: @c -------------------------------------------------------------
8720: @node Basic Mini-OOF Usage, Mini-OOF Example, , Mini-OOF
8721: @subsubsection Basic @file{mini-oof.fs} Usage
8722: @cindex mini-oof usage
8723:
8724: There is a base class (@code{class}, which allocates one cell for the
8725: object pointer) plus seven other words: to define a method, a variable,
8726: a class; to end a class, to resolve binding, to allocate an object and
8727: to compile a class method.
8728: @comment TODO better description of the last one
8729:
8730:
8731: doc-object
8732: doc-method
8733: doc-var
8734: doc-class
8735: doc-end-class
8736: doc-defines
8737: doc-new
8738: doc-::
8739:
8740:
8741:
8742: @c -------------------------------------------------------------
8743: @node Mini-OOF Example, Mini-OOF Implementation, Basic Mini-OOF Usage, Mini-OOF
8744: @subsubsection Mini-OOF Example
8745: @cindex mini-oof example
8746:
8747: A short example shows how to use this package. This example, in slightly
8748: extended form, is supplied as @file{moof-exm.fs}
8749: @comment TODO could flesh this out with some comments from the Forthwrite article
8750:
8751: @example
8752: object class
8753: method init
8754: method draw
8755: end-class graphical
8756: @end example
8757:
8758: This code defines a class @code{graphical} with an
8759: operation @code{draw}. We can perform the operation
8760: @code{draw} on any @code{graphical} object, e.g.:
8761:
8762: @example
8763: 100 100 t-rex draw
8764: @end example
8765:
8766: where @code{t-rex} is an object or object pointer, created with e.g.
8767: @code{graphical new Constant t-rex}.
8768:
8769: For concrete graphical objects, we define child classes of the
8770: class @code{graphical}, e.g.:
8771:
8772: @example
8773: graphical class
8774: cell var circle-radius
8775: end-class circle \ "graphical" is the parent class
8776:
8777: :noname ( x y -- )
8778: circle-radius @@ draw-circle ; circle defines draw
8779: :noname ( r -- )
8780: circle-radius ! ; circle defines init
8781: @end example
8782:
8783: There is no implicit init method, so we have to define one. The creation
8784: code of the object now has to call init explicitely.
8785:
8786: @example
8787: circle new Constant my-circle
8788: 50 my-circle init
8789: @end example
8790:
8791: It is also possible to add a function to create named objects with
8792: automatic call of @code{init}, given that all objects have @code{init}
8793: on the same place:
8794:
8795: @example
8796: : new: ( .. o "name" -- )
8797: new dup Constant init ;
8798: 80 circle new: large-circle
8799: @end example
8800:
8801: We can draw this new circle at (100,100) with:
8802:
8803: @example
8804: 100 100 my-circle draw
8805: @end example
8806:
8807: @node Mini-OOF Implementation, , Mini-OOF Example, Mini-OOF
8808: @subsubsection @file{mini-oof.fs} Implementation
8809:
8810: Object-oriented systems with late binding typically use a
8811: ``vtable''-approach: the first variable in each object is a pointer to a
8812: table, which contains the methods as function pointers. The vtable
8813: may also contain other information.
8814:
8815: So first, let's declare methods:
8816:
8817: @example
8818: : method ( m v -- m' v ) Create over , swap cell+ swap
8819: DOES> ( ... o -- ... ) @ over @ + @ execute ;
8820: @end example
8821:
8822: During method declaration, the number of methods and instance
8823: variables is on the stack (in address units). @code{method} creates
8824: one method and increments the method number. To execute a method, it
8825: takes the object, fetches the vtable pointer, adds the offset, and
8826: executes the @i{xt} stored there. Each method takes the object it is
8827: invoked from as top of stack parameter. The method itself should
8828: consume that object.
8829:
8830: Now, we also have to declare instance variables
8831:
8832: @example
8833: : var ( m v size -- m v' ) Create over , +
8834: DOES> ( o -- addr ) @ + ;
8835: @end example
8836:
8837: As before, a word is created with the current offset. Instance
8838: variables can have different sizes (cells, floats, doubles, chars), so
8839: all we do is take the size and add it to the offset. If your machine
8840: has alignment restrictions, put the proper @code{aligned} or
8841: @code{faligned} before the variable, to adjust the variable
8842: offset. That's why it is on the top of stack.
8843:
8844: We need a starting point (the base object) and some syntactic sugar:
8845:
8846: @example
8847: Create object 1 cells , 2 cells ,
8848: : class ( class -- class methods vars ) dup 2@ ;
8849: @end example
8850:
8851: For inheritance, the vtable of the parent object has to be
8852: copied when a new, derived class is declared. This gives all the
8853: methods of the parent class, which can be overridden, though.
8854:
8855: @example
8856: : end-class ( class methods vars -- )
8857: Create here >r , dup , 2 cells ?DO ['] noop , 1 cells +LOOP
8858: cell+ dup cell+ r> rot @ 2 cells /string move ;
8859: @end example
8860:
8861: The first line creates the vtable, initialized with
8862: @code{noop}s. The second line is the inheritance mechanism, it
8863: copies the xts from the parent vtable.
8864:
8865: We still have no way to define new methods, let's do that now:
8866:
8867: @example
8868: : defines ( xt class -- ) ' >body @ + ! ;
8869: @end example
8870:
8871: To allocate a new object, we need a word, too:
8872:
8873: @example
8874: : new ( class -- o ) here over @ allot swap over ! ;
8875: @end example
8876:
8877: Sometimes derived classes want to access the method of the
8878: parent object. There are two ways to achieve this with Mini-OOF:
8879: first, you could use named words, and second, you could look up the
8880: vtable of the parent object.
8881:
8882: @example
8883: : :: ( class "name" -- ) ' >body @ + @ compile, ;
8884: @end example
8885:
8886:
8887: Nothing can be more confusing than a good example, so here is
8888: one. First let's declare a text object (called
8889: @code{button}), that stores text and position:
8890:
8891: @example
8892: object class
8893: cell var text
8894: cell var len
8895: cell var x
8896: cell var y
8897: method init
8898: method draw
8899: end-class button
8900: @end example
8901:
8902: @noindent
8903: Now, implement the two methods, @code{draw} and @code{init}:
8904:
8905: @example
8906: :noname ( o -- )
8907: >r r@ x @ r@ y @ at-xy r@ text @ r> len @ type ;
8908: button defines draw
8909: :noname ( addr u o -- )
8910: >r 0 r@ x ! 0 r@ y ! r@ len ! r> text ! ;
8911: button defines init
8912: @end example
8913:
8914: @noindent
8915: To demonstrate inheritance, we define a class @code{bold-button}, with no
8916: new data and no new methods:
8917:
8918: @example
8919: button class
8920: end-class bold-button
8921:
8922: : bold 27 emit ." [1m" ;
8923: : normal 27 emit ." [0m" ;
8924: @end example
8925:
8926: @noindent
8927: The class @code{bold-button} has a different draw method to
8928: @code{button}, but the new method is defined in terms of the draw method
8929: for @code{button}:
8930:
8931: @example
8932: :noname bold [ button :: draw ] normal ; bold-button defines draw
8933: @end example
8934:
8935: @noindent
8936: Finally, create two objects and apply methods:
8937:
8938: @example
8939: button new Constant foo
8940: s" thin foo" foo init
8941: page
8942: foo draw
8943: bold-button new Constant bar
8944: s" fat bar" bar init
8945: 1 bar y !
8946: bar draw
8947: @end example
8948:
8949:
8950: @node Comparison with other object models, , Mini-OOF, Object-oriented Forth
8951: @subsubsection Comparison with other object models
8952: @cindex comparison of object models
8953: @cindex object models, comparison
8954:
8955: Many object-oriented Forth extensions have been proposed (@cite{A survey
8956: of object-oriented Forths} (SIGPLAN Notices, April 1996) by Bradford
8957: J. Rodriguez and W. F. S. Poehlman lists 17). This section discusses the
8958: relation of the object models described here to two well-known and two
8959: closely-related (by the use of method maps) models.
8960:
8961: @cindex Neon model
8962: The most popular model currently seems to be the Neon model (see
8963: @cite{Object-oriented programming in ANS Forth} (Forth Dimensions, March
8964: 1997) by Andrew McKewan) but this model has a number of limitations
8965: @footnote{A longer version of this critique can be
8966: found in @cite{On Standardizing Object-Oriented Forth Extensions} (Forth
8967: Dimensions, May 1997) by Anton Ertl.}:
8968:
8969: @itemize @bullet
8970: @item
8971: It uses a @code{@emph{selector
8972: object}} syntax, which makes it unnatural to pass objects on the
8973: stack.
8974:
8975: @item
8976: It requires that the selector parses the input stream (at
8977: compile time); this leads to reduced extensibility and to bugs that are+
8978: hard to find.
8979:
8980: @item
8981: It allows using every selector to every object;
8982: this eliminates the need for classes, but makes it harder to create
8983: efficient implementations.
8984: @end itemize
8985:
8986: @cindex Pountain's object-oriented model
8987: Another well-known publication is @cite{Object-Oriented Forth} (Academic
8988: Press, London, 1987) by Dick Pountain. However, it is not really about
8989: object-oriented programming, because it hardly deals with late
8990: binding. Instead, it focuses on features like information hiding and
8991: overloading that are characteristic of modular languages like Ada (83).
8992:
8993: @cindex Zsoter's object-oriented model
8994: In @cite{Does late binding have to be slow?} (Forth Dimensions 18(1) 1996, pages 31-35)
8995: Andras Zsoter describes a model that makes heavy use of an active object
8996: (like @code{this} in @file{objects.fs}): The active object is not only
8997: used for accessing all fields, but also specifies the receiving object
8998: of every selector invocation; you have to change the active object
8999: explicitly with @code{@{ ... @}}, whereas in @file{objects.fs} it
9000: changes more or less implicitly at @code{m: ... ;m}. Such a change at
9001: the method entry point is unnecessary with the Zsoter's model, because
9002: the receiving object is the active object already. On the other hand, the explicit
9003: change is absolutely necessary in that model, because otherwise no one
9004: could ever change the active object. An ANS Forth implementation of this
9005: model is available at @uref{http://www.forth.org/fig/oopf.html}.
9006:
9007: @cindex @file{oof.fs}, differences to other models
9008: The @file{oof.fs} model combines information hiding and overloading
9009: resolution (by keeping names in various word lists) with object-oriented
9010: programming. It sets the active object implicitly on method entry, but
9011: also allows explicit changing (with @code{>o...o>} or with
9012: @code{with...endwith}). It uses parsing and state-smart objects and
9013: classes for resolving overloading and for early binding: the object or
9014: class parses the selector and determines the method from this. If the
9015: selector is not parsed by an object or class, it performs a call to the
9016: selector for the active object (late binding), like Zsoter's model.
9017: Fields are always accessed through the active object. The big
9018: disadvantage of this model is the parsing and the state-smartness, which
9019: reduces extensibility and increases the opportunities for subtle bugs;
9020: essentially, you are only safe if you never tick or @code{postpone} an
9021: object or class (Bernd disagrees, but I (Anton) am not convinced).
9022:
9023: @cindex @file{mini-oof.fs}, differences to other models
9024: The @file{mini-oof.fs} model is quite similar to a very stripped-down version of
9025: the @file{objects.fs} model, but syntactically it is a mixture of the @file{objects.fs} and
9026: @file{oof.fs} models.
9027:
9028: @c -------------------------------------------------------------
9029: @node Passing Commands to the OS, Keeping track of Time, Object-oriented Forth, Words
9030: @section Passing Commands to the Operating System
9031: @cindex operating system - passing commands
9032: @cindex shell commands
9033:
9034: Gforth allows you to pass an arbitrary string to the host operating
9035: system shell (if such a thing exists) for execution.
9036:
9037:
9038: doc-sh
9039: doc-system
9040: doc-$?
9041: doc-getenv
9042:
9043:
9044: @c -------------------------------------------------------------
9045: @node Keeping track of Time, Miscellaneous Words, Passing Commands to the OS, Words
9046: @section Keeping track of Time
9047: @cindex time-related words
9048:
9049: Gforth implements time-related operations by making calls to the C
9050: library function, @code{gettimeofday}.
9051:
9052: doc-ms
9053: doc-time&date
9054:
9055:
9056:
9057: @c -------------------------------------------------------------
9058: @node Miscellaneous Words, , Keeping track of Time, Words
9059: @section Miscellaneous Words
9060: @cindex miscellaneous words
9061:
9062: @comment TODO find homes for these
9063:
9064: These section lists the ANS Forth words that are not documented
9065: elsewhere in this manual. Ultimately, they all need proper homes.
9066:
9067: doc-[compile]
9068:
9069:
9070: The following ANS Forth words are not currently supported by Gforth
9071: (@pxref{ANS conformance}):
9072:
9073: @code{EDITOR}
9074: @code{EMIT?}
9075: @code{FORGET}
9076:
9077: @c ******************************************************************
9078: @node Error messages, Tools, Words, Top
9079: @chapter Error messages
9080: @cindex error messages
9081: @cindex backtrace
9082:
9083: A typical Gforth error message looks like this:
9084:
9085: @example
9086: in file included from :-1
9087: in file included from ./yyy.fs:1
9088: ./xxx.fs:4: Invalid memory address
9089: bar
9090: ^^^
9091: $400E664C @@
9092: $400E6664 foo
9093: @end example
9094:
9095: The message identifying the error is @code{Invalid memory address}. The
9096: error happened when text-interpreting line 4 of the file
9097: @file{./xxx.fs}. This line is given (it contains @code{bar}), and the
9098: word on the line where the error happened, is pointed out (with
9099: @code{^^^}).
9100:
9101: The file containing the error was included in line 1 of @file{./yyy.fs},
9102: and @file{yyy.fs} was included from a non-file (in this case, by giving
9103: @file{yyy.fs} as command-line parameter to Gforth).
9104:
9105: At the end of the error message you find a return stack dump that can be
9106: interpreted as a backtrace (possibly empty). On top you find the top of
9107: the return stack when the @code{throw} happened, and at the bottom you
9108: find the return stack entry just above the return stack of the topmost
9109: text interpreter.
9110:
9111: To the right of most return stack entries you see a guess for the word
9112: that pushed that return stack entry as its return address. This gives a
9113: backtrace. In our case we see that @code{bar} called @code{foo}, and
9114: @code{foo} called @code{@@} (and @code{@@} had an @emph{Invalid memory
9115: address} exception).
9116:
9117: Note that the backtrace is not perfect: We don't know which return stack
9118: entries are return addresses (so we may get false positives); and in
9119: some cases (e.g., for @code{abort"}) we cannot determine from the return
9120: address the word that pushed the return address, so for some return
9121: addresses you see no names in the return stack dump.
9122:
9123: @cindex @code{catch} and backtraces
9124: The return stack dump represents the return stack at the time when a
9125: specific @code{throw} was executed. In programs that make use of
9126: @code{catch}, it is not necessarily clear which @code{throw} should be
9127: used for the return stack dump (e.g., consider one @code{throw} that
9128: indicates an error, which is caught, and during recovery another error
9129: happens; which @code{throw} should be used for the stack dump?). Gforth
9130: presents the return stack dump for the first @code{throw} after the last
9131: executed (not returned-to) @code{catch}; this works well in the usual
9132: case.
9133:
9134: @cindex @code{gforth-fast} and backtraces
9135: @cindex @code{gforth-fast}, difference from @code{gforth}
9136: @cindex backtraces with @code{gforth-fast}
9137: @cindex return stack dump with @code{gforth-fast}
9138: @code{gforth} is able to do a return stack dump for throws generated
9139: from primitives (e.g., invalid memory address, stack empty etc.);
9140: @code{gforth-fast} is only able to do a return stack dump from a
9141: directly called @code{throw} (including @code{abort} etc.). This is the
9142: only difference (apart from a speed factor of between 1.15 (K6-2) and
9143: 1.6 (21164A)) between @code{gforth} and @code{gforth-fast}. Given an
9144: exception caused by a primitive in @code{gforth-fast}, you will
9145: typically see no return stack dump at all; however, if the exception is
9146: caught by @code{catch} (e.g., for restoring some state), and then
9147: @code{throw}n again, the return stack dump will be for the first such
9148: @code{throw}.
9149:
9150: @c ******************************************************************
9151: @node Tools, ANS conformance, Error messages, Top
9152: @chapter Tools
9153:
9154: @menu
9155: * ANS Report:: Report the words used, sorted by wordset.
9156: @end menu
9157:
9158: See also @ref{Emacs and Gforth}.
9159:
9160: @node ANS Report, , Tools, Tools
9161: @section @file{ans-report.fs}: Report the words used, sorted by wordset
9162: @cindex @file{ans-report.fs}
9163: @cindex report the words used in your program
9164: @cindex words used in your program
9165:
9166: If you want to label a Forth program as ANS Forth Program, you must
9167: document which wordsets the program uses; for extension wordsets, it is
9168: helpful to list the words the program requires from these wordsets
9169: (because Forth systems are allowed to provide only some words of them).
9170:
9171: The @file{ans-report.fs} tool makes it easy for you to determine which
9172: words from which wordset and which non-ANS words your application
9173: uses. You simply have to include @file{ans-report.fs} before loading the
9174: program you want to check. After loading your program, you can get the
9175: report with @code{print-ans-report}. A typical use is to run this as
9176: batch job like this:
9177: @example
9178: gforth ans-report.fs myprog.fs -e "print-ans-report bye"
9179: @end example
9180:
9181: The output looks like this (for @file{compat/control.fs}):
9182: @example
9183: The program uses the following words
9184: from CORE :
9185: : POSTPONE THEN ; immediate ?dup IF 0=
9186: from BLOCK-EXT :
9187: \
9188: from FILE :
9189: (
9190: @end example
9191:
9192: @subsection Caveats
9193:
9194: Note that @file{ans-report.fs} just checks which words are used, not whether
9195: they are used in an ANS Forth conforming way!
9196:
9197: Some words are defined in several wordsets in the
9198: standard. @file{ans-report.fs} reports them for only one of the
9199: wordsets, and not necessarily the one you expect. It depends on usage
9200: which wordset is the right one to specify. E.g., if you only use the
9201: compilation semantics of @code{S"}, it is a Core word; if you also use
9202: its interpretation semantics, it is a File word.
9203:
9204: @c ******************************************************************
9205: @node ANS conformance, Model, Tools, Top
9206: @chapter ANS conformance
9207: @cindex ANS conformance of Gforth
9208:
9209: To the best of our knowledge, Gforth is an
9210:
9211: ANS Forth System
9212: @itemize @bullet
9213: @item providing the Core Extensions word set
9214: @item providing the Block word set
9215: @item providing the Block Extensions word set
9216: @item providing the Double-Number word set
9217: @item providing the Double-Number Extensions word set
9218: @item providing the Exception word set
9219: @item providing the Exception Extensions word set
9220: @item providing the Facility word set
9221: @item providing @code{EKEY}, @code{EKEY>CHAR}, @code{EKEY?}, @code{MS} and @code{TIME&DATE} from the Facility Extensions word set
9222: @item providing the File Access word set
9223: @item providing the File Access Extensions word set
9224: @item providing the Floating-Point word set
9225: @item providing the Floating-Point Extensions word set
9226: @item providing the Locals word set
9227: @item providing the Locals Extensions word set
9228: @item providing the Memory-Allocation word set
9229: @item providing the Memory-Allocation Extensions word set (that one's easy)
9230: @item providing the Programming-Tools word set
9231: @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
9232: @item providing the Search-Order word set
9233: @item providing the Search-Order Extensions word set
9234: @item providing the String word set
9235: @item providing the String Extensions word set (another easy one)
9236: @end itemize
9237:
9238: @cindex system documentation
9239: In addition, ANS Forth systems are required to document certain
9240: implementation choices. This chapter tries to meet these
9241: requirements. In many cases it gives a way to ask the system for the
9242: information instead of providing the information directly, in
9243: particular, if the information depends on the processor, the operating
9244: system or the installation options chosen, or if they are likely to
9245: change during the maintenance of Gforth.
9246:
9247: @comment The framework for the rest has been taken from pfe.
9248:
9249: @menu
9250: * The Core Words::
9251: * The optional Block word set::
9252: * The optional Double Number word set::
9253: * The optional Exception word set::
9254: * The optional Facility word set::
9255: * The optional File-Access word set::
9256: * The optional Floating-Point word set::
9257: * The optional Locals word set::
9258: * The optional Memory-Allocation word set::
9259: * The optional Programming-Tools word set::
9260: * The optional Search-Order word set::
9261: @end menu
9262:
9263:
9264: @c =====================================================================
9265: @node The Core Words, The optional Block word set, ANS conformance, ANS conformance
9266: @comment node-name, next, previous, up
9267: @section The Core Words
9268: @c =====================================================================
9269: @cindex core words, system documentation
9270: @cindex system documentation, core words
9271:
9272: @menu
9273: * core-idef:: Implementation Defined Options
9274: * core-ambcond:: Ambiguous Conditions
9275: * core-other:: Other System Documentation
9276: @end menu
9277:
9278: @c ---------------------------------------------------------------------
9279: @node core-idef, core-ambcond, The Core Words, The Core Words
9280: @subsection Implementation Defined Options
9281: @c ---------------------------------------------------------------------
9282: @cindex core words, implementation-defined options
9283: @cindex implementation-defined options, core words
9284:
9285:
9286: @table @i
9287: @item (Cell) aligned addresses:
9288: @cindex cell-aligned addresses
9289: @cindex aligned addresses
9290: processor-dependent. Gforth's alignment words perform natural alignment
9291: (e.g., an address aligned for a datum of size 8 is divisible by
9292: 8). Unaligned accesses usually result in a @code{-23 THROW}.
9293:
9294: @item @code{EMIT} and non-graphic characters:
9295: @cindex @code{EMIT} and non-graphic characters
9296: @cindex non-graphic characters and @code{EMIT}
9297: The character is output using the C library function (actually, macro)
9298: @code{putc}.
9299:
9300: @item character editing of @code{ACCEPT} and @code{EXPECT}:
9301: @cindex character editing of @code{ACCEPT} and @code{EXPECT}
9302: @cindex editing in @code{ACCEPT} and @code{EXPECT}
9303: @cindex @code{ACCEPT}, editing
9304: @cindex @code{EXPECT}, editing
9305: This is modeled on the GNU readline library (@pxref{Readline
9306: Interaction, , Command Line Editing, readline, The GNU Readline
9307: Library}) with Emacs-like key bindings. @kbd{Tab} deviates a little by
9308: producing a full word completion every time you type it (instead of
9309: producing the common prefix of all completions). @xref{Command-line editing}.
9310:
9311: @item character set:
9312: @cindex character set
9313: The character set of your computer and display device. Gforth is
9314: 8-bit-clean (but some other component in your system may make trouble).
9315:
9316: @item Character-aligned address requirements:
9317: @cindex character-aligned address requirements
9318: installation-dependent. Currently a character is represented by a C
9319: @code{unsigned char}; in the future we might switch to @code{wchar_t}
9320: (Comments on that requested).
9321:
9322: @item character-set extensions and matching of names:
9323: @cindex character-set extensions and matching of names
9324: @cindex case-sensitivity for name lookup
9325: @cindex name lookup, case-sensitivity
9326: @cindex locale and case-sensitivity
9327: Any character except the ASCII NUL character can be used in a
9328: name. Matching is case-insensitive (except in @code{TABLE}s). The
9329: matching is performed using the C library function @code{strncasecmp}, whose
9330: function is probably influenced by the locale. E.g., the @code{C} locale
9331: does not know about accents and umlauts, so they are matched
9332: case-sensitively in that locale. For portability reasons it is best to
9333: write programs such that they work in the @code{C} locale. Then one can
9334: use libraries written by a Polish programmer (who might use words
9335: containing ISO Latin-2 encoded characters) and by a French programmer
9336: (ISO Latin-1) in the same program (of course, @code{WORDS} will produce
9337: funny results for some of the words (which ones, depends on the font you
9338: are using)). Also, the locale you prefer may not be available in other
9339: operating systems. Hopefully, Unicode will solve these problems one day.
9340:
9341: @item conditions under which control characters match a space delimiter:
9342: @cindex space delimiters
9343: @cindex control characters as delimiters
9344: If @code{WORD} is called with the space character as a delimiter, all
9345: white-space characters (as identified by the C macro @code{isspace()})
9346: are delimiters. @code{PARSE}, on the other hand, treats space like other
9347: delimiters. @code{SWORD} treats space like @code{WORD}, but behaves
9348: like @code{PARSE} otherwise. @code{(NAME)}, which is used by the outer
9349: interpreter (aka text interpreter) by default, treats all white-space
9350: characters as delimiters.
9351:
9352: @item format of the control-flow stack:
9353: @cindex control-flow stack, format
9354: The data stack is used as control-flow stack. The size of a control-flow
9355: stack item in cells is given by the constant @code{cs-item-size}. At the
9356: time of this writing, an item consists of a (pointer to a) locals list
9357: (third), an address in the code (second), and a tag for identifying the
9358: item (TOS). The following tags are used: @code{defstart},
9359: @code{live-orig}, @code{dead-orig}, @code{dest}, @code{do-dest},
9360: @code{scopestart}.
9361:
9362: @item conversion of digits > 35
9363: @cindex digits > 35
9364: The characters @code{[\]^_'} are the digits with the decimal value
9365: 36@minus{}41. There is no way to input many of the larger digits.
9366:
9367: @item display after input terminates in @code{ACCEPT} and @code{EXPECT}:
9368: @cindex @code{EXPECT}, display after end of input
9369: @cindex @code{ACCEPT}, display after end of input
9370: The cursor is moved to the end of the entered string. If the input is
9371: terminated using the @kbd{Return} key, a space is typed.
9372:
9373: @item exception abort sequence of @code{ABORT"}:
9374: @cindex exception abort sequence of @code{ABORT"}
9375: @cindex @code{ABORT"}, exception abort sequence
9376: The error string is stored into the variable @code{"error} and a
9377: @code{-2 throw} is performed.
9378:
9379: @item input line terminator:
9380: @cindex input line terminator
9381: @cindex line terminator on input
9382: @cindex newline character on input
9383: For interactive input, @kbd{C-m} (CR) and @kbd{C-j} (LF) terminate
9384: lines. One of these characters is typically produced when you type the
9385: @kbd{Enter} or @kbd{Return} key.
9386:
9387: @item maximum size of a counted string:
9388: @cindex maximum size of a counted string
9389: @cindex counted string, maximum size
9390: @code{s" /counted-string" environment? drop .}. Currently 255 characters
9391: on all ports, but this may change.
9392:
9393: @item maximum size of a parsed string:
9394: @cindex maximum size of a parsed string
9395: @cindex parsed string, maximum size
9396: Given by the constant @code{/line}. Currently 255 characters.
9397:
9398: @item maximum size of a definition name, in characters:
9399: @cindex maximum size of a definition name, in characters
9400: @cindex name, maximum length
9401: 31
9402:
9403: @item maximum string length for @code{ENVIRONMENT?}, in characters:
9404: @cindex maximum string length for @code{ENVIRONMENT?}, in characters
9405: @cindex @code{ENVIRONMENT?} string length, maximum
9406: 31
9407:
9408: @item method of selecting the user input device:
9409: @cindex user input device, method of selecting
9410: The user input device is the standard input. There is currently no way to
9411: change it from within Gforth. However, the input can typically be
9412: redirected in the command line that starts Gforth.
9413:
9414: @item method of selecting the user output device:
9415: @cindex user output device, method of selecting
9416: @code{EMIT} and @code{TYPE} output to the file-id stored in the value
9417: @code{outfile-id} (@code{stdout} by default). Gforth uses unbuffered
9418: output when the user output device is a terminal, otherwise the output
9419: is buffered.
9420:
9421: @item methods of dictionary compilation:
9422: What are we expected to document here?
9423:
9424: @item number of bits in one address unit:
9425: @cindex number of bits in one address unit
9426: @cindex address unit, size in bits
9427: @code{s" address-units-bits" environment? drop .}. 8 in all current
9428: ports.
9429:
9430: @item number representation and arithmetic:
9431: @cindex number representation and arithmetic
9432: Processor-dependent. Binary two's complement on all current ports.
9433:
9434: @item ranges for integer types:
9435: @cindex ranges for integer types
9436: @cindex integer types, ranges
9437: Installation-dependent. Make environmental queries for @code{MAX-N},
9438: @code{MAX-U}, @code{MAX-D} and @code{MAX-UD}. The lower bounds for
9439: unsigned (and positive) types is 0. The lower bound for signed types on
9440: two's complement and one's complement machines machines can be computed
9441: by adding 1 to the upper bound.
9442:
9443: @item read-only data space regions:
9444: @cindex read-only data space regions
9445: @cindex data-space, read-only regions
9446: The whole Forth data space is writable.
9447:
9448: @item size of buffer at @code{WORD}:
9449: @cindex size of buffer at @code{WORD}
9450: @cindex @code{WORD} buffer size
9451: @code{PAD HERE - .}. 104 characters on 32-bit machines. The buffer is
9452: shared with the pictured numeric output string. If overwriting
9453: @code{PAD} is acceptable, it is as large as the remaining dictionary
9454: space, although only as much can be sensibly used as fits in a counted
9455: string.
9456:
9457: @item size of one cell in address units:
9458: @cindex cell size
9459: @code{1 cells .}.
9460:
9461: @item size of one character in address units:
9462: @cindex char size
9463: @code{1 chars .}. 1 on all current ports.
9464:
9465: @item size of the keyboard terminal buffer:
9466: @cindex size of the keyboard terminal buffer
9467: @cindex terminal buffer, size
9468: Varies. You can determine the size at a specific time using @code{lp@@
9469: tib - .}. It is shared with the locals stack and TIBs of files that
9470: include the current file. You can change the amount of space for TIBs
9471: and locals stack at Gforth startup with the command line option
9472: @code{-l}.
9473:
9474: @item size of the pictured numeric output buffer:
9475: @cindex size of the pictured numeric output buffer
9476: @cindex pictured numeric output buffer, size
9477: @code{PAD HERE - .}. 104 characters on 32-bit machines. The buffer is
9478: shared with @code{WORD}.
9479:
9480: @item size of the scratch area returned by @code{PAD}:
9481: @cindex size of the scratch area returned by @code{PAD}
9482: @cindex @code{PAD} size
9483: The remainder of dictionary space. @code{unused pad here - - .}.
9484:
9485: @item system case-sensitivity characteristics:
9486: @cindex case-sensitivity characteristics
9487: Dictionary searches are case-insensitive (except in
9488: @code{TABLE}s). However, as explained above under @i{character-set
9489: extensions}, the matching for non-ASCII characters is determined by the
9490: locale you are using. In the default @code{C} locale all non-ASCII
9491: characters are matched case-sensitively.
9492:
9493: @item system prompt:
9494: @cindex system prompt
9495: @cindex prompt
9496: @code{ ok} in interpret state, @code{ compiled} in compile state.
9497:
9498: @item division rounding:
9499: @cindex division rounding
9500: installation dependent. @code{s" floored" environment? drop .}. We leave
9501: the choice to @code{gcc} (what to use for @code{/}) and to you (whether
9502: to use @code{fm/mod}, @code{sm/rem} or simply @code{/}).
9503:
9504: @item values of @code{STATE} when true:
9505: @cindex @code{STATE} values
9506: -1.
9507:
9508: @item values returned after arithmetic overflow:
9509: On two's complement machines, arithmetic is performed modulo
9510: 2**bits-per-cell for single arithmetic and 4**bits-per-cell for double
9511: arithmetic (with appropriate mapping for signed types). Division by zero
9512: typically results in a @code{-55 throw} (Floating-point unidentified
9513: fault), although a @code{-10 throw} (divide by zero) would be more
9514: appropriate.
9515:
9516: @item whether the current definition can be found after @t{DOES>}:
9517: @cindex @t{DOES>}, visibility of current definition
9518: No.
9519:
9520: @end table
9521:
9522: @c ---------------------------------------------------------------------
9523: @node core-ambcond, core-other, core-idef, The Core Words
9524: @subsection Ambiguous conditions
9525: @c ---------------------------------------------------------------------
9526: @cindex core words, ambiguous conditions
9527: @cindex ambiguous conditions, core words
9528:
9529: @table @i
9530:
9531: @item a name is neither a word nor a number:
9532: @cindex name not found
9533: @cindex undefined word
9534: @code{-13 throw} (Undefined word). Actually, @code{-13 bounce}, which
9535: preserves the data and FP stack, so you don't lose more work than
9536: necessary.
9537:
9538: @item a definition name exceeds the maximum length allowed:
9539: @cindex word name too long
9540: @code{-19 throw} (Word name too long)
9541:
9542: @item addressing a region not inside the various data spaces of the forth system:
9543: @cindex Invalid memory address
9544: The stacks, code space and header space are accessible. Machine code space is
9545: typically readable. Accessing other addresses gives results dependent on
9546: the operating system. On decent systems: @code{-9 throw} (Invalid memory
9547: address).
9548:
9549: @item argument type incompatible with parameter:
9550: @cindex argument type mismatch
9551: This is usually not caught. Some words perform checks, e.g., the control
9552: flow words, and issue a @code{ABORT"} or @code{-12 THROW} (Argument type
9553: mismatch).
9554:
9555: @item attempting to obtain the execution token of a word with undefined execution semantics:
9556: @cindex Interpreting a compile-only word, for @code{'} etc.
9557: @cindex execution token of words with undefined execution semantics
9558: @code{-14 throw} (Interpreting a compile-only word). In some cases, you
9559: get an execution token for @code{compile-only-error} (which performs a
9560: @code{-14 throw} when executed).
9561:
9562: @item dividing by zero:
9563: @cindex dividing by zero
9564: @cindex floating point unidentified fault, integer division
9565: On better platforms, this produces a @code{-10 throw} (Division by
9566: zero); on other systems, this typically results in a @code{-55 throw}
9567: (Floating-point unidentified fault).
9568:
9569: @item insufficient data stack or return stack space:
9570: @cindex insufficient data stack or return stack space
9571: @cindex stack overflow
9572: @cindex address alignment exception, stack overflow
9573: @cindex Invalid memory address, stack overflow
9574: Depending on the operating system, the installation, and the invocation
9575: of Gforth, this is either checked by the memory management hardware, or
9576: it is not checked. If it is checked, you typically get a @code{-3 throw}
9577: (Stack overflow), @code{-5 throw} (Return stack overflow), or @code{-9
9578: throw} (Invalid memory address) (depending on the platform and how you
9579: achieved the overflow) as soon as the overflow happens. If it is not
9580: checked, overflows typically result in mysterious illegal memory
9581: accesses, producing @code{-9 throw} (Invalid memory address) or
9582: @code{-23 throw} (Address alignment exception); they might also destroy
9583: the internal data structure of @code{ALLOCATE} and friends, resulting in
9584: various errors in these words.
9585:
9586: @item insufficient space for loop control parameters:
9587: @cindex insufficient space for loop control parameters
9588: like other return stack overflows.
9589:
9590: @item insufficient space in the dictionary:
9591: @cindex insufficient space in the dictionary
9592: @cindex dictionary overflow
9593: If you try to allot (either directly with @code{allot}, or indirectly
9594: with @code{,}, @code{create} etc.) more memory than available in the
9595: dictionary, you get a @code{-8 throw} (Dictionary overflow). If you try
9596: to access memory beyond the end of the dictionary, the results are
9597: similar to stack overflows.
9598:
9599: @item interpreting a word with undefined interpretation semantics:
9600: @cindex interpreting a word with undefined interpretation semantics
9601: @cindex Interpreting a compile-only word
9602: For some words, we have defined interpretation semantics. For the
9603: others: @code{-14 throw} (Interpreting a compile-only word).
9604:
9605: @item modifying the contents of the input buffer or a string literal:
9606: @cindex modifying the contents of the input buffer or a string literal
9607: These are located in writable memory and can be modified.
9608:
9609: @item overflow of the pictured numeric output string:
9610: @cindex overflow of the pictured numeric output string
9611: @cindex pictured numeric output string, overflow
9612: @code{-17 throw} (Pictured numeric ouput string overflow).
9613:
9614: @item parsed string overflow:
9615: @cindex parsed string overflow
9616: @code{PARSE} cannot overflow. @code{WORD} does not check for overflow.
9617:
9618: @item producing a result out of range:
9619: @cindex result out of range
9620: On two's complement machines, arithmetic is performed modulo
9621: 2**bits-per-cell for single arithmetic and 4**bits-per-cell for double
9622: arithmetic (with appropriate mapping for signed types). Division by zero
9623: typically results in a @code{-10 throw} (divide by zero) or @code{-55
9624: throw} (floating point unidentified fault). @code{convert} and
9625: @code{>number} currently overflow silently.
9626:
9627: @item reading from an empty data or return stack:
9628: @cindex stack empty
9629: @cindex stack underflow
9630: @cindex return stack underflow
9631: The data stack is checked by the outer (aka text) interpreter after
9632: every word executed. If it has underflowed, a @code{-4 throw} (Stack
9633: underflow) is performed. Apart from that, stacks may be checked or not,
9634: depending on operating system, installation, and invocation. If they are
9635: caught by a check, they typically result in @code{-4 throw} (Stack
9636: underflow), @code{-6 throw} (Return stack underflow) or @code{-9 throw}
9637: (Invalid memory address), depending on the platform and which stack
9638: underflows and by how much. Note that even if the system uses checking
9639: (through the MMU), your program may have to underflow by a significant
9640: number of stack items to trigger the reaction (the reason for this is
9641: that the MMU, and therefore the checking, works with a page-size
9642: granularity). If there is no checking, the symptoms resulting from an
9643: underflow are similar to those from an overflow. Unbalanced return
9644: stack errors result in a variaty of symptoms, including @code{-9 throw}
9645: (Invalid memory address) and Illegal Instruction (typically @code{-260
9646: throw}).
9647:
9648: @item unexpected end of the input buffer, resulting in an attempt to use a zero-length string as a name:
9649: @cindex unexpected end of the input buffer
9650: @cindex zero-length string as a name
9651: @cindex Attempt to use zero-length string as a name
9652: @code{Create} and its descendants perform a @code{-16 throw} (Attempt to
9653: use zero-length string as a name). Words like @code{'} probably will not
9654: find what they search. Note that it is possible to create zero-length
9655: names with @code{nextname} (should it not?).
9656:
9657: @item @code{>IN} greater than input buffer:
9658: @cindex @code{>IN} greater than input buffer
9659: The next invocation of a parsing word returns a string with length 0.
9660:
9661: @item @code{RECURSE} appears after @code{DOES>}:
9662: @cindex @code{RECURSE} appears after @code{DOES>}
9663: Compiles a recursive call to the defining word, not to the defined word.
9664:
9665: @item argument input source different than current input source for @code{RESTORE-INPUT}:
9666: @cindex argument input source different than current input source for @code{RESTORE-INPUT}
9667: @cindex argument type mismatch, @code{RESTORE-INPUT}
9668: @cindex @code{RESTORE-INPUT}, Argument type mismatch
9669: @code{-12 THROW}. Note that, once an input file is closed (e.g., because
9670: the end of the file was reached), its source-id may be
9671: reused. Therefore, restoring an input source specification referencing a
9672: closed file may lead to unpredictable results instead of a @code{-12
9673: THROW}.
9674:
9675: In the future, Gforth may be able to restore input source specifications
9676: from other than the current input source.
9677:
9678: @item data space containing definitions gets de-allocated:
9679: @cindex data space containing definitions gets de-allocated
9680: Deallocation with @code{allot} is not checked. This typically results in
9681: memory access faults or execution of illegal instructions.
9682:
9683: @item data space read/write with incorrect alignment:
9684: @cindex data space read/write with incorrect alignment
9685: @cindex alignment faults
9686: @cindex address alignment exception
9687: Processor-dependent. Typically results in a @code{-23 throw} (Address
9688: alignment exception). Under Linux-Intel on a 486 or later processor with
9689: alignment turned on, incorrect alignment results in a @code{-9 throw}
9690: (Invalid memory address). There are reportedly some processors with
9691: alignment restrictions that do not report violations.
9692:
9693: @item data space pointer not properly aligned, @code{,}, @code{C,}:
9694: @cindex data space pointer not properly aligned, @code{,}, @code{C,}
9695: Like other alignment errors.
9696:
9697: @item less than u+2 stack items (@code{PICK} and @code{ROLL}):
9698: Like other stack underflows.
9699:
9700: @item loop control parameters not available:
9701: @cindex loop control parameters not available
9702: Not checked. The counted loop words simply assume that the top of return
9703: stack items are loop control parameters and behave accordingly.
9704:
9705: @item most recent definition does not have a name (@code{IMMEDIATE}):
9706: @cindex most recent definition does not have a name (@code{IMMEDIATE})
9707: @cindex last word was headerless
9708: @code{abort" last word was headerless"}.
9709:
9710: @item name not defined by @code{VALUE} used by @code{TO}:
9711: @cindex name not defined by @code{VALUE} used by @code{TO}
9712: @cindex @code{TO} on non-@code{VALUE}s
9713: @cindex Invalid name argument, @code{TO}
9714: @code{-32 throw} (Invalid name argument) (unless name is a local or was
9715: defined by @code{CONSTANT}; in the latter case it just changes the constant).
9716:
9717: @item name not found (@code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]}):
9718: @cindex name not found (@code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]})
9719: @cindex undefined word, @code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]}
9720: @code{-13 throw} (Undefined word)
9721:
9722: @item parameters are not of the same type (@code{DO}, @code{?DO}, @code{WITHIN}):
9723: @cindex parameters are not of the same type (@code{DO}, @code{?DO}, @code{WITHIN})
9724: Gforth behaves as if they were of the same type. I.e., you can predict
9725: the behaviour by interpreting all parameters as, e.g., signed.
9726:
9727: @item @code{POSTPONE} or @code{[COMPILE]} applied to @code{TO}:
9728: @cindex @code{POSTPONE} or @code{[COMPILE]} applied to @code{TO}
9729: Assume @code{: X POSTPONE TO ; IMMEDIATE}. @code{X} performs the
9730: compilation semantics of @code{TO}.
9731:
9732: @item String longer than a counted string returned by @code{WORD}:
9733: @cindex string longer than a counted string returned by @code{WORD}
9734: @cindex @code{WORD}, string overflow
9735: Not checked. The string will be ok, but the count will, of course,
9736: contain only the least significant bits of the length.
9737:
9738: @item u greater than or equal to the number of bits in a cell (@code{LSHIFT}, @code{RSHIFT}):
9739: @cindex @code{LSHIFT}, large shift counts
9740: @cindex @code{RSHIFT}, large shift counts
9741: Processor-dependent. Typical behaviours are returning 0 and using only
9742: the low bits of the shift count.
9743:
9744: @item word not defined via @code{CREATE}:
9745: @cindex @code{>BODY} of non-@code{CREATE}d words
9746: @code{>BODY} produces the PFA of the word no matter how it was defined.
9747:
9748: @cindex @code{DOES>} of non-@code{CREATE}d words
9749: @code{DOES>} changes the execution semantics of the last defined word no
9750: matter how it was defined. E.g., @code{CONSTANT DOES>} is equivalent to
9751: @code{CREATE , DOES>}.
9752:
9753: @item words improperly used outside @code{<#} and @code{#>}:
9754: Not checked. As usual, you can expect memory faults.
9755:
9756: @end table
9757:
9758:
9759: @c ---------------------------------------------------------------------
9760: @node core-other, , core-ambcond, The Core Words
9761: @subsection Other system documentation
9762: @c ---------------------------------------------------------------------
9763: @cindex other system documentation, core words
9764: @cindex core words, other system documentation
9765:
9766: @table @i
9767: @item nonstandard words using @code{PAD}:
9768: @cindex @code{PAD} use by nonstandard words
9769: None.
9770:
9771: @item operator's terminal facilities available:
9772: @cindex operator's terminal facilities available
9773: After processing the command line, Gforth goes into interactive mode,
9774: and you can give commands to Gforth interactively. The actual facilities
9775: available depend on how you invoke Gforth.
9776:
9777: @item program data space available:
9778: @cindex program data space available
9779: @cindex data space available
9780: @code{UNUSED .} gives the remaining dictionary space. The total
9781: dictionary space can be specified with the @code{-m} switch
9782: (@pxref{Invoking Gforth}) when Gforth starts up.
9783:
9784: @item return stack space available:
9785: @cindex return stack space available
9786: You can compute the total return stack space in cells with
9787: @code{s" RETURN-STACK-CELLS" environment? drop .}. You can specify it at
9788: startup time with the @code{-r} switch (@pxref{Invoking Gforth}).
9789:
9790: @item stack space available:
9791: @cindex stack space available
9792: You can compute the total data stack space in cells with
9793: @code{s" STACK-CELLS" environment? drop .}. You can specify it at
9794: startup time with the @code{-d} switch (@pxref{Invoking Gforth}).
9795:
9796: @item system dictionary space required, in address units:
9797: @cindex system dictionary space required, in address units
9798: Type @code{here forthstart - .} after startup. At the time of this
9799: writing, this gives 80080 (bytes) on a 32-bit system.
9800: @end table
9801:
9802:
9803: @c =====================================================================
9804: @node The optional Block word set, The optional Double Number word set, The Core Words, ANS conformance
9805: @section The optional Block word set
9806: @c =====================================================================
9807: @cindex system documentation, block words
9808: @cindex block words, system documentation
9809:
9810: @menu
9811: * block-idef:: Implementation Defined Options
9812: * block-ambcond:: Ambiguous Conditions
9813: * block-other:: Other System Documentation
9814: @end menu
9815:
9816:
9817: @c ---------------------------------------------------------------------
9818: @node block-idef, block-ambcond, The optional Block word set, The optional Block word set
9819: @subsection Implementation Defined Options
9820: @c ---------------------------------------------------------------------
9821: @cindex implementation-defined options, block words
9822: @cindex block words, implementation-defined options
9823:
9824: @table @i
9825: @item the format for display by @code{LIST}:
9826: @cindex @code{LIST} display format
9827: First the screen number is displayed, then 16 lines of 64 characters,
9828: each line preceded by the line number.
9829:
9830: @item the length of a line affected by @code{\}:
9831: @cindex length of a line affected by @code{\}
9832: @cindex @code{\}, line length in blocks
9833: 64 characters.
9834: @end table
9835:
9836:
9837: @c ---------------------------------------------------------------------
9838: @node block-ambcond, block-other, block-idef, The optional Block word set
9839: @subsection Ambiguous conditions
9840: @c ---------------------------------------------------------------------
9841: @cindex block words, ambiguous conditions
9842: @cindex ambiguous conditions, block words
9843:
9844: @table @i
9845: @item correct block read was not possible:
9846: @cindex block read not possible
9847: Typically results in a @code{throw} of some OS-derived value (between
9848: -512 and -2048). If the blocks file was just not long enough, blanks are
9849: supplied for the missing portion.
9850:
9851: @item I/O exception in block transfer:
9852: @cindex I/O exception in block transfer
9853: @cindex block transfer, I/O exception
9854: Typically results in a @code{throw} of some OS-derived value (between
9855: -512 and -2048).
9856:
9857: @item invalid block number:
9858: @cindex invalid block number
9859: @cindex block number invalid
9860: @code{-35 throw} (Invalid block number)
9861:
9862: @item a program directly alters the contents of @code{BLK}:
9863: @cindex @code{BLK}, altering @code{BLK}
9864: The input stream is switched to that other block, at the same
9865: position. If the storing to @code{BLK} happens when interpreting
9866: non-block input, the system will get quite confused when the block ends.
9867:
9868: @item no current block buffer for @code{UPDATE}:
9869: @cindex @code{UPDATE}, no current block buffer
9870: @code{UPDATE} has no effect.
9871:
9872: @end table
9873:
9874: @c ---------------------------------------------------------------------
9875: @node block-other, , block-ambcond, The optional Block word set
9876: @subsection Other system documentation
9877: @c ---------------------------------------------------------------------
9878: @cindex other system documentation, block words
9879: @cindex block words, other system documentation
9880:
9881: @table @i
9882: @item any restrictions a multiprogramming system places on the use of buffer addresses:
9883: No restrictions (yet).
9884:
9885: @item the number of blocks available for source and data:
9886: depends on your disk space.
9887:
9888: @end table
9889:
9890:
9891: @c =====================================================================
9892: @node The optional Double Number word set, The optional Exception word set, The optional Block word set, ANS conformance
9893: @section The optional Double Number word set
9894: @c =====================================================================
9895: @cindex system documentation, double words
9896: @cindex double words, system documentation
9897:
9898: @menu
9899: * double-ambcond:: Ambiguous Conditions
9900: @end menu
9901:
9902:
9903: @c ---------------------------------------------------------------------
9904: @node double-ambcond, , The optional Double Number word set, The optional Double Number word set
9905: @subsection Ambiguous conditions
9906: @c ---------------------------------------------------------------------
9907: @cindex double words, ambiguous conditions
9908: @cindex ambiguous conditions, double words
9909:
9910: @table @i
9911: @item @i{d} outside of range of @i{n} in @code{D>S}:
9912: @cindex @code{D>S}, @i{d} out of range of @i{n}
9913: The least significant cell of @i{d} is produced.
9914:
9915: @end table
9916:
9917:
9918: @c =====================================================================
9919: @node The optional Exception word set, The optional Facility word set, The optional Double Number word set, ANS conformance
9920: @section The optional Exception word set
9921: @c =====================================================================
9922: @cindex system documentation, exception words
9923: @cindex exception words, system documentation
9924:
9925: @menu
9926: * exception-idef:: Implementation Defined Options
9927: @end menu
9928:
9929:
9930: @c ---------------------------------------------------------------------
9931: @node exception-idef, , The optional Exception word set, The optional Exception word set
9932: @subsection Implementation Defined Options
9933: @c ---------------------------------------------------------------------
9934: @cindex implementation-defined options, exception words
9935: @cindex exception words, implementation-defined options
9936:
9937: @table @i
9938: @item @code{THROW}-codes used in the system:
9939: @cindex @code{THROW}-codes used in the system
9940: The codes -256@minus{}-511 are used for reporting signals. The mapping
9941: from OS signal numbers to throw codes is -256@minus{}@i{signal}. The
9942: codes -512@minus{}-2047 are used for OS errors (for file and memory
9943: allocation operations). The mapping from OS error numbers to throw codes
9944: is -512@minus{}@code{errno}. One side effect of this mapping is that
9945: undefined OS errors produce a message with a strange number; e.g.,
9946: @code{-1000 THROW} results in @code{Unknown error 488} on my system.
9947: @end table
9948:
9949: @c =====================================================================
9950: @node The optional Facility word set, The optional File-Access word set, The optional Exception word set, ANS conformance
9951: @section The optional Facility word set
9952: @c =====================================================================
9953: @cindex system documentation, facility words
9954: @cindex facility words, system documentation
9955:
9956: @menu
9957: * facility-idef:: Implementation Defined Options
9958: * facility-ambcond:: Ambiguous Conditions
9959: @end menu
9960:
9961:
9962: @c ---------------------------------------------------------------------
9963: @node facility-idef, facility-ambcond, The optional Facility word set, The optional Facility word set
9964: @subsection Implementation Defined Options
9965: @c ---------------------------------------------------------------------
9966: @cindex implementation-defined options, facility words
9967: @cindex facility words, implementation-defined options
9968:
9969: @table @i
9970: @item encoding of keyboard events (@code{EKEY}):
9971: @cindex keyboard events, encoding in @code{EKEY}
9972: @cindex @code{EKEY}, encoding of keyboard events
9973: Keys corresponding to ASCII characters are encoded as ASCII characters.
9974: Other keys are encoded with the constants @code{k-left}, @code{k-right},
9975: @code{k-up}, @code{k-down}, @code{k-home}, @code{k-end}, @code{k1},
9976: @code{k2}, @code{k3}, @code{k4}, @code{k5}, @code{k6}, @code{k7},
9977: @code{k8}, @code{k9}, @code{k10}, @code{k11}, @code{k12}.
9978:
9979:
9980: @item duration of a system clock tick:
9981: @cindex duration of a system clock tick
9982: @cindex clock tick duration
9983: System dependent. With respect to @code{MS}, the time is specified in
9984: microseconds. How well the OS and the hardware implement this, is
9985: another question.
9986:
9987: @item repeatability to be expected from the execution of @code{MS}:
9988: @cindex repeatability to be expected from the execution of @code{MS}
9989: @cindex @code{MS}, repeatability to be expected
9990: System dependent. On Unix, a lot depends on load. If the system is
9991: lightly loaded, and the delay is short enough that Gforth does not get
9992: swapped out, the performance should be acceptable. Under MS-DOS and
9993: other single-tasking systems, it should be good.
9994:
9995: @end table
9996:
9997:
9998: @c ---------------------------------------------------------------------
9999: @node facility-ambcond, , facility-idef, The optional Facility word set
10000: @subsection Ambiguous conditions
10001: @c ---------------------------------------------------------------------
10002: @cindex facility words, ambiguous conditions
10003: @cindex ambiguous conditions, facility words
10004:
10005: @table @i
10006: @item @code{AT-XY} can't be performed on user output device:
10007: @cindex @code{AT-XY} can't be performed on user output device
10008: Largely terminal dependent. No range checks are done on the arguments.
10009: No errors are reported. You may see some garbage appearing, you may see
10010: simply nothing happen.
10011:
10012: @end table
10013:
10014:
10015: @c =====================================================================
10016: @node The optional File-Access word set, The optional Floating-Point word set, The optional Facility word set, ANS conformance
10017: @section The optional File-Access word set
10018: @c =====================================================================
10019: @cindex system documentation, file words
10020: @cindex file words, system documentation
10021:
10022: @menu
10023: * file-idef:: Implementation Defined Options
10024: * file-ambcond:: Ambiguous Conditions
10025: @end menu
10026:
10027: @c ---------------------------------------------------------------------
10028: @node file-idef, file-ambcond, The optional File-Access word set, The optional File-Access word set
10029: @subsection Implementation Defined Options
10030: @c ---------------------------------------------------------------------
10031: @cindex implementation-defined options, file words
10032: @cindex file words, implementation-defined options
10033:
10034: @table @i
10035: @item file access methods used:
10036: @cindex file access methods used
10037: @code{R/O}, @code{R/W} and @code{BIN} work as you would
10038: expect. @code{W/O} translates into the C file opening mode @code{w} (or
10039: @code{wb}): The file is cleared, if it exists, and created, if it does
10040: not (with both @code{open-file} and @code{create-file}). Under Unix
10041: @code{create-file} creates a file with 666 permissions modified by your
10042: umask.
10043:
10044: @item file exceptions:
10045: @cindex file exceptions
10046: The file words do not raise exceptions (except, perhaps, memory access
10047: faults when you pass illegal addresses or file-ids).
10048:
10049: @item file line terminator:
10050: @cindex file line terminator
10051: System-dependent. Gforth uses C's newline character as line
10052: terminator. What the actual character code(s) of this are is
10053: system-dependent.
10054:
10055: @item file name format:
10056: @cindex file name format
10057: System dependent. Gforth just uses the file name format of your OS.
10058:
10059: @item information returned by @code{FILE-STATUS}:
10060: @cindex @code{FILE-STATUS}, returned information
10061: @code{FILE-STATUS} returns the most powerful file access mode allowed
10062: for the file: Either @code{R/O}, @code{W/O} or @code{R/W}. If the file
10063: cannot be accessed, @code{R/O BIN} is returned. @code{BIN} is applicable
10064: along with the returned mode.
10065:
10066: @item input file state after an exception when including source:
10067: @cindex exception when including source
10068: All files that are left via the exception are closed.
10069:
10070: @item @i{ior} values and meaning:
10071: @cindex @i{ior} values and meaning
10072: The @i{ior}s returned by the file and memory allocation words are
10073: intended as throw codes. They typically are in the range
10074: -512@minus{}-2047 of OS errors. The mapping from OS error numbers to
10075: @i{ior}s is -512@minus{}@i{errno}.
10076:
10077: @item maximum depth of file input nesting:
10078: @cindex maximum depth of file input nesting
10079: @cindex file input nesting, maximum depth
10080: limited by the amount of return stack, locals/TIB stack, and the number
10081: of open files available. This should not give you troubles.
10082:
10083: @item maximum size of input line:
10084: @cindex maximum size of input line
10085: @cindex input line size, maximum
10086: @code{/line}. Currently 255.
10087:
10088: @item methods of mapping block ranges to files:
10089: @cindex mapping block ranges to files
10090: @cindex files containing blocks
10091: @cindex blocks in files
10092: By default, blocks are accessed in the file @file{blocks.fb} in the
10093: current working directory. The file can be switched with @code{USE}.
10094:
10095: @item number of string buffers provided by @code{S"}:
10096: @cindex @code{S"}, number of string buffers
10097: 1
10098:
10099: @item size of string buffer used by @code{S"}:
10100: @cindex @code{S"}, size of string buffer
10101: @code{/line}. currently 255.
10102:
10103: @end table
10104:
10105: @c ---------------------------------------------------------------------
10106: @node file-ambcond, , file-idef, The optional File-Access word set
10107: @subsection Ambiguous conditions
10108: @c ---------------------------------------------------------------------
10109: @cindex file words, ambiguous conditions
10110: @cindex ambiguous conditions, file words
10111:
10112: @table @i
10113: @item attempting to position a file outside its boundaries:
10114: @cindex @code{REPOSITION-FILE}, outside the file's boundaries
10115: @code{REPOSITION-FILE} is performed as usual: Afterwards,
10116: @code{FILE-POSITION} returns the value given to @code{REPOSITION-FILE}.
10117:
10118: @item attempting to read from file positions not yet written:
10119: @cindex reading from file positions not yet written
10120: End-of-file, i.e., zero characters are read and no error is reported.
10121:
10122: @item @i{file-id} is invalid (@code{INCLUDE-FILE}):
10123: @cindex @code{INCLUDE-FILE}, @i{file-id} is invalid
10124: An appropriate exception may be thrown, but a memory fault or other
10125: problem is more probable.
10126:
10127: @item I/O exception reading or closing @i{file-id} (@code{INCLUDE-FILE}, @code{INCLUDED}):
10128: @cindex @code{INCLUDE-FILE}, I/O exception reading or closing @i{file-id}
10129: @cindex @code{INCLUDED}, I/O exception reading or closing @i{file-id}
10130: The @i{ior} produced by the operation, that discovered the problem, is
10131: thrown.
10132:
10133: @item named file cannot be opened (@code{INCLUDED}):
10134: @cindex @code{INCLUDED}, named file cannot be opened
10135: The @i{ior} produced by @code{open-file} is thrown.
10136:
10137: @item requesting an unmapped block number:
10138: @cindex unmapped block numbers
10139: There are no unmapped legal block numbers. On some operating systems,
10140: writing a block with a large number may overflow the file system and
10141: have an error message as consequence.
10142:
10143: @item using @code{source-id} when @code{blk} is non-zero:
10144: @cindex @code{SOURCE-ID}, behaviour when @code{BLK} is non-zero
10145: @code{source-id} performs its function. Typically it will give the id of
10146: the source which loaded the block. (Better ideas?)
10147:
10148: @end table
10149:
10150:
10151: @c =====================================================================
10152: @node The optional Floating-Point word set, The optional Locals word set, The optional File-Access word set, ANS conformance
10153: @section The optional Floating-Point word set
10154: @c =====================================================================
10155: @cindex system documentation, floating-point words
10156: @cindex floating-point words, system documentation
10157:
10158: @menu
10159: * floating-idef:: Implementation Defined Options
10160: * floating-ambcond:: Ambiguous Conditions
10161: @end menu
10162:
10163:
10164: @c ---------------------------------------------------------------------
10165: @node floating-idef, floating-ambcond, The optional Floating-Point word set, The optional Floating-Point word set
10166: @subsection Implementation Defined Options
10167: @c ---------------------------------------------------------------------
10168: @cindex implementation-defined options, floating-point words
10169: @cindex floating-point words, implementation-defined options
10170:
10171: @table @i
10172: @item format and range of floating point numbers:
10173: @cindex format and range of floating point numbers
10174: @cindex floating point numbers, format and range
10175: System-dependent; the @code{double} type of C.
10176:
10177: @item results of @code{REPRESENT} when @i{float} is out of range:
10178: @cindex @code{REPRESENT}, results when @i{float} is out of range
10179: System dependent; @code{REPRESENT} is implemented using the C library
10180: function @code{ecvt()} and inherits its behaviour in this respect.
10181:
10182: @item rounding or truncation of floating-point numbers:
10183: @cindex rounding of floating-point numbers
10184: @cindex truncation of floating-point numbers
10185: @cindex floating-point numbers, rounding or truncation
10186: System dependent; the rounding behaviour is inherited from the hosting C
10187: compiler. IEEE-FP-based (i.e., most) systems by default round to
10188: nearest, and break ties by rounding to even (i.e., such that the last
10189: bit of the mantissa is 0).
10190:
10191: @item size of floating-point stack:
10192: @cindex floating-point stack size
10193: @code{s" FLOATING-STACK" environment? drop .} gives the total size of
10194: the floating-point stack (in floats). You can specify this on startup
10195: with the command-line option @code{-f} (@pxref{Invoking Gforth}).
10196:
10197: @item width of floating-point stack:
10198: @cindex floating-point stack width
10199: @code{1 floats}.
10200:
10201: @end table
10202:
10203:
10204: @c ---------------------------------------------------------------------
10205: @node floating-ambcond, , floating-idef, The optional Floating-Point word set
10206: @subsection Ambiguous conditions
10207: @c ---------------------------------------------------------------------
10208: @cindex floating-point words, ambiguous conditions
10209: @cindex ambiguous conditions, floating-point words
10210:
10211: @table @i
10212: @item @code{df@@} or @code{df!} used with an address that is not double-float aligned:
10213: @cindex @code{df@@} or @code{df!} used with an address that is not double-float aligned
10214: System-dependent. Typically results in a @code{-23 THROW} like other
10215: alignment violations.
10216:
10217: @item @code{f@@} or @code{f!} used with an address that is not float aligned:
10218: @cindex @code{f@@} used with an address that is not float aligned
10219: @cindex @code{f!} used with an address that is not float aligned
10220: System-dependent. Typically results in a @code{-23 THROW} like other
10221: alignment violations.
10222:
10223: @item floating-point result out of range:
10224: @cindex floating-point result out of range
10225: System-dependent. Can result in a @code{-55 THROW} (Floating-point
10226: unidentified fault), or can produce a special value representing, e.g.,
10227: Infinity.
10228:
10229: @item @code{sf@@} or @code{sf!} used with an address that is not single-float aligned:
10230: @cindex @code{sf@@} or @code{sf!} used with an address that is not single-float aligned
10231: System-dependent. Typically results in an alignment fault like other
10232: alignment violations.
10233:
10234: @item @code{base} is not decimal (@code{REPRESENT}, @code{F.}, @code{FE.}, @code{FS.}):
10235: @cindex @code{base} is not decimal (@code{REPRESENT}, @code{F.}, @code{FE.}, @code{FS.})
10236: The floating-point number is converted into decimal nonetheless.
10237:
10238: @item Both arguments are equal to zero (@code{FATAN2}):
10239: @cindex @code{FATAN2}, both arguments are equal to zero
10240: System-dependent. @code{FATAN2} is implemented using the C library
10241: function @code{atan2()}.
10242:
10243: @item Using @code{FTAN} on an argument @i{r1} where cos(@i{r1}) is zero:
10244: @cindex @code{FTAN} on an argument @i{r1} where cos(@i{r1}) is zero
10245: System-dependent. Anyway, typically the cos of @i{r1} will not be zero
10246: because of small errors and the tan will be a very large (or very small)
10247: but finite number.
10248:
10249: @item @i{d} cannot be presented precisely as a float in @code{D>F}:
10250: @cindex @code{D>F}, @i{d} cannot be presented precisely as a float
10251: The result is rounded to the nearest float.
10252:
10253: @item dividing by zero:
10254: @cindex dividing by zero, floating-point
10255: @cindex floating-point dividing by zero
10256: @cindex floating-point unidentified fault, FP divide-by-zero
10257: @code{-55 throw} (Floating-point unidentified fault)
10258:
10259: @item exponent too big for conversion (@code{DF!}, @code{DF@@}, @code{SF!}, @code{SF@@}):
10260: @cindex exponent too big for conversion (@code{DF!}, @code{DF@@}, @code{SF!}, @code{SF@@})
10261: System dependent. On IEEE-FP based systems the number is converted into
10262: an infinity.
10263:
10264: @item @i{float}<1 (@code{FACOSH}):
10265: @cindex @code{FACOSH}, @i{float}<1
10266: @cindex floating-point unidentified fault, @code{FACOSH}
10267: @code{-55 throw} (Floating-point unidentified fault)
10268:
10269: @item @i{float}=<-1 (@code{FLNP1}):
10270: @cindex @code{FLNP1}, @i{float}=<-1
10271: @cindex floating-point unidentified fault, @code{FLNP1}
10272: @code{-55 throw} (Floating-point unidentified fault). On IEEE-FP systems
10273: negative infinity is typically produced for @i{float}=-1.
10274:
10275: @item @i{float}=<0 (@code{FLN}, @code{FLOG}):
10276: @cindex @code{FLN}, @i{float}=<0
10277: @cindex @code{FLOG}, @i{float}=<0
10278: @cindex floating-point unidentified fault, @code{FLN} or @code{FLOG}
10279: @code{-55 throw} (Floating-point unidentified fault). On IEEE-FP systems
10280: negative infinity is typically produced for @i{float}=0.
10281:
10282: @item @i{float}<0 (@code{FASINH}, @code{FSQRT}):
10283: @cindex @code{FASINH}, @i{float}<0
10284: @cindex @code{FSQRT}, @i{float}<0
10285: @cindex floating-point unidentified fault, @code{FASINH} or @code{FSQRT}
10286: @code{-55 throw} (Floating-point unidentified fault). @code{fasinh}
10287: produces values for these inputs on my Linux box (Bug in the C library?)
10288:
10289: @item |@i{float}|>1 (@code{FACOS}, @code{FASIN}, @code{FATANH}):
10290: @cindex @code{FACOS}, |@i{float}|>1
10291: @cindex @code{FASIN}, |@i{float}|>1
10292: @cindex @code{FATANH}, |@i{float}|>1
10293: @cindex floating-point unidentified fault, @code{FACOS}, @code{FASIN} or @code{FATANH}
10294: @code{-55 throw} (Floating-point unidentified fault).
10295:
10296: @item integer part of float cannot be represented by @i{d} in @code{F>D}:
10297: @cindex @code{F>D}, integer part of float cannot be represented by @i{d}
10298: @cindex floating-point unidentified fault, @code{F>D}
10299: @code{-55 throw} (Floating-point unidentified fault).
10300:
10301: @item string larger than pictured numeric output area (@code{f.}, @code{fe.}, @code{fs.}):
10302: @cindex string larger than pictured numeric output area (@code{f.}, @code{fe.}, @code{fs.})
10303: This does not happen.
10304: @end table
10305:
10306: @c =====================================================================
10307: @node The optional Locals word set, The optional Memory-Allocation word set, The optional Floating-Point word set, ANS conformance
10308: @section The optional Locals word set
10309: @c =====================================================================
10310: @cindex system documentation, locals words
10311: @cindex locals words, system documentation
10312:
10313: @menu
10314: * locals-idef:: Implementation Defined Options
10315: * locals-ambcond:: Ambiguous Conditions
10316: @end menu
10317:
10318:
10319: @c ---------------------------------------------------------------------
10320: @node locals-idef, locals-ambcond, The optional Locals word set, The optional Locals word set
10321: @subsection Implementation Defined Options
10322: @c ---------------------------------------------------------------------
10323: @cindex implementation-defined options, locals words
10324: @cindex locals words, implementation-defined options
10325:
10326: @table @i
10327: @item maximum number of locals in a definition:
10328: @cindex maximum number of locals in a definition
10329: @cindex locals, maximum number in a definition
10330: @code{s" #locals" environment? drop .}. Currently 15. This is a lower
10331: bound, e.g., on a 32-bit machine there can be 41 locals of up to 8
10332: characters. The number of locals in a definition is bounded by the size
10333: of locals-buffer, which contains the names of the locals.
10334:
10335: @end table
10336:
10337:
10338: @c ---------------------------------------------------------------------
10339: @node locals-ambcond, , locals-idef, The optional Locals word set
10340: @subsection Ambiguous conditions
10341: @c ---------------------------------------------------------------------
10342: @cindex locals words, ambiguous conditions
10343: @cindex ambiguous conditions, locals words
10344:
10345: @table @i
10346: @item executing a named local in interpretation state:
10347: @cindex local in interpretation state
10348: @cindex Interpreting a compile-only word, for a local
10349: Locals have no interpretation semantics. If you try to perform the
10350: interpretation semantics, you will get a @code{-14 throw} somewhere
10351: (Interpreting a compile-only word). If you perform the compilation
10352: semantics, the locals access will be compiled (irrespective of state).
10353:
10354: @item @i{name} not defined by @code{VALUE} or @code{(LOCAL)} (@code{TO}):
10355: @cindex name not defined by @code{VALUE} or @code{(LOCAL)} used by @code{TO}
10356: @cindex @code{TO} on non-@code{VALUE}s and non-locals
10357: @cindex Invalid name argument, @code{TO}
10358: @code{-32 throw} (Invalid name argument)
10359:
10360: @end table
10361:
10362:
10363: @c =====================================================================
10364: @node The optional Memory-Allocation word set, The optional Programming-Tools word set, The optional Locals word set, ANS conformance
10365: @section The optional Memory-Allocation word set
10366: @c =====================================================================
10367: @cindex system documentation, memory-allocation words
10368: @cindex memory-allocation words, system documentation
10369:
10370: @menu
10371: * memory-idef:: Implementation Defined Options
10372: @end menu
10373:
10374:
10375: @c ---------------------------------------------------------------------
10376: @node memory-idef, , The optional Memory-Allocation word set, The optional Memory-Allocation word set
10377: @subsection Implementation Defined Options
10378: @c ---------------------------------------------------------------------
10379: @cindex implementation-defined options, memory-allocation words
10380: @cindex memory-allocation words, implementation-defined options
10381:
10382: @table @i
10383: @item values and meaning of @i{ior}:
10384: @cindex @i{ior} values and meaning
10385: The @i{ior}s returned by the file and memory allocation words are
10386: intended as throw codes. They typically are in the range
10387: -512@minus{}-2047 of OS errors. The mapping from OS error numbers to
10388: @i{ior}s is -512@minus{}@i{errno}.
10389:
10390: @end table
10391:
10392: @c =====================================================================
10393: @node The optional Programming-Tools word set, The optional Search-Order word set, The optional Memory-Allocation word set, ANS conformance
10394: @section The optional Programming-Tools word set
10395: @c =====================================================================
10396: @cindex system documentation, programming-tools words
10397: @cindex programming-tools words, system documentation
10398:
10399: @menu
10400: * programming-idef:: Implementation Defined Options
10401: * programming-ambcond:: Ambiguous Conditions
10402: @end menu
10403:
10404:
10405: @c ---------------------------------------------------------------------
10406: @node programming-idef, programming-ambcond, The optional Programming-Tools word set, The optional Programming-Tools word set
10407: @subsection Implementation Defined Options
10408: @c ---------------------------------------------------------------------
10409: @cindex implementation-defined options, programming-tools words
10410: @cindex programming-tools words, implementation-defined options
10411:
10412: @table @i
10413: @item ending sequence for input following @code{;CODE} and @code{CODE}:
10414: @cindex @code{;CODE} ending sequence
10415: @cindex @code{CODE} ending sequence
10416: @code{END-CODE}
10417:
10418: @item manner of processing input following @code{;CODE} and @code{CODE}:
10419: @cindex @code{;CODE}, processing input
10420: @cindex @code{CODE}, processing input
10421: The @code{ASSEMBLER} vocabulary is pushed on the search order stack, and
10422: the input is processed by the text interpreter, (starting) in interpret
10423: state.
10424:
10425: @item search order capability for @code{EDITOR} and @code{ASSEMBLER}:
10426: @cindex @code{ASSEMBLER}, search order capability
10427: The ANS Forth search order word set.
10428:
10429: @item source and format of display by @code{SEE}:
10430: @cindex @code{SEE}, source and format of output
10431: The source for @code{see} is the intermediate code used by the inner
10432: interpreter. The current @code{see} tries to output Forth source code
10433: as well as possible.
10434:
10435: @end table
10436:
10437: @c ---------------------------------------------------------------------
10438: @node programming-ambcond, , programming-idef, The optional Programming-Tools word set
10439: @subsection Ambiguous conditions
10440: @c ---------------------------------------------------------------------
10441: @cindex programming-tools words, ambiguous conditions
10442: @cindex ambiguous conditions, programming-tools words
10443:
10444: @table @i
10445:
10446: @item deleting the compilation word list (@code{FORGET}):
10447: @cindex @code{FORGET}, deleting the compilation word list
10448: Not implemented (yet).
10449:
10450: @item fewer than @i{u}+1 items on the control-flow stack (@code{CS-PICK}, @code{CS-ROLL}):
10451: @cindex @code{CS-PICK}, fewer than @i{u}+1 items on the control flow-stack
10452: @cindex @code{CS-ROLL}, fewer than @i{u}+1 items on the control flow-stack
10453: @cindex control-flow stack underflow
10454: This typically results in an @code{abort"} with a descriptive error
10455: message (may change into a @code{-22 throw} (Control structure mismatch)
10456: in the future). You may also get a memory access error. If you are
10457: unlucky, this ambiguous condition is not caught.
10458:
10459: @item @i{name} can't be found (@code{FORGET}):
10460: @cindex @code{FORGET}, @i{name} can't be found
10461: Not implemented (yet).
10462:
10463: @item @i{name} not defined via @code{CREATE}:
10464: @cindex @code{;CODE}, @i{name} not defined via @code{CREATE}
10465: @code{;CODE} behaves like @code{DOES>} in this respect, i.e., it changes
10466: the execution semantics of the last defined word no matter how it was
10467: defined.
10468:
10469: @item @code{POSTPONE} applied to @code{[IF]}:
10470: @cindex @code{POSTPONE} applied to @code{[IF]}
10471: @cindex @code{[IF]} and @code{POSTPONE}
10472: After defining @code{: X POSTPONE [IF] ; IMMEDIATE}. @code{X} is
10473: equivalent to @code{[IF]}.
10474:
10475: @item reaching the end of the input source before matching @code{[ELSE]} or @code{[THEN]}:
10476: @cindex @code{[IF]}, end of the input source before matching @code{[ELSE]} or @code{[THEN]}
10477: Continue in the same state of conditional compilation in the next outer
10478: input source. Currently there is no warning to the user about this.
10479:
10480: @item removing a needed definition (@code{FORGET}):
10481: @cindex @code{FORGET}, removing a needed definition
10482: Not implemented (yet).
10483:
10484: @end table
10485:
10486:
10487: @c =====================================================================
10488: @node The optional Search-Order word set, , The optional Programming-Tools word set, ANS conformance
10489: @section The optional Search-Order word set
10490: @c =====================================================================
10491: @cindex system documentation, search-order words
10492: @cindex search-order words, system documentation
10493:
10494: @menu
10495: * search-idef:: Implementation Defined Options
10496: * search-ambcond:: Ambiguous Conditions
10497: @end menu
10498:
10499:
10500: @c ---------------------------------------------------------------------
10501: @node search-idef, search-ambcond, The optional Search-Order word set, The optional Search-Order word set
10502: @subsection Implementation Defined Options
10503: @c ---------------------------------------------------------------------
10504: @cindex implementation-defined options, search-order words
10505: @cindex search-order words, implementation-defined options
10506:
10507: @table @i
10508: @item maximum number of word lists in search order:
10509: @cindex maximum number of word lists in search order
10510: @cindex search order, maximum depth
10511: @code{s" wordlists" environment? drop .}. Currently 16.
10512:
10513: @item minimum search order:
10514: @cindex minimum search order
10515: @cindex search order, minimum
10516: @code{root root}.
10517:
10518: @end table
10519:
10520: @c ---------------------------------------------------------------------
10521: @node search-ambcond, , search-idef, The optional Search-Order word set
10522: @subsection Ambiguous conditions
10523: @c ---------------------------------------------------------------------
10524: @cindex search-order words, ambiguous conditions
10525: @cindex ambiguous conditions, search-order words
10526:
10527: @table @i
10528: @item changing the compilation word list (during compilation):
10529: @cindex changing the compilation word list (during compilation)
10530: @cindex compilation word list, change before definition ends
10531: The word is entered into the word list that was the compilation word list
10532: at the start of the definition. Any changes to the name field (e.g.,
10533: @code{immediate}) or the code field (e.g., when executing @code{DOES>})
10534: are applied to the latest defined word (as reported by @code{last} or
10535: @code{lastxt}), if possible, irrespective of the compilation word list.
10536:
10537: @item search order empty (@code{previous}):
10538: @cindex @code{previous}, search order empty
10539: @cindex vocstack empty, @code{previous}
10540: @code{abort" Vocstack empty"}.
10541:
10542: @item too many word lists in search order (@code{also}):
10543: @cindex @code{also}, too many word lists in search order
10544: @cindex vocstack full, @code{also}
10545: @code{abort" Vocstack full"}.
10546:
10547: @end table
10548:
10549: @c ***************************************************************
10550: @node Model, Integrating Gforth, ANS conformance, Top
10551: @chapter Model
10552:
10553: This chapter has yet to be written. It will contain information, on
10554: which internal structures you can rely.
10555:
10556: @c ***************************************************************
10557: @node Integrating Gforth, Emacs and Gforth, Model, Top
10558: @chapter Integrating Gforth into C programs
10559:
10560: This is not yet implemented.
10561:
10562: Several people like to use Forth as scripting language for applications
10563: that are otherwise written in C, C++, or some other language.
10564:
10565: The Forth system ATLAST provides facilities for embedding it into
10566: applications; unfortunately it has several disadvantages: most
10567: importantly, it is not based on ANS Forth, and it is apparently dead
10568: (i.e., not developed further and not supported). The facilities
10569: provided by Gforth in this area are inspired by ATLAST's facilities, so
10570: making the switch should not be hard.
10571:
10572: We also tried to design the interface such that it can easily be
10573: implemented by other Forth systems, so that we may one day arrive at a
10574: standardized interface. Such a standard interface would allow you to
10575: replace the Forth system without having to rewrite C code.
10576:
10577: You embed the Gforth interpreter by linking with the library
10578: @code{libgforth.a} (give the compiler the option @code{-lgforth}). All
10579: global symbols in this library that belong to the interface, have the
10580: prefix @code{forth_}. (Global symbols that are used internally have the
10581: prefix @code{gforth_}).
10582:
10583: You can include the declarations of Forth types and the functions and
10584: variables of the interface with @code{#include <forth.h>}.
10585:
10586: Types.
10587:
10588: Variables.
10589:
10590: Data and FP Stack pointer. Area sizes.
10591:
10592: functions.
10593:
10594: forth_init(imagefile)
10595: forth_evaluate(string) exceptions?
10596: forth_goto(address) (or forth_execute(xt)?)
10597: forth_continue() (a corountining mechanism)
10598:
10599: Adding primitives.
10600:
10601: No checking.
10602:
10603: Signals?
10604:
10605: Accessing the Stacks
10606:
10607: @c ******************************************************************
10608: @node Emacs and Gforth, Image Files, Integrating Gforth, Top
10609: @chapter Emacs and Gforth
10610: @cindex Emacs and Gforth
10611:
10612: @cindex @file{gforth.el}
10613: @cindex @file{forth.el}
10614: @cindex Rydqvist, Goran
10615: @cindex comment editing commands
10616: @cindex @code{\}, editing with Emacs
10617: @cindex debug tracer editing commands
10618: @cindex @code{~~}, removal with Emacs
10619: @cindex Forth mode in Emacs
10620: Gforth comes with @file{gforth.el}, an improved version of
10621: @file{forth.el} by Goran Rydqvist (included in the TILE package). The
10622: improvements are:
10623:
10624: @itemize @bullet
10625: @item
10626: A better (but still not perfect) handling of indentation.
10627: @item
10628: Comment paragraph filling (@kbd{M-q})
10629: @item
10630: Commenting (@kbd{C-x \}) and uncommenting (@kbd{C-u C-x \}) of regions
10631: @item
10632: Removal of debugging tracers (@kbd{C-x ~}, @pxref{Debugging}).
10633: @item
10634: Support of the @code{info-lookup} feature for looking up the
10635: documentation of a word.
10636: @end itemize
10637:
10638: I left the stuff I do not use alone, even though some of it only makes
10639: sense for TILE. To get a description of these features, enter Forth mode
10640: and type @kbd{C-h m}.
10641:
10642: @cindex source location of error or debugging output in Emacs
10643: @cindex error output, finding the source location in Emacs
10644: @cindex debugging output, finding the source location in Emacs
10645: In addition, Gforth supports Emacs quite well: The source code locations
10646: given in error messages, debugging output (from @code{~~}) and failed
10647: assertion messages are in the right format for Emacs' compilation mode
10648: (@pxref{Compilation, , Running Compilations under Emacs, emacs, Emacs
10649: Manual}) so the source location corresponding to an error or other
10650: message is only a few keystrokes away (@kbd{C-x `} for the next error,
10651: @kbd{C-c C-c} for the error under the cursor).
10652:
10653: @cindex @file{TAGS} file
10654: @cindex @file{etags.fs}
10655: @cindex viewing the source of a word in Emacs
10656: @cindex @code{require}, placement in files
10657: @cindex @code{include}, placement in files
10658: Also, if you @code{require} @file{etags.fs}, a new @file{TAGS} file will
10659: be produced (@pxref{Tags, , Tags Tables, emacs, Emacs Manual}) that
10660: contains the definitions of all words defined afterwards. You can then
10661: find the source for a word using @kbd{M-.}. Note that emacs can use
10662: several tags files at the same time (e.g., one for the Gforth sources
10663: and one for your program, @pxref{Select Tags Table,,Selecting a Tags
10664: Table,emacs, Emacs Manual}). The TAGS file for the preloaded words is
10665: @file{$(datadir)/gforth/$(VERSION)/TAGS} (e.g.,
10666: @file{/usr/local/share/gforth/0.2.0/TAGS}). To get the best behaviour
10667: with @file{etags.fs}, you should avoid putting definitions both before
10668: and after @code{require} etc., otherwise you will see the same file
10669: visited several times by commands like @code{tags-search}.
10670:
10671: @cindex viewing the documentation of a word in Emacs
10672: @cindex context-sensitive help
10673: Moreover, for words documented in this manual, you can look up the
10674: glossary entry quickly by using @kbd{C-h TAB}
10675: (@code{info-lookup-symbol}, see @pxref{Documentation, ,Documentation
10676: Commands, emacs, Emacs Manual}). This feature requires Emacs 20.3 or
10677: later and does not work for words containing @code{:}.
10678:
10679:
10680: @cindex @file{.emacs}
10681: To get all these benefits, add the following lines to your @file{.emacs}
10682: file:
10683:
10684: @example
10685: (autoload 'forth-mode "gforth.el")
10686: (setq auto-mode-alist (cons '("\\.fs\\'" . forth-mode) auto-mode-alist))
10687: @end example
10688:
10689: @c ******************************************************************
10690: @node Image Files, Engine, Emacs and Gforth, Top
10691: @chapter Image Files
10692: @cindex image file
10693: @cindex @file{.fi} files
10694: @cindex precompiled Forth code
10695: @cindex dictionary in persistent form
10696: @cindex persistent form of dictionary
10697:
10698: An image file is a file containing an image of the Forth dictionary,
10699: i.e., compiled Forth code and data residing in the dictionary. By
10700: convention, we use the extension @code{.fi} for image files.
10701:
10702: @menu
10703: * Image Licensing Issues:: Distribution terms for images.
10704: * Image File Background:: Why have image files?
10705: * Non-Relocatable Image Files:: don't always work.
10706: * Data-Relocatable Image Files:: are better.
10707: * Fully Relocatable Image Files:: better yet.
10708: * Stack and Dictionary Sizes:: Setting the default sizes for an image.
10709: * Running Image Files:: @code{gforth -i @i{file}} or @i{file}.
10710: * Modifying the Startup Sequence:: and turnkey applications.
10711: @end menu
10712:
10713: @node Image Licensing Issues, Image File Background, Image Files, Image Files
10714: @section Image Licensing Issues
10715: @cindex license for images
10716: @cindex image license
10717:
10718: An image created with @code{gforthmi} (@pxref{gforthmi}) or
10719: @code{savesystem} (@pxref{Non-Relocatable Image Files}) includes the
10720: original image; i.e., according to copyright law it is a derived work of
10721: the original image.
10722:
10723: Since Gforth is distributed under the GNU GPL, the newly created image
10724: falls under the GNU GPL, too. In particular, this means that if you
10725: distribute the image, you have to make all of the sources for the image
10726: available, including those you wrote. For details see @ref{License, ,
10727: GNU General Public License (Section 3)}.
10728:
10729: If you create an image with @code{cross} (@pxref{cross.fs}), the image
10730: contains only code compiled from the sources you gave it; if none of
10731: these sources is under the GPL, the terms discussed above do not apply
10732: to the image. However, if your image needs an engine (a gforth binary)
10733: that is under the GPL, you should make sure that you distribute both in
10734: a way that is at most a @emph{mere aggregation}, if you don't want the
10735: terms of the GPL to apply to the image.
10736:
10737: @node Image File Background, Non-Relocatable Image Files, Image Licensing Issues, Image Files
10738: @section Image File Background
10739: @cindex image file background
10740:
10741: Our Forth system consists not only of primitives, but also of
10742: definitions written in Forth. Since the Forth compiler itself belongs to
10743: those definitions, it is not possible to start the system with the
10744: primitives and the Forth source alone. Therefore we provide the Forth
10745: code as an image file in nearly executable form. When Gforth starts up,
10746: a C routine loads the image file into memory, optionally relocates the
10747: addresses, then sets up the memory (stacks etc.) according to
10748: information in the image file, and (finally) starts executing Forth
10749: code.
10750:
10751: The image file variants represent different compromises between the
10752: goals of making it easy to generate image files and making them
10753: portable.
10754:
10755: @cindex relocation at run-time
10756: Win32Forth 3.4 and Mitch Bradley's @code{cforth} use relocation at
10757: run-time. This avoids many of the complications discussed below (image
10758: files are data relocatable without further ado), but costs performance
10759: (one addition per memory access).
10760:
10761: @cindex relocation at load-time
10762: By contrast, the Gforth loader performs relocation at image load time. The
10763: loader also has to replace tokens that represent primitive calls with the
10764: appropriate code-field addresses (or code addresses in the case of
10765: direct threading).
10766:
10767: There are three kinds of image files, with different degrees of
10768: relocatability: non-relocatable, data-relocatable, and fully relocatable
10769: image files.
10770:
10771: @cindex image file loader
10772: @cindex relocating loader
10773: @cindex loader for image files
10774: These image file variants have several restrictions in common; they are
10775: caused by the design of the image file loader:
10776:
10777: @itemize @bullet
10778: @item
10779: There is only one segment; in particular, this means, that an image file
10780: cannot represent @code{ALLOCATE}d memory chunks (and pointers to
10781: them). The contents of the stacks are not represented, either.
10782:
10783: @item
10784: The only kinds of relocation supported are: adding the same offset to
10785: all cells that represent data addresses; and replacing special tokens
10786: with code addresses or with pieces of machine code.
10787:
10788: If any complex computations involving addresses are performed, the
10789: results cannot be represented in the image file. Several applications that
10790: use such computations come to mind:
10791: @itemize @minus
10792: @item
10793: Hashing addresses (or data structures which contain addresses) for table
10794: lookup. If you use Gforth's @code{table}s or @code{wordlist}s for this
10795: purpose, you will have no problem, because the hash tables are
10796: recomputed automatically when the system is started. If you use your own
10797: hash tables, you will have to do something similar.
10798:
10799: @item
10800: There's a cute implementation of doubly-linked lists that uses
10801: @code{XOR}ed addresses. You could represent such lists as singly-linked
10802: in the image file, and restore the doubly-linked representation on
10803: startup.@footnote{In my opinion, though, you should think thrice before
10804: using a doubly-linked list (whatever implementation).}
10805:
10806: @item
10807: The code addresses of run-time routines like @code{docol:} cannot be
10808: represented in the image file (because their tokens would be replaced by
10809: machine code in direct threaded implementations). As a workaround,
10810: compute these addresses at run-time with @code{>code-address} from the
10811: executions tokens of appropriate words (see the definitions of
10812: @code{docol:} and friends in @file{kernel.fs}).
10813:
10814: @item
10815: On many architectures addresses are represented in machine code in some
10816: shifted or mangled form. You cannot put @code{CODE} words that contain
10817: absolute addresses in this form in a relocatable image file. Workarounds
10818: are representing the address in some relative form (e.g., relative to
10819: the CFA, which is present in some register), or loading the address from
10820: a place where it is stored in a non-mangled form.
10821: @end itemize
10822: @end itemize
10823:
10824: @node Non-Relocatable Image Files, Data-Relocatable Image Files, Image File Background, Image Files
10825: @section Non-Relocatable Image Files
10826: @cindex non-relocatable image files
10827: @cindex image file, non-relocatable
10828:
10829: These files are simple memory dumps of the dictionary. They are specific
10830: to the executable (i.e., @file{gforth} file) they were created
10831: with. What's worse, they are specific to the place on which the
10832: dictionary resided when the image was created. Now, there is no
10833: guarantee that the dictionary will reside at the same place the next
10834: time you start Gforth, so there's no guarantee that a non-relocatable
10835: image will work the next time (Gforth will complain instead of crashing,
10836: though).
10837:
10838: You can create a non-relocatable image file with
10839:
10840:
10841: doc-savesystem
10842:
10843:
10844: @node Data-Relocatable Image Files, Fully Relocatable Image Files, Non-Relocatable Image Files, Image Files
10845: @section Data-Relocatable Image Files
10846: @cindex data-relocatable image files
10847: @cindex image file, data-relocatable
10848:
10849: These files contain relocatable data addresses, but fixed code addresses
10850: (instead of tokens). They are specific to the executable (i.e.,
10851: @file{gforth} file) they were created with. For direct threading on some
10852: architectures (e.g., the i386), data-relocatable images do not work. You
10853: get a data-relocatable image, if you use @file{gforthmi} with a
10854: Gforth binary that is not doubly indirect threaded (@pxref{Fully
10855: Relocatable Image Files}).
10856:
10857: @node Fully Relocatable Image Files, Stack and Dictionary Sizes, Data-Relocatable Image Files, Image Files
10858: @section Fully Relocatable Image Files
10859: @cindex fully relocatable image files
10860: @cindex image file, fully relocatable
10861:
10862: @cindex @file{kern*.fi}, relocatability
10863: @cindex @file{gforth.fi}, relocatability
10864: These image files have relocatable data addresses, and tokens for code
10865: addresses. They can be used with different binaries (e.g., with and
10866: without debugging) on the same machine, and even across machines with
10867: the same data formats (byte order, cell size, floating point
10868: format). However, they are usually specific to the version of Gforth
10869: they were created with. The files @file{gforth.fi} and @file{kernl*.fi}
10870: are fully relocatable.
10871:
10872: There are two ways to create a fully relocatable image file:
10873:
10874: @menu
10875: * gforthmi:: The normal way
10876: * cross.fs:: The hard way
10877: @end menu
10878:
10879: @node gforthmi, cross.fs, Fully Relocatable Image Files, Fully Relocatable Image Files
10880: @subsection @file{gforthmi}
10881: @cindex @file{comp-i.fs}
10882: @cindex @file{gforthmi}
10883:
10884: You will usually use @file{gforthmi}. If you want to create an
10885: image @i{file} that contains everything you would load by invoking
10886: Gforth with @code{gforth @i{options}}, you simply say:
10887: @example
10888: gforthmi @i{file} @i{options}
10889: @end example
10890:
10891: E.g., if you want to create an image @file{asm.fi} that has the file
10892: @file{asm.fs} loaded in addition to the usual stuff, you could do it
10893: like this:
10894:
10895: @example
10896: gforthmi asm.fi asm.fs
10897: @end example
10898:
10899: @file{gforthmi} is implemented as a sh script and works like this: It
10900: produces two non-relocatable images for different addresses and then
10901: compares them. Its output reflects this: first you see the output (if
10902: any) of the two Gforth invocations that produce the nonrelocatable image
10903: files, then you see the output of the comparing program: It displays the
10904: offset used for data addresses and the offset used for code addresses;
10905: moreover, for each cell that cannot be represented correctly in the
10906: image files, it displays a line like this:
10907:
10908: @example
10909: 78DC BFFFFA50 BFFFFA40
10910: @end example
10911:
10912: This means that at offset $78dc from @code{forthstart}, one input image
10913: contains $bffffa50, and the other contains $bffffa40. Since these cells
10914: cannot be represented correctly in the output image, you should examine
10915: these places in the dictionary and verify that these cells are dead
10916: (i.e., not read before they are written).
10917:
10918: @cindex --application, @code{gforthmi} option
10919: If you insert the option @code{--application} in front of the image file
10920: name, you will get an image that uses the @code{--appl-image} option
10921: instead of the @code{--image-file} option (@pxref{Invoking
10922: Gforth}). When you execute such an image on Unix (by typing the image
10923: name as command), the Gforth engine will pass all options to the image
10924: instead of trying to interpret them as engine options.
10925:
10926: If you type @file{gforthmi} with no arguments, it prints some usage
10927: instructions.
10928:
10929: @cindex @code{savesystem} during @file{gforthmi}
10930: @cindex @code{bye} during @file{gforthmi}
10931: @cindex doubly indirect threaded code
10932: @cindex environment variables
10933: @cindex @code{GFORTHD} -- environment variable
10934: @cindex @code{GFORTH} -- environment variable
10935: @cindex @code{gforth-ditc}
10936: There are a few wrinkles: After processing the passed @i{options}, the
10937: words @code{savesystem} and @code{bye} must be visible. A special doubly
10938: indirect threaded version of the @file{gforth} executable is used for
10939: creating the nonrelocatable images; you can pass the exact filename of
10940: this executable through the environment variable @code{GFORTHD}
10941: (default: @file{gforth-ditc}); if you pass a version that is not doubly
10942: indirect threaded, you will not get a fully relocatable image, but a
10943: data-relocatable image (because there is no code address offset). The
10944: normal @file{gforth} executable is used for creating the relocatable
10945: image; you can pass the exact filename of this executable through the
10946: environment variable @code{GFORTH}.
10947:
10948: @node cross.fs, , gforthmi, Fully Relocatable Image Files
10949: @subsection @file{cross.fs}
10950: @cindex @file{cross.fs}
10951: @cindex cross-compiler
10952: @cindex metacompiler
10953: @cindex target compiler
10954:
10955: You can also use @code{cross}, a batch compiler that accepts a Forth-like
10956: programming language (@pxref{Cross Compiler}).
10957:
10958: @code{cross} allows you to create image files for machines with
10959: different data sizes and data formats than the one used for generating
10960: the image file. You can also use it to create an application image that
10961: does not contain a Forth compiler. These features are bought with
10962: restrictions and inconveniences in programming. E.g., addresses have to
10963: be stored in memory with special words (@code{A!}, @code{A,}, etc.) in
10964: order to make the code relocatable.
10965:
10966:
10967: @node Stack and Dictionary Sizes, Running Image Files, Fully Relocatable Image Files, Image Files
10968: @section Stack and Dictionary Sizes
10969: @cindex image file, stack and dictionary sizes
10970: @cindex dictionary size default
10971: @cindex stack size default
10972:
10973: If you invoke Gforth with a command line flag for the size
10974: (@pxref{Invoking Gforth}), the size you specify is stored in the
10975: dictionary. If you save the dictionary with @code{savesystem} or create
10976: an image with @file{gforthmi}, this size will become the default
10977: for the resulting image file. E.g., the following will create a
10978: fully relocatable version of @file{gforth.fi} with a 1MB dictionary:
10979:
10980: @example
10981: gforthmi gforth.fi -m 1M
10982: @end example
10983:
10984: In other words, if you want to set the default size for the dictionary
10985: and the stacks of an image, just invoke @file{gforthmi} with the
10986: appropriate options when creating the image.
10987:
10988: @cindex stack size, cache-friendly
10989: Note: For cache-friendly behaviour (i.e., good performance), you should
10990: make the sizes of the stacks modulo, say, 2K, somewhat different. E.g.,
10991: the default stack sizes are: data: 16k (mod 2k=0); fp: 15.5k (mod
10992: 2k=1.5k); return: 15k(mod 2k=1k); locals: 14.5k (mod 2k=0.5k).
10993:
10994: @node Running Image Files, Modifying the Startup Sequence, Stack and Dictionary Sizes, Image Files
10995: @section Running Image Files
10996: @cindex running image files
10997: @cindex invoking image files
10998: @cindex image file invocation
10999:
11000: @cindex -i, invoke image file
11001: @cindex --image file, invoke image file
11002: You can invoke Gforth with an image file @i{image} instead of the
11003: default @file{gforth.fi} with the @code{-i} flag (@pxref{Invoking Gforth}):
11004: @example
11005: gforth -i @i{image}
11006: @end example
11007:
11008: @cindex executable image file
11009: @cindex image file, executable
11010: If your operating system supports starting scripts with a line of the
11011: form @code{#! ...}, you just have to type the image file name to start
11012: Gforth with this image file (note that the file extension @code{.fi} is
11013: just a convention). I.e., to run Gforth with the image file @i{image},
11014: you can just type @i{image} instead of @code{gforth -i @i{image}}.
11015: This works because every @code{.fi} file starts with a line of this
11016: format:
11017:
11018: @example
11019: #! /usr/local/bin/gforth-0.4.0 -i
11020: @end example
11021:
11022: The file and pathname for the Gforth engine specified on this line is
11023: the specific Gforth executable that it was built against; i.e. the value
11024: of the environment variable @code{GFORTH} at the time that
11025: @file{gforthmi} was executed.
11026:
11027: You can make use of the same shell capability to make a Forth source
11028: file into an executable. For example, if you place this text in a file:
11029:
11030: @example
11031: #! /usr/local/bin/gforth
11032:
11033: ." Hello, world" CR
11034: bye
11035: @end example
11036:
11037: @noindent
11038: and then make the file executable (chmod +x in Unix), you can run it
11039: directly from the command line. The sequence @code{#!} is used in two
11040: ways; firstly, it is recognised as a ``magic sequence'' by the operating
11041: system@footnote{The Unix kernel actually recognises two types of files:
11042: executable files and files of data, where the data is processed by an
11043: interpreter that is specified on the ``interpreter line'' -- the first
11044: line of the file, starting with the sequence #!. There may be a small
11045: limit (e.g., 32) on the number of characters that may be specified on
11046: the interpreter line.} secondly it is treated as a comment character by
11047: Gforth. Because of the second usage, a space is required between
11048: @code{#!} and the path to the executable.
11049:
11050: The disadvantage of this latter technique, compared with using
11051: @file{gforthmi}, is that it is slower; the Forth source code is compiled
11052: on-the-fly, each time the program is invoked.
11053:
11054:
11055: doc-#!
11056:
11057:
11058: @node Modifying the Startup Sequence, , Running Image Files, Image Files
11059: @section Modifying the Startup Sequence
11060: @cindex startup sequence for image file
11061: @cindex image file initialization sequence
11062: @cindex initialization sequence of image file
11063:
11064: You can add your own initialization to the startup sequence through the
11065: deferred word @code{'cold}. @code{'cold} is invoked just before the
11066: image-specific command line processing (by default, loading files and
11067: evaluating (@code{-e}) strings) starts.
11068:
11069: A sequence for adding your initialization usually looks like this:
11070:
11071: @example
11072: :noname
11073: Defers 'cold \ do other initialization stuff (e.g., rehashing wordlists)
11074: ... \ your stuff
11075: ; IS 'cold
11076: @end example
11077:
11078: @cindex turnkey image files
11079: @cindex image file, turnkey applications
11080: You can make a turnkey image by letting @code{'cold} execute a word
11081: (your turnkey application) that never returns; instead, it exits Gforth
11082: via @code{bye} or @code{throw}.
11083:
11084: @cindex command-line arguments, access
11085: @cindex arguments on the command line, access
11086: You can access the (image-specific) command-line arguments through the
11087: variables @code{argc} and @code{argv}. @code{arg} provides convenient
11088: access to @code{argv}.
11089:
11090: If @code{'cold} exits normally, Gforth processes the command-line
11091: arguments as files to be loaded and strings to be evaluated. Therefore,
11092: @code{'cold} should remove the arguments it has used in this case.
11093:
11094:
11095:
11096: doc-'cold
11097: doc-argc
11098: doc-argv
11099: doc-arg
11100:
11101:
11102:
11103: @c ******************************************************************
11104: @node Engine, Binding to System Library, Image Files, Top
11105: @chapter Engine
11106: @cindex engine
11107: @cindex virtual machine
11108:
11109: Reading this chapter is not necessary for programming with Gforth. It
11110: may be helpful for finding your way in the Gforth sources.
11111:
11112: The ideas in this section have also been published in the papers
11113: @cite{ANS fig/GNU/??? Forth} (in German) by Bernd Paysan, presented at
11114: the Forth-Tagung '93 and @cite{A Portable Forth Engine} by M. Anton
11115: Ertl, presented at EuroForth '93; the latter is available at
11116: @*@uref{http://www.complang.tuwien.ac.at/papers/ertl93.ps.Z}.
11117:
11118: @menu
11119: * Portability::
11120: * Threading::
11121: * Primitives::
11122: * Performance::
11123: @end menu
11124:
11125: @node Portability, Threading, Engine, Engine
11126: @section Portability
11127: @cindex engine portability
11128:
11129: An important goal of the Gforth Project is availability across a wide
11130: range of personal machines. fig-Forth, and, to a lesser extent, F83,
11131: achieved this goal by manually coding the engine in assembly language
11132: for several then-popular processors. This approach is very
11133: labor-intensive and the results are short-lived due to progress in
11134: computer architecture.
11135:
11136: @cindex C, using C for the engine
11137: Others have avoided this problem by coding in C, e.g., Mitch Bradley
11138: (cforth), Mikael Patel (TILE) and Dirk Zoller (pfe). This approach is
11139: particularly popular for UNIX-based Forths due to the large variety of
11140: architectures of UNIX machines. Unfortunately an implementation in C
11141: does not mix well with the goals of efficiency and with using
11142: traditional techniques: Indirect or direct threading cannot be expressed
11143: in C, and switch threading, the fastest technique available in C, is
11144: significantly slower. Another problem with C is that it is very
11145: cumbersome to express double integer arithmetic.
11146:
11147: @cindex GNU C for the engine
11148: @cindex long long
11149: Fortunately, there is a portable language that does not have these
11150: limitations: GNU C, the version of C processed by the GNU C compiler
11151: (@pxref{C Extensions, , Extensions to the C Language Family, gcc.info,
11152: GNU C Manual}). Its labels as values feature (@pxref{Labels as Values, ,
11153: Labels as Values, gcc.info, GNU C Manual}) makes direct and indirect
11154: threading possible, its @code{long long} type (@pxref{Long Long, ,
11155: Double-Word Integers, gcc.info, GNU C Manual}) corresponds to Forth's
11156: double numbers@footnote{Unfortunately, long longs are not implemented
11157: properly on all machines (e.g., on alpha-osf1, long longs are only 64
11158: bits, the same size as longs (and pointers), but they should be twice as
11159: long according to @pxref{Long Long, , Double-Word Integers, gcc.info, GNU
11160: C Manual}). So, we had to implement doubles in C after all. Still, on
11161: most machines we can use long longs and achieve better performance than
11162: with the emulation package.}. GNU C is available for free on all
11163: important (and many unimportant) UNIX machines, VMS, 80386s running
11164: MS-DOS, the Amiga, and the Atari ST, so a Forth written in GNU C can run
11165: on all these machines.
11166:
11167: Writing in a portable language has the reputation of producing code that
11168: is slower than assembly. For our Forth engine we repeatedly looked at
11169: the code produced by the compiler and eliminated most compiler-induced
11170: inefficiencies by appropriate changes in the source code.
11171:
11172: @cindex explicit register declarations
11173: @cindex --enable-force-reg, configuration flag
11174: @cindex -DFORCE_REG
11175: However, register allocation cannot be portably influenced by the
11176: programmer, leading to some inefficiencies on register-starved
11177: machines. We use explicit register declarations (@pxref{Explicit Reg
11178: Vars, , Variables in Specified Registers, gcc.info, GNU C Manual}) to
11179: improve the speed on some machines. They are turned on by using the
11180: configuration flag @code{--enable-force-reg} (@code{gcc} switch
11181: @code{-DFORCE_REG}). Unfortunately, this feature not only depends on the
11182: machine, but also on the compiler version: On some machines some
11183: compiler versions produce incorrect code when certain explicit register
11184: declarations are used. So by default @code{-DFORCE_REG} is not used.
11185:
11186: @node Threading, Primitives, Portability, Engine
11187: @section Threading
11188: @cindex inner interpreter implementation
11189: @cindex threaded code implementation
11190:
11191: @cindex labels as values
11192: GNU C's labels as values extension (available since @code{gcc-2.0},
11193: @pxref{Labels as Values, , Labels as Values, gcc.info, GNU C Manual})
11194: makes it possible to take the address of @i{label} by writing
11195: @code{&&@i{label}}. This address can then be used in a statement like
11196: @code{goto *@i{address}}. I.e., @code{goto *&&x} is the same as
11197: @code{goto x}.
11198:
11199: @cindex @code{NEXT}, indirect threaded
11200: @cindex indirect threaded inner interpreter
11201: @cindex inner interpreter, indirect threaded
11202: With this feature an indirect threaded @code{NEXT} looks like:
11203: @example
11204: cfa = *ip++;
11205: ca = *cfa;
11206: goto *ca;
11207: @end example
11208: @cindex instruction pointer
11209: For those unfamiliar with the names: @code{ip} is the Forth instruction
11210: pointer; the @code{cfa} (code-field address) corresponds to ANS Forths
11211: execution token and points to the code field of the next word to be
11212: executed; The @code{ca} (code address) fetched from there points to some
11213: executable code, e.g., a primitive or the colon definition handler
11214: @code{docol}.
11215:
11216: @cindex @code{NEXT}, direct threaded
11217: @cindex direct threaded inner interpreter
11218: @cindex inner interpreter, direct threaded
11219: Direct threading is even simpler:
11220: @example
11221: ca = *ip++;
11222: goto *ca;
11223: @end example
11224:
11225: Of course we have packaged the whole thing neatly in macros called
11226: @code{NEXT} and @code{NEXT1} (the part of @code{NEXT} after fetching the cfa).
11227:
11228: @menu
11229: * Scheduling::
11230: * Direct or Indirect Threaded?::
11231: * DOES>::
11232: @end menu
11233:
11234: @node Scheduling, Direct or Indirect Threaded?, Threading, Threading
11235: @subsection Scheduling
11236: @cindex inner interpreter optimization
11237:
11238: There is a little complication: Pipelined and superscalar processors,
11239: i.e., RISC and some modern CISC machines can process independent
11240: instructions while waiting for the results of an instruction. The
11241: compiler usually reorders (schedules) the instructions in a way that
11242: achieves good usage of these delay slots. However, on our first tries
11243: the compiler did not do well on scheduling primitives. E.g., for
11244: @code{+} implemented as
11245: @example
11246: n=sp[0]+sp[1];
11247: sp++;
11248: sp[0]=n;
11249: NEXT;
11250: @end example
11251: the @code{NEXT} comes strictly after the other code, i.e., there is nearly no
11252: scheduling. After a little thought the problem becomes clear: The
11253: compiler cannot know that @code{sp} and @code{ip} point to different
11254: addresses (and the version of @code{gcc} we used would not know it even
11255: if it was possible), so it could not move the load of the cfa above the
11256: store to the TOS. Indeed the pointers could be the same, if code on or
11257: very near the top of stack were executed. In the interest of speed we
11258: chose to forbid this probably unused ``feature'' and helped the compiler
11259: in scheduling: @code{NEXT} is divided into the loading part (@code{NEXT_P1})
11260: and the goto part (@code{NEXT_P2}). @code{+} now looks like:
11261: @example
11262: n=sp[0]+sp[1];
11263: sp++;
11264: NEXT_P1;
11265: sp[0]=n;
11266: NEXT_P2;
11267: @end example
11268: This can be scheduled optimally by the compiler.
11269:
11270: This division can be turned off with the switch @code{-DCISC_NEXT}. This
11271: switch is on by default on machines that do not profit from scheduling
11272: (e.g., the 80386), in order to preserve registers.
11273:
11274: @node Direct or Indirect Threaded?, DOES>, Scheduling, Threading
11275: @subsection Direct or Indirect Threaded?
11276: @cindex threading, direct or indirect?
11277:
11278: @cindex -DDIRECT_THREADED
11279: Both! After packaging the nasty details in macro definitions we
11280: realized that we could switch between direct and indirect threading by
11281: simply setting a compilation flag (@code{-DDIRECT_THREADED}) and
11282: defining a few machine-specific macros for the direct-threading case.
11283: On the Forth level we also offer access words that hide the
11284: differences between the threading methods (@pxref{Threading Words}).
11285:
11286: Indirect threading is implemented completely machine-independently.
11287: Direct threading needs routines for creating jumps to the executable
11288: code (e.g. to @code{docol} or @code{dodoes}). These routines are inherently
11289: machine-dependent, but they do not amount to many source lines. Therefore,
11290: even porting direct threading to a new machine requires little effort.
11291:
11292: @cindex --enable-indirect-threaded, configuration flag
11293: @cindex --enable-direct-threaded, configuration flag
11294: The default threading method is machine-dependent. You can enforce a
11295: specific threading method when building Gforth with the configuration
11296: flag @code{--enable-direct-threaded} or
11297: @code{--enable-indirect-threaded}. Note that direct threading is not
11298: supported on all machines.
11299:
11300: @node DOES>, , Direct or Indirect Threaded?, Threading
11301: @subsection DOES>
11302: @cindex @code{DOES>} implementation
11303:
11304: @cindex @code{dodoes} routine
11305: @cindex @code{DOES>}-code
11306: One of the most complex parts of a Forth engine is @code{dodoes}, i.e.,
11307: the chunk of code executed by every word defined by a
11308: @code{CREATE}...@code{DOES>} pair. The main problem here is: How to find
11309: the Forth code to be executed, i.e. the code after the
11310: @code{DOES>} (the @code{DOES>}-code)? There are two solutions:
11311:
11312: In fig-Forth the code field points directly to the @code{dodoes} and the
11313: @code{DOES>}-code address is stored in the cell after the code address (i.e. at
11314: @code{@i{CFA} cell+}). It may seem that this solution is illegal in
11315: the Forth-79 and all later standards, because in fig-Forth this address
11316: lies in the body (which is illegal in these standards). However, by
11317: making the code field larger for all words this solution becomes legal
11318: again. We use this approach for the indirect threaded version and for
11319: direct threading on some machines. Leaving a cell unused in most words
11320: is a bit wasteful, but on the machines we are targeting this is hardly a
11321: problem. The other reason for having a code field size of two cells is
11322: to avoid having different image files for direct and indirect threaded
11323: systems (direct threaded systems require two-cell code fields on many
11324: machines).
11325:
11326: @cindex @code{DOES>}-handler
11327: The other approach is that the code field points or jumps to the cell
11328: after @code{DOES>}. In this variant there is a jump to @code{dodoes} at
11329: this address (the @code{DOES>}-handler). @code{dodoes} can then get the
11330: @code{DOES>}-code address by computing the code address, i.e., the address of
11331: the jump to @code{dodoes}, and add the length of that jump field. A variant of
11332: this is to have a call to @code{dodoes} after the @code{DOES>}; then the
11333: return address (which can be found in the return register on RISCs) is
11334: the @code{DOES>}-code address. Since the two cells available in the code field
11335: are used up by the jump to the code address in direct threading on many
11336: architectures, we use this approach for direct threading on these
11337: architectures. We did not want to add another cell to the code field.
11338:
11339: @node Primitives, Performance, Threading, Engine
11340: @section Primitives
11341: @cindex primitives, implementation
11342: @cindex virtual machine instructions, implementation
11343:
11344: @menu
11345: * Automatic Generation::
11346: * TOS Optimization::
11347: * Produced code::
11348: @end menu
11349:
11350: @node Automatic Generation, TOS Optimization, Primitives, Primitives
11351: @subsection Automatic Generation
11352: @cindex primitives, automatic generation
11353:
11354: @cindex @file{prims2x.fs}
11355: Since the primitives are implemented in a portable language, there is no
11356: longer any need to minimize the number of primitives. On the contrary,
11357: having many primitives has an advantage: speed. In order to reduce the
11358: number of errors in primitives and to make programming them easier, we
11359: provide a tool, the primitive generator (@file{prims2x.fs}), that
11360: automatically generates most (and sometimes all) of the C code for a
11361: primitive from the stack effect notation. The source for a primitive
11362: has the following form:
11363:
11364: @cindex primitive source format
11365: @format
11366: @i{Forth-name} @i{stack-effect} @i{category} [@i{pronounc.}]
11367: [@code{""}@i{glossary entry}@code{""}]
11368: @i{C code}
11369: [@code{:}
11370: @i{Forth code}]
11371: @end format
11372:
11373: The items in brackets are optional. The category and glossary fields
11374: are there for generating the documentation, the Forth code is there
11375: for manual implementations on machines without GNU C. E.g., the source
11376: for the primitive @code{+} is:
11377: @example
11378: + n1 n2 -- n core plus
11379: n = n1+n2;
11380: @end example
11381:
11382: This looks like a specification, but in fact @code{n = n1+n2} is C
11383: code. Our primitive generation tool extracts a lot of information from
11384: the stack effect notations@footnote{We use a one-stack notation, even
11385: though we have separate data and floating-point stacks; The separate
11386: notation can be generated easily from the unified notation.}: The number
11387: of items popped from and pushed on the stack, their type, and by what
11388: name they are referred to in the C code. It then generates a C code
11389: prelude and postlude for each primitive. The final C code for @code{+}
11390: looks like this:
11391:
11392: @example
11393: I_plus: /* + ( n1 n2 -- n ) */ /* label, stack effect */
11394: /* */ /* documentation */
11395: @{
11396: DEF_CA /* definition of variable ca (indirect threading) */
11397: Cell n1; /* definitions of variables */
11398: Cell n2;
11399: Cell n;
11400: n1 = (Cell) sp[1]; /* input */
11401: n2 = (Cell) TOS;
11402: sp += 1; /* stack adjustment */
11403: NAME("+") /* debugging output (with -DDEBUG) */
11404: @{
11405: n = n1+n2; /* C code taken from the source */
11406: @}
11407: NEXT_P1; /* NEXT part 1 */
11408: TOS = (Cell)n; /* output */
11409: NEXT_P2; /* NEXT part 2 */
11410: @}
11411: @end example
11412:
11413: This looks long and inefficient, but the GNU C compiler optimizes quite
11414: well and produces optimal code for @code{+} on, e.g., the R3000 and the
11415: HP RISC machines: Defining the @code{n}s does not produce any code, and
11416: using them as intermediate storage also adds no cost.
11417:
11418: There are also other optimizations that are not illustrated by this
11419: example: assignments between simple variables are usually for free (copy
11420: propagation). If one of the stack items is not used by the primitive
11421: (e.g. in @code{drop}), the compiler eliminates the load from the stack
11422: (dead code elimination). On the other hand, there are some things that
11423: the compiler does not do, therefore they are performed by
11424: @file{prims2x.fs}: The compiler does not optimize code away that stores
11425: a stack item to the place where it just came from (e.g., @code{over}).
11426:
11427: While programming a primitive is usually easy, there are a few cases
11428: where the programmer has to take the actions of the generator into
11429: account, most notably @code{?dup}, but also words that do not (always)
11430: fall through to @code{NEXT}.
11431:
11432: @node TOS Optimization, Produced code, Automatic Generation, Primitives
11433: @subsection TOS Optimization
11434: @cindex TOS optimization for primitives
11435: @cindex primitives, keeping the TOS in a register
11436:
11437: An important optimization for stack machine emulators, e.g., Forth
11438: engines, is keeping one or more of the top stack items in
11439: registers. If a word has the stack effect @i{in1}...@i{inx} @code{--}
11440: @i{out1}...@i{outy}, keeping the top @i{n} items in registers
11441: @itemize @bullet
11442: @item
11443: is better than keeping @i{n-1} items, if @i{x>=n} and @i{y>=n},
11444: due to fewer loads from and stores to the stack.
11445: @item is slower than keeping @i{n-1} items, if @i{x<>y} and @i{x<n} and
11446: @i{y<n}, due to additional moves between registers.
11447: @end itemize
11448:
11449: @cindex -DUSE_TOS
11450: @cindex -DUSE_NO_TOS
11451: In particular, keeping one item in a register is never a disadvantage,
11452: if there are enough registers. Keeping two items in registers is a
11453: disadvantage for frequent words like @code{?branch}, constants,
11454: variables, literals and @code{i}. Therefore our generator only produces
11455: code that keeps zero or one items in registers. The generated C code
11456: covers both cases; the selection between these alternatives is made at
11457: C-compile time using the switch @code{-DUSE_TOS}. @code{TOS} in the C
11458: code for @code{+} is just a simple variable name in the one-item case,
11459: otherwise it is a macro that expands into @code{sp[0]}. Note that the
11460: GNU C compiler tries to keep simple variables like @code{TOS} in
11461: registers, and it usually succeeds, if there are enough registers.
11462:
11463: @cindex -DUSE_FTOS
11464: @cindex -DUSE_NO_FTOS
11465: The primitive generator performs the TOS optimization for the
11466: floating-point stack, too (@code{-DUSE_FTOS}). For floating-point
11467: operations the benefit of this optimization is even larger:
11468: floating-point operations take quite long on most processors, but can be
11469: performed in parallel with other operations as long as their results are
11470: not used. If the FP-TOS is kept in a register, this works. If
11471: it is kept on the stack, i.e., in memory, the store into memory has to
11472: wait for the result of the floating-point operation, lengthening the
11473: execution time of the primitive considerably.
11474:
11475: The TOS optimization makes the automatic generation of primitives a
11476: bit more complicated. Just replacing all occurrences of @code{sp[0]} by
11477: @code{TOS} is not sufficient. There are some special cases to
11478: consider:
11479: @itemize @bullet
11480: @item In the case of @code{dup ( w -- w w )} the generator must not
11481: eliminate the store to the original location of the item on the stack,
11482: if the TOS optimization is turned on.
11483: @item Primitives with stack effects of the form @code{--}
11484: @i{out1}...@i{outy} must store the TOS to the stack at the start.
11485: Likewise, primitives with the stack effect @i{in1}...@i{inx} @code{--}
11486: must load the TOS from the stack at the end. But for the null stack
11487: effect @code{--} no stores or loads should be generated.
11488: @end itemize
11489:
11490: @node Produced code, , TOS Optimization, Primitives
11491: @subsection Produced code
11492: @cindex primitives, assembly code listing
11493:
11494: @cindex @file{engine.s}
11495: To see what assembly code is produced for the primitives on your machine
11496: with your compiler and your flag settings, type @code{make engine.s} and
11497: look at the resulting file @file{engine.s}.
11498:
11499: @node Performance, , Primitives, Engine
11500: @section Performance
11501: @cindex performance of some Forth interpreters
11502: @cindex engine performance
11503: @cindex benchmarking Forth systems
11504: @cindex Gforth performance
11505:
11506: On RISCs the Gforth engine is very close to optimal; i.e., it is usually
11507: impossible to write a significantly faster engine.
11508:
11509: On register-starved machines like the 386 architecture processors
11510: improvements are possible, because @code{gcc} does not utilize the
11511: registers as well as a human, even with explicit register declarations;
11512: e.g., Bernd Beuster wrote a Forth system fragment in assembly language
11513: and hand-tuned it for the 486; this system is 1.19 times faster on the
11514: Sieve benchmark on a 486DX2/66 than Gforth compiled with
11515: @code{gcc-2.6.3} with @code{-DFORCE_REG}. The situation has improved
11516: with gcc-2.95 and gforth-0.4.9; now the most important virtual machine
11517: registers fit in real registers (and we can even afford to use the TOS
11518: optimization), resulting in a speedup of 1.14 on the sieve over the
11519: earlier results.
11520:
11521: @cindex Win32Forth performance
11522: @cindex NT Forth performance
11523: @cindex eforth performance
11524: @cindex ThisForth performance
11525: @cindex PFE performance
11526: @cindex TILE performance
11527: The potential advantage of assembly language implementations
11528: is not necessarily realized in complete Forth systems: We compared
11529: Gforth-0.4.9 (direct threaded, compiled with @code{gcc-2.95.1} and
11530: @code{-DFORCE_REG}) with Win32Forth 1.2093, LMI's NT Forth (Beta, May
11531: 1994) and Eforth (with and without peephole (aka pinhole) optimization
11532: of the threaded code); all these systems were written in assembly
11533: language. We also compared Gforth with three systems written in C:
11534: PFE-0.9.14 (compiled with @code{gcc-2.6.3} with the default
11535: configuration for Linux: @code{-O2 -fomit-frame-pointer -DUSE_REGS
11536: -DUNROLL_NEXT}), ThisForth Beta (compiled with @code{gcc-2.6.3 -O3
11537: -fomit-frame-pointer}; ThisForth employs peephole optimization of the
11538: threaded code) and TILE (compiled with @code{make opt}). We benchmarked
11539: Gforth, PFE, ThisForth and TILE on a 486DX2/66 under Linux. Kenneth
11540: O'Heskin kindly provided the results for Win32Forth and NT Forth on a
11541: 486DX2/66 with similar memory performance under Windows NT. Marcel
11542: Hendrix ported Eforth to Linux, then extended it to run the benchmarks,
11543: added the peephole optimizer, ran the benchmarks and reported the
11544: results.
11545:
11546: We used four small benchmarks: the ubiquitous Sieve; bubble-sorting and
11547: matrix multiplication come from the Stanford integer benchmarks and have
11548: been translated into Forth by Martin Fraeman; we used the versions
11549: included in the TILE Forth package, but with bigger data set sizes; and
11550: a recursive Fibonacci number computation for benchmarking calling
11551: performance. The following table shows the time taken for the benchmarks
11552: scaled by the time taken by Gforth (in other words, it shows the speedup
11553: factor that Gforth achieved over the other systems).
11554:
11555: @example
11556: relative Win32- NT eforth This-
11557: time Gforth Forth Forth eforth +opt PFE Forth TILE
11558: sieve 1.00 1.58 1.30 1.58 0.97 1.80 3.63 9.79
11559: bubble 1.00 1.55 1.67 1.75 1.04 1.78 4.59
11560: matmul 1.00 1.67 1.53 1.66 0.84 1.79 4.63
11561: fib 1.00 1.75 1.53 1.40 0.99 1.99 3.43 4.93
11562: @end example
11563:
11564: You may be quite surprised by the good performance of Gforth when
11565: compared with systems written in assembly language. One important reason
11566: for the disappointing performance of these other systems is probably
11567: that they are not written optimally for the 486 (e.g., they use the
11568: @code{lods} instruction). In addition, Win32Forth uses a comfortable,
11569: but costly method for relocating the Forth image: like @code{cforth}, it
11570: computes the actual addresses at run time, resulting in two address
11571: computations per @code{NEXT} (@pxref{Image File Background}).
11572:
11573: Only Eforth with the peephole optimizer performs comparable to
11574: Gforth. The speedups achieved with peephole optimization of threaded
11575: code are quite remarkable. Adding a peephole optimizer to Gforth should
11576: cause similar speedups.
11577:
11578: The speedup of Gforth over PFE, ThisForth and TILE can be easily
11579: explained with the self-imposed restriction of the latter systems to
11580: standard C, which makes efficient threading impossible (however, the
11581: measured implementation of PFE uses a GNU C extension: @pxref{Global Reg
11582: Vars, , Defining Global Register Variables, gcc.info, GNU C Manual}).
11583: Moreover, current C compilers have a hard time optimizing other aspects
11584: of the ThisForth and the TILE source.
11585:
11586: The performance of Gforth on 386 architecture processors varies widely
11587: with the version of @code{gcc} used. E.g., @code{gcc-2.5.8} failed to
11588: allocate any of the virtual machine registers into real machine
11589: registers by itself and would not work correctly with explicit register
11590: declarations, giving a 1.5 times slower engine (on a 486DX2/66 running
11591: the Sieve) than the one measured above.
11592:
11593: Note that there have been several releases of Win32Forth since the
11594: release presented here, so the results presented above may have little
11595: predictive value for the performance of Win32Forth today (results for
11596: the current release on an i486DX2/66 are welcome).
11597:
11598: @cindex @file{Benchres}
11599: In @cite{Translating Forth to Efficient C} by M. Anton Ertl and Martin
11600: Maierhofer (presented at EuroForth '95), an indirect threaded version of
11601: Gforth is compared with Win32Forth, NT Forth, PFE, and ThisForth; that
11602: version of Gforth is slower on a 486 than the direct threaded version
11603: used here. The paper available at
11604: @*@uref{http://www.complang.tuwien.ac.at/papers/ertl&maierhofer95.ps.gz};
11605: it also contains numbers for some native code systems. You can find a
11606: newer version of these measurements at
11607: @uref{http://www.complang.tuwien.ac.at/forth/performance.html}. You can
11608: find numbers for Gforth on various machines in @file{Benchres}.
11609:
11610: @c ******************************************************************
11611: @node Binding to System Library, Cross Compiler, Engine, Top
11612: @chapter Binding to System Library
11613:
11614: @node Cross Compiler, Bugs, Binding to System Library, Top
11615: @chapter Cross Compiler
11616: @cindex @file{cross.fs}
11617: @cindex cross-compiler
11618: @cindex metacompiler
11619: @cindex target compiler
11620:
11621: The cross compiler is used to bootstrap a Forth kernel. Since Gforth is
11622: mostly written in Forth, including crucial parts like the outer
11623: interpreter and compiler, it needs compiled Forth code to get
11624: started. The cross compiler allows to create new images for other
11625: architectures, even running under another Forth system.
11626:
11627: @menu
11628: * Using the Cross Compiler::
11629: * How the Cross Compiler Works::
11630: @end menu
11631:
11632: @node Using the Cross Compiler, How the Cross Compiler Works, Cross Compiler, Cross Compiler
11633: @section Using the Cross Compiler
11634:
11635: The cross compiler uses a language that resembles Forth, but isn't. The
11636: main difference is that you can execute Forth code after definition,
11637: while you usually can't execute the code compiled by cross, because the
11638: code you are compiling is typically for a different computer than the
11639: one you are compiling on.
11640:
11641: The Makefile is already set up to allow you to create kernels for new
11642: architectures with a simple make command. The generic kernels using the
11643: GCC compiled virtual machine are created in the normal build process
11644: with @code{make}. To create a embedded Gforth executable for e.g. the
11645: 8086 processor (running on a DOS machine), type
11646:
11647: @example
11648: make kernl-8086.fi
11649: @end example
11650:
11651: This will use the machine description from the @file{arch/8086}
11652: directory to create a new kernel. A machine file may look like that:
11653:
11654: @example
11655: \ Parameter for target systems 06oct92py
11656:
11657: 4 Constant cell \ cell size in bytes
11658: 2 Constant cell<< \ cell shift to bytes
11659: 5 Constant cell>bit \ cell shift to bits
11660: 8 Constant bits/char \ bits per character
11661: 8 Constant bits/byte \ bits per byte [default: 8]
11662: 8 Constant float \ bytes per float
11663: 8 Constant /maxalign \ maximum alignment in bytes
11664: false Constant bigendian \ byte order
11665: ( true=big, false=little )
11666:
11667: include machpc.fs \ feature list
11668: @end example
11669:
11670: This part is obligatory for the cross compiler itself, the feature list
11671: is used by the kernel to conditionally compile some features in and out,
11672: depending on whether the target supports these features.
11673:
11674: There are some optional features, if you define your own primitives,
11675: have an assembler, or need special, nonstandard preparation to make the
11676: boot process work. @code{asm-include} include an assembler,
11677: @code{prims-include} includes primitives, and @code{>boot} prepares for
11678: booting.
11679:
11680: @example
11681: : asm-include ." Include assembler" cr
11682: s" arch/8086/asm.fs" included ;
11683:
11684: : prims-include ." Include primitives" cr
11685: s" arch/8086/prim.fs" included ;
11686:
11687: : >boot ." Prepare booting" cr
11688: s" ' boot >body into-forth 1+ !" evaluate ;
11689: @end example
11690:
11691: These words are used as sort of macro during the cross compilation in
11692: the file @file{kernel/main.fs}. Instead of using this macros, it would
11693: be possible --- but more complicated --- to write a new kernel project
11694: file, too.
11695:
11696: @file{kernel/main.fs} expects the machine description file name on the
11697: stack; the cross compiler itself (@file{cross.fs}) assumes that either
11698: @code{mach-file} leaves a counted string on the stack, or
11699: @code{machine-file} leaves an address, count pair of the filename on the
11700: stack.
11701:
11702: The feature list is typically controlled using @code{SetValue}, generic
11703: files that are used by several projects can use @code{DefaultValue}
11704: instead. Both functions work like @code{Value}, when the value isn't
11705: defined, but @code{SetValue} works like @code{to} if the value is
11706: defined, and @code{DefaultValue} doesn't set anything, if the value is
11707: defined.
11708:
11709: @example
11710: \ generic mach file for pc gforth 03sep97jaw
11711:
11712: true DefaultValue NIL \ relocating
11713:
11714: >ENVIRON
11715:
11716: true DefaultValue file \ controls the presence of the
11717: \ file access wordset
11718: true DefaultValue OS \ flag to indicate a operating system
11719:
11720: true DefaultValue prims \ true: primitives are c-code
11721:
11722: true DefaultValue floating \ floating point wordset is present
11723:
11724: true DefaultValue glocals \ gforth locals are present
11725: \ will be loaded
11726: true DefaultValue dcomps \ double number comparisons
11727:
11728: true DefaultValue hash \ hashing primitives are loaded/present
11729:
11730: true DefaultValue xconds \ used together with glocals,
11731: \ special conditionals supporting gforths'
11732: \ local variables
11733: true DefaultValue header \ save a header information
11734:
11735: true DefaultValue backtrace \ enables backtrace code
11736:
11737: false DefaultValue ec
11738: false DefaultValue crlf
11739:
11740: cell 2 = [IF] &32 [ELSE] &256 [THEN] KB DefaultValue kernel-size
11741:
11742: &16 KB DefaultValue stack-size
11743: &15 KB &512 + DefaultValue fstack-size
11744: &15 KB DefaultValue rstack-size
11745: &14 KB &512 + DefaultValue lstack-size
11746: @end example
11747:
11748: @node How the Cross Compiler Works, , Using the Cross Compiler, Cross Compiler
11749: @section How the Cross Compiler Works
11750:
11751: @node Bugs, Origin, Cross Compiler, Top
11752: @appendix Bugs
11753: @cindex bug reporting
11754:
11755: Known bugs are described in the file @file{BUGS} in the Gforth distribution.
11756:
11757: If you find a bug, please send a bug report to
11758: @email{bug-gforth@@gnu.org}. A bug report should include this
11759: information:
11760:
11761: @itemize @bullet
11762: @item
11763: The Gforth version used (it is announced at the start of an
11764: interactive Gforth session).
11765: @item
11766: The machine and operating system (on Unix
11767: systems @code{uname -a} will report this information).
11768: @item
11769: The installation options (send the file @file{config.status}).
11770: @item
11771: A complete list of changes (if any) you (or your installer) have made to the
11772: Gforth sources.
11773: @item
11774: A program (or a sequence of keyboard commands) that reproduces the bug.
11775: @item
11776: A description of what you think constitutes the buggy behaviour.
11777: @end itemize
11778:
11779: For a thorough guide on reporting bugs read @ref{Bug Reporting, , How
11780: to Report Bugs, gcc.info, GNU C Manual}.
11781:
11782:
11783: @node Origin, Forth-related information, Bugs, Top
11784: @appendix Authors and Ancestors of Gforth
11785:
11786: @section Authors and Contributors
11787: @cindex authors of Gforth
11788: @cindex contributors to Gforth
11789:
11790: The Gforth project was started in mid-1992 by Bernd Paysan and Anton
11791: Ertl. The third major author was Jens Wilke. Lennart Benschop (who was
11792: one of Gforth's first users, in mid-1993) and Stuart Ramsden inspired us
11793: with their continuous feedback. Lennart Benshop contributed
11794: @file{glosgen.fs}, while Stuart Ramsden has been working on automatic
11795: support for calling C libraries. Helpful comments also came from Paul
11796: Kleinrubatscher, Christian Pirker, Dirk Zoller, Marcel Hendrix, John
11797: Wavrik, Barrie Stott, Marc de Groot, and Jorge Acerada. Since the
11798: release of Gforth-0.2.1 there were also helpful comments from many
11799: others; thank you all, sorry for not listing you here (but digging
11800: through my mailbox to extract your names is on my to-do list). Since the
11801: release of Gforth-0.4.0 Neal Crook worked on the manual.
11802:
11803: Gforth also owes a lot to the authors of the tools we used (GCC, CVS,
11804: and autoconf, among others), and to the creators of the Internet: Gforth
11805: was developed across the Internet, and its authors did not meet
11806: physically for the first 4 years of development.
11807:
11808: @section Pedigree
11809: @cindex pedigree of Gforth
11810:
11811: Gforth descends from bigFORTH (1993) and fig-Forth. Gforth and PFE (by
11812: Dirk Zoller) will cross-fertilize each other. Of course, a significant
11813: part of the design of Gforth was prescribed by ANS Forth.
11814:
11815: Bernd Paysan wrote bigFORTH, a descendent from TurboForth, an unreleased
11816: 32 bit native code version of VolksForth for the Atari ST, written
11817: mostly by Dietrich Weineck.
11818:
11819: VolksForth descends from F83. It was written by Klaus Schleisiek, Bernd
11820: Pennemann, Georg Rehfeld and Dietrich Weineck for the C64 (called
11821: UltraForth there) in the mid-80s and ported to the Atari ST in 1986.
11822:
11823: Henry Laxen and Mike Perry wrote F83 as a model implementation of the
11824: Forth-83 standard. !! Pedigree? When?
11825:
11826: A team led by Bill Ragsdale implemented fig-Forth on many processors in
11827: 1979. Robert Selzer and Bill Ragsdale developed the original
11828: implementation of fig-Forth for the 6502 based on microForth.
11829:
11830: The principal architect of microForth was Dean Sanderson. microForth was
11831: FORTH, Inc.'s first off-the-shelf product. It was developed in 1976 for
11832: the 1802, and subsequently implemented on the 8080, the 6800 and the
11833: Z80.
11834:
11835: All earlier Forth systems were custom-made, usually by Charles Moore,
11836: who discovered (as he puts it) Forth during the late 60s. The first full
11837: Forth existed in 1971.
11838:
11839: A part of the information in this section comes from @cite{The Evolution
11840: of Forth} by Elizabeth D. Rather, Donald R. Colburn and Charles
11841: H. Moore, presented at the HOPL-II conference and preprinted in SIGPLAN
11842: Notices 28(3), 1993. You can find more historical and genealogical
11843: information about Forth there.
11844:
11845: @node Forth-related information, Word Index, Origin, Top
11846: @appendix Other Forth-related information
11847: @cindex Forth-related information
11848:
11849: @menu
11850: * Internet resources::
11851: * Books::
11852: * The Forth Interest Group::
11853: * Conferences::
11854: @end menu
11855:
11856:
11857: @node Internet resources, Books, Forth-related information, Forth-related information
11858: @section Internet resources
11859: @cindex internet resources
11860:
11861: @cindex comp.lang.forth
11862: @cindex frequently asked questions
11863: There is an active news group (comp.lang.forth) discussing Forth and
11864: Forth-related issues. A frequently-asked-questions (FAQ) list
11865: is posted to the news group regularly, and archived at these sites:
11866:
11867: @itemize @bullet
11868: @item
11869: @uref{ftp://rtfm.mit.edu/pub/usenet-by-group/comp.lang.forth/}
11870: @item
11871: @uref{ftp://ftp.forth.org/pub/Forth/FAQ/}
11872: @end itemize
11873:
11874: The FAQ list should be considered mandatory reading before posting to
11875: the news group.
11876:
11877: Here are some other web sites holding Forth-related material:
11878:
11879: @itemize @bullet
11880: @item
11881: @uref{http://www.taygeta.com/forth.html} -- Skip Carter's Forth pages.
11882: @item
11883: @uref{http://www.jwdt.com/~paysan/gforth.html} -- the Gforth home page.
11884: @item
11885: @uref{http://www.minerva.com/uathena.htm} -- home of ANS Forth Standard.
11886: @item
11887: @uref{http://dec.bournemouth.ac.uk/forth/index.html} -- the Forth
11888: Research page, including links to the Journal of Forth Application and
11889: Research (JFAR) and a searchable Forth bibliography.
11890: @end itemize
11891:
11892:
11893: @node Books, The Forth Interest Group, Internet resources, Forth-related information
11894: @section Books
11895: @cindex books on Forth
11896:
11897: As the Standard is relatively new, there are not many books out yet. It
11898: is not recommended to learn Forth by using Gforth and a book that is not
11899: written for ANS Forth, as you will not know your mistakes from the
11900: deviations of the book. However, books based on the Forth-83 standard
11901: should be ok, because ANS Forth is primarily an extension of Forth-83.
11902: Refer to the Forth FAQ for details of Forth-related books.
11903:
11904: @cindex standard document for ANS Forth
11905: @cindex ANS Forth document
11906: The definite reference if you want to write ANS Forth programs is, of
11907: course, the ANS Forth document. It is available in printed form from the
11908: National Standards Institute Sales Department (Tel.: USA (212) 642-4900;
11909: Fax.: USA (212) 302-1286) as document @cite{X3.215-1994} for about
11910: $200. You can also get it from Global Engineering Documents (Tel.: USA
11911: (800) 854-7179; Fax.: (303) 843-9880) for about $300.
11912:
11913: @cite{dpANS6}, the last draft of the standard, which was then submitted
11914: to ANSI for publication is available electronically and for free in some
11915: MS Word format, and it has been converted to HTML
11916: (@uref{http://www.taygeta.com/forth/dpans.html}; this HTML version also
11917: includes the answers to Requests for Interpretation (RFIs). Some
11918: pointers to these versions can be found through
11919: @*@uref{http://www.complang.tuwien.ac.at/projects/forth.html}.
11920:
11921:
11922: @node The Forth Interest Group, Conferences, Books, Forth-related information
11923: @section The Forth Interest Group
11924: @cindex Forth interest group (FIG)
11925:
11926: The Forth Interest Group (FIG) is a world-wide, non-profit,
11927: member-supported organisation. It publishes a regular magazine,
11928: @var{FORTH Dimensions}, and offers other benefits of membership. You can
11929: contact the FIG through their office email address:
11930: @email{office@@forth.org} or by visiting their web site at
11931: @uref{http://www.forth.org/}. This web site also includes links to FIG
11932: chapters in other countries and American cities
11933: (@uref{http://www.forth.org/chapters.html}).
11934:
11935: @node Conferences, , The Forth Interest Group, Forth-related information
11936: @section Conferences
11937: @cindex Conferences
11938:
11939: There are several regular conferences related to Forth. They are all
11940: well-publicised in @var{FORTH Dimensions} and on the comp.lang.forth
11941: news group:
11942:
11943: @itemize @bullet
11944: @item
11945: FORML -- the Forth modification laboratory convenes every year near
11946: Monterey, California.
11947: @item
11948: The Rochester Forth Conference -- an annual conference traditionally
11949: held in Rochester, New York.
11950: @item
11951: EuroForth -- this European conference takes place annually.
11952: @end itemize
11953:
11954:
11955: @node Word Index, Name Index, Forth-related information, Top
11956: @unnumbered Word Index
11957:
11958: This index is a list of Forth words that have ``glossary'' entries
11959: within this manual. Each word is listed with its stack effect and
11960: wordset.
11961:
11962: @printindex fn
11963:
11964: @node Name Index, Concept Index, Word Index, Top
11965: @unnumbered Name Index
11966:
11967: This index is a list of Forth words that have ``glossary'' entries
11968: within this manual.
11969:
11970: @printindex ky
11971:
11972: @node Concept Index, , Name Index, Top
11973: @unnumbered Concept and Word Index
11974:
11975: Not all entries listed in this index are present verbatim in the
11976: text. This index also duplicates, in abbreviated form, all of the words
11977: listed in the Word Index (only the names are listed for the words here).
11978:
11979: @printindex cp
11980:
11981: @contents
11982: @bye
11983:
FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>
![[BACK]](/cvsweb/icons/back.gif){kind=icon}
Return to gforth.ds CVS log ![[TXT]](/cvsweb/icons/text.gif){kind=icon}
![[DIR]](/cvsweb/icons/dir.gif){kind=icon}
Up to [gforth] / gforth / doc