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 04-May-1999
99:
100: @comment The following two commands start the copyright page.
101: @page
102: @vskip 0pt plus 1filll
103: Copyright @copyright{} 1995--1999 Free Software Foundation, Inc.
104:
105: @comment !! Published by ... or You can get a copy of this manual ...
106:
107: Permission is granted to make and distribute verbatim copies of
108: this manual provided the copyright notice and this permission notice
109: are preserved on all copies.
110:
111: Permission is granted to copy and distribute modified versions of this
112: manual under the conditions for verbatim copying, provided also that the
113: sections entitled "Distribution" and "General Public License" are
114: included exactly as in the original, and provided that the entire
115: resulting derived work is distributed under the terms of a permission
116: notice identical to this one.
117:
118: Permission is granted to copy and distribute translations of this manual
119: into another language, under the above conditions for modified versions,
120: except that the sections entitled "Distribution" and "General Public
121: License" may be included in a translation approved by the author instead
122: of in the original English.
123: @end titlepage
124:
125:
126: @node Top, License, (dir), (dir)
127: @ifinfo
128: Gforth is a free implementation of ANS Forth available on many
129: personal machines. This manual corresponds to version @value{VERSION}.
130: @end ifinfo
131:
132: @menu
133: * License:: The GPL
134: * Goals:: About the Gforth Project
135: * Gforth Environment:: Starting (and exiting) Gforth
136: * Introduction:: An introduction to ANS Forth
137: * Words:: Forth words available in Gforth
138: * Error messages:: How to interpret them
139: * Tools:: Programming tools
140: * ANS conformance:: Implementation-defined options etc.
141: * Model:: The abstract machine of Gforth
142: * Integrating Gforth:: Forth as scripting language for applications
143: * Emacs and Gforth:: The Gforth Mode
144: * Image Files:: @code{.fi} files contain compiled code
145: * Engine:: The inner interpreter and the primitives
146: * Binding to System Library::
147: * Cross Compiler:: The Cross Compiler
148: * Bugs:: How to report them
149: * Origin:: Authors and ancestors of Gforth
150: * Forth-related information:: Books and places to look on the WWW
151: * Word Index:: An item for each Forth word
152: * Concept Index:: A menu covering many topics
153:
154: @detailmenu --- The Detailed Node Listing ---
155:
156: Goals of Gforth
157:
158: * Gforth Extensions Sinful?::
159:
160: Gforth Environment
161:
162: * Invoking Gforth:: Getting in
163: * Leaving Gforth:: Getting out
164: * Command-line editing::
165: * Upper and lower case::
166: * Environment variables:: ..that affect how Gforth starts up
167: * Gforth Files:: What gets installed and where
168:
169: An Introduction to ANS Forth
170:
171: * Introducing the Text Interpreter::
172: * Stacks and Postfix notation::
173: * Your first definition::
174: * How does that work?::
175: * Forth is written in Forth::
176: * Review - elements of a Forth system::
177: * Where to go next::
178: * Exercises::
179:
180: Forth Words
181:
182: * Notation::
183: * Comments::
184: * Boolean Flags::
185: * Arithmetic::
186: * Stack Manipulation::
187: * Memory::
188: * Control Structures::
189: * Defining Words::
190: * The Text Interpreter::
191: * Tokens for Words::
192: * Word Lists::
193: * Environmental Queries::
194: * Files::
195: * Blocks::
196: * Other I/O::
197: * Programming Tools::
198: * Assembler and Code Words::
199: * Threading Words::
200: * Locals::
201: * Structures::
202: * Object-oriented Forth::
203: * Passing Commands to the OS::
204: * Miscellaneous Words::
205:
206: Arithmetic
207:
208: * Single precision::
209: * Bitwise operations::
210: * Double precision:: Double-cell integer arithmetic
211: * Numeric comparison::
212: * Mixed precision:: Operations with single and double-cell integers
213: * Floating Point::
214:
215: Stack Manipulation
216:
217: * Data stack::
218: * Floating point stack::
219: * Return stack::
220: * Locals stack::
221: * Stack pointer manipulation::
222:
223: Memory
224:
225: * Memory model::
226: * Dictionary allocation::
227: * Heap Allocation::
228: * Memory Access::
229: * Address arithmetic::
230: * Memory Blocks::
231:
232: Control Structures
233:
234: * Selection:: IF.. ELSE.. ENDIF
235: * Simple Loops:: BEGIN..
236: * Counted Loops:: DO
237: * Arbitrary control structures::
238: * Calls and returns::
239: * Exception Handling::
240:
241: Defining Words
242:
243: * Simple Defining Words:: Variables, values and constants
244: * Colon Definitions::
245: * User-defined Defining Words::
246: * Supplying names::
247: * Interpretation and Compilation Semantics::
248:
249: The Text Interpreter
250:
251: * Input Sources::
252: * Number Conversion::
253: * Interpret/Compile states::
254: * Literals::
255: * Interpreter Directives::
256:
257: Word Lists
258:
259: * Why use word lists?::
260: * Word list examples::
261:
262: Files
263:
264: * Forth source files::
265: * General files::
266: * Search Paths::
267: * Forth Search Paths::
268: * General Search Paths::
269:
270: Other I/O
271:
272: * Simple numeric output:: Predefined formats
273: * Formatted numeric output:: Formatted (pictured) output
274: * String Formats:: How Forth stores strings in memory
275: * Displaying characters and strings:: Other stuff
276: * Input:: Input
277:
278: Programming Tools
279:
280: * Debugging:: Simple and quick.
281: * Assertions:: Making your programs self-checking.
282: * Singlestep Debugger:: Executing your program word by word.
283:
284: Locals
285:
286: * Gforth locals::
287: * ANS Forth locals::
288:
289: Gforth locals
290:
291: * Where are locals visible by name?::
292: * How long do locals live?::
293: * Programming Style::
294: * Implementation::
295:
296: Structures
297:
298: * Why explicit structure support?::
299: * Structure Usage::
300: * Structure Naming Convention::
301: * Structure Implementation::
302: * Structure Glossary::
303:
304: Object-oriented Forth
305:
306: * Why object-oriented programming?::
307: * Object-Oriented Terminology::
308: * Objects::
309: * OOF::
310: * Mini-OOF::
311: * Comparison with other object models::
312:
313: The @file{objects.fs} model
314:
315: * Properties of the Objects model::
316: * Basic Objects Usage::
317: * The Objects base class::
318: * Creating objects::
319: * Object-Oriented Programming Style::
320: * Class Binding::
321: * Method conveniences::
322: * Classes and Scoping::
323: * Object Interfaces::
324: * Objects Implementation::
325: * Objects Glossary::
326:
327: The @file{oof.fs} model
328:
329: * Properties of the OOF model::
330: * Basic OOF Usage::
331: * The OOF base class::
332: * Class Declaration::
333: * Class Implementation::
334:
335: The @file{mini-oof.fs} model
336:
337: * Basic Mini-OOF Usage::
338: * Mini-OOF Example::
339: * Mini-OOF Implementation::
340:
341: Tools
342:
343: * ANS Report:: Report the words used, sorted by wordset.
344:
345: ANS conformance
346:
347: * The Core Words::
348: * The optional Block word set::
349: * The optional Double Number word set::
350: * The optional Exception word set::
351: * The optional Facility word set::
352: * The optional File-Access word set::
353: * The optional Floating-Point word set::
354: * The optional Locals word set::
355: * The optional Memory-Allocation word set::
356: * The optional Programming-Tools word set::
357: * The optional Search-Order word set::
358:
359: The Core Words
360:
361: * core-idef:: Implementation Defined Options
362: * core-ambcond:: Ambiguous Conditions
363: * core-other:: Other System Documentation
364:
365: The optional Block word set
366:
367: * block-idef:: Implementation Defined Options
368: * block-ambcond:: Ambiguous Conditions
369: * block-other:: Other System Documentation
370:
371: The optional Double Number word set
372:
373: * double-ambcond:: Ambiguous Conditions
374:
375: The optional Exception word set
376:
377: * exception-idef:: Implementation Defined Options
378:
379: The optional Facility word set
380:
381: * facility-idef:: Implementation Defined Options
382: * facility-ambcond:: Ambiguous Conditions
383:
384: The optional File-Access word set
385:
386: * file-idef:: Implementation Defined Options
387: * file-ambcond:: Ambiguous Conditions
388:
389: The optional Floating-Point word set
390:
391: * floating-idef:: Implementation Defined Options
392: * floating-ambcond:: Ambiguous Conditions
393:
394: The optional Locals word set
395:
396: * locals-idef:: Implementation Defined Options
397: * locals-ambcond:: Ambiguous Conditions
398:
399: The optional Memory-Allocation word set
400:
401: * memory-idef:: Implementation Defined Options
402:
403: The optional Programming-Tools word set
404:
405: * programming-idef:: Implementation Defined Options
406: * programming-ambcond:: Ambiguous Conditions
407:
408: The optional Search-Order word set
409:
410: * search-idef:: Implementation Defined Options
411: * search-ambcond:: Ambiguous Conditions
412:
413: Image Files
414:
415: * Image Licensing Issues:: Distribution terms for images.
416: * Image File Background:: Why have image files?
417: * Non-Relocatable Image Files:: don't always work.
418: * Data-Relocatable Image Files:: are better.
419: * Fully Relocatable Image Files:: better yet.
420: * Stack and Dictionary Sizes:: Setting the default sizes for an image.
421: * Running Image Files:: @code{gforth -i @i{file}} or @i{file}.
422: * Modifying the Startup Sequence:: and turnkey applications.
423:
424: Fully Relocatable Image Files
425:
426: * gforthmi:: The normal way
427: * cross.fs:: The hard way
428:
429: Engine
430:
431: * Portability::
432: * Threading::
433: * Primitives::
434: * Performance::
435:
436: Threading
437:
438: * Scheduling::
439: * Direct or Indirect Threaded?::
440: * DOES>::
441:
442: Primitives
443:
444: * Automatic Generation::
445: * TOS Optimization::
446: * Produced code::
447:
448: Cross Compiler
449:
450: * Using the Cross Compiler::
451: * How the Cross Compiler Works::
452:
453: Other Forth-related information
454:
455: * Internet resources::
456: * Books::
457: * The Forth Interest Group::
458: * Conferences::
459:
460: @end detailmenu
461: @end menu
462:
463: @node License, Goals, Top, Top
464: @unnumbered GNU GENERAL PUBLIC LICENSE
465: @center Version 2, June 1991
466:
467: @display
468: Copyright @copyright{} 1989, 1991 Free Software Foundation, Inc.
469: 675 Mass Ave, Cambridge, MA 02139, USA
470:
471: Everyone is permitted to copy and distribute verbatim copies
472: of this license document, but changing it is not allowed.
473: @end display
474:
475: @unnumberedsec Preamble
476:
477: The licenses for most software are designed to take away your
478: freedom to share and change it. By contrast, the GNU General Public
479: License is intended to guarantee your freedom to share and change free
480: software---to make sure the software is free for all its users. This
481: General Public License applies to most of the Free Software
482: Foundation's software and to any other program whose authors commit to
483: using it. (Some other Free Software Foundation software is covered by
484: the GNU Library General Public License instead.) You can apply it to
485: your programs, too.
486:
487: When we speak of free software, we are referring to freedom, not
488: price. Our General Public Licenses are designed to make sure that you
489: have the freedom to distribute copies of free software (and charge for
490: this service if you wish), that you receive source code or can get it
491: if you want it, that you can change the software or use pieces of it
492: in new free programs; and that you know you can do these things.
493:
494: To protect your rights, we need to make restrictions that forbid
495: anyone to deny you these rights or to ask you to surrender the rights.
496: These restrictions translate to certain responsibilities for you if you
497: distribute copies of the software, or if you modify it.
498:
499: For example, if you distribute copies of such a program, whether
500: gratis or for a fee, you must give the recipients all the rights that
501: you have. You must make sure that they, too, receive or can get the
502: source code. And you must show them these terms so they know their
503: rights.
504:
505: We protect your rights with two steps: (1) copyright the software, and
506: (2) offer you this license which gives you legal permission to copy,
507: distribute and/or modify the software.
508:
509: Also, for each author's protection and ours, we want to make certain
510: that everyone understands that there is no warranty for this free
511: software. If the software is modified by someone else and passed on, we
512: want its recipients to know that what they have is not the original, so
513: that any problems introduced by others will not reflect on the original
514: authors' reputations.
515:
516: Finally, any free program is threatened constantly by software
517: patents. We wish to avoid the danger that redistributors of a free
518: program will individually obtain patent licenses, in effect making the
519: program proprietary. To prevent this, we have made it clear that any
520: patent must be licensed for everyone's free use or not licensed at all.
521:
522: The precise terms and conditions for copying, distribution and
523: modification follow.
524:
525: @iftex
526: @unnumberedsec TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION
527: @end iftex
528: @ifinfo
529: @center TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION
530: @end ifinfo
531:
532: @enumerate 0
533: @item
534: This License applies to any program or other work which contains
535: a notice placed by the copyright holder saying it may be distributed
536: under the terms of this General Public License. The ``Program'', below,
537: refers to any such program or work, and a ``work based on the Program''
538: means either the Program or any derivative work under copyright law:
539: that is to say, a work containing the Program or a portion of it,
540: either verbatim or with modifications and/or translated into another
541: language. (Hereinafter, translation is included without limitation in
542: the term ``modification''.) Each licensee is addressed as ``you''.
543:
544: Activities other than copying, distribution and modification are not
545: covered by this License; they are outside its scope. The act of
546: running the Program is not restricted, and the output from the Program
547: is covered only if its contents constitute a work based on the
548: Program (independent of having been made by running the Program).
549: Whether that is true depends on what the Program does.
550:
551: @item
552: You may copy and distribute verbatim copies of the Program's
553: source code as you receive it, in any medium, provided that you
554: conspicuously and appropriately publish on each copy an appropriate
555: copyright notice and disclaimer of warranty; keep intact all the
556: notices that refer to this License and to the absence of any warranty;
557: and give any other recipients of the Program a copy of this License
558: along with the Program.
559:
560: You may charge a fee for the physical act of transferring a copy, and
561: you may at your option offer warranty protection in exchange for a fee.
562:
563: @item
564: You may modify your copy or copies of the Program or any portion
565: of it, thus forming a work based on the Program, and copy and
566: distribute such modifications or work under the terms of Section 1
567: above, provided that you also meet all of these conditions:
568:
569: @enumerate a
570: @item
571: You must cause the modified files to carry prominent notices
572: stating that you changed the files and the date of any change.
573:
574: @item
575: You must cause any work that you distribute or publish, that in
576: whole or in part contains or is derived from the Program or any
577: part thereof, to be licensed as a whole at no charge to all third
578: parties under the terms of this License.
579:
580: @item
581: If the modified program normally reads commands interactively
582: when run, you must cause it, when started running for such
583: interactive use in the most ordinary way, to print or display an
584: announcement including an appropriate copyright notice and a
585: notice that there is no warranty (or else, saying that you provide
586: a warranty) and that users may redistribute the program under
587: these conditions, and telling the user how to view a copy of this
588: License. (Exception: if the Program itself is interactive but
589: does not normally print such an announcement, your work based on
590: the Program is not required to print an announcement.)
591: @end enumerate
592:
593: These requirements apply to the modified work as a whole. If
594: identifiable sections of that work are not derived from the Program,
595: and can be reasonably considered independent and separate works in
596: themselves, then this License, and its terms, do not apply to those
597: sections when you distribute them as separate works. But when you
598: distribute the same sections as part of a whole which is a work based
599: on the Program, the distribution of the whole must be on the terms of
600: this License, whose permissions for other licensees extend to the
601: entire whole, and thus to each and every part regardless of who wrote it.
602:
603: Thus, it is not the intent of this section to claim rights or contest
604: your rights to work written entirely by you; rather, the intent is to
605: exercise the right to control the distribution of derivative or
606: collective works based on the Program.
607:
608: In addition, mere aggregation of another work not based on the Program
609: with the Program (or with a work based on the Program) on a volume of
610: a storage or distribution medium does not bring the other work under
611: the scope of this License.
612:
613: @item
614: You may copy and distribute the Program (or a work based on it,
615: under Section 2) in object code or executable form under the terms of
616: Sections 1 and 2 above provided that you also do one of the following:
617:
618: @enumerate a
619: @item
620: Accompany it with the complete corresponding machine-readable
621: source code, which must be distributed under the terms of Sections
622: 1 and 2 above on a medium customarily used for software interchange; or,
623:
624: @item
625: Accompany it with a written offer, valid for at least three
626: years, to give any third party, for a charge no more than your
627: cost of physically performing source distribution, a complete
628: machine-readable copy of the corresponding source code, to be
629: distributed under the terms of Sections 1 and 2 above on a medium
630: customarily used for software interchange; or,
631:
632: @item
633: Accompany it with the information you received as to the offer
634: to distribute corresponding source code. (This alternative is
635: allowed only for noncommercial distribution and only if you
636: received the program in object code or executable form with such
637: an offer, in accord with Subsection b above.)
638: @end enumerate
639:
640: The source code for a work means the preferred form of the work for
641: making modifications to it. For an executable work, complete source
642: code means all the source code for all modules it contains, plus any
643: associated interface definition files, plus the scripts used to
644: control compilation and installation of the executable. However, as a
645: special exception, the source code distributed need not include
646: anything that is normally distributed (in either source or binary
647: form) with the major components (compiler, kernel, and so on) of the
648: operating system on which the executable runs, unless that component
649: itself accompanies the executable.
650:
651: If distribution of executable or object code is made by offering
652: access to copy from a designated place, then offering equivalent
653: access to copy the source code from the same place counts as
654: distribution of the source code, even though third parties are not
655: compelled to copy the source along with the object code.
656:
657: @item
658: You may not copy, modify, sublicense, or distribute the Program
659: except as expressly provided under this License. Any attempt
660: otherwise to copy, modify, sublicense or distribute the Program is
661: void, and will automatically terminate your rights under this License.
662: However, parties who have received copies, or rights, from you under
663: this License will not have their licenses terminated so long as such
664: parties remain in full compliance.
665:
666: @item
667: You are not required to accept this License, since you have not
668: signed it. However, nothing else grants you permission to modify or
669: distribute the Program or its derivative works. These actions are
670: prohibited by law if you do not accept this License. Therefore, by
671: modifying or distributing the Program (or any work based on the
672: Program), you indicate your acceptance of this License to do so, and
673: all its terms and conditions for copying, distributing or modifying
674: the Program or works based on it.
675:
676: @item
677: Each time you redistribute the Program (or any work based on the
678: Program), the recipient automatically receives a license from the
679: original licensor to copy, distribute or modify the Program subject to
680: these terms and conditions. You may not impose any further
681: restrictions on the recipients' exercise of the rights granted herein.
682: You are not responsible for enforcing compliance by third parties to
683: this License.
684:
685: @item
686: If, as a consequence of a court judgment or allegation of patent
687: infringement or for any other reason (not limited to patent issues),
688: conditions are imposed on you (whether by court order, agreement or
689: otherwise) that contradict the conditions of this License, they do not
690: excuse you from the conditions of this License. If you cannot
691: distribute so as to satisfy simultaneously your obligations under this
692: License and any other pertinent obligations, then as a consequence you
693: may not distribute the Program at all. For example, if a patent
694: license would not permit royalty-free redistribution of the Program by
695: all those who receive copies directly or indirectly through you, then
696: the only way you could satisfy both it and this License would be to
697: refrain entirely from distribution of the Program.
698:
699: If any portion of this section is held invalid or unenforceable under
700: any particular circumstance, the balance of the section is intended to
701: apply and the section as a whole is intended to apply in other
702: circumstances.
703:
704: It is not the purpose of this section to induce you to infringe any
705: patents or other property right claims or to contest validity of any
706: such claims; this section has the sole purpose of protecting the
707: integrity of the free software distribution system, which is
708: implemented by public license practices. Many people have made
709: generous contributions to the wide range of software distributed
710: through that system in reliance on consistent application of that
711: system; it is up to the author/donor to decide if he or she is willing
712: to distribute software through any other system and a licensee cannot
713: impose that choice.
714:
715: This section is intended to make thoroughly clear what is believed to
716: be a consequence of the rest of this License.
717:
718: @item
719: If the distribution and/or use of the Program is restricted in
720: certain countries either by patents or by copyrighted interfaces, the
721: original copyright holder who places the Program under this License
722: may add an explicit geographical distribution limitation excluding
723: those countries, so that distribution is permitted only in or among
724: countries not thus excluded. In such case, this License incorporates
725: the limitation as if written in the body of this License.
726:
727: @item
728: The Free Software Foundation may publish revised and/or new versions
729: of the General Public License from time to time. Such new versions will
730: be similar in spirit to the present version, but may differ in detail to
731: address new problems or concerns.
732:
733: Each version is given a distinguishing version number. If the Program
734: specifies a version number of this License which applies to it and ``any
735: later version'', you have the option of following the terms and conditions
736: either of that version or of any later version published by the Free
737: Software Foundation. If the Program does not specify a version number of
738: this License, you may choose any version ever published by the Free Software
739: Foundation.
740:
741: @item
742: If you wish to incorporate parts of the Program into other free
743: programs whose distribution conditions are different, write to the author
744: to ask for permission. For software which is copyrighted by the Free
745: Software Foundation, write to the Free Software Foundation; we sometimes
746: make exceptions for this. Our decision will be guided by the two goals
747: of preserving the free status of all derivatives of our free software and
748: of promoting the sharing and reuse of software generally.
749:
750: @iftex
751: @heading NO WARRANTY
752: @end iftex
753: @ifinfo
754: @center NO WARRANTY
755: @end ifinfo
756:
757: @item
758: BECAUSE THE PROGRAM IS LICENSED FREE OF CHARGE, THERE IS NO WARRANTY
759: FOR THE PROGRAM, TO THE EXTENT PERMITTED BY APPLICABLE LAW. EXCEPT WHEN
760: OTHERWISE STATED IN WRITING THE COPYRIGHT HOLDERS AND/OR OTHER PARTIES
761: PROVIDE THE PROGRAM ``AS IS'' WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESSED
762: OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
763: MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. THE ENTIRE RISK AS
764: TO THE QUALITY AND PERFORMANCE OF THE PROGRAM IS WITH YOU. SHOULD THE
765: PROGRAM PROVE DEFECTIVE, YOU ASSUME THE COST OF ALL NECESSARY SERVICING,
766: REPAIR OR CORRECTION.
767:
768: @item
769: IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING
770: WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MAY MODIFY AND/OR
771: REDISTRIBUTE THE PROGRAM AS PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES,
772: INCLUDING ANY GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING
773: OUT OF THE USE OR INABILITY TO USE THE PROGRAM (INCLUDING BUT NOT LIMITED
774: TO LOSS OF DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY
775: YOU OR THIRD PARTIES OR A FAILURE OF THE PROGRAM TO OPERATE WITH ANY OTHER
776: PROGRAMS), EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE
777: POSSIBILITY OF SUCH DAMAGES.
778: @end enumerate
779:
780: @iftex
781: @heading END OF TERMS AND CONDITIONS
782: @end iftex
783: @ifinfo
784: @center END OF TERMS AND CONDITIONS
785: @end ifinfo
786:
787: @page
788: @unnumberedsec How to Apply These Terms to Your New Programs
789:
790: If you develop a new program, and you want it to be of the greatest
791: possible use to the public, the best way to achieve this is to make it
792: free software which everyone can redistribute and change under these terms.
793:
794: To do so, attach the following notices to the program. It is safest
795: to attach them to the start of each source file to most effectively
796: convey the exclusion of warranty; and each file should have at least
797: the ``copyright'' line and a pointer to where the full notice is found.
798:
799: @smallexample
800: @var{one line to give the program's name and a brief idea of what it does.}
801: Copyright (C) 19@var{yy} @var{name of author}
802:
803: This program is free software; you can redistribute it and/or modify
804: it under the terms of the GNU General Public License as published by
805: the Free Software Foundation; either version 2 of the License, or
806: (at your option) any later version.
807:
808: This program is distributed in the hope that it will be useful,
809: but WITHOUT ANY WARRANTY; without even the implied warranty of
810: MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
811: GNU General Public License for more details.
812:
813: You should have received a copy of the GNU General Public License
814: along with this program; if not, write to the Free Software
815: Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
816: @end smallexample
817:
818: Also add information on how to contact you by electronic and paper mail.
819:
820: If the program is interactive, make it output a short notice like this
821: when it starts in an interactive mode:
822:
823: @smallexample
824: Gnomovision version 69, Copyright (C) 19@var{yy} @var{name of author}
825: Gnomovision comes with ABSOLUTELY NO WARRANTY; for details
826: type `show w'.
827: This is free software, and you are welcome to redistribute it
828: under certain conditions; type `show c' for details.
829: @end smallexample
830:
831: The hypothetical commands @samp{show w} and @samp{show c} should show
832: the appropriate parts of the General Public License. Of course, the
833: commands you use may be called something other than @samp{show w} and
834: @samp{show c}; they could even be mouse-clicks or menu items---whatever
835: suits your program.
836:
837: You should also get your employer (if you work as a programmer) or your
838: school, if any, to sign a ``copyright disclaimer'' for the program, if
839: necessary. Here is a sample; alter the names:
840:
841: @smallexample
842: Yoyodyne, Inc., hereby disclaims all copyright interest in the program
843: `Gnomovision' (which makes passes at compilers) written by James Hacker.
844:
845: @var{signature of Ty Coon}, 1 April 1989
846: Ty Coon, President of Vice
847: @end smallexample
848:
849: This General Public License does not permit incorporating your program into
850: proprietary programs. If your program is a subroutine library, you may
851: consider it more useful to permit linking proprietary applications with the
852: library. If this is what you want to do, use the GNU Library General
853: Public License instead of this License.
854:
855: @iftex
856: @unnumbered Preface
857: @cindex Preface
858: This manual documents Gforth. Some introductory material is provided for
859: readers who are unfamiliar with Forth or who are migrating to Gforth
860: from other Forth compilers. However, this manual is primarily a
861: reference manual.
862: @end iftex
863:
864: @comment TODO much more blurb here.
865:
866: @c ******************************************************************
867: @node Goals, Gforth Environment, License, Top
868: @comment node-name, next, previous, up
869: @chapter Goals of Gforth
870: @cindex goals of the Gforth project
871: The goal of the Gforth Project is to develop a standard model for
872: ANS Forth. This can be split into several subgoals:
873:
874: @itemize @bullet
875: @item
876: Gforth should conform to the ANS Forth Standard.
877: @item
878: It should be a model, i.e. it should define all the
879: implementation-dependent things.
880: @item
881: It should become standard, i.e. widely accepted and used. This goal
882: is the most difficult one.
883: @end itemize
884:
885: To achieve these goals Gforth should be
886: @itemize @bullet
887: @item
888: Similar to previous models (fig-Forth, F83)
889: @item
890: Powerful. It should provide for all the things that are considered
891: necessary today and even some that are not yet considered necessary.
892: @item
893: Efficient. It should not get the reputation of being exceptionally
894: slow.
895: @item
896: Free.
897: @item
898: Available on many machines/easy to port.
899: @end itemize
900:
901: Have we achieved these goals? Gforth conforms to the ANS Forth
902: standard. It may be considered a model, but we have not yet documented
903: which parts of the model are stable and which parts we are likely to
904: change. It certainly has not yet become a de facto standard, but it
905: appears to be quite popular. It has some similarities to and some
906: differences from previous models. It has some powerful features, but not
907: yet everything that we envisioned. We certainly have achieved our
908: execution speed goals (@pxref{Performance}). It is free and available
909: on many machines.
910:
911: @menu
912: * Gforth Extensions Sinful?::
913: @end menu
914:
915: @node Gforth Extensions Sinful?, , Goals, Goals
916: @comment node-name, next, previous, up
917: @section Is it a Sin to use Gforth Extensions?
918: @cindex Gforth extensions
919:
920: If you've been paying attention, you will have realised that there is an
921: ANS (American National Standard) for Forth. As you read through the rest
922: of this manual, you will see documentation for @i{Standard} words, and
923: documentation for some appealing Gforth @i{extensions}. You might ask
924: yourself the question: @i{``Given that there is a standard, would I be
925: committing a sin to use (non-Standard) Gforth extensions?''}
926:
927: The answer to that question is somewhat pragmatic and somewhat
928: philosophical. Consider these points:
929:
930: @itemize @bullet
931: @item
932: A number of the Gforth extensions can be implemented in ANS Forth using
933: files provided in the @file{compat/} directory. These are mentioned in
934: the text in passing.
935: @item
936: Forth has a rich historical precedent for programmers taking advantage
937: of implementation-dependent features of their tools (for example,
938: relying on a knowledge of the dictionary structure). Sometimes these
939: techniques are necessary to extract every last bit of performance from
940: the hardware, sometimes they are just a programming shorthand.
941: @item
942: The best way to break the rules is to know what the rules are. To learn
943: the rules, there is no substitute for studying the text of the Standard
944: itself. In particular, Appendix A of the Standard (@var{Rationale})
945: provides a valuable insight into the thought processes of the technical
946: committee.
947: @item
948: The best reason to break a rule is because you have to; because it's
949: more productive to do that, because it makes your code run fast enough
950: or because you can see no Standard way to achieve what you want to
951: achieve.
952: @end itemize
953:
954: The tool @file{ans-report.fs} (@pxref{ANS Report}) makes it easy to
955: analyse your program and determine what non-Standard definitions it
956: relies upon.
957:
958:
959: @c ******************************************************************
960: @node Gforth Environment, Introduction, Goals, Top
961: @chapter Gforth Environment
962: @cindex Gforth environment
963:
964: Note: ultimately, the gforth man page will be auto-generated from the
965: material in this chapter.
966:
967: @menu
968: * Invoking Gforth:: Getting in
969: * Leaving Gforth:: Getting out
970: * Command-line editing::
971: * Upper and lower case::
972: * Environment variables:: ..that affect how Gforth starts up
973: * Gforth Files:: What gets installed and where
974: @end menu
975:
976: @xref{Image Files} for related information about the creation of images.
977:
978: @comment ----------------------------------------------
979: @node Invoking Gforth, Leaving Gforth, ,Gforth Environment
980: @section Invoking Gforth
981: @cindex invoking Gforth
982: @cindex running Gforth
983: @cindex command-line options
984: @cindex options on the command line
985: @cindex flags on the command line
986:
987: Gforth is made up of two parts; an executable ``engine'' (named
988: @file{gforth} or @file{gforth-fast}) and an image file. To start it, you
989: will usually just say @code{gforth} -- this automatically loads the
990: default image file @file{gforth.fi}. In many other cases the default
991: Gforth image will be invoked like this:
992: @example
993: gforth [file | -e forth-code] ...
994: @end example
995: @noindent
996: This interprets the contents of the files and the Forth code in the order they
997: are given.
998:
999: In addition to the @file{gforth} engine, there is also an engine called
1000: @file{gforth-fast}, which is faster, but gives less informative error
1001: messages (@pxref{Error messages}).
1002:
1003: In general, the command line looks like this:
1004:
1005: @example
1006: gforth[-fast] [engine options] [image options]
1007: @end example
1008:
1009: The engine options must come before the rest of the command
1010: line. They are:
1011:
1012: @table @code
1013: @cindex -i, command-line option
1014: @cindex --image-file, command-line option
1015: @item --image-file @i{file}
1016: @itemx -i @i{file}
1017: Loads the Forth image @i{file} instead of the default
1018: @file{gforth.fi} (@pxref{Image Files}).
1019:
1020: @cindex --appl-image, command-line option
1021: @item --appl-image @i{file}
1022: Loads the image @i{file} and leaves all further command-line arguments
1023: to the image (instead of processing them as options). This is useful
1024: for building executable application images on Unix, built with
1025: @code{gforthmi --application ...}.
1026:
1027: @cindex --path, command-line option
1028: @cindex -p, command-line option
1029: @item --path @i{path}
1030: @itemx -p @i{path}
1031: Uses @i{path} for searching the image file and Forth source code files
1032: instead of the default in the environment variable @code{GFORTHPATH} or
1033: the path specified at installation time (e.g.,
1034: @file{/usr/local/share/gforth/0.2.0:.}). A path is given as a list of
1035: directories, separated by @samp{:} (on Unix) or @samp{;} (on other OSs).
1036:
1037: @cindex --dictionary-size, command-line option
1038: @cindex -m, command-line option
1039: @cindex @i{size} parameters for command-line options
1040: @cindex size of the dictionary and the stacks
1041: @item --dictionary-size @i{size}
1042: @itemx -m @i{size}
1043: Allocate @i{size} space for the Forth dictionary space instead of
1044: using the default specified in the image (typically 256K). The
1045: @i{size} specification for this and subsequent options consists of
1046: an integer and a unit (e.g.,
1047: @code{4M}). The unit can be one of @code{b} (bytes), @code{e} (element
1048: size, in this case Cells), @code{k} (kilobytes), @code{M} (Megabytes),
1049: @code{G} (Gigabytes), and @code{T} (Terabytes). If no unit is specified,
1050: @code{e} is used.
1051:
1052: @cindex --data-stack-size, command-line option
1053: @cindex -d, command-line option
1054: @item --data-stack-size @i{size}
1055: @itemx -d @i{size}
1056: Allocate @i{size} space for the data stack instead of using the
1057: default specified in the image (typically 16K).
1058:
1059: @cindex --return-stack-size, command-line option
1060: @cindex -r, command-line option
1061: @item --return-stack-size @i{size}
1062: @itemx -r @i{size}
1063: Allocate @i{size} space for the return stack instead of using the
1064: default specified in the image (typically 15K).
1065:
1066: @cindex --fp-stack-size, command-line option
1067: @cindex -f, command-line option
1068: @item --fp-stack-size @i{size}
1069: @itemx -f @i{size}
1070: Allocate @i{size} space for the floating point stack instead of
1071: using the default specified in the image (typically 15.5K). In this case
1072: the unit specifier @code{e} refers to floating point numbers.
1073:
1074: @cindex --locals-stack-size, command-line option
1075: @cindex -l, command-line option
1076: @item --locals-stack-size @i{size}
1077: @itemx -l @i{size}
1078: Allocate @i{size} space for the locals stack instead of using the
1079: default specified in the image (typically 14.5K).
1080:
1081: @cindex -h, command-line option
1082: @cindex --help, command-line option
1083: @item --help
1084: @itemx -h
1085: Print a message about the command-line options
1086:
1087: @cindex -v, command-line option
1088: @cindex --version, command-line option
1089: @item --version
1090: @itemx -v
1091: Print version and exit
1092:
1093: @cindex --debug, command-line option
1094: @item --debug
1095: Print some information useful for debugging on startup.
1096:
1097: @cindex --offset-image, command-line option
1098: @item --offset-image
1099: Start the dictionary at a slightly different position than would be used
1100: otherwise (useful for creating data-relocatable images,
1101: @pxref{Data-Relocatable Image Files}).
1102:
1103: @cindex --no-offset-im, command-line option
1104: @item --no-offset-im
1105: Start the dictionary at the normal position.
1106:
1107: @cindex --clear-dictionary, command-line option
1108: @item --clear-dictionary
1109: Initialize all bytes in the dictionary to 0 before loading the image
1110: (@pxref{Data-Relocatable Image Files}).
1111:
1112: @cindex --die-on-signal, command-line-option
1113: @item --die-on-signal
1114: Normally Gforth handles most signals (e.g., the user interrupt SIGINT,
1115: or the segmentation violation SIGSEGV) by translating it into a Forth
1116: @code{THROW}. With this option, Gforth exits if it receives such a
1117: signal. This option is useful when the engine and/or the image might be
1118: severely broken (such that it causes another signal before recovering
1119: from the first); this option avoids endless loops in such cases.
1120: @end table
1121:
1122: @cindex loading files at startup
1123: @cindex executing code on startup
1124: @cindex batch processing with Gforth
1125: As explained above, the image-specific command-line arguments for the
1126: default image @file{gforth.fi} consist of a sequence of filenames and
1127: @code{-e @var{forth-code}} options that are interpreted in the sequence
1128: in which they are given. The @code{-e @var{forth-code}} or
1129: @code{--evaluate @var{forth-code}} option evaluates the Forth
1130: code. This option takes only one argument; if you want to evaluate more
1131: Forth words, you have to quote them or use @code{-e} several times. To exit
1132: after processing the command line (instead of entering interactive mode)
1133: append @code{-e bye} to the command line.
1134:
1135: @cindex versions, invoking other versions of Gforth
1136: If you have several versions of Gforth installed, @code{gforth} will
1137: invoke the version that was installed last. @code{gforth-@i{version}}
1138: invokes a specific version. You may want to use the option
1139: @code{--path}, if your environment contains the variable
1140: @code{GFORTHPATH}.
1141:
1142: Not yet implemented:
1143: On startup the system first executes the system initialization file
1144: (unless the option @code{--no-init-file} is given; note that the system
1145: resulting from using this option may not be ANS Forth conformant). Then
1146: the user initialization file @file{.gforth.fs} is executed, unless the
1147: option @code{--no-rc} is given; this file is first searched in @file{.},
1148: then in @file{~}, then in the normal path (see above).
1149:
1150:
1151:
1152: @comment ----------------------------------------------
1153: @node Leaving Gforth, Command-line editing, Invoking Gforth, Gforth Environment
1154: @section Leaving Gforth
1155: @cindex Gforth - leaving
1156: @cindex leaving Gforth
1157:
1158: You can leave Gforth by typing @code{bye} or @kbd{Ctrl-d} (at the start
1159: of a line) or (if you invoked Gforth with the @code{--die-on-signal}
1160: option) @kbd{Ctrl-c}. When you leave Gforth, all of your definitions and
1161: data are discarded. @xref{Image Files} for ways of saving the state of
1162: the system before leaving Gforth.
1163:
1164: doc-bye
1165:
1166: @comment ----------------------------------------------
1167: @node Command-line editing, Upper and lower case,Leaving Gforth,Gforth Environment
1168: @section Command-line editing
1169: @cindex command-line editing
1170:
1171: Gforth maintains a history file that records every line that you type to
1172: the text interpreter. This file is preserved between sessions, and is
1173: used to provide a command-line recall facility; if you type ctrl-P
1174: repeatedly you can recall successively older commands from this (or
1175: previous) session(s). The full list of command-line editing facilities is:
1176:
1177: @comment use @table? - anton
1178: @itemize @bullet
1179: @item
1180: @kbd{Ctrl-p} (``previous'') (or up-arrow) to recall successively older
1181: commands from the history buffer.
1182: @item
1183: @kbd{Ctrl-n} (``next'') (or down-arrow) to recall successively newer commands
1184: from the history buffer.
1185: @item
1186: @kbd{Ctrl-f} (or right-arrow) to move the cursor right, non-destructively.
1187: @item
1188: @kbd{Ctrl-b} (or left-arrow) to move the cursor left, non-destructively.
1189: @item
1190: @kbd{Ctrl-h} (backspace) to delete the character to the left of the cursor,
1191: closing up the line.
1192: @item
1193: @kbd{Ctrl-k} to delete (``kill'') from the cursor to the end of the line.
1194: @item
1195: @kbd{Ctrl-a} to move the cursor to the start of the line.
1196: @item
1197: @kbd{Ctrl-e} to move the cursor to the end of the line.
1198: @item
1199: @key{RET} (@kbd{Ctrl-m}) or @key{LFD} (@kbd{Ctrl-j}) to submit the current
1200: line.
1201: @item
1202: @key{TAB} to step through all possible full-word completions of the word
1203: currently being typed.
1204: @item
1205: @kbd{Ctrl-d} at the start of the line to terminate Gforth (gracefully,
1206: using @code{bye}).
1207: @end itemize
1208:
1209: When editing, displayable characters are inserted to the left of the
1210: cursor position; the line is always in ``insert'' (as opposed to
1211: ``overstrike'') mode.
1212:
1213: @cindex history file
1214: @cindex @file{.gforth-history}
1215: On Unix systems, the history file is @file{~/.gforth-history} by
1216: default@footnote{i.e. it is stored in the user's home directory.}. You
1217: can find out the name and location of your history file using:
1218:
1219: @example
1220: history-file type \ Unix-class systems
1221:
1222: history-file type \ Other systems
1223: history-dir type
1224: @end example
1225:
1226: If you enter long definitions by hand, you can use a text editor to
1227: paste them out of the history file into a Forth source file for reuse at
1228: a later time.
1229:
1230: Gforth never trims the size of the history file, so you should do this
1231: periodically, if necessary.
1232:
1233: @comment this is all defined in history.fs
1234:
1235:
1236:
1237: @comment ----------------------------------------------
1238: @node Upper and lower case, Environment variables,Command-line editing,Gforth Environment
1239: @section Upper and lower case
1240: @cindex case-sensitivity
1241: @cindex upper and lower case
1242:
1243: Gforth is case-insensitive, so you can enter definitions and invoke
1244: Standard words using upper, lower or mixed case (however,
1245: @pxref{core-idef, Implementation-defined options, Implementation-defined
1246: options}).
1247:
1248: ANS Forth only @i{requires} implementations to recognise Standard words
1249: when they are typed entirely in upper case. Therefore, a Standard
1250: program must use upper case for all Standard words. You can use whatever
1251: case you like for words that you define, but in a standard program you
1252: have to use the words in the same case that you defined them.
1253:
1254: Gforth supports case sensitivity through @code{table}s (case-sensitive
1255: wordlists, @pxref{Word Lists}).
1256:
1257: Two people have asked how to convert Gforth to case sensitivity; while
1258: we think this is a bad idea, you can change all wordlists into tables
1259: like this:
1260:
1261: @example
1262: ' table-find forth-wordlist wordlist-map @ !
1263: @end example
1264:
1265: Note that you now have to type the predefined words in the same case
1266: that we defined them, which are varying. You may want to convert them
1267: to your favourite case before doing this operation (I won't explain how,
1268: because if you are even contemplating to do this, you'd better have
1269: enough knowledge of Forth systems to know this already).
1270:
1271: @comment ----------------------------------------------
1272: @node Environment variables, Gforth Files, Upper and lower case,Gforth Environment
1273: @section Environment variables
1274: @cindex environment variables
1275:
1276: Gforth uses these environment variables:
1277:
1278: @itemize @bullet
1279: @item
1280: @cindex GFORTHHIST - environment variable
1281: GFORTHHIST - (Unix systems only) specifies the directory in which to
1282: open/create the history file, @file{.gforth-history}. Default:
1283: @code{$HOME}.
1284:
1285: @item
1286: @cindex GFORTHPATH - environment variable
1287: GFORTHPATH - specifies the path used when searching for the gforth image file and
1288: for Forth source-code files.
1289:
1290: @item
1291: @cindex GFORTH - environment variable
1292: GFORTH - used by @file{gforthmi} @xref{gforthmi}.
1293:
1294: @item
1295: @cindex GFORTHD - environment variable
1296: GFORTHD - used by @file{gforthmi} @xref{gforthmi}.
1297:
1298: @item
1299: @cindex TMP, TEMP - environment variable
1300: TMP, TEMP - (non-Unix systems only) used as a potential location for the
1301: history file.
1302: @end itemize
1303:
1304: @comment also POSIXELY_CORRECT LINES COLUMNS HOME but no interest in
1305: @comment mentioning these.
1306:
1307: All the Gforth environment variables default to sensible values if they
1308: are not set.
1309:
1310:
1311: @comment ----------------------------------------------
1312: @node Gforth Files, ,Environment variables,Gforth Environment
1313: @section Gforth files
1314: @cindex Gforth files
1315:
1316: When you Gforth on a Unix system in the default places, it installs
1317: files in these locations:
1318:
1319: @itemize @bullet
1320: @item
1321: @file{/usr/local/bin/gforth}
1322: @item
1323: @file{/usr/local/bin/gforthmi}
1324: @item
1325: @file{/usr/local/man/man1/gforth.1} - man page.
1326: @item
1327: @file{/usr/local/info} - the Info version of this manual.
1328: @item
1329: @file{/usr/local/lib/gforth/<version>/...} - Gforth @file{.fi} files.
1330: @item
1331: @file{/usr/local/share/gforth/<version>/TAGS} - Emacs TAGS file.
1332: @item
1333: @file{/usr/local/share/gforth/<version>/...} - Gforth source files.
1334: @item
1335: @file{.../emacs/site-lisp/gforth.el} - Emacs gforth mode.
1336: @end itemize
1337:
1338: You can select different places for installation by using
1339: @code{configure} options (listed with @code{configure --help}).
1340:
1341: @c ******************************************************************
1342: @node Introduction, Words, Gforth Environment, Top
1343: @comment node-name, next, previous, up
1344: @chapter An Introduction to ANS Forth
1345: @cindex Forth - an introduction
1346:
1347: The primary purpose of this manual is to document Gforth. However, since
1348: Forth is not a widely-known language and there is a lack of up-to-date
1349: teaching material, it seems worthwhile to provide some introductory
1350: material. @xref{Forth-related information} for other sources of Forth-related
1351: information.
1352:
1353: The examples in this section should work on any ANS Forth; the
1354: output shown was produced using Gforth. Each example attempts to
1355: reproduce the exact output that Gforth produces. If you try out the
1356: examples (and you should), what you should type is shown @kbd{like this}
1357: and Gforth's response is shown @code{like this}. The single exception is
1358: that, where the example shows @key{RET} it means that you should
1359: press the ``carriage return'' key. Unfortunately, some output formats for
1360: this manual cannot show the difference between @kbd{this} and
1361: @code{this} which will make trying out the examples harder (but not
1362: impossible).
1363:
1364: Forth is an unusual language. It provides an interactive development
1365: environment which includes both an interpreter and compiler. Forth
1366: programming style encourages you to break a problem down into many
1367: @cindex factoring
1368: small fragments (@dfn{factoring}), and then to develop and test each
1369: fragment interactively. Forth advocates assert that breaking the
1370: edit-compile-test cycle used by conventional programming languages can
1371: lead to great productivity improvements.
1372:
1373: @menu
1374: * Introducing the Text Interpreter::
1375: * Stacks and Postfix notation::
1376: * Your first definition::
1377: * How does that work?::
1378: * Forth is written in Forth::
1379: * Review - elements of a Forth system::
1380: * Where to go next::
1381: * Exercises::
1382: @end menu
1383:
1384: @comment ----------------------------------------------
1385: @node Introducing the Text Interpreter, Stacks and Postfix notation, Introduction, Introduction
1386: @section Introducing the Text Interpreter
1387: @cindex text interpreter
1388: @cindex outer interpreter
1389:
1390: @c IMO this is too detailed and the pace is too slow for
1391: @c an introduction. If you know German, take a look at
1392: @c http://www.complang.tuwien.ac.at/anton/lvas/skriptum-stack.html
1393: @c to see how I do it - anton
1394:
1395: When you invoke the Forth image, you will see a startup banner printed
1396: and nothing else (if you have Gforth installed on your system, try
1397: invoking it now, by typing @kbd{gforth@key{RET}}). Forth is now running
1398: its command line interpreter, which is called the @dfn{Text Interpreter}
1399: (also known as the @dfn{Outer Interpreter}). (You will learn a lot
1400: about the text interpreter as you read through this chapter, but
1401: @pxref{The Text Interpreter} for more detail).
1402:
1403: Although it's not obvious, Forth is actually waiting for your
1404: input. Type a number and press the @key{RET} key:
1405:
1406: @example
1407: @kbd{45@key{RET}} ok
1408: @end example
1409:
1410: Rather than give you a prompt to invite you to input something, the text
1411: interpreter prints a status message @i{after} it has processed a line
1412: of input. The status message in this case (``@code{ ok}'' followed by
1413: carriage-return) indicates that the text interpreter was able to process
1414: all of your input successfully. Now type something illegal:
1415:
1416: @example
1417: @kbd{qwer341@key{RET}}
1418: :1: Undefined word
1419: qwer341
1420: ^^^^^^^
1421: $400D2BA8 Bounce
1422: $400DBDA8 no.extensions
1423: @end example
1424:
1425: The exact text, other than the ``Undefined word'' may differ slightly on
1426: your system, but the effect is the same; when the text interpreter
1427: detects an error, it discards any remaining text on a line, resets
1428: certain internal state and prints an error message. @xref{Error
1429: messages} for a detailed description of error messages.
1430:
1431: The text interpreter waits for you to press carriage-return, and then
1432: processes your input line. Starting at the beginning of the line, it
1433: breaks the line into groups of characters separated by spaces. For each
1434: group of characters in turn, it makes two attempts to do something:
1435:
1436: @itemize @bullet
1437: @item
1438: It tries to treat it as a command. It does this by searching a @dfn{name
1439: dictionary}. If the group of characters matches an entry in the name
1440: dictionary, the name dictionary provides the text interpreter with
1441: information that allows the text interpreter perform some actions. In
1442: Forth jargon, we say that the group
1443: @cindex word
1444: @cindex definition
1445: @cindex execution token
1446: @cindex xt
1447: of characters names a @dfn{word}, that the dictionary search returns an
1448: @dfn{execution token (xt)} corresponding to the @dfn{definition} of the
1449: word, and that the text interpreter executes the xt. Often, the terms
1450: @dfn{word} and @dfn{definition} are used interchangeably.
1451: @item
1452: If the text interpreter fails to find a match in the name dictionary, it
1453: tries to treat the group of characters as a number in the current number
1454: base (when you start up Forth, the current number base is base 10). If
1455: the group of characters legitimately represents a number, the text
1456: interpreter pushes the number onto a stack (we'll learn more about that
1457: in the next section).
1458: @end itemize
1459:
1460: If the text interpreter is unable to do either of these things with any
1461: group of characters, it discards the group of characters and the rest of
1462: the line, then prints an error message. If the text interpreter reaches
1463: the end of the line without error, it prints the status message ``@code{ ok}''
1464: followed by carriage-return.
1465:
1466: This is the simplest command we can give to the text interpreter:
1467:
1468: @example
1469: @key{RET} ok
1470: @end example
1471:
1472: The text interpreter did everything we asked it to do (nothing) without
1473: an error, so it said that everything is ``@code{ ok}''. Try a slightly longer
1474: command:
1475:
1476: @example
1477: @kbd{12 dup fred dup@key{RET}}
1478: :1: Undefined word
1479: 12 dup fred dup
1480: ^^^^
1481: $400D2BA8 Bounce
1482: $400DBDA8 no.extensions
1483: @end example
1484:
1485: When you press the carriage-return key, the text interpreter starts to
1486: work its way along the line:
1487:
1488: @itemize @bullet
1489: @item
1490: When it gets to the space after the @code{2}, it takes the group of
1491: characters @code{12} and looks them up in the name
1492: dictionary@footnote{We can't tell if it found them or not, but assume
1493: for now that it did not}. There is no match for this group of characters
1494: in the name dictionary, so it tries to treat them as a number. It is
1495: able to do this successfully, so it puts the number, 12, ``on the stack''
1496: (whatever that means).
1497: @item
1498: The text interpreter resumes scanning the line and gets the next group
1499: of characters, @code{dup}. It looks it up in the name dictionary and
1500: (you'll have to take my word for this) finds it, and executes the word
1501: @code{dup} (whatever that means).
1502: @item
1503: Once again, the text interpreter resumes scanning the line and gets the
1504: group of characters @code{fred}. It looks them up in the name
1505: dictionary, but can't find them. It tries to treat them as a number, but
1506: they don't represent any legal number.
1507: @end itemize
1508:
1509: At this point, the text interpreter gives up and prints an error
1510: message. The error message shows exactly how far the text interpreter
1511: got in processing the line. In particular, it shows that the text
1512: interpreter made no attempt to do anything with the final character
1513: group, @code{dup}, even though we have good reason to believe that the
1514: text interpreter would have no problem looking that word up and
1515: executing it a second time.
1516:
1517:
1518: @comment ----------------------------------------------
1519: @node Stacks and Postfix notation, Your first definition, Introducing the Text Interpreter, Introduction
1520: @section Stacks, postfix notation and parameter passing
1521: @cindex text interpreter
1522: @cindex outer interpreter
1523:
1524: In procedural programming languages (like C and Pascal), the
1525: building-block of programs is the @dfn{function} or @dfn{procedure}. These
1526: functions or procedures are called with @dfn{explicit parameters}. For
1527: example, in C we might write:
1528:
1529: @example
1530: total = total + new_volume(length,height,depth);
1531: @end example
1532:
1533: @noindent
1534: where new_volume is a function-call to another piece of code, and total,
1535: length, height and depth are all variables. length, height and depth are
1536: parameters to the function-call.
1537:
1538: In Forth, the equivalent of the function or procedure is the
1539: @dfn{definition} and parameters are implicitly passed between
1540: definitions using a shared stack that is visible to the
1541: programmer. Although Forth does support variables, the existence of the
1542: stack means that they are used far less often than in most other
1543: programming languages. When the text interpreter encounters a number, it
1544: will place (@dfn{push}) it on the stack. There are several stacks (the
1545: actual number is implementation-dependent ...) and the particular stack
1546: used for any operation is implied unambiguously by the operation being
1547: performed. The stack used for all integer operations is called the @dfn{data
1548: stack} and, since this is the stack used most commonly, references to
1549: ``the data stack'' are often abbreviated to ``the stack''.
1550:
1551: The stacks have a last-in, first-out (LIFO) organisation. If you type:
1552:
1553: @example
1554: @kbd{1 2 3@key{RET}} ok
1555: @end example
1556:
1557: Then this instructs the text interpreter to placed three numbers on the
1558: (data) stack. An analogy for the behaviour of the stack is to take a
1559: pack of playing cards and deal out the ace (1), 2 and 3 into a pile on
1560: the table. The 3 was the last card onto the pile (``last-in'') and if
1561: you take a card off the pile then, unless you're prepared to fiddle a
1562: bit, the card that you take off will be the 3 (``first-out''). The
1563: number that will be first-out of the stack is called the @dfn{top of
1564: stack}, which
1565: @cindex TOS definition
1566: is often abbreviated to @dfn{TOS}.
1567:
1568: To understand how parameters are passed in Forth, consider the
1569: behaviour of the definition @code{+} (pronounced ``plus''). You will not
1570: be surprised to learn that this definition performs addition. More
1571: precisely, it adds two number together and produces a result. Where does
1572: it get the two numbers from? It takes the top two numbers off the
1573: stack. Where does it place the result? On the stack. You can act-out the
1574: behaviour of @code{+} with your playing cards like this:
1575:
1576: @itemize @bullet
1577: @item
1578: Pick up two cards from the stack on the table
1579: @item
1580: Stare at them intently and ask yourself ``what @i{is} the sum of these two
1581: numbers''
1582: @item
1583: Decide that the answer is 5
1584: @item
1585: Shuffle the two cards back into the pack and find a 5
1586: @item
1587: Put a 5 on the remaining ace that's on the table.
1588: @end itemize
1589:
1590: If you don't have a pack of cards handy but you do have Forth running,
1591: you can use the definition @code{.s} to show the current state of the stack,
1592: without affecting the stack. Type:
1593:
1594: @example
1595: @kbd{clearstack 1 2 3@key{RET}} ok
1596: @kbd{.s@key{RET}} <3> 1 2 3 ok
1597: @end example
1598:
1599: The text interpreter looks up the word @code{clearstack} and executes
1600: it; it tidies up the stack and removes any entries that may have been
1601: left on it by earlier examples. The text interpreter pushes each of the
1602: three numbers in turn onto the stack. Finally, the text interpreter
1603: looks up the word @code{.s} and executes it. The effect of executing
1604: @code{.s} is to print the ``<3>'' (the total number of items on the stack)
1605: followed by a list of all the items on the stack; the item on the far
1606: right-hand side is the TOS.
1607:
1608: You can now type:
1609:
1610: @example
1611: @kbd{+ .s@key{RET}} <2> 1 5 ok
1612: @end example
1613:
1614: @noindent
1615: which is correct; there are now 2 items on the stack and the result of
1616: the addition is 5.
1617:
1618: If you're playing with cards, try doing a second addition: pick up the
1619: two cards, work out that their sum is 6, shuffle them into the pack,
1620: look for a 6 and place that on the table. You now have just one item on
1621: the stack. What happens if you try to do a third addition? Pick up the
1622: first card, pick up the second card -- ah! There is no second card. This
1623: is called a @dfn{stack underflow} and consitutes an error. If you try to
1624: do the same thing with Forth it will report an error (probably a Stack
1625: Underflow or an Invalid Memory Address error).
1626:
1627: The opposite situation to a stack underflow is a @dfn{stack overflow},
1628: which simply accepts that there is a finite amount of storage space
1629: reserved for the stack. To stretch the playing card analogy, if you had
1630: enough packs of cards and you piled the cards up on the table, you would
1631: eventually be unable to add another card; you'd hit the ceiling. Gforth
1632: allows you to set the maximum size of the stacks. In general, the only
1633: time that you will get a stack overflow is because a definition has a
1634: bug in it and is generating data on the stack uncontrollably.
1635:
1636: There's one final use for the playing card analogy. If you model your
1637: stack using a pack of playing cards, the maximum number of items on
1638: your stack will be 52 (I assume you didn't use the Joker). The maximum
1639: @i{value} of any item on the stack is 13 (the King). In fact, the only
1640: possible numbers are positive integer numbers 1 through 13; you can't
1641: have (for example) 0 or 27 or 3.52 or -2. If you change the way you
1642: think about some of the cards, you can accommodate different
1643: numbers. For example, you could think of the Jack as representing 0,
1644: the Queen as representing -1 and the King as representing -2. Your
1645: *range* remains unchanged (you can still only represent a total of 13
1646: numbers) but the numbers that you can represent are -2 through 10.
1647:
1648: In that analogy, the limit was the amount of information that a single
1649: stack entry could hold, and Forth has a similar limit. In Forth, the
1650: size of a stack entry is called a @dfn{cell}. The actual size of a cell is
1651: implementation dependent and affects the maximum value that a stack
1652: entry can hold. A Standard Forth provides a cell size of at least
1653: 16-bits, and most desktop systems use a cell size of 32-bits.
1654:
1655: Forth does not do any type checking for you, so you are free to
1656: manipulate and combine stack items in any way you wish. A convenient way
1657: of treating stack items is as 2's complement signed integers, and that
1658: is what Standard words like @code{+} do. Therefore you can type:
1659:
1660: @example
1661: @kbd{-5 12 + .s@key{RET}} <1> 7 ok
1662: @end example
1663:
1664: If you use numbers and definitions like @code{+} in order to turn Forth
1665: into a great big pocket calculator, you will realise that it's rather
1666: different from a normal calculator. Rather than typing 2 + 3 = you had
1667: to type 2 3 + (ignore the fact that you had to use @code{.s} to see the
1668: result). The terminology used to describe this difference is to say that
1669: your calculator uses @dfn{Infix Notation} (parameters and operators are
1670: mixed) whilst Forth uses @dfn{Postfix Notation} (parameters and
1671: operators are separate), also called @dfn{Reverse Polish Notation}.
1672:
1673: Whilst postfix notation might look confusing to begin with, it has
1674: several important advantages:
1675:
1676: @itemize @bullet
1677: @item
1678: it is unambiguous
1679: @item
1680: it is more concise
1681: @item
1682: it fits naturally with a stack-based system
1683: @end itemize
1684:
1685: To examine these claims in more detail, consider these sums:
1686:
1687: @example
1688: 6 + 5 * 4 =
1689: 4 * 5 + 6 =
1690: @end example
1691:
1692: If you're just learning maths or your maths is very rusty, you will
1693: probably come up with the answer 44 for the first and 26 for the
1694: second. If you are a bit of a whizz at maths you will remember the
1695: @i{convention} that multiplication takes precendence over addition, and
1696: you'd come up with the answer 26 both times. To explain the answer 26
1697: to someone who got the answer 44, you'd probably rewrite the first sum
1698: like this:
1699:
1700: @example
1701: 6 + (5 * 4) =
1702: @end example
1703:
1704: If what you really wanted was to perform the addition before the
1705: multiplication, you would have to use parentheses to force it.
1706:
1707: If you did the first two sums on a pocket calculator you would probably
1708: get the right answers, unless you were very cautious and entered them using
1709: these keystroke sequences:
1710:
1711: 6 + 5 = * 4 =
1712: 4 * 5 = + 6 =
1713:
1714: Postfix notation is unambiguous because the order that the operators
1715: are applied is always explicit; that also means that parentheses are
1716: never required. The operators are @i{active} (the act of quoting the
1717: operator makes the operation occur) which removes the need for ``=''.
1718:
1719: The sum 6 + 5 * 4 can be written (in postfix notation) in two
1720: equivalent ways:
1721:
1722: @example
1723: 6 5 4 * + or:
1724: 5 4 * 6 +
1725: @end example
1726:
1727: An important thing that you should notice about this notation is that
1728: the @i{order} of the numbers does not change; if you want to subtract
1729: 2 from 10 you type @code{10 2 -}.
1730:
1731: The reason that Forth uses postfix notation is very simple to explain: it
1732: makes the implementation extremely simple, and it follows naturally from
1733: using the stack as a mechanism for passing parameters. Another way of
1734: thinking about this is to realise that all Forth definitions are
1735: @i{active}; they execute as they are encountered by the text
1736: interpreter. The result of this is that the syntax of Forth is trivially
1737: simple.
1738:
1739:
1740:
1741: @comment ----------------------------------------------
1742: @node Your first definition, How does that work?, Stacks and Postfix notation, Introduction
1743: @section Your first Forth definition
1744: @cindex first definition
1745:
1746: Until now, the examples we've seen have been trivial; we've just been
1747: using Forth as a bigger-than-pocket calculator. Also, each calculation
1748: we've shown has been a ``one-off'' -- to repeat it we'd need to type it in
1749: again@footnote{That's not quite true. If you press the up-arrow key on
1750: your keyboard you should be able to scroll back to any earlier command,
1751: edit it and re-enter it.} In this section we'll see how to add new
1752: words to Forth's vocabulary.
1753:
1754: The easiest way to create a new word is to use a @dfn{colon
1755: definition}. We'll define a few and try them out before worrying too
1756: much about how they work. Try typing in these examples; be careful to
1757: copy the spaces accurately:
1758:
1759: @example
1760: : add-two 2 + . ;
1761: : greet ." Hello and welcome" ;
1762: : demo 5 add-two ;
1763: @end example
1764:
1765: @noindent
1766: Now try them out:
1767:
1768: @example
1769: @kbd{greet@key{RET}} Hello and welcome ok
1770: @kbd{greet greet@key{RET}} Hello and welcomeHello and welcome ok
1771: @kbd{4 add-two@key{RET}} 6 ok
1772: @kbd{demo@key{RET}} 7 ok
1773: @kbd{9 greet demo add-two@key{RET}} Hello and welcome7 11 ok
1774: @end example
1775:
1776: The first new thing that we've introduced here is the pair of words
1777: @code{:} and @code{;}. These are used to start and terminate a new
1778: definition, respectively. The first word after the @code{:} is the name
1779: for the new definition.
1780:
1781: As you can see from the examples, a definition is built up of words that
1782: have already been defined; Forth makes no distinction between
1783: definitions that existed when you started the system up, and those that
1784: you define yourself.
1785:
1786: The examples also introduce the words @code{.} (dot), @code{."}
1787: (dot-quote) and @code{dup} (dewp). Dot takes the value from the top of
1788: the stack and displays it. It's like @code{.s} except that it only
1789: displays the top item of the stack and it is destructive; after it has
1790: executed, the number is no longer on the stack. There is always one
1791: space printed after the number, and no spaces before it. Dot-quote
1792: defines a string (a sequence of characters) that will be printed when
1793: the word is executed. The string can contain any printable characters
1794: except @code{"}. A @code{"} has a special function; it is not a Forth
1795: word but it acts as a delimiter (the way that delimiters work is
1796: described in the next section). Finally, @code{dup} duplicates the value
1797: at the top of the stack. Try typing @code{5 dup .s} to see what it does.
1798:
1799: We already know that the text interpreter searches through the
1800: dictionary to locate names. If you've followed the examples earlier, you
1801: will already have a definition called @code{add-two}. Lets try modifying
1802: it by typing in a new definition:
1803:
1804: @example
1805: @kbd{: add-two dup . ." + 2 =" 2 + . ;@key{RET}} redefined add-two ok
1806: @end example
1807:
1808: Forth recognised that we were defining a word that already exists, and
1809: printed a message to warn us of that fact. Let's try out the new
1810: definition:
1811:
1812: @example
1813: @kbd{9 add-two@key{RET}} 9 + 2 =11 ok
1814: @end example
1815:
1816: @noindent
1817: All that we've actually done here, though, is to create a new
1818: definition, with a particular name. The fact that there was already a
1819: definition with the same name did not make any difference to the way
1820: that the new definition was created (except that Forth printed a warning
1821: message). The old definition of add-two still exists (try @code{demo}
1822: again to see that this is true). Any new definition will use the new
1823: definition of @code{add-two}, but old definitions continue to use the
1824: version that already existed at the time that they were @code{compiled}.
1825:
1826: Before you go on to the next section, try defining and redefining some
1827: words of your own.
1828:
1829: @comment ----------------------------------------------
1830: @node How does that work?, Forth is written in Forth, Your first definition, Introduction
1831: @section How does that work?
1832: @cindex parsing words
1833:
1834: @c That's pretty deep (IMO way too deep) for an introduction. - anton
1835:
1836: @c Is it a good idea to talk about the interpretation semantics of a
1837: @c number? We don't have an xt to go along with it. - anton
1838:
1839: @c Now that I have eliminated execution semantics, I wonder if it would not
1840: @c be better to keep them (or add run-time semantics), to make it easier to
1841: @c explain what compilation semantics usually does. - anton
1842:
1843: Now we're going to take another look at the definition of @code{add-two}
1844: from the previous section. From our knowledge of the way that the text
1845: interpreter works, we would have expected this result when we tried to
1846: define @code{add-two}:
1847:
1848: @example
1849: @kbd{: add-two 2 + . " ;@key{RET}}
1850: ^^^^^^^
1851: Error: Undefined word
1852: @end example
1853:
1854: The reason that this didn't happen is bound up in the way that @code{:}
1855: works. The word @code{:} does two special things. The first special
1856: thing that it does prevents the text interpreter from ever seeing the
1857: characters @code{add-two}. The text interpreter uses a variable called
1858: @cindex modifying >IN
1859: @code{>IN} (pronounced ''to-in'') to keep track of where it is in the
1860: input line. When it encounters the word @code{:} it behaves in exactly
1861: the same way as it does for any other word; it looks it up in the name
1862: dictionary, finds its xt and executes it. When @code{:} executes, it
1863: looks at the input buffer, finds the word @code{add-two} and advances the
1864: value of @code{>IN} to point past it. It then does some other stuff
1865: associated with creating the new definition (including creating an entry
1866: for @code{add-two} in the name dictionary). When the execution of @code{:}
1867: completes, control returns to the text interpreter, which is oblivious
1868: to the fact that it has been tricked into ignoring part of the input
1869: line.
1870:
1871: @cindex parsing words
1872: Words like @code{:} -- words that advance the value of @code{>IN} and so
1873: prevent the text interpreter from acting on the whole of the input line
1874: -- are called @dfn{parsing words}.
1875:
1876: @cindex @code{state} - effect on the text interpreter
1877: @cindex text interpreter - effect of state
1878: The second special thing that @code{:} does is change the value of a
1879: variable called @code{state}, which affects the way that the text
1880: interpreter behaves. When Gforth starts up, @code{state} has the value
1881: 0, and the text interpreter is said to be @dfn{interpreting}. During a
1882: colon definition (started with @code{:}), @code{state} is set to -1 and
1883: the text interpreter is said to be @dfn{compiling}. The word @code{;}
1884: ends the definition -- one of the things that it does is to change the
1885: value of @code{state} back to 0.
1886:
1887: We have already seen how the text interpreter behaves when it is
1888: interpreting; it looks for each character sequence in the dictionary,
1889: finds its xt and executes it, or it converts it to a number and pushes
1890: it onto the stack, or it fails to do either and generates an error.
1891:
1892: When the text interpreter is compiling, its behaviour is slightly
1893: different; it still looks for each character sequence in the dictionary
1894: and finds it, or converts it to a number, or fails to do either and
1895: generates an error. But instead of the execution token of a word it
1896: finds and executes the compilation token. For most words executing the
1897: compilation token results in laying down (@dfn{compiling}) the execution
1898: token, i.e., some magic to make that xt or number get executed or pushed
1899: at a later time; at the time that @code{add-two} is
1900: @dfn{executed}. Therefore, when you execute @code{add-two} its
1901: @dfn{run-time effect} is exactly the same as if you had typed @code{2 +
1902: .} outside of a definition, and pressed carriage-return.
1903:
1904: In Forth, every word or number can be described in terms of two
1905: properties:
1906:
1907: @itemize @bullet
1908: @item
1909: Its @dfn{interpretation semantics}, represented by the execution token.
1910: @item
1911: Its @dfn{compilation semantics}, represented by the compilation token.
1912: @end itemize
1913:
1914: The value of @code{state} determines whether the text interpreter will
1915: use the compilation or interpretation semantics of a word or number that
1916: it encounters.
1917:
1918: @itemize @bullet
1919: @item
1920: @cindex interpretation semantics
1921: When the text interpreter encounters a word or number in @dfn{interpret}
1922: state, it performs the @dfn{interpretation semantics} of the word or
1923: number.
1924: @item
1925: @cindex compilation semantics
1926: When the text interpreter encounters a word or number in @dfn{compile}
1927: state, it performs the @dfn{compilation semantics} of the word or
1928: number.
1929: @end itemize
1930:
1931: @noindent
1932: Numbers are always treated in a fixed way:
1933:
1934: @itemize @bullet
1935: @item
1936: When the number is @dfn{interpreted}, its behaviour is to push the number onto the stack.
1937: @item
1938: When the number is @dfn{compiled}, a piece of code is appended to the
1939: current definition that pushes the number when it runs. (In other words,
1940: the compilation semantics of a number are to postpone its interpretation
1941: semantics until the run-time of the definition that it is being compiled
1942: into.)
1943: @end itemize
1944:
1945: The behaviour of a word is not so regular, but most have @i{default
1946: compilation semantics} which means that they behave like this:
1947:
1948: @itemize @bullet
1949: @item
1950: The @dfn{interpretation semantics} of the word are to do something useful.
1951: @item
1952: The @dfn{compilation semantics} of the word are to append its
1953: @dfn{interpretation semantics} to the current definition (so that its
1954: run-time behaviour is to do something useful).
1955: @end itemize
1956:
1957: @cindex immediate words
1958: The actual behaviour of any particular word depends upon the way in
1959: which it was defined. When the text interpreter finds the word in the
1960: name dictionary, it not only retrieves the xt for the word, it also
1961: retrieves some flags: the @dfn{compile-only} flag and the @dfn{immediate
1962: flag}. The compile-only flag indicates that the word has no
1963: interpretation semantics (the run-time behaviour for the default
1964: compilation semantics is not affected by this flag, however); any
1965: attempt to interpret a word that has the compile-only flag set will
1966: generate an error (for example, @code{IF} has no interpretation
1967: semantics). The immediate flag changes the compilation semantics of the
1968: word; if it is set, the compilation semantics are equal to the
1969: interpretation semantics (again ignoring the compile-only flag). it. In
1970: other words, these so-called @dfn{immediate} words behave like this:
1971:
1972: @itemize @bullet
1973: @item
1974: The @dfn{interpretation semantics} of the word are to do something useful.
1975: @item
1976: The @dfn{compilation semantics} of the word are to do something useful
1977: (and actually the same thing); i.e., it is executed during compilation.
1978: @end itemize
1979:
1980: This example shows the difference between an immediate and a
1981: non-immediate word:
1982:
1983: @example
1984: : show-state state @@ . ;
1985: : show-state-now show-state ; immediate
1986: : word1 show-state ;
1987: : word2 show-state-now ;
1988: @end example
1989:
1990: The word @code{immediate} after the definition of @code{show-state-now}
1991: makes that word an immediate word. These definitions introduce a new
1992: word: @code{@@} (pronounced ``fetch''). This word fetches the value of a
1993: variable, and leaves it on the stack. Therefore, the behaviour of
1994: @code{show-state} is to print a number that represents the current value
1995: of @code{state}.
1996:
1997: When you execute @code{word1}, it prints the number 0, indicating that
1998: the system is interpreting. When the text interpreter compiled the
1999: definition of @code{word1}, it encountered @code{show-state} whose
2000: compilation semantics are to append its interpretation semantics to the
2001: current definition. When you execute @code{word1}, it performs the
2002: interpretation semantics of @code{show-state}. At the time that @code{word1}
2003: (and therefore @code{show-state}) are executed, the system is
2004: interpreting.
2005:
2006: When you pressed @key{RET} after entering the definition of @code{word2},
2007: you should have seen the number -1 printed, followed by ``@code{
2008: ok}''. When the text interpreter compiled the definition of
2009: @code{word2}, it encountered @code{show-state-now}, an immediate word,
2010: whose compilation semantics are therefore to perform its interpretation
2011: semantics. It is executed straight away (even before the text
2012: interpreter has moved on to process another group of characters; the
2013: @code{;} in this example). The effect of executing it are to display the
2014: value of @code{state} @i{at the time that the definition of}
2015: @code{word2} @i{is being defined}. Printing -1 demonstrates that the
2016: system is compiling at this time. If you execute @code{word2} it does
2017: nothing at all.
2018:
2019: @cindex @code{."}, how it works
2020: Before leaving the subject of immediate words, consider the behaviour of
2021: @code{."} in the definition of @code{greet}, in the previous
2022: section. This word is both a parsing word and an immediate word. Notice
2023: that there is a space between @code{."} and the start of the text
2024: @code{Hello and welcome}, but that there is no space between the last
2025: letter of @code{welcome} and the @code{"} character. The reason for this
2026: is that @code{."} is a Forth word; it must have a space after it so that
2027: the text interpreter can identify it. The @code{"} is not a Forth word;
2028: it is a @dfn{delimiter}. The examples earlier show that, when the string
2029: is displayed, there is neither a space before the @code{H} nor after the
2030: @code{e}. Since @code{."} is an immediate word, it executes at the time
2031: that @code{greet} is defined. When it executes, its behaviour is to
2032: search forward in the input line looking for the delimiter. When it
2033: finds the delimiter, it updates @code{>IN} to point past the
2034: delimiter. It also compiles some magic code into the definition of
2035: @code{greet}; the xt of a run-time routine that prints a text string. It
2036: compiles the string @code{Hello and welcome} into memory so that it is
2037: available to be printed later. When the text interpreter gains control,
2038: the next word it finds in the input stream is @code{;} and so it
2039: terminates the definition of @code{greet}.
2040:
2041:
2042: @comment ----------------------------------------------
2043: @node Forth is written in Forth, Review - elements of a Forth system, How does that work?, Introduction
2044: @section Forth is written in Forth
2045: @cindex structure of Forth programs
2046:
2047: When you start up a Forth compiler, a large number of definitions
2048: already exist. In Forth, you develop a new application using bottom-up
2049: programming techniques to create new definitions that are defined in
2050: terms of existing definitions. As you create each definition you can
2051: test and debug it interactively.
2052:
2053: If you have tried out the examples in this section, you will probably
2054: have typed them in by hand; when you leave Gforth, your definitions will
2055: be lost. You can avoid this by using a text editor to enter Forth source
2056: code into a file, and then loading code from the file using
2057: @code{include} (@xref{Forth source files}). A Forth source file is
2058: processed by the text interpreter, just as though you had typed it in by
2059: hand@footnote{Actually, there are some subtle differences -- see
2060: @ref{The Text Interpreter}.}.
2061:
2062: Gforth also supports the traditional Forth alternative to using text
2063: files for program entry (@xref{Blocks}).
2064:
2065: In common with many, if not most, Forth compilers, most of Gforth is
2066: actually written in Forth. All of the @file{.fs} files in the
2067: installation directory@footnote{For example,
2068: @file{/usr/local/share/gforth...}} are Forth source files, which you can
2069: study to see examples of Forth programming.
2070:
2071: Gforth maintains a history file that records every line that you type to
2072: the text interpreter. This file is preserved between sessions, and is
2073: used to provide a command-line recall facility. If you enter long
2074: definitions by hand, you can use a text editor to paste them out of the
2075: history file into a Forth source file for reuse at a later time
2076: (@pxref{Command-line editing} for more information).
2077:
2078:
2079: @comment ----------------------------------------------
2080: @node Review - elements of a Forth system, Where to go next, Forth is written in Forth, Introduction
2081: @section Review - elements of a Forth system
2082: @cindex elements of a Forth system
2083:
2084: To summarise this chapter:
2085:
2086: @itemize @bullet
2087: @item
2088: Forth programs use @dfn{factoring} to break a problem down into small
2089: fragments called @dfn{words} or @dfn{definitions}.
2090: @item
2091: Forth program development is an interactive process.
2092: @item
2093: The main command loop that accepts input, and controls both
2094: interpretation and compilation, is called the @dfn{text interpreter}
2095: (also known as the @dfn{outer interpreter}).
2096: @item
2097: Forth has a very simple syntax, consisting of words and numbers
2098: separated by spaces or carriage-return characters. Any additional syntax
2099: is imposed by @dfn{parsing words}.
2100: @item
2101: Forth uses a stack to pass parameters between words. As a result, it
2102: uses postfix notation.
2103: @item
2104: To use a word that has previously been defined, the text interpreter
2105: searches for the word in the @dfn{name dictionary}.
2106: @item
2107: Words have @dfn{interpretation semantics} and @dfn{compilation semantics}.
2108: @item
2109: The text interpreter uses the value of @code{state} to select between
2110: the use of the @dfn{interpretation semantics} and the @dfn{compilation
2111: semantics} of a word that it encounters.
2112: @item
2113: The relationship between the @dfn{interpretation semantics} and
2114: @dfn{compilation semantics} for a word
2115: depend upon the way in which the word was defined (for example, whether
2116: it is an @dfn{immediate} word).
2117: @item
2118: Forth definitions can be implemented in Forth (called @dfn{high-level
2119: definitions}) or in some other way (usually a lower-level language and
2120: as a result often called @dfn{low-level definitions}, @dfn{code
2121: definitions} or @dfn{primitives}).
2122: @item
2123: Many Forth systems are implemented mainly in Forth.
2124: @end itemize
2125:
2126:
2127: @comment ----------------------------------------------
2128: @node Where to go next,Exercises,Review - elements of a Forth system, Introduction
2129: @section Where To Go Next
2130: @cindex where to go next
2131:
2132: Amazing as it may seem, if you have read (and understood) this far, you
2133: know almost all the fundamentals about the inner workings of a Forth
2134: system. You certainly know enough to be able to read and understand the
2135: rest of this manual and the ANS Forth document, to learn more about the
2136: facilities that Forth in general and Gforth in particular provide. Even
2137: scarier, you know almost enough to implement your own Forth system.
2138: However, that's not a good idea just yet... better to try writing some
2139: programs in Gforth.
2140:
2141: Forth has such a rich vocabulary that it can be hard to know where to
2142: start in learning it. This section suggests a few sets of words that are
2143: enough to write small but useful programs. Use the word index in this
2144: document to learn more about each word, then try it out and try to write
2145: small definitions using it. Start by experimenting with these words:
2146:
2147: @itemize @bullet
2148: @item
2149: Arithmetic: @code{+ - * / /MOD */ ABS INVERT}
2150: @item
2151: Comparison: @code{MIN MAX =}
2152: @item
2153: Logic: @code{AND OR XOR NOT}
2154: @item
2155: Stack manipulation: @code{DUP DROP SWAP OVER}
2156: @item
2157: Loops and decisions: @code{IF ELSE ENDIF ?DO I LOOP}
2158: @item
2159: Input/Output: @code{. ." EMIT CR KEY}
2160: @item
2161: Defining words: @code{: ; CREATE}
2162: @item
2163: Memory allocation words: @code{ALLOT ,}
2164: @item
2165: Tools: @code{SEE WORDS .S MARKER}
2166: @end itemize
2167:
2168: When you have mastered those, go on to:
2169:
2170: @itemize @bullet
2171: @item
2172: More defining words: @code{VARIABLE CONSTANT VALUE TO CREATE DOES>}
2173: @item
2174: Memory access: @code{@@ !}
2175: @end itemize
2176:
2177: When you have mastered these, there's nothing for it but to read through
2178: the whole of this manual and find out what you've missed.
2179:
2180: @comment ----------------------------------------------
2181: @node Exercises, ,Where to go next, Introduction
2182: @section Exercises
2183: @cindex exercises
2184:
2185: TODO: provide a set of programming excercises linked into the stuff done
2186: already and into other sections of the manual. Provide solutions to all
2187: the exercises in a .fs file in the distribution.
2188:
2189: @c Get some inspiration from Starting Forth and Kelly&Spies.
2190:
2191: @c excercises:
2192: @c 1. take inches and convert to feet and inches.
2193: @c 2. take temperature and convert from fahrenheight to celcius;
2194: @c may need to care about symmetric vs floored??
2195: @c 3. take input line and do character substitution
2196: @c to encipher or decipher
2197: @c 4. as above but work on a file for in and out
2198: @c 5. take input line and convert to pig-latin
2199: @c
2200: @c thing of sets of things to exercise then come up with
2201: @c problems that need those things.
2202:
2203:
2204: @c ******************************************************************
2205: @node Words, Error messages, Introduction, Top
2206: @chapter Forth Words
2207: @cindex words
2208:
2209: @menu
2210: * Notation::
2211: * Comments::
2212: * Boolean Flags::
2213: * Arithmetic::
2214: * Stack Manipulation::
2215: * Memory::
2216: * Control Structures::
2217: * Defining Words::
2218: * The Text Interpreter::
2219: * Tokens for Words::
2220: * Word Lists::
2221: * Environmental Queries::
2222: * Files::
2223: * Blocks::
2224: * Other I/O::
2225: * Programming Tools::
2226: * Assembler and Code Words::
2227: * Threading Words::
2228: * Locals::
2229: * Structures::
2230: * Object-oriented Forth::
2231: * Passing Commands to the OS::
2232: * Miscellaneous Words::
2233: @end menu
2234:
2235: @node Notation, Comments, Words, Words
2236: @section Notation
2237: @cindex notation of glossary entries
2238: @cindex format of glossary entries
2239: @cindex glossary notation format
2240: @cindex word glossary entry format
2241:
2242: The Forth words are described in this section in the glossary notation
2243: that has become a de-facto standard for Forth texts, i.e.,
2244:
2245: @format
2246: @i{word} @i{Stack effect} @i{wordset} @i{pronunciation}
2247: @end format
2248: @i{Description}
2249:
2250: @table @var
2251: @item word
2252: The name of the word.
2253:
2254: @item Stack effect
2255: @cindex stack effect
2256: The stack effect is written in the notation @code{@i{before} --
2257: @i{after}}, where @i{before} and @i{after} describe the top of
2258: stack entries before and after the execution of the word. The rest of
2259: the stack is not touched by the word. The top of stack is rightmost,
2260: i.e., a stack sequence is written as it is typed in. Note that Gforth
2261: uses a separate floating point stack, but a unified stack
2262: notation. Also, return stack effects are not shown in @i{stack
2263: effect}, but in @i{Description}. The name of a stack item describes
2264: the type and/or the function of the item. See below for a discussion of
2265: the types.
2266:
2267: All words have two stack effects: A compile-time stack effect and a
2268: run-time stack effect. The compile-time stack-effect of most words is
2269: @i{ -- }. If the compile-time stack-effect of a word deviates from
2270: this standard behaviour, or the word does other unusual things at
2271: compile time, both stack effects are shown; otherwise only the run-time
2272: stack effect is shown.
2273:
2274: @cindex pronounciation of words
2275: @item pronunciation
2276: How the word is pronounced.
2277:
2278: @cindex wordset
2279: @item wordset
2280: The ANS Forth standard is divided into several word sets. A standard
2281: system need not support all of them. Therefore, in theory, the fewer
2282: word sets your program uses the more portable it will be. However, we
2283: suspect that most ANS Forth systems on personal machines will feature
2284: all word sets. Words that are not defined in ANS Forth have
2285: @code{gforth} or @code{gforth-internal} as word set. @code{gforth}
2286: describes words that will work in future releases of Gforth;
2287: @code{gforth-internal} words are more volatile. Environmental query
2288: strings are also displayed like words; you can recognize them by the
2289: @code{environment} in the word set field.
2290:
2291: @item Description
2292: A description of the behaviour of the word.
2293: @end table
2294:
2295: @cindex types of stack items
2296: @cindex stack item types
2297: The type of a stack item is specified by the character(s) the name
2298: starts with:
2299:
2300: @table @code
2301: @item f
2302: @cindex @code{f}, stack item type
2303: Boolean flags, i.e. @code{false} or @code{true}.
2304: @item c
2305: @cindex @code{c}, stack item type
2306: Char
2307: @item w
2308: @cindex @code{w}, stack item type
2309: Cell, can contain an integer or an address
2310: @item n
2311: @cindex @code{n}, stack item type
2312: signed integer
2313: @item u
2314: @cindex @code{u}, stack item type
2315: unsigned integer
2316: @item d
2317: @cindex @code{d}, stack item type
2318: double sized signed integer
2319: @item ud
2320: @cindex @code{ud}, stack item type
2321: double sized unsigned integer
2322: @item r
2323: @cindex @code{r}, stack item type
2324: Float (on the FP stack)
2325: @item a-
2326: @cindex @code{a_}, stack item type
2327: Cell-aligned address
2328: @item c-
2329: @cindex @code{c_}, stack item type
2330: Char-aligned address (note that a Char may have two bytes in Windows NT)
2331: @item f-
2332: @cindex @code{f_}, stack item type
2333: Float-aligned address
2334: @item df-
2335: @cindex @code{df_}, stack item type
2336: Address aligned for IEEE double precision float
2337: @item sf-
2338: @cindex @code{sf_}, stack item type
2339: Address aligned for IEEE single precision float
2340: @item xt
2341: @cindex @code{xt}, stack item type
2342: Execution token, same size as Cell
2343: @item wid
2344: @cindex @code{wid}, stack item type
2345: Word list ID, same size as Cell
2346: @item f83name
2347: @cindex @code{f83name}, stack item type
2348: Pointer to a name structure
2349: @item "
2350: @cindex @code{"}, stack item type
2351: string in the input stream (not on the stack). The terminating character
2352: is a blank by default. If it is not a blank, it is shown in @code{<>}
2353: quotes.
2354: @end table
2355:
2356: @node Comments, Boolean Flags, Notation, Words
2357: @section Comments
2358: @cindex comments
2359:
2360: Forth supports two styles of comment; the traditional @i{in-line} comment,
2361: @code{(} and its modern cousin, the @i{comment to end of line}; @code{\}.
2362:
2363: doc-(
2364: doc-\
2365: doc-\G
2366:
2367: @node Boolean Flags, Arithmetic, Comments, Words
2368: @section Boolean Flags
2369: @cindex Boolean flags
2370:
2371: A Boolean flag is cell-sized. A cell with all bits clear represents the
2372: flag @code{false} and a flag with all bits set represents the flag
2373: @code{true}. Words that check a flag (for example, @code{IF}) will treat
2374: a cell that has @i{any} bit set as @code{true}.
2375:
2376: doc-true
2377: doc-false
2378: doc-on
2379: doc-off
2380:
2381: @node Arithmetic, Stack Manipulation, Boolean Flags, Words
2382: @section Arithmetic
2383: @cindex arithmetic words
2384:
2385: @cindex division with potentially negative operands
2386: Forth arithmetic is not checked, i.e., you will not hear about integer
2387: overflow on addition or multiplication, you may hear about division by
2388: zero if you are lucky. The operator is written after the operands, but
2389: the operands are still in the original order. I.e., the infix @code{2-1}
2390: corresponds to @code{2 1 -}. Forth offers a variety of division
2391: operators. If you perform division with potentially negative operands,
2392: you do not want to use @code{/} or @code{/mod} with its undefined
2393: behaviour, but rather @code{fm/mod} or @code{sm/mod} (probably the
2394: former, @pxref{Mixed precision}).
2395: @comment TODO discuss the different division forms and the std approach
2396:
2397: @menu
2398: * Single precision::
2399: * Bitwise operations::
2400: * Double precision:: Double-cell integer arithmetic
2401: * Numeric comparison::
2402: * Mixed precision:: Operations with single and double-cell integers
2403: * Floating Point::
2404: @end menu
2405:
2406: @node Single precision, Bitwise operations, Arithmetic, Arithmetic
2407: @subsection Single precision
2408: @cindex single precision arithmetic words
2409:
2410: By default, numbers in Forth are single-precision integers that are 1
2411: cell in size. They can be signed or unsigned, depending upon how you
2412: treat them. @xref{Number Conversion} for the rules used by the text
2413: interpreter for recognising single-precision integers.
2414:
2415: doc-+
2416: doc-1+
2417: doc--
2418: doc-1-
2419: doc-*
2420: doc-/
2421: doc-mod
2422: doc-/mod
2423: doc-negate
2424: doc-abs
2425: doc-min
2426: doc-max
2427: doc-d>s
2428: doc-floored
2429:
2430: @node Bitwise operations, Double precision, Single precision, Arithmetic
2431: @subsection Bitwise operations
2432: @cindex bitwise operation words
2433:
2434: doc-and
2435: doc-or
2436: doc-xor
2437: doc-invert
2438: doc-lshift
2439: doc-rshift
2440: doc-2*
2441: doc-d2*
2442: doc-2/
2443: doc-d2/
2444:
2445: @node Double precision, Numeric comparison, Bitwise operations, Arithmetic
2446: @subsection Double precision
2447: @cindex double precision arithmetic words
2448:
2449: @xref{Number Conversion} for the rules used by the text interpreter for
2450: recognising double-precision integers.
2451:
2452: A double precision number is represented by a cell pair, with the most
2453: significant cell at the TOS. It is trivial to convert an unsigned
2454: single to an (unsigned) double; simply push a @code{0} onto the
2455: TOS. Since numbers are represented by Gforth using 2's complement
2456: arithmetic, converting a signed single to a (signed) double requires
2457: sign-extension across the most significant cell. This can be achieved
2458: using @code{s>d}. The moral of the story is that you cannot convert a
2459: number without knowing whether it represents an unsigned or a
2460: signed number.
2461:
2462: doc-s>d
2463: doc-d+
2464: doc-d-
2465: doc-dnegate
2466: doc-dabs
2467: doc-dmin
2468: doc-dmax
2469:
2470: @node Numeric comparison, Mixed precision, Double precision, Arithmetic
2471: @subsection Numeric comparison
2472: @cindex numeric comparison words
2473:
2474: doc-<
2475: doc-<=
2476: doc-<>
2477: doc-=
2478: doc->
2479: doc->=
2480:
2481: doc-0<
2482: doc-0<=
2483: doc-0<>
2484: doc-0=
2485: doc-0>
2486: doc-0>=
2487:
2488: doc-u<
2489: doc-u<=
2490: @c TODO why u<> and u= ... they are the same as <> and =
2491: @c commented them out because they are unnecessary
2492: @c doc-u<>
2493: @c doc-u=
2494: doc-u>
2495: doc-u>=
2496:
2497: doc-within
2498:
2499: doc-d<
2500: doc-d<=
2501: doc-d<>
2502: doc-d=
2503: doc-d>
2504: doc-d>=
2505:
2506: doc-d0<
2507: doc-d0<=
2508: doc-d0<>
2509: doc-d0=
2510: doc-d0>
2511: doc-d0>=
2512:
2513: doc-du<
2514: doc-du<=
2515: @c doc-du<>
2516: @c doc-du=
2517: doc-du>
2518: doc-du>=
2519:
2520: @node Mixed precision, Floating Point, Numeric comparison, Arithmetic
2521: @subsection Mixed precision
2522: @cindex mixed precision arithmetic words
2523:
2524: doc-m+
2525: doc-*/
2526: doc-*/mod
2527: doc-m*
2528: doc-um*
2529: doc-m*/
2530: doc-um/mod
2531: doc-fm/mod
2532: doc-sm/rem
2533:
2534: @node Floating Point, , Mixed precision, Arithmetic
2535: @subsection Floating Point
2536: @cindex floating point arithmetic words
2537:
2538: @xref{Number Conversion} for the rules used by the text interpreter for
2539: recognising floating-point numbers.
2540:
2541: Gforth has a separate floating point
2542: stack, but the documentation uses the unified notation.
2543:
2544: @cindex floating-point arithmetic, pitfalls
2545: Floating point numbers have a number of unpleasant surprises for the
2546: unwary (e.g., floating point addition is not associative) and even a few
2547: for the wary. You should not use them unless you know what you are doing
2548: or you don't care that the results you get are totally bogus. If you
2549: want to learn about the problems of floating point numbers (and how to
2550: avoid them), you might start with @cite{David Goldberg, What Every
2551: Computer Scientist Should Know About Floating-Point Arithmetic, ACM
2552: Computing Surveys 23(1):5@minus{}48, March 1991}
2553: (@url{http://www.validgh.com/goldberg/paper.ps}).
2554:
2555: doc-d>f
2556: doc-f>d
2557: doc-f+
2558: doc-f-
2559: doc-f*
2560: doc-f/
2561: doc-fnegate
2562: doc-fabs
2563: doc-fmax
2564: doc-fmin
2565: doc-floor
2566: doc-fround
2567: doc-f**
2568: doc-fsqrt
2569: doc-fexp
2570: doc-fexpm1
2571: doc-fln
2572: doc-flnp1
2573: doc-flog
2574: doc-falog
2575: doc-f2*
2576: doc-f2/
2577: doc-1/f
2578: doc-precision
2579: doc-set-precision
2580:
2581: @cindex angles in trigonometric operations
2582: @cindex trigonometric operations
2583: Angles in floating point operations are given in radians (a full circle
2584: has 2 pi radians).
2585:
2586: doc-fsin
2587: doc-fcos
2588: doc-fsincos
2589: doc-ftan
2590: doc-fasin
2591: doc-facos
2592: doc-fatan
2593: doc-fatan2
2594: doc-fsinh
2595: doc-fcosh
2596: doc-ftanh
2597: doc-fasinh
2598: doc-facosh
2599: doc-fatanh
2600: doc-pi
2601:
2602: @cindex equality of floats
2603: @cindex floating-point comparisons
2604: One particular problem with floating-point arithmetic is that comparison
2605: for equality often fails when you would expect it to succeed. For this
2606: reason approximate equality is often preferred (but you still have to
2607: know what you are doing). The comparison words are:
2608:
2609: doc-f~rel
2610: doc-f~abs
2611: doc-f=
2612: doc-f~
2613: doc-f<>
2614:
2615: doc-f<
2616: doc-f<=
2617: doc-f>
2618: doc-f>=
2619:
2620: doc-f0<
2621: doc-f0<=
2622: doc-f0<>
2623: doc-f0=
2624: doc-f0>
2625: doc-f0>=
2626:
2627:
2628: @node Stack Manipulation, Memory, Arithmetic, Words
2629: @section Stack Manipulation
2630: @cindex stack manipulation words
2631:
2632: @cindex floating-point stack in the standard
2633: Gforth maintains a number of separate stacks:
2634:
2635: @cindex data stack
2636: @cindex parameter stack
2637: @itemize @bullet
2638: @item
2639: A data stack (also known as the @dfn{parameter stack}) -- for
2640: characters, cells, addresses, and double cells.
2641:
2642: @cindex floating-point stack
2643: @item
2644: A floating point stack -- for floating point numbers.
2645:
2646: @cindex return stack
2647: @item
2648: A return stack -- for storing the return addresses of colon
2649: definitions and other (non-FP) data.
2650:
2651: @cindex locals stack
2652: @item
2653: A locals stack for storing local variables.
2654: @end itemize
2655:
2656: @menu
2657: * Data stack::
2658: * Floating point stack::
2659: * Return stack::
2660: * Locals stack::
2661: * Stack pointer manipulation::
2662: @end menu
2663:
2664: @node Data stack, Floating point stack, Stack Manipulation, Stack Manipulation
2665: @subsection Data stack
2666: @cindex data stack manipulation words
2667: @cindex stack manipulations words, data stack
2668:
2669: doc-drop
2670: doc-nip
2671: doc-dup
2672: doc-over
2673: doc-tuck
2674: doc-swap
2675: doc-pick
2676: doc-rot
2677: doc--rot
2678: doc-?dup
2679: doc-roll
2680: doc-2drop
2681: doc-2nip
2682: doc-2dup
2683: doc-2over
2684: doc-2tuck
2685: doc-2swap
2686: doc-2rot
2687:
2688: @node Floating point stack, Return stack, Data stack, Stack Manipulation
2689: @subsection Floating point stack
2690: @cindex floating-point stack manipulation words
2691: @cindex stack manipulation words, floating-point stack
2692:
2693: Whilst every sane Forth has a separate floating-point stack, it is not
2694: strictly required; an ANS Forth system could theoretically keep
2695: floating-point numbers on the data stack. As an additional difficulty,
2696: you don't know how many cells a floating-point number takes. It is
2697: reportedly possible to write words in a way that they work also for a
2698: unified stack model, but we do not recommend trying it. Instead, just
2699: say that your program has an environmental dependency on a separate
2700: floating-point stack.
2701:
2702: doc-floating-stack
2703:
2704: doc-fdrop
2705: doc-fnip
2706: doc-fdup
2707: doc-fover
2708: doc-ftuck
2709: doc-fswap
2710: doc-fpick
2711: doc-frot
2712:
2713: @node Return stack, Locals stack, Floating point stack, Stack Manipulation
2714: @subsection Return stack
2715: @cindex return stack manipulation words
2716: @cindex stack manipulation words, return stack
2717:
2718: @cindex return stack and locals
2719: @cindex locals and return stack
2720: A Forth system is allowed to keep local variables on the
2721: return stack. This is reasonable, as local variables usually eliminate
2722: the need to use the return stack explicitly. So, if you want to produce
2723: a standard compliant program and you are using local variables in a
2724: word, forget about return stack manipulations in that word (refer to the
2725: standard document for the exact rules).
2726:
2727: doc->r
2728: doc-r>
2729: doc-r@
2730: doc-rdrop
2731: doc-2>r
2732: doc-2r>
2733: doc-2r@
2734: doc-2rdrop
2735:
2736: @node Locals stack, Stack pointer manipulation, Return stack, Stack Manipulation
2737: @subsection Locals stack
2738:
2739: @comment TODO
2740:
2741: @node Stack pointer manipulation, , Locals stack, Stack Manipulation
2742: @subsection Stack pointer manipulation
2743: @cindex stack pointer manipulation words
2744:
2745: doc-sp0
2746: doc-sp@
2747: doc-sp!
2748: doc-fp0
2749: doc-fp@
2750: doc-fp!
2751: doc-rp0
2752: doc-rp@
2753: doc-rp!
2754: doc-lp0
2755: doc-lp@
2756: doc-lp!
2757:
2758: @node Memory, Control Structures, Stack Manipulation, Words
2759: @section Memory
2760: @cindex memory words
2761:
2762: @menu
2763: * Memory model::
2764: * Dictionary allocation::
2765: * Heap Allocation::
2766: * Memory Access::
2767: * Address arithmetic::
2768: * Memory Blocks::
2769: @end menu
2770:
2771: @node Memory model, Dictionary allocation, Memory, Memory
2772: @subsection ANS Forth and Gforth memory models
2773:
2774: @c The ANS Forth description is a mess (e.g., is the heap part of
2775: @c the dictionary?), so let's not stick to closely with it.
2776:
2777: ANS Forth considers a Forth system as consisting of several memories, of
2778: which only @dfn{data space} is managed and accessible with the memory
2779: words. Memory not necessarily in data space includes the stacks, the
2780: code (called code space) and the headers (called name space). In Gforth
2781: everything is in data space, but the code for the primitives is usually
2782: read-only.
2783:
2784: Data space is divided into a number of areas: The (data space portion of
2785: the) dictionary@footnote{Sometimes, the term @dfn{dictionary} is used to
2786: refer to the search data structure embodied in word lists and headers,
2787: because it is used for looking up names, just as you would in a
2788: conventional dictionary.}, the heap, and a number of system-allocated
2789: buffers.
2790:
2791: In ANS Forth data space is also divided into contiguous regions. You
2792: can only use address arithmetic within a contiguous region, not between
2793: them. Usually each allocation gives you one contiguous region, but the
2794: dictionary allocation words have additional rules (@pxref{Dictionary
2795: allocation}).
2796:
2797: Gforth provides one big address space, and address arithmetic can be
2798: performed between any addresses. However, in the dictionary headers or
2799: code are interleaved with data, so almost the only contiguous data space
2800: regions there are those described by ANS Forth as contiguous; but you
2801: can be sure that the dictionary is allocated towards increasing
2802: addresses even between contiguous regions. The memory order of
2803: allocations in the heap is platform-dependent (and possibly different
2804: from one run to the next).
2805:
2806: @subsubsection ANS Forth dictionary details
2807:
2808: @c !! I have deleted some of the stuff this section refers to - anton
2809:
2810: This section is just informative, you can skip it if you are in a hurry.
2811:
2812: When you create a colon definition, the text interpreter compiles the
2813: code for the definition into the code space and compiles the name
2814: of the definition into the header space, together with other
2815: information about the definition (such as its execution token).
2816:
2817: When you create a variable, the execution of @code{variable} will
2818: compile some code, assign one cell in data space, and compile the name
2819: of the variable into the header space.
2820:
2821: @cindex memory regions - relationship between them
2822: ANS Forth does not specify the relationship between the three memory
2823: regions, and specifies that a Standard program must not access code or
2824: data space directly -- it may only access data space directly. In
2825: addition, the Standard defines what relationships you may and may not
2826: rely on when allocating regions in data space. These constraints are
2827: simply a reflection of the many diverse techniques that are used to
2828: implement Forth systems; understanding and following the requirements of
2829: the Standard allows you to write portable programs -- programs that run
2830: in the same way on any of these diverse systems. Another way of looking
2831: at this is to say that ANS Forth was designed to permit compliant Forth
2832: systems to be implemented in many diverse ways.
2833:
2834: @cindex memory regions - how they are assigned
2835: Here are some examples of ways in which name, code and data spaces
2836: might be assigned in different Forth implementations:
2837:
2838: @itemize @bullet
2839: @item
2840: For a Forth system that runs from RAM under a general-purpose operating
2841: system, it can be convenient to interleave name, code and data spaces in
2842: a single contiguous memory region. This organisation can be
2843: memory-efficient (for example, because the relationship between the name
2844: dictionary entry and the associated code space entry can be
2845: implicit, rather than requiring an explicit memory pointer to reference
2846: from the header space and the code space). This is the
2847: organisation used by Gforth, as this example@footnote{The addresses
2848: in the example have been truncated to fit it onto the page, and the
2849: addresses and data shown will not match the output from your system} shows:
2850: @example
2851: hex
2852: variable fred 123456 fred !
2853: variable jim abcd jim !
2854: : foo + / - ;
2855: ' fred 10 - 50 dump
2856: ..80: 5C 46 0E 40 84 66 72 65 - 64 20 20 20 20 20 20 20 \F.@.fred
2857: ..90: D0 9B 04 08 00 00 00 00 - 56 34 12 00 80 46 0E 40 ........V4...F.@
2858: ..A0: 83 6A 69 6D 20 20 20 20 - D0 9B 04 08 00 00 00 00 .jim ........
2859: ..B0: CD AB 00 00 9C 46 0E 40 - 83 66 6F 6F 20 20 20 20 .....F.@.foo
2860: ..C0: 80 9B 04 08 00 00 00 00 - E4 2E 05 08 0C 2F 05 08 ............./..
2861: @end example
2862:
2863: @item
2864: For a high-performance system running on a modern RISC processor with a
2865: modified Harvard architecture (one that has a unified main memory but
2866: separate instruction and data caches), it is desirable to separate
2867: processor instructions from processor data. This encourages a high cache
2868: density and therefore a high cache hit rate. The Forth code space
2869: is not necessarily made up entirely of processor instructions; its
2870: nature is dependent upon the Forth implementation.
2871:
2872: @item
2873: A Forth compiler that runs on a segmented 8086 processor could be
2874: designed to interleave the name, code and data spaces within a single
2875: 64Kbyte segment. A more common implementation choice is to use a
2876: separate 64Kbyte segment for each region, which provides more memory
2877: overall but provides an address map in which only the data space is
2878: accessible.
2879:
2880: @item
2881: Microprocessors exist that run Forth (or many of the primitives required
2882: to implement the Forth virtual machine efficiently) directly. On these
2883: processors, the relationship between name, code and data spaces may be
2884: imposed as a side-effect of the architecture of the processor.
2885:
2886: @item
2887: A Forth compiler that executes from ROM on an embedded system needs its
2888: data space separated from the name and code spaces so that the data
2889: space can be mapped to a RAM area.
2890:
2891: @item
2892: A Forth compiler that runs on an embedded system may have a requirement
2893: for a small memory footprint. On such a system it can be useful to
2894: separate the header space from the data and code spaces; once the
2895: application has been compiled, the header space is no longer
2896: required@footnote{more strictly speaking, most applications can be
2897: designed so that this is the case}. The header space can be deleted
2898: entirely, or could be stored in memory on a remote @i{host} system for
2899: debug and development purposes. In the latter case, the compiler running
2900: on the @i{target} system could implement a protocol across a
2901: communication link that would allow it to interrogate the header space.
2902: @end itemize
2903:
2904:
2905: @node Dictionary allocation, Heap Allocation, Memory model, Memory
2906: @subsection Dictionary allocation
2907: @cindex reserving data space
2908: @cindex data space - reserving some
2909:
2910: Dictionary allocation is a stack-oriented allocation scheme, i.e., if
2911: you want to deallocate X, you also deallocate everything
2912: allocated after X.
2913:
2914: The allocations using the words below are contiguous and grow the region
2915: towards increasing addresses. Other words that allocate dictionary
2916: memory of any kind (i.e., defining words including @code{:noname}) end
2917: the contiguous region and start a new one.
2918:
2919: In ANS Forth only @code{create}d words are guaranteed to produce an
2920: address that is the start of the following contiguous region. In
2921: particular, the cell allocated by @code{variable} is not guaranteed to
2922: be contiguous with following @code{allot}ed memory.
2923:
2924: You can deallocate memory by using @code{allot} with a negative argument
2925: (with some restrictions, see @code{allot}). For larger deallocations use
2926: @code{marker}.
2927:
2928:
2929: doc-here
2930: doc-unused
2931: doc-allot
2932: doc-c,
2933: doc-f,
2934: doc-,
2935: doc-2,
2936: @cindex user space
2937: doc-udp
2938: doc-uallot
2939:
2940: Memory accesses have to be aligned (@pxref{Address arithmetic}). So of
2941: course you should allocate memory in an aligned way, too. I.e., before
2942: allocating allocating a cell, @code{here} must be cell-aligned, etc.
2943: The words below align @code{here} if it is not already. Basically it is
2944: only already aligned for a type, if the last allocation was a multiple
2945: of the size of this type and if @code{here} was aligned for this type
2946: before.
2947:
2948: After freshly @code{create}ing a word, @code{here} is @code{align}ed in
2949: ANS Forth (@code{maxalign}ed in Gforth).
2950:
2951: doc-align
2952: doc-falign
2953: doc-sfalign
2954: doc-dfalign
2955: doc-maxalign
2956: doc-cfalign
2957:
2958:
2959: @node Heap Allocation, Memory Access, Dictionary allocation, Memory
2960: @subsection Heap allocation
2961: @cindex heap allocation
2962: @cindex dynamic allocation of memory
2963: @cindex memory-allocation word set
2964:
2965: Heap allocation supports deallocation of allocated memory in any
2966: order. Dictionary allocation is not affected by it (i.e., it does not
2967: end a contiguous region). In Gforth, these words are implemented using
2968: the standard C library calls malloc(), free() and resize().
2969:
2970: doc-allocate
2971: doc-free
2972: doc-resize
2973:
2974:
2975: @node Memory Access, Address arithmetic, Heap Allocation, Memory
2976: @subsection Memory Access
2977: @cindex memory access words
2978:
2979: doc-@
2980: doc-!
2981: doc-+!
2982: doc-c@
2983: doc-c!
2984: doc-2@
2985: doc-2!
2986: doc-f@
2987: doc-f!
2988: doc-sf@
2989: doc-sf!
2990: doc-df@
2991: doc-df!
2992:
2993: @node Address arithmetic, Memory Blocks, Memory Access, Memory
2994: @subsection Address arithmetic
2995: @cindex address arithmetic words
2996:
2997: Address arithmetic is the foundation on which data structures like
2998: arrays, records (@pxref{Structures}) and objects (@pxref{Object-oriented
2999: Forth}) are built.
3000:
3001: ANS Forth does not specify the sizes of the data types. Instead, it
3002: offers a number of words for computing sizes and doing address
3003: arithmetic. Address arithmetic is performed in terms of address units
3004: (aus); on most systems the address unit is one byte. Note that a
3005: character may have more than one au, so @code{chars} is no noop (on
3006: systems where it is a noop, it compiles to nothing).
3007:
3008: @cindex alignment of addresses for types
3009: ANS Forth also defines words for aligning addresses for specific
3010: types. Many computers require that accesses to specific data types
3011: must only occur at specific addresses; e.g., that cells may only be
3012: accessed at addresses divisible by 4. Even if a machine allows unaligned
3013: accesses, it can usually perform aligned accesses faster.
3014:
3015: For the performance-conscious: alignment operations are usually only
3016: necessary during the definition of a data structure, not during the
3017: (more frequent) accesses to it.
3018:
3019: ANS Forth defines no words for character-aligning addresses. This is not
3020: an oversight, but reflects the fact that addresses that are not
3021: char-aligned have no use in the standard and therefore will not be
3022: created.
3023:
3024: @cindex @code{CREATE} and alignment
3025: ANS Forth guarantees that addresses returned by @code{CREATE}d words
3026: are cell-aligned; in addition, Gforth guarantees that these addresses
3027: are aligned for all purposes.
3028:
3029: Note that the ANS Forth word @code{char} has nothing to do with address
3030: arithmetic.
3031:
3032: doc-chars
3033: doc-char+
3034: doc-cells
3035: doc-cell+
3036: doc-cell
3037: doc-aligned
3038: doc-floats
3039: doc-float+
3040: doc-float
3041: doc-faligned
3042: doc-sfloats
3043: doc-sfloat+
3044: doc-sfaligned
3045: doc-dfloats
3046: doc-dfloat+
3047: doc-dfaligned
3048: doc-maxaligned
3049: doc-cfaligned
3050: doc-address-unit-bits
3051:
3052: @node Memory Blocks, , Address arithmetic, Memory
3053: @subsection Memory Blocks
3054: @cindex memory block words
3055: @cindex character strings - moving and copying
3056:
3057: Memory blocks often represent character strings; @xref{String Formats}
3058: for ways of storing character strings in memory. @xref{Displaying
3059: characters and strings} for other string-processing words.
3060:
3061: Some of these words work on address units. Others work on character
3062: units (increments of @code{CHAR}), and expect a @code{CHAR}-aligned
3063: address. Choose the correct operation depending upon your data type.
3064:
3065: When copying characters between overlapping memory regions, choose
3066: carefully between @code{cmove} and @code{cmove>}.
3067:
3068: You can only use any of these words @i{portably} to access data space.
3069:
3070: @comment TODO - think the naming of the arguments is wrong for move
3071: @comment well, really it seems to be the Standard that's wrong; it
3072: @comment describes MOVE as a word that requires a CELL-aligned source
3073: @comment and destination address but a xtranfer count that need not
3074: @comment be a multiple of CELL.
3075: doc-move
3076: doc-erase
3077: doc-cmove
3078: doc-cmove>
3079: doc-fill
3080: doc-blank
3081: doc-compare
3082: doc-search
3083: doc--trailing
3084: doc-/string
3085:
3086: @comment TODO examples
3087:
3088:
3089: @node Control Structures, Defining Words, Memory, Words
3090: @section Control Structures
3091: @cindex control structures
3092:
3093: Control structures in Forth cannot be used interpretively, only in a
3094: colon definition@footnote{To be precise, they have no interpretation
3095: semantics (@pxref{Interpretation and Compilation Semantics}).}. We do
3096: not like this limitation, but have not seen a satisfying way around it
3097: yet, although many schemes have been proposed.
3098:
3099: @menu
3100: * Selection:: IF ... ELSE ... ENDIF
3101: * Simple Loops:: BEGIN ...
3102: * Counted Loops:: DO
3103: * Arbitrary control structures::
3104: * Calls and returns::
3105: * Exception Handling::
3106: @end menu
3107:
3108: @node Selection, Simple Loops, Control Structures, Control Structures
3109: @subsection Selection
3110: @cindex selection control structures
3111: @cindex control structures for selection
3112:
3113: @c what's the purpose of all these @i? Maybe we should define a macro
3114: @c so we can produce logical markup. - anton
3115:
3116: @cindex @code{IF} control structure
3117: @example
3118: @i{flag}
3119: IF
3120: @i{code}
3121: ENDIF
3122: @end example
3123: @noindent
3124:
3125: @var{code} is executed if @var{flag} is non-zero (that's truth as far as
3126: @code{IF} etc. are concerned).
3127:
3128: @example
3129: @i{flag}
3130: IF
3131: @i{code1}
3132: ELSE
3133: @i{code2}
3134: ENDIF
3135: @end example
3136:
3137: If @var{flag} is true, perform @var{code1}, otherwise @var{code2}.
3138:
3139: You can use @code{THEN} instead of @code{ENDIF}. Indeed, @code{THEN} is
3140: standard, and @code{ENDIF} is not, although it is quite popular. We
3141: recommend using @code{ENDIF}, because it is less confusing for people
3142: who also know other languages (and is not prone to reinforcing negative
3143: prejudices against Forth in these people). Adding @code{ENDIF} to a
3144: system that only supplies @code{THEN} is simple:
3145: @example
3146: : ENDIF POSTPONE THEN ; immediate
3147: @end example
3148:
3149: [According to @cite{Webster's New Encyclopedic Dictionary}, @dfn{then
3150: (adv.)} has the following meanings:
3151: @quotation
3152: ... 2b: following next after in order ... 3d: as a necessary consequence
3153: (if you were there, then you saw them).
3154: @end quotation
3155: Forth's @code{THEN} has the meaning 2b, whereas @code{THEN} in Pascal
3156: and many other programming languages has the meaning 3d.]
3157:
3158: Gforth also provides the words @code{?DUP-IF} and @code{?DUP-0=-IF}, so
3159: you can avoid using @code{?dup}. Using these alternatives is also more
3160: efficient than using @code{?dup}. Definitions in ANS Forth
3161: for @code{ENDIF}, @code{?DUP-IF} and @code{?DUP-0=-IF} are provided in
3162: @file{compat/control.fs}.
3163:
3164: @cindex @code{CASE} control structure
3165: @example
3166: @i{n}
3167: CASE
3168: @i{n1} OF @i{code1} ENDOF
3169: @i{n2} OF @i{code2} ENDOF
3170: @dots{}
3171: ENDCASE
3172: @end example
3173:
3174: Executes the first @i{codei}, where the @i{ni} is equal to
3175: @i{n}. A default case can be added by simply writing the code after
3176: the last @code{ENDOF}. It may use @i{n}, which is on top of the stack,
3177: but must not consume it.
3178:
3179: @node Simple Loops, Counted Loops, Selection, Control Structures
3180: @subsection Simple Loops
3181: @cindex simple loops
3182: @cindex loops without count
3183:
3184: @cindex @code{WHILE} loop
3185: @example
3186: BEGIN
3187: @i{code1}
3188: @i{flag}
3189: WHILE
3190: @i{code2}
3191: REPEAT
3192: @end example
3193:
3194: @i{code1} is executed and @i{flag} is computed. If it is true,
3195: @i{code2} is executed and the loop is restarted; If @i{flag} is
3196: false, execution continues after the @code{REPEAT}.
3197:
3198: @cindex @code{UNTIL} loop
3199: @example
3200: BEGIN
3201: @i{code}
3202: @i{flag}
3203: UNTIL
3204: @end example
3205:
3206: @i{code} is executed. The loop is restarted if @code{flag} is false.
3207:
3208: @cindex endless loop
3209: @cindex loops, endless
3210: @example
3211: BEGIN
3212: @i{code}
3213: AGAIN
3214: @end example
3215:
3216: This is an endless loop.
3217:
3218: @node Counted Loops, Arbitrary control structures, Simple Loops, Control Structures
3219: @subsection Counted Loops
3220: @cindex counted loops
3221: @cindex loops, counted
3222: @cindex @code{DO} loops
3223:
3224: The basic counted loop is:
3225: @example
3226: @i{limit} @i{start}
3227: ?DO
3228: @i{body}
3229: LOOP
3230: @end example
3231:
3232: This performs one iteration for every integer, starting from @i{start}
3233: and up to, but excluding @i{limit}. The counter, or @i{index}, can be
3234: accessed with @code{i}. For example, the loop:
3235: @example
3236: 10 0 ?DO
3237: i .
3238: LOOP
3239: @end example
3240: @noindent
3241: prints @code{0 1 2 3 4 5 6 7 8 9}
3242:
3243: The index of the innermost loop can be accessed with @code{i}, the index
3244: of the next loop with @code{j}, and the index of the third loop with
3245: @code{k}.
3246:
3247: doc-i
3248: doc-j
3249: doc-k
3250:
3251: The loop control data are kept on the return stack, so there are some
3252: restrictions on mixing return stack accesses and counted loop words. In
3253: particuler, if you put values on the return stack outside the loop, you
3254: cannot read them inside the loop@footnote{well, not in a way that is
3255: portable.}. If you put values on the return stack within a loop, you
3256: have to remove them before the end of the loop and before accessing the
3257: index of the loop.
3258:
3259: There are several variations on the counted loop:
3260:
3261: @itemize @bullet
3262: @item
3263: @code{LEAVE} leaves the innermost counted loop immediately; execution
3264: continues after the associated @code{LOOP} or @code{NEXT}. For example:
3265:
3266: @example
3267: 10 0 ?DO i DUP . 3 = IF LEAVE THEN LOOP
3268: @end example
3269: prints @code{0 1 2 3}
3270:
3271:
3272: @item
3273: @code{UNLOOP} prepares for an abnormal loop exit, e.g., via
3274: @code{EXIT}. @code{UNLOOP} removes the loop control parameters from the
3275: return stack so @code{EXIT} can get to its return address. For example:
3276:
3277: @example
3278: : demo 10 0 ?DO i DUP . 3 = IF UNLOOP EXIT THEN LOOP ." Done" ;
3279: @end example
3280: prints @code{0 1 2 3}
3281:
3282:
3283: @item
3284: If @i{start} is greater than @i{limit}, a @code{?DO} loop is entered
3285: (and @code{LOOP} iterates until they become equal by wrap-around
3286: arithmetic). This behaviour is usually not what you want. Therefore,
3287: Gforth offers @code{+DO} and @code{U+DO} (as replacements for
3288: @code{?DO}), which do not enter the loop if @i{start} is greater than
3289: @i{limit}; @code{+DO} is for signed loop parameters, @code{U+DO} for
3290: unsigned loop parameters.
3291:
3292: @item
3293: @code{?DO} can be replaced by @code{DO}. @code{DO} always enters
3294: the loop, independent of the loop parameters. Do not use @code{DO}, even
3295: if you know that the loop is entered in any case. Such knowledge tends
3296: to become invalid during maintenance of a program, and then the
3297: @code{DO} will make trouble.
3298:
3299: @item
3300: @code{LOOP} can be replaced with @code{@i{n} +LOOP}; this updates the
3301: index by @i{n} instead of by 1. The loop is terminated when the border
3302: between @i{limit-1} and @i{limit} is crossed. E.g.:
3303:
3304: @example
3305: 4 0 +DO i . 2 +LOOP
3306: @end example
3307: @noindent
3308: prints @code{0 2}
3309:
3310: @example
3311: 4 1 +DO i . 2 +LOOP
3312: @end example
3313: @noindent
3314: prints @code{1 3}
3315:
3316:
3317: @cindex negative increment for counted loops
3318: @cindex counted loops with negative increment
3319: The behaviour of @code{@i{n} +LOOP} is peculiar when @i{n} is negative:
3320:
3321: @example
3322: -1 0 ?DO i . -1 +LOOP
3323: @end example
3324: @noindent
3325: prints @code{0 -1}
3326:
3327: @example
3328: 0 0 ?DO i . -1 +LOOP
3329: @end example
3330: prints nothing.
3331:
3332: Therefore we recommend avoiding @code{@i{n} +LOOP} with negative
3333: @i{n}. One alternative is @code{@i{u} -LOOP}, which reduces the
3334: index by @i{u} each iteration. The loop is terminated when the border
3335: between @i{limit+1} and @i{limit} is crossed. Gforth also provides
3336: @code{-DO} and @code{U-DO} for down-counting loops. E.g.:
3337:
3338: @example
3339: -2 0 -DO i . 1 -LOOP
3340: @end example
3341: @noindent
3342: prints @code{0 -1}
3343:
3344: @example
3345: -1 0 -DO i . 1 -LOOP
3346: @end example
3347: @noindent
3348: prints @code{0}
3349:
3350: @example
3351: 0 0 -DO i . 1 -LOOP
3352: @end example
3353: @noindent
3354: prints nothing.
3355:
3356: @end itemize
3357:
3358: Unfortunately, @code{+DO}, @code{U+DO}, @code{-DO}, @code{U-DO} and
3359: @code{-LOOP} are not defined in ANS Forth. However, an implementation
3360: for these words that uses only standard words is provided in
3361: @file{compat/loops.fs}.
3362:
3363:
3364: @cindex @code{FOR} loops
3365: Another counted loop is:
3366: @example
3367: @i{n}
3368: FOR
3369: @i{body}
3370: NEXT
3371: @end example
3372: This is the preferred loop of native code compiler writers who are too
3373: lazy to optimize @code{?DO} loops properly. This loop structure is not
3374: defined in ANS Forth. In Gforth, this loop iterates @i{n+1} times;
3375: @code{i} produces values starting with @i{n} and ending with 0. Other
3376: Forth systems may behave differently, even if they support @code{FOR}
3377: loops. To avoid problems, don't use @code{FOR} loops.
3378:
3379: @node Arbitrary control structures, Calls and returns, Counted Loops, Control Structures
3380: @subsection Arbitrary control structures
3381: @cindex control structures, user-defined
3382:
3383: @cindex control-flow stack
3384: ANS Forth permits and supports using control structures in a non-nested
3385: way. Information about incomplete control structures is stored on the
3386: control-flow stack. This stack may be implemented on the Forth data
3387: stack, and this is what we have done in Gforth.
3388:
3389: @cindex @code{orig}, control-flow stack item
3390: @cindex @code{dest}, control-flow stack item
3391: An @i{orig} entry represents an unresolved forward branch, a @i{dest}
3392: entry represents a backward branch target. A few words are the basis for
3393: building any control structure possible (except control structures that
3394: need storage, like calls, coroutines, and backtracking).
3395:
3396: doc-if
3397: doc-ahead
3398: doc-then
3399: doc-begin
3400: doc-until
3401: doc-again
3402: doc-cs-pick
3403: doc-cs-roll
3404:
3405: The Standard words @code{CS-PICK} and @code{CS-ROLL} allow you to
3406: manipulate the control-flow stack in a portable way. Without them, you
3407: would need to know how many stack items are occupied by a control-flow
3408: entry (many systems use one cell. In Gforth they currently take three,
3409: but this may change in the future).
3410:
3411: Some standard control structure words are built from these words:
3412:
3413: doc-else
3414: doc-while
3415: doc-repeat
3416:
3417: Gforth adds some more control-structure words:
3418:
3419: doc-endif
3420: doc-?dup-if
3421: doc-?dup-0=-if
3422:
3423: Counted loop words constitute a separate group of words:
3424:
3425: doc-?do
3426: doc-+do
3427: doc-u+do
3428: doc--do
3429: doc-u-do
3430: doc-do
3431: doc-for
3432: doc-loop
3433: doc-+loop
3434: doc--loop
3435: doc-next
3436: doc-leave
3437: doc-?leave
3438: doc-unloop
3439: doc-done
3440:
3441: The standard does not allow using @code{CS-PICK} and @code{CS-ROLL} on
3442: @i{do-sys}. Gforth allows it, but it's your job to ensure that for
3443: every @code{?DO} etc. there is exactly one @code{UNLOOP} on any path
3444: through the definition (@code{LOOP} etc. compile an @code{UNLOOP} on the
3445: fall-through path). Also, you have to ensure that all @code{LEAVE}s are
3446: resolved (by using one of the loop-ending words or @code{DONE}).
3447:
3448: Another group of control structure words are:
3449:
3450: doc-case
3451: doc-endcase
3452: doc-of
3453: doc-endof
3454:
3455: @i{case-sys} and @i{of-sys} cannot be processed using @code{CS-PICK} and
3456: @code{CS-ROLL}.
3457:
3458: @subsubsection Programming Style
3459:
3460: In order to ensure readability we recommend that you do not create
3461: arbitrary control structures directly, but define new control structure
3462: words for the control structure you want and use these words in your
3463: program. For example, instead of writing:
3464:
3465: @example
3466: BEGIN
3467: ...
3468: IF [ 1 CS-ROLL ]
3469: ...
3470: AGAIN THEN
3471: @end example
3472:
3473: @noindent
3474: we recommend defining control structure words, e.g.,
3475:
3476: @example
3477: : WHILE ( DEST -- ORIG DEST )
3478: POSTPONE IF
3479: 1 CS-ROLL ; immediate
3480:
3481: : REPEAT ( orig dest -- )
3482: POSTPONE AGAIN
3483: POSTPONE THEN ; immediate
3484: @end example
3485:
3486: @noindent
3487: and then using these to create the control structure:
3488:
3489: @example
3490: BEGIN
3491: ...
3492: WHILE
3493: ...
3494: REPEAT
3495: @end example
3496:
3497: That's much easier to read, isn't it? Of course, @code{REPEAT} and
3498: @code{WHILE} are predefined, so in this example it would not be
3499: necessary to define them.
3500:
3501: @node Calls and returns, Exception Handling, Arbitrary control structures, Control Structures
3502: @subsection Calls and returns
3503: @cindex calling a definition
3504: @cindex returning from a definition
3505:
3506: @cindex recursive definitions
3507: A definition can be called simply be writing the name of the definition
3508: to be called. Normally a definition is invisible during its own
3509: definition. If you want to write a directly recursive definition, you
3510: can use @code{recursive} to make the current definition visible, or
3511: @code{recurse} to call the current definition directly.
3512:
3513: doc-recursive
3514: doc-recurse
3515:
3516: @comment TODO add example of the two recursion methods
3517: @quotation
3518: @progstyle
3519: I prefer using @code{recursive} to @code{recurse}, because calling the
3520: definition by name is more descriptive (if the name is well-chosen) than
3521: the somewhat cryptic @code{recurse}. E.g., in a quicksort
3522: implementation, it is much better to read (and think) ``now sort the
3523: partitions'' than to read ``now do a recursive call''.
3524: @end quotation
3525:
3526: For mutual recursion, use @code{Defer}red words, like this:
3527:
3528: @example
3529: Defer foo
3530:
3531: : bar ( ... -- ... )
3532: ... foo ... ;
3533:
3534: :noname ( ... -- ... )
3535: ... bar ... ;
3536: IS foo
3537: @end example
3538:
3539: Deferred words are discussed in more detail in @ref{Simple
3540: Defining Words}.
3541:
3542: The current definition returns control to the calling definition when
3543: the end of the definition is reached or @code{EXIT} is encountered.
3544:
3545: doc-exit
3546: doc-;s
3547:
3548: @node Exception Handling, , Calls and returns, Control Structures
3549: @subsection Exception Handling
3550: @cindex exceptions
3551:
3552: If your program detects a fatal error condition, the simplest action
3553: that it can take is to @code{quit}. This resets the return stack and
3554: restarts the text interpreter, but does not print any error message.
3555:
3556: The next stage in severity is to execute @code{abort}, which has the
3557: same effect as @code{quit}, with the addition that it resets the data
3558: stack.
3559:
3560: A slightly more sophisticated approach is use use @code{abort"}, which
3561: compiles a string to be used as an error message and does a conditional
3562: @code{abort} at run-time. For example:
3563:
3564: @example
3565: @kbd{: checker abort" That flag was true" ." A false flag" ;@key{RET}} ok
3566: @kbd{0 checker@key{RET}} A false flag ok
3567: @kbd{1 checker@key{RET}}
3568: :1: That flag was true
3569: 1 checker
3570: ^^^^^^^
3571: $400D1648 throw
3572: $400E4660
3573: @end example
3574:
3575: These simple techniques allow a program to react to a fatal error
3576: condition, but they are not exactly user-friendly. The ANS Forth
3577: Exception word set provides the pair of words @code{throw} and
3578: @code{catch}, which can be used to provide sophisticated error-handling.
3579:
3580: @code{catch} has a similar behaviour to @code{execute}, in that it takes
3581: an @i{xt} as a parameter and starts execution of the xt. However,
3582: before passing control to the xt, @code{catch} pushes an
3583: @dfn{exception frame} onto the @dfn{exception stack}. This exception
3584: frame is used to restore the system to a known state if a detected error
3585: occurs during the execution of the xt. A typical way to use @code{catch}
3586: would be:
3587:
3588: @example
3589: ... ['] foo catch IF ...
3590: @end example
3591:
3592: @c TOS is undefined. - anton
3593: Whilst @code{foo} executes, it can call other words to any level of
3594: nesting, as usual. If @code{foo} (and all the words that it calls)
3595: execute successfully, control will ultimately pass to the word following
3596: the @code{catch}, and there will be a 0 at TOS. However, if any word
3597: detects an error, it can terminate the execution of @code{foo} by
3598: pushing a non-zero error code onto the stack and then performing a
3599: @code{throw}. The execution of @code{throw} will pass control to the
3600: word following the @code{catch}, but this time the TOS will hold the
3601: error code. Therefore, the @code{IF} in the example can be used to
3602: determine whether @code{foo} executed successfully.
3603:
3604: This simple example shows how you can use @code{throw} and @code{catch}
3605: to ``take over'' exception handling from the system:
3606: @example
3607: : my-div ['] / catch if ." DIVIDE ERROR" else ." OK.. " . then ;
3608: @end example
3609:
3610: The next example is more sophisticated and shows a multi-level
3611: @code{throw} and @code{catch}. To understand this example, start at the
3612: definition of @code{top-level} and work backwards:
3613:
3614: @example
3615: : lowest-level ( -- c )
3616: key dup 27 = if
3617: 1 throw \ ESCAPE key pressed
3618: else
3619: ." lowest-level successfull" CR
3620: then
3621: ;
3622:
3623: : lower-level ( -- c )
3624: lowest-level
3625: \ at this level consider a CTRL-U to be a fatal error
3626: dup 21 = if \ CTRL-U
3627: 2 throw
3628: else
3629: ." lower-level successfull" CR
3630: then
3631: ;
3632:
3633: : low-level ( -- c )
3634: ['] lower-level catch
3635: ?dup if
3636: \ error occurred - do we recognise it?
3637: dup 1 = if
3638: \ ESCAPE key pressed.. pretend it was an E
3639: [char] E
3640: else throw \ propogate the error upwards
3641: then
3642: then
3643: ." low-level successfull" CR
3644: ;
3645:
3646: : top-level ( -- )
3647: CR ['] low-level catch \ CATCH is used like EXECUTE
3648: ?dup if \ error occurred..
3649: ." Error " . ." occurred - contact your supplier"
3650: else
3651: ." The '" emit ." ' key was pressed" CR
3652: then
3653: ;
3654: @end example
3655:
3656: The ANS Forth document assigns @code{throw} codes thus:
3657:
3658: @itemize @bullet
3659: @item
3660: codes in the range -1 -- -255 are reserved to be assigned by the
3661: Standard. Assignments for codes in the range -1 -- -58 are currently
3662: documented in the Standard. In particular, @code{-1 throw} is equivalent
3663: to @code{abort} and @code{-2 throw} is equivalent to @code{abort"}.
3664: @item
3665: codes in the range -256 -- -4095 are reserved to be assigned by the system.
3666: @item
3667: all other codes may be assigned by programs.
3668: @end itemize
3669:
3670: Gforth provides the word @code{exception} as a mechanism for assigning
3671: system throw codes to applications. This allows multiple applications to
3672: co-exist in memory without any clash of @code{throw} codes. A definition
3673: of @code{exception} in ANS Forth is provided in
3674: @file{compat/exception.fs}.
3675:
3676: doc-quit
3677: doc-abort
3678: doc-abort"
3679:
3680: doc-catch
3681: doc-throw
3682: doc---exception-exception
3683:
3684:
3685: @c -------------------------------------------------------------
3686: @node Defining Words, The Text Interpreter, Control Structures, Words
3687: @section Defining Words
3688: @cindex defining words
3689:
3690: @menu
3691: * Simple Defining Words:: Variables, values and constants
3692: * Colon Definitions::
3693: * User-defined Defining Words::
3694: * Supplying names::
3695: * Interpretation and Compilation Semantics::
3696: @end menu
3697:
3698: @node Simple Defining Words, Colon Definitions, Defining Words, Defining Words
3699: @subsection Simple Defining Words
3700: @cindex simple defining words
3701: @cindex defining words, simple
3702:
3703: @c split this section?
3704:
3705: Defining words are used to create new entries in the dictionary. The
3706: simplest defining word is @code{CREATE}. @code{CREATE} is used like
3707: this:
3708:
3709: @example
3710: CREATE new-word1
3711: @end example
3712:
3713: @code{CREATE} is a parsing word that generates a dictionary entry for
3714: @code{new-word1}. When @code{new-word1} is executed, all that it does is
3715: leave an address on the stack. The address represents the value of
3716: the data space pointer (@code{HERE}) at the time that @code{new-word1}
3717: was defined. Therefore, @code{CREATE} is a way of associating a name
3718: with the address of a region of memory.
3719:
3720: doc-create
3721:
3722: By extending this example to reserve some memory in data space, we end
3723: up with a @i{variable}. Here are two different ways to do it:
3724:
3725: @example
3726: CREATE new-word2 1 cells allot \ reserve 1 cell - initial value undefined
3727: CREATE new-word3 4 , \ reserve 1 cell and initialise it (to 4)
3728: @end example
3729:
3730: The variable can be examined and modified using @code{@@} (``fetch'') and
3731: @code{!} (``store'') like this:
3732:
3733: @example
3734: new-word2 @@ . \ get address, fetch from it and display
3735: 1234 new-word2 ! \ new value, get address, store to it
3736: @end example
3737:
3738: As a final refinement, the whole code sequence can be wrapped up in a
3739: defining word (pre-empting the subject of the next section), making it
3740: easier to create new variables:
3741:
3742: @example
3743: : myvariable ( "name" -- a-addr ) CREATE 0 , ;
3744:
3745: myvariable foo
3746: myvariable joe
3747:
3748: 45 3 * foo ! \ set foo to 135
3749: 1234 joe ! \ set joe to 1234
3750: 3 joe +! \ increment joe by 3.. to 1237
3751: @end example
3752:
3753: Not surprisingly, there is no need to define @code{myvariable}, since
3754: Forth already has a definition @code{Variable}. It behaves in exactly
3755: the same way as @code{myvariable}. Forth also provides @code{2Variable}
3756: and @code{fvariable} for double and floating-point variables,
3757: respectively.
3758:
3759: doc-variable
3760: doc-2variable
3761: doc-fvariable
3762:
3763: @cindex arrays
3764: A similar mechanism can be used to create arrays. For example, an
3765: 80-character text input buffer:
3766:
3767: @example
3768: CREATE text-buf 80 chars allot
3769:
3770: text-buf 0 chars c@@ \ the 1st character (offset 0)
3771: text-buf 3 chars c@@ \ the 4th character (offset 3)
3772: @end example
3773:
3774: You can build arbitrarily complex data structures by allocating
3775: appropriate areas of memory. @xref{Structures} for further discussions
3776: of this, and to learn about some Gforth tools that make it easier.
3777:
3778: @cindex user variables
3779: @cindex user space
3780: The defining word @code{User} behaves in the same way as @code{Variable}.
3781: The difference is that it reserves space in @i{user (data) space} rather
3782: than normal data space. In a Forth system that has a multi-tasker, each
3783: task has its own set of user variables.
3784:
3785: doc-user
3786:
3787: @comment TODO is that stuff about user variables strictly correct? Is it
3788: @comment just terminal tasks that have user variables?
3789: @comment should document tasker.fs (with some examples) elsewhere
3790: @comment in this manual, then expand on user space and user variables.
3791:
3792: After @code{CREATE} and @code{Variable}s, the next defining word to
3793: consider is @code{Constant}. @code{Constant} allows you to declare a
3794: fixed value and refer to it by name. For example:
3795:
3796: @example
3797: 12 Constant INCHES-PER-FOOT
3798: 3E+08 fconstant SPEED-O-LIGHT
3799: @end example
3800:
3801: A @code{Variable} can be both read and written, so its run-time
3802: behaviour is to supply an address through which its current value can be
3803: manipulated. In contrast, the value of a @code{Constant} cannot be
3804: changed once it has been declared@footnote{Well, often it can be -- but
3805: not in a Standard, portable way. It's safer to use a @code{Value} (read
3806: on).} so it's not necessary to supply the address -- it is more
3807: efficient to return the value of the constant directly. That's exactly
3808: what happens; the run-time effect of a constant is to put its value on
3809: the top of the stack (@ref{User-defined Defining Words} describes one
3810: way of implementing @code{Constant}).
3811:
3812: Gforth also provides @code{2Constant} and @code{fconstant} for defining
3813: double and floating-point constants, respectively.
3814:
3815: doc-constant
3816: doc-2constant
3817: doc-fconstant
3818:
3819: @c that's too deep, and it's not necessarily true for all ANS Forths. - anton
3820: Constants in Forth behave differently from their equivalents in other
3821: programming languages. In other languages, a constant (such as an EQU in
3822: assembler or a #define in C) only exists at compile-time; in the
3823: executable program the constant has been translated into an absolute
3824: number and, unless you are using a symbolic debugger, it's impossible to
3825: know what abstract thing that number represents. In Forth a constant has
3826: an entry in the header space and remains there after the code that
3827: uses it has been defined. In fact, it must remain in the dictionary
3828: since it has run-time duties to perform. For example:
3829:
3830: @example
3831: 12 Constant INCHES-PER-FOOT
3832: : FEET-TO-INCHES ( n1 -- n2 ) INCHES-PER-FOOT * ;
3833: @end example
3834:
3835: @cindex in-lining of constants
3836: When @code{FEET-TO-INCHES} is executed, it will in turn execute the xt
3837: associated with the constant @code{INCHES-PER-FOOT}. If you use
3838: @code{see} to decompile the definition of @code{FEET-TO-INCHES}, you can
3839: see that it makes a call to @code{INCHES-PER-FOOT}. Some Forth compilers
3840: attempt to optimise constants by in-lining them where they are used. You
3841: can force Gforth to in-line a constant like this:
3842:
3843: @example
3844: : FEET-TO-INCHES ( n1 -- n2 ) [ INCHES-PER-FOOT ] LITERAL * ;
3845: @end example
3846:
3847: If you use @code{see} to decompile @i{this} version of
3848: @code{FEET-TO-INCHES}, you can see that @code{INCHES-PER-FOOT} is no
3849: longer present. @xref{Interpret/Compile states} and @ref{Literals} on
3850: how this works.
3851:
3852: In-lining constants in this way might improve execution time
3853: fractionally, and can ensure that a constant is now only referenced at
3854: compile-time. However, the definition of the constant still remains in
3855: the dictionary. Some Forth compilers provide a mechanism for controlling
3856: a second dictionary for holding transient words such that this second
3857: dictionary can be deleted later in order to recover memory
3858: space. However, there is no standard way of doing this.
3859:
3860: One aspect of constants and variables that can sometimes be confusing is
3861: that they have different stack effects; one returns its value whilst the
3862: other returns the address of its value. The defining word @code{Value}
3863: provides an alternative to @code{Variable}, and has the same stack
3864: effect as a constant. A @code{Value} needs an additional word, @code{TO}
3865: to allow its value to be changed. Here are some examples:
3866:
3867: @example
3868: 12 Value APPLES \ a Value is initialised when it is declared.. like a
3869: \ constant but unlike a variable
3870: 34 TO APPLES \ Change the value of APPLES. TO is a parsing word
3871: APPLES \ puts 34 on the top of the stack.
3872: @end example
3873:
3874: doc-value
3875: doc-to
3876:
3877: The defining word @code{Defer} allows you to define a word by name
3878: without defining its behaviour; the definition of its behaviour is
3879: deferred. Here are two situation where this can be useful:
3880:
3881: @itemize @bullet
3882: @item
3883: Where you want to allow the behaviour of a word to be altered later, and
3884: for all precompiled references to the word to change when its behaviour
3885: is changed.
3886: @item
3887: For mutual recursion; @xref{Calls and returns}.
3888: @end itemize
3889:
3890: In the following example, @code{foo} always invokes the version of
3891: @code{greet} that prints ``@code{Good morning}'' whilst @code{bar}
3892: always invokes the version that prints ``@code{Hello}''. There is no way
3893: of getting @code{foo} to use the later version without re-ordering the
3894: source code and recompilng it.
3895:
3896: @example
3897: : greet ." Good morning" ;
3898: : foo ... greet ... ;
3899: : greet ." Hello" ;
3900: : bar ... greet ... ;
3901: @end example
3902:
3903: This problem can be solved by defining @code{greet} as a @code{Defer}red
3904: word. The behaviour of a @code{Defer}red word can be defined and
3905: redefined at any time by using @code{IS} to associate the xt of a
3906: previously-defined word with it. The previous example becomes:
3907:
3908: @example
3909: Defer greet
3910: : foo ... greet ... ;
3911: : bar ... greet ... ;
3912: : greet1 ." Good morning" ;
3913: : greet2 ." Hello" ;
3914: ' greet2 <IS> greet \ make greet behave like greet2
3915: @end example
3916:
3917: One thing to note is that @code{<IS>} consumes it's name when it is
3918: executed. If you want to specify the name at compile time, use
3919: @code{[IS]}:
3920:
3921: @example
3922: : set-greet ( xt -- )
3923: [IS] greet ;
3924:
3925: ' greet1 set-greet
3926: @end example
3927:
3928: A deferred word can only inherit default semantics from the xt (because
3929: that is all that an xt can represent -- @pxref{Tokens for Words} for
3930: more discussion of this). However, the semantics of the deferred word
3931: itself can be modified at the time that it is defined. For example:
3932:
3933: @example
3934: : bar .... ; compile-only
3935: Defer fred immediate
3936: Defer jim
3937:
3938: ' bar <IS> jim \ jim has default semantics
3939: ' bar <IS> fred \ fred is immediate
3940: @end example
3941:
3942: doc-defer
3943: doc-<is>
3944: doc-[is]
3945: @comment TODO document these: what's defers [is]
3946: doc-what's
3947: doc-defers
3948:
3949: Definitions in ANS Forth for @code{defer}, @code{<is>} and
3950: @code{[is]} are provided in @file{compat/defer.fs}.
3951:
3952: The defining word @code{Alias} allows you to define a word by name that
3953: has the same behaviour as some other word. Here are two situation where
3954: this can be useful:
3955:
3956: @itemize @bullet
3957: @item
3958: When you want access to a word's definition from a different word list
3959: (for an example of this, see the definition of the @code{Root} word list
3960: in the Gforth source).
3961: @item
3962: When you want to create a synonym; a definition that can be known by
3963: either of two names (for example, @code{THEN} and @code{ENDIF} are
3964: aliases).
3965: @end itemize
3966:
3967: The word whose behaviour the alias is to inherit is represented by an
3968: xt. Therefore, the alias only inherits default semantics from its
3969: ancestor. The semantics of the alias itself can be modified at the time
3970: that it is defined. For example:
3971:
3972: @example
3973: : foo ... ; immediate
3974:
3975: ' foo Alias bar \ bar is not an immediate word
3976: ' foo Alias fooby immediate \ fooby is an immediate word
3977: @end example
3978:
3979: @c "combined words" is an undefined term
3980: Words that are aliases have the same xt, different headers in the
3981: dictionary, and consequently different name tokens (@pxref{Tokens for
3982: Words}) and possibly different immediate flags. An alias can only have
3983: default or immediate compilation semantics; you can define aliases for
3984: combined words with @code{interpret/compile:}.
3985:
3986: @c distribute this to the appropriate paragraphs? - anton
3987: doc-alias
3988:
3989: @node Colon Definitions, User-defined Defining Words, Simple Defining Words, Defining Words
3990: @subsection Colon Definitions
3991: @cindex colon definitions
3992:
3993: @example
3994: : name ( ... -- ... )
3995: word1 word2 word3 ;
3996: @end example
3997:
3998: @noindent
3999: Creates a word called @code{name} that, upon execution, executes
4000: @code{word1 word2 word3}. @code{name} is a @dfn{(colon) definition}.
4001:
4002: The explanation above is somewhat superficial. @xref{Your first
4003: definition} for simple examples of colon definitions, then
4004: @xref{Interpretation and Compilation Semantics} for an in-depth
4005: discussion of some of the issues involved.
4006:
4007: doc-:
4008: doc-;
4009:
4010: @node User-defined Defining Words, Supplying names, Colon Definitions, Defining Words
4011: @subsection User-defined Defining Words
4012: @cindex user-defined defining words
4013: @cindex defining words, user-defined
4014:
4015: You can create a new defining word by wrapping defining-time code around
4016: an existing defining word and putting the sequence in a colon
4017: definition. For example, suppose that you have a word @code{stats} that
4018: gathers statistics about colon definitions given the @i{xt} of the
4019: definition, and you want every colon definition in your application to
4020: make a call to @code{stats}. You can define and use a new version of
4021: @code{:} like this:
4022:
4023: @example
4024: : stats ( xt -- ) DUP ." (Gathering statistics for " . ." )"
4025: ... ; \ other code
4026:
4027: : my: : lastxt postpone literal ['] stats compile, ;
4028:
4029: my: foo + - ;
4030: @end example
4031:
4032: When @code{foo} is defined using @code{my:} these steps occur:
4033:
4034: @itemize @bullet
4035: @item
4036: @code{my:} is executed.
4037: @item
4038: The @code{:} within the definition (the one between @code{my:} and
4039: @code{lastxt}) is executed, and does just what it always does; it parses
4040: the input stream for a name, builds a dictionary header for the name
4041: @code{foo} and switches @code{state} from interpret to compile.
4042: @item
4043: The word @code{lastxt} is executed. It puts the @i{xt} for the word that is
4044: being defined -- @code{foo} -- onto the stack.
4045: @item
4046: The code that was produced by @code{postpone literal} is executed; this
4047: causes the value on the stack to be compiled as a literal in the code
4048: area of @code{foo}.
4049: @item
4050: The code @code{['] stats} compiles a literal into the definition of
4051: @code{my:}. When @code{compile,} is executed, that literal -- the
4052: execution token for @code{stats} -- is layed down in the code area of
4053: @code{foo} , following the literal@footnote{Strictly speaking, the
4054: mechanism that @code{compile,} uses to convert an @i{xt} into something
4055: in the code area is implementation-dependent. A threaded implementation
4056: might spit out the execution token directly whilst another
4057: implementation might spit out a native code sequence.}.
4058: @item
4059: At this point, the execution of @code{my:} is complete, and control
4060: returns to the text interpreter. The text interpreter is in compile
4061: state, so subsequent text @code{+ -} is compiled into the definition of
4062: @code{foo} and the @code{;} terminates the definition as always.
4063: @end itemize
4064:
4065: You can use @code{see} to decompile a word that was defined using
4066: @code{my:} and see how it is different from a normal @code{:}
4067: definition. For example:
4068:
4069: @example
4070: : bar + - ; \ like foo but using : rather than my:
4071: see bar
4072: : bar
4073: + - ;
4074: see foo
4075: : foo
4076: 107645672 stats + - ;
4077:
4078: \ use ' stats . to show that 107645672 is the xt for stats
4079: @end example
4080:
4081:
4082: @c a deferred word is not neccessary for these examples. - anton
4083: Rather than edit your application's source code to change every @code{:}
4084: to a @code{my:}, use a deferred word:
4085:
4086: @example
4087: : real: : ; \ retain access to the original
4088: defer : \ redefine as a deferred word
4089: ' my: IS : \ use special version of :
4090: \
4091: \ load application here
4092: \
4093: ' real: IS : \ go back to the original
4094: @end example
4095:
4096: You can use techniques like this to make new defining words in terms of
4097: @i{any} existing defining word.
4098:
4099:
4100: @cindex defining defining words
4101: @cindex @code{CREATE} ... @code{DOES>}
4102: If you want the words defined with your defining words to behave
4103: differently from words defined with standard defining words, you can
4104: write your defining word like this:
4105:
4106: @example
4107: : def-word ( "name" -- )
4108: CREATE @i{code1}
4109: DOES> ( ... -- ... )
4110: @i{code2} ;
4111:
4112: def-word name
4113: @end example
4114:
4115: @cindex child words
4116: This fragment defines a @dfn{defining word} @code{def-word} and then
4117: executes it. When @code{def-word} executes, it @code{CREATE}s a new
4118: word, @code{name}, and executes the code @i{code1}. The code @i{code2}
4119: is not executed at this time. The word @code{name} is sometimes called a
4120: @dfn{child} of @code{def-word}.
4121:
4122: When you execute @code{name}, the address of the body of @code{name} is
4123: put on the data stack and @i{code2} is executed (the address of the body
4124: of @code{name} is the address @code{HERE} returns immediately after the
4125: @code{CREATE}).
4126:
4127: @cindex atavism in child words
4128: You can use @code{def-word} to define a set of child words that behave
4129: differently, though atavistically; they all have a common run-time
4130: behaviour determined by @i{code2}. Typically, the @i{code1} sequence
4131: builds a data area in the body of the child word. The structure of the
4132: data is common to all children of @code{def-word}, but the data values
4133: are specific -- and private -- to each child word. When a child word is
4134: executed, the address of its private data area is passed as a parameter
4135: on TOS to be used and manipulated@footnote{It is legitimate both to read
4136: and write to this data area.} by @i{code2}.
4137:
4138: The two fragments of code that make up the defining words act (are
4139: executed) at two completely separate times:
4140:
4141: @itemize @bullet
4142: @item
4143: At @i{define time}, the defining word executes @i{code1} to generate a
4144: child word
4145: @item
4146: At @i{child execution time}, when a child word is invoked, @i{code2}
4147: is executed, using parameters (data) that are private and specific to
4148: the child word.
4149: @end itemize
4150:
4151: @c NAC I think this is a really bad example, because it diminishes
4152: @c rather than emphasising the fact that some important stuff happens
4153: @c at define time, and other important stuff happens at child-invocation
4154: @c time, and that those two times are potentially very different.
4155:
4156: @c Well, IMO CREATE-DOES> is usually presented with much ado, making
4157: @c people think that it's hard to understand, and making those people who
4158: @c understand it easily think that it's hyped. I prefer presenting it in a
4159: @c diminished way and only emphasize the special issues later. - anton
4160:
4161: In other words, if you make the following definitions:
4162: @example
4163: : def-word1 ( "name" -- )
4164: CREATE @i{code1} ;
4165:
4166: : action1 ( ... -- ... )
4167: @i{code2} ;
4168:
4169: def-word1 name1
4170: @end example
4171:
4172: Using @code{name1 action1} is equivalent to using @code{name}.
4173:
4174: The classic example is that you can define @code{CONSTANT} in this way:
4175:
4176: @example
4177: : CONSTANT ( w "name" -- )
4178: CREATE ,
4179: DOES> ( -- w )
4180: @@ ;
4181: @end example
4182:
4183: @comment There is a beautiful description of how this works and what
4184: @comment it does in the Forthwrite 100th edition.. as well as an elegant
4185: @comment commentary on the Counting Fruits problem.
4186:
4187: When you create a constant with @code{5 CONSTANT five}, a set of
4188: define-time actions take place; first a new word @code{five} is created,
4189: then the value 5 is laid down in the body of @code{five} with
4190: @code{,}. When @code{five} is invoked, the address of the body is put on
4191: the stack, and @code{@@} retrieves the value 5. The word @code{five} has
4192: no code of its own; it simply contains a data field and a pointer to the
4193: code that follows @code{DOES>} in its defining word. That makes words
4194: created in this way very compact.
4195:
4196: The final example in this section is intended to remind you that space
4197: reserved in @code{CREATE}d words is @i{data} space and therefore can be
4198: both read and written by a Standard program@footnote{Exercise: use this
4199: example as a starting point for your own implementation of @code{Value}
4200: and @code{TO} -- if you get stuck, investigate the behaviour of @code{'} and
4201: @code{[']}.}:
4202:
4203: @example
4204: : foo ( "name" -- )
4205: CREATE -1 ,
4206: DOES> ( -- )
4207: @@ . ;
4208:
4209: foo first-word
4210: foo second-word
4211:
4212: 123 ' first-word >BODY !
4213: @end example
4214:
4215: If @code{first-word} had been a @code{CREATE}d word, we could simply
4216: have executed it to get the address of its data field. However, since it
4217: was defined to have @code{DOES>} actions, its execution semantics are to
4218: perform those @code{DOES>} actions. To get the address of its data field
4219: it's necessary to use @code{'} to get its xt, then @code{>BODY} to
4220: translate the xt into the address of the data field. When you execute
4221: @code{first-word}, it will display @code{123}. When you execute
4222: @code{second-word} it will display @code{-1}.
4223:
4224: @cindex stack effect of @code{DOES>}-parts
4225: @cindex @code{DOES>}-parts, stack effect
4226: In the examples above the stack comment after the @code{DOES>} specifies
4227: the stack effect of the defined words, not the stack effect of the
4228: following code (the following code expects the address of the body on
4229: the top of stack, which is not reflected in the stack comment). This is
4230: the convention that I use and recommend (it clashes a bit with using
4231: locals declarations for stack effect specification, though).
4232:
4233: @subsubsection Applications of @code{CREATE..DOES>}
4234: @cindex @code{CREATE} ... @code{DOES>}, applications
4235:
4236: You may wonder how to use this feature. Here are some usage patterns:
4237:
4238: @cindex factoring similar colon definitions
4239: When you see a sequence of code occurring several times, and you can
4240: identify a meaning, you will factor it out as a colon definition. When
4241: you see similar colon definitions, you can factor them using
4242: @code{CREATE..DOES>}. E.g., an assembler usually defines several words
4243: that look very similar:
4244: @example
4245: : ori, ( reg-target reg-source n -- )
4246: 0 asm-reg-reg-imm ;
4247: : andi, ( reg-target reg-source n -- )
4248: 1 asm-reg-reg-imm ;
4249: @end example
4250:
4251: @noindent
4252: This could be factored with:
4253: @example
4254: : reg-reg-imm ( op-code -- )
4255: CREATE ,
4256: DOES> ( reg-target reg-source n -- )
4257: @@ asm-reg-reg-imm ;
4258:
4259: 0 reg-reg-imm ori,
4260: 1 reg-reg-imm andi,
4261: @end example
4262:
4263: @cindex currying
4264: Another view of @code{CREATE..DOES>} is to consider it as a crude way to
4265: supply a part of the parameters for a word (known as @dfn{currying} in
4266: the functional language community). E.g., @code{+} needs two
4267: parameters. Creating versions of @code{+} with one parameter fixed can
4268: be done like this:
4269: @example
4270: : curry+ ( n1 -- )
4271: CREATE ,
4272: DOES> ( n2 -- n1+n2 )
4273: @@ + ;
4274:
4275: 3 curry+ 3+
4276: -2 curry+ 2-
4277: @end example
4278:
4279: @subsubsection The gory details of @code{CREATE..DOES>}
4280: @cindex @code{CREATE} ... @code{DOES>}, details
4281:
4282: doc-does>
4283:
4284: @cindex @code{DOES>} in a separate definition
4285: This means that you need not use @code{CREATE} and @code{DOES>} in the
4286: same definition; you can put the @code{DOES>}-part in a separate
4287: definition. This allows us to, e.g., select among different @code{DOES>}-parts:
4288: @example
4289: : does1
4290: DOES> ( ... -- ... )
4291: ... ;
4292:
4293: : does2
4294: DOES> ( ... -- ... )
4295: ... ;
4296:
4297: : def-word ( ... -- ... )
4298: create ...
4299: IF
4300: does1
4301: ELSE
4302: does2
4303: ENDIF ;
4304: @end example
4305:
4306: In this example, the selection of whether to use @code{does1} or
4307: @code{does2} is made at compile-time; at the time that the child word is
4308: @code{CREATE}d.
4309:
4310: @cindex @code{DOES>} in interpretation state
4311: In a standard program you can apply a @code{DOES>}-part only if the last
4312: word was defined with @code{CREATE}. In Gforth, the @code{DOES>}-part
4313: will override the behaviour of the last word defined in any case. In a
4314: standard program, you can use @code{DOES>} only in a colon
4315: definition. In Gforth, you can also use it in interpretation state, in a
4316: kind of one-shot mode; for example:
4317: @example
4318: CREATE name ( ... -- ... )
4319: @i{initialization}
4320: DOES>
4321: @i{code} ;
4322: @end example
4323:
4324: @noindent
4325: is equivalent to the standard:
4326: @example
4327: :noname
4328: DOES>
4329: @i{code} ;
4330: CREATE name EXECUTE ( ... -- ... )
4331: @i{initialization}
4332: @end example
4333:
4334: You can get the address of the body of a word with:
4335:
4336: doc->body
4337:
4338: @node Supplying names, Interpretation and Compilation Semantics, User-defined Defining Words, Defining Words
4339: @subsection Supplying the name of a defined word
4340: @cindex names for defined words
4341: @cindex defining words, name parameter
4342:
4343: @cindex defining words, name given in a string
4344: By default, a defining word takes the name for the defined word from the
4345: input stream. Sometimes you want to supply the name from a string. You
4346: can do this with:
4347:
4348: doc-nextname
4349:
4350: For example:
4351:
4352: @example
4353: s" foo" nextname create
4354: @end example
4355: @noindent
4356: is equivalent to:
4357: @example
4358: create foo
4359: @end example
4360:
4361: @cindex defining words without name
4362: Sometimes you want to define an @dfn{anonymous word}; a word without a
4363: name. You can do this with:
4364:
4365: doc-:noname
4366:
4367: This leaves the execution token for the word on the stack after the
4368: closing @code{;}. Here's an example in which a deferred word is
4369: initialised with an @code{xt} from an anonymous colon definition:
4370: @example
4371: Defer deferred
4372: :noname ( ... -- ... )
4373: ... ;
4374: IS deferred
4375: @end example
4376:
4377: @noindent
4378: Gforth provides an alternative way of doing this, using two separate
4379: words:
4380:
4381: doc-noname
4382: @cindex execution token of last defined word
4383: doc-lastxt
4384:
4385: @noindent
4386: The previous example can be rewritten using @code{noname} and
4387: @code{lastxt}:
4388:
4389: @example
4390: Defer deferred
4391: noname : ( ... -- ... )
4392: ... ;
4393: lastxt IS deferred
4394: @end example
4395:
4396: @noindent
4397: @code{noname} and @code{nextname} work with any defining word, not just
4398: @code{:}.
4399:
4400: @code{lastxt} also works when the last word was not defined as
4401: @code{noname}. It also has the useful property that is is valid as soon
4402: as the header for a definition has been build. Thus:
4403:
4404: @example
4405: lastxt . : foo [ lastxt . ] ; ' foo .
4406: @end example
4407:
4408: @noindent
4409: prints 3 numbers; the last two are the same.
4410:
4411:
4412: @node Interpretation and Compilation Semantics, , Supplying names, Defining Words
4413: @subsection Interpretation and Compilation Semantics
4414: @cindex semantics, interpretation and compilation
4415:
4416: @cindex interpretation semantics
4417: The @dfn{interpretation semantics} of a word are what the text
4418: interpreter does when it encounters the word in interpret state. It also
4419: appears in some other contexts, e.g., the execution token returned by
4420: @code{' @i{word}} identifies the interpretation semantics of
4421: @i{word} (in other words, @code{' @i{word} execute} is equivalent to
4422: interpret-state text interpretation of @code{@i{word}}).
4423:
4424: @cindex compilation semantics
4425: The @dfn{compilation semantics} of a word are what the text interpreter
4426: does when it encounters the word in compile state. It also appears in
4427: other contexts, e.g, @code{POSTPONE @i{word}} compiles@footnote{In
4428: standard terminology, ``appends to the current definition''.} the
4429: compilation semantics of @i{word}.
4430:
4431: @cindex execution semantics
4432: The standard also talks about @dfn{execution semantics}. They are used
4433: only for defining the interpretation and compilation semantics of many
4434: words. By default, the interpretation semantics of a word are to
4435: @code{execute} its execution semantics, and the compilation semantics of
4436: a word are to @code{compile,} its execution semantics.@footnote{In
4437: standard terminology: The default interpretation semantics are its
4438: execution semantics; the default compilation semantics are to append its
4439: execution semantics to the execution semantics of the current
4440: definition.}
4441:
4442: @comment TODO expand, make it co-operate with new sections on text interpreter.
4443:
4444: @cindex immediate words
4445: @cindex compile-only words
4446: You can change the semantics of the most-recently defined word:
4447:
4448: doc-immediate
4449: doc-compile-only
4450: doc-restrict
4451:
4452: Note that ticking (@code{'}) a compile-only word gives an error
4453: (``Interpreting a compile-only word'').
4454:
4455: Gforth also allows you to define words with arbitrary combinations of
4456: interpretation and compilation semantics.
4457:
4458: doc-interpret/compile:
4459:
4460: This feature was introduced for implementing @code{TO} and @code{S"}. I
4461: recommend that you do not define such words, as cute as they may be:
4462: they make it hard to get at both parts of the word in some contexts.
4463: E.g., assume you want to get an execution token for the compilation
4464: part. Instead, define two words, one that embodies the interpretation
4465: part, and one that embodies the compilation part. Once you have done
4466: that, you can define a combined word with @code{interpret/compile:} for
4467: the convenience of your users.
4468:
4469: You might try to use this feature to provide an optimizing
4470: implementation of the default compilation semantics of a word. For
4471: example, by defining:
4472: @example
4473: :noname
4474: foo bar ;
4475: :noname
4476: POSTPONE foo POSTPONE bar ;
4477: interpret/compile: opti-foobar
4478: @end example
4479:
4480: @noindent
4481: as an optimizing version of:
4482:
4483: @example
4484: : foobar
4485: foo bar ;
4486: @end example
4487:
4488: Unfortunately, this does not work correctly with @code{[compile]},
4489: because @code{[compile]} assumes that the compilation semantics of all
4490: @code{interpret/compile:} words are non-default. I.e., @code{[compile]
4491: opti-foobar} would compile compilation semantics, whereas
4492: @code{[compile] foobar} would compile interpretation semantics.
4493:
4494: @cindex state-smart words (are a bad idea)
4495: Some people try to use @dfn{state-smart} words to emulate the feature provided
4496: by @code{interpret/compile:} (words are state-smart if they check
4497: @code{STATE} during execution). E.g., they would try to code
4498: @code{foobar} like this:
4499:
4500: @example
4501: : foobar
4502: STATE @@
4503: IF ( compilation state )
4504: POSTPONE foo POSTPONE bar
4505: ELSE
4506: foo bar
4507: ENDIF ; immediate
4508: @end example
4509:
4510: Although this works if @code{foobar} is only processed by the text
4511: interpreter, it does not work in other contexts (like @code{'} or
4512: @code{POSTPONE}). E.g., @code{' foobar} will produce an execution token
4513: for a state-smart word, not for the interpretation semantics of the
4514: original @code{foobar}; when you execute this execution token (directly
4515: with @code{EXECUTE} or indirectly through @code{COMPILE,}) in compile
4516: state, the result will not be what you expected (i.e., it will not
4517: perform @code{foo bar}). State-smart words are a bad idea. Simply don't
4518: write them@footnote{For a more detailed discussion of this topic, see
4519: @cite{@code{State}-smartness -- Why it is Evil and How to Exorcise it} by Anton
4520: Ertl; presented at EuroForth '98 and available from
4521: @url{http://www.complang.tuwien.ac.at/papers/ertl98.ps.gz}}!
4522:
4523: @cindex defining words with arbitrary semantics combinations
4524: It is also possible to write defining words that define words with
4525: arbitrary combinations of interpretation and compilation semantics. In
4526: general, they look like this:
4527:
4528: @example
4529: : def-word
4530: create-interpret/compile
4531: @i{code1}
4532: interpretation>
4533: @i{code2}
4534: <interpretation
4535: compilation>
4536: @i{code3}
4537: <compilation ;
4538: @end example
4539:
4540: For a @i{word} defined with @code{def-word}, the interpretation
4541: semantics are to push the address of the body of @i{word} and perform
4542: @i{code2}, and the compilation semantics are to push the address of
4543: the body of @i{word} and perform @i{code3}. E.g., @code{constant}
4544: can also be defined like this (except that the defined constants don't
4545: behave correctly when @code{[compile]}d):
4546:
4547: @example
4548: : constant ( n "name" -- )
4549: create-interpret/compile
4550: ,
4551: interpretation> ( -- n )
4552: @@
4553: <interpretation
4554: compilation> ( compilation. -- ; run-time. -- n )
4555: @@ postpone literal
4556: <compilation ;
4557: @end example
4558:
4559: doc-create-interpret/compile
4560: doc-interpretation>
4561: doc-<interpretation
4562: doc-compilation>
4563: doc-<compilation
4564:
4565: Words defined with @code{interpret/compile:} and
4566: @code{create-interpret/compile} have an extended header structure that
4567: differs from other words; however, unless you try to access them with
4568: plain address arithmetic, you should not notice this. Words for
4569: accessing the header structure usually know how to deal with this; e.g.,
4570: @code{'} @i{word} @code{>body} also gives you the body of a word created
4571: with @code{create-interpret/compile}.
4572:
4573: doc-postpone
4574: @comment TODO -- expand glossary text for POSTPONE
4575:
4576: @c ----------------------------------------------------------
4577: @node The Text Interpreter, Tokens for Words, Defining Words, Words
4578: @section The Text Interpreter
4579: @cindex interpreter - outer
4580: @cindex text interpreter
4581: @cindex outer interpreter
4582:
4583: @c Should we really describe all these ugly details? IMO the text
4584: @c interpreter should be much cleaner, but that may not be possible within
4585: @c ANS Forth. - anton
4586:
4587: The text interpreter@footnote{This is an expanded version of the
4588: material in @ref{Introducing the Text Interpreter}.} is an endless loop
4589: that processes input from the current input device. It is also called
4590: the outer interpreter, in contrast to the inner interpreter
4591: (@pxref{Engine}) which executes the compiled Forth code on interpretive
4592: implementations.
4593:
4594: @cindex interpret state
4595: @cindex compile state
4596: The text interpreter operates in one of two states: @dfn{interpret
4597: state} and @dfn{compile state}. The current state is defined by the
4598: aptly-named variable, @code{state}.
4599:
4600: This section starts by describing how the text interpreter behaves when
4601: it is in interpret state, processing input from the user input device --
4602: the keyboard. This is the mode that a Forth system is in after it starts
4603: up.
4604:
4605: @cindex input buffer
4606: @cindex terminal input buffer
4607: The text interpreter works from an area of memory called the @dfn{input
4608: buffer}@footnote{When the text interpreter is processing input from the
4609: keyboard, this area of memory is called the @dfn{terminal input buffer}
4610: (TIB) and is addressed by the (obsolescent) words @code{TIB} and
4611: @code{#TIB}.}, which stores your keyboard input when you press the
4612: @key{RET} key. Starting at the beginning of the input buffer, it skips
4613: leading spaces (called @dfn{delimiters}) then parses a string (a
4614: sequence of non-space characters) until it reaches either a space
4615: character or the end of the buffer. Having parsed a string, it makes two
4616: attempts to process it:
4617:
4618: @cindex dictionary
4619: @itemize @bullet
4620: @item
4621: It looks for the string in a @dfn{dictionary} of definitions. If the
4622: string is found, the string names a @dfn{definition} (also known as a
4623: @dfn{word}) and the dictionary search returns information that allows
4624: the text interpreter to perform the word's @dfn{interpretation
4625: semantics}. In most cases, this simply means that the word will be
4626: executed.
4627: @item
4628: If the string is not found in the dictionary, the text interpreter
4629: attempts to treat it as a number, using the rules described in
4630: @ref{Number Conversion}. If the string represents a legal number in the
4631: current radix, the number is pushed onto a parameter stack (the data
4632: stack for integers, the floating-point stack for floating-point
4633: numbers).
4634: @end itemize
4635:
4636: If both attempts fail, or if the word is found in the dictionary but has
4637: no interpretation semantics@footnote{This happens if the word was
4638: defined as @code{COMPILE-ONLY}.} the text interpreter discards the
4639: remainder of the input buffer, issues an error message and waits for
4640: more input. If one of the attempts succeeds, the text interpreter
4641: repeats the parsing process until the whole of the input buffer has been
4642: processed, at which point it prints the status message ``@code{ ok}''
4643: and waits for more input.
4644:
4645: @cindex parse area
4646: The text interpreter keeps track of its position in the input buffer by
4647: updating a variable called @code{>IN} (pronounced ``to-in''). The value
4648: of @code{>IN} starts out as 0, indicating an offset of 0 from the start
4649: of the input buffer. The region from offset @code{>IN @@} to the end of
4650: the input buffer is called the @dfn{parse area}@footnote{In other words,
4651: the text interpreter processes the contents of the input buffer by
4652: parsing strings from the parse area until the parse area is empty.}.
4653: This example shows how @code{>IN} changes as the text interpreter parses
4654: the input buffer:
4655:
4656: @example
4657: : remaining >IN @@ SOURCE 2 PICK - -ROT + SWAP
4658: CR ." ->" TYPE ." <-" ; IMMEDIATE
4659:
4660: 1 2 3 remaining + remaining .
4661:
4662: : foo 1 2 3 remaining SWAP remaining ;
4663: @end example
4664:
4665: @noindent
4666: The result is:
4667:
4668: @example
4669: ->+ remaining .<-
4670: ->.<-5 ok
4671:
4672: ->SWAP remaining ;-<
4673: ->;<- ok
4674: @end example
4675:
4676: @cindex parsing words
4677: The value of @code{>IN} can also be modified by a word in the input
4678: buffer that is executed by the text interpreter. This means that a word
4679: can ``trick'' the text interpreter into either skipping a section of the
4680: input buffer@footnote{This is how parsing words work.} or into parsing a
4681: section twice. For example:
4682:
4683: @example
4684: : lat ." <<lat>>" ;
4685: : flat ." <<flat>>" >IN DUP @@ 3 - SWAP ! ;
4686: @end example
4687:
4688: @noindent
4689: When @code{flat} is executed, this output is produced@footnote{Exercise
4690: for the reader: what would happen if the @code{3} were replaced with
4691: @code{4}?}:
4692:
4693: @example
4694: <<flat>><<lat>>
4695: @end example
4696:
4697: @noindent
4698: Two important notes about the behaviour of the text interpreter:
4699:
4700: @itemize @bullet
4701: @item
4702: It processes each input string to completion before parsing additional
4703: characters from the input buffer.
4704: @item
4705: It treats the input buffer as a read-only region (and so must your code).
4706: @end itemize
4707:
4708: @noindent
4709: When the text interpreter is in compile state, its behaviour changes in
4710: these ways:
4711:
4712: @itemize @bullet
4713: @item
4714: If a parsed string is found in the dictionary, the text interpreter will
4715: perform the word's @dfn{compilation semantics}. In most cases, this
4716: simply means that the execution semantics of the word will be appended
4717: to the current definition.
4718: @item
4719: When a number is encountered, it is compiled into the current definition
4720: (as a literal) rather than being pushed onto a parameter stack.
4721: @item
4722: If an error occurs, @code{state} is modified to put the text interpreter
4723: back into interpret state.
4724: @item
4725: Each time a line is entered from the keyboard, Gforth prints
4726: ``@code{ compiled}'' rather than `` @code{ok}''.
4727: @end itemize
4728:
4729: @cindex text interpreter - input sources
4730: When the text interpreter is using an input device other than the
4731: keyboard, its behaviour changes in these ways:
4732:
4733: @itemize @bullet
4734: @item
4735: When the parse area is empty, the text interpreter attempts to refill
4736: the input buffer from the input source. When the input source is
4737: exhausted, the input source is set back to the user input device.
4738: @item
4739: It doesn't print out ``@code{ ok}'' or ``@code{ compiled}'' messages each
4740: time the parse area is emptied.
4741: @item
4742: If an error occurs, the input source is set back to the user input
4743: device.
4744: @end itemize
4745:
4746: @ref{Input Sources} describes this in more detail.
4747:
4748: doc->in
4749: doc-source
4750:
4751: doc-tib
4752: doc-#tib
4753:
4754: @menu
4755: * Input Sources::
4756: * Number Conversion::
4757: * Interpret/Compile states::
4758: * Literals::
4759: * Interpreter Directives::
4760: @end menu
4761:
4762: @node Input Sources, Number Conversion, The Text Interpreter, The Text Interpreter
4763: @subsection Input Sources
4764: @cindex input sources
4765: @cindex text interpreter - input sources
4766:
4767: By default, the text interpreter accepts input from the user input
4768: device (the keyboard) when Forth starts up. The text interpreter can
4769: process input from any of these sources:
4770:
4771: @itemize @bullet
4772: @item
4773: The user input device -- the keyboard.
4774: @item
4775: A file, using the words described in @ref{Forth source files}.
4776: @item
4777: A block, using the words described in @ref{Blocks}.
4778: @item
4779: A text string, using @code{evaluate}.
4780: @end itemize
4781:
4782: A program can identify the current input device from the values of
4783: @code{source-id} and @code{blk}.
4784:
4785: doc-source-id
4786: doc-blk
4787:
4788: doc-save-input
4789: doc-restore-input
4790:
4791: doc-evaluate
4792:
4793:
4794: @node Number Conversion, Interpret/Compile states, Input Sources, The Text Interpreter
4795: @subsection Number Conversion
4796: @cindex number conversion
4797: @cindex double-cell numbers, input format
4798: @cindex input format for double-cell numbers
4799: @cindex single-cell numbers, input format
4800: @cindex input format for single-cell numbers
4801: @cindex floating-point numbers, input format
4802: @cindex input format for floating-point numbers
4803:
4804: This section describes the rules that the text interpreter uses when it
4805: tries to convert a string into a number.
4806:
4807: Let <digit> represent any character that is a legal digit in the current
4808: number base@footnote{For example, 0-9 when the number base is decimal or
4809: 0-9, A-F when the number base is hexadecimal.}.
4810:
4811: Let <decimal digit> represent any character in the range 0-9.
4812:
4813: Let @{@i{a b}@} represent the @i{optional} presence of any of the characters
4814: in the braces (@i{a} or @i{b} or neither).
4815:
4816: Let * represent any number of instances of the previous character
4817: (including none).
4818:
4819: Let any other character represent itself.
4820:
4821: @noindent
4822: Now, the conversion rules are:
4823:
4824: @itemize @bullet
4825: @item
4826: A string of the form <digit><digit>* is treated as a single-precision
4827: (cell-sized) positive integer. Examples are 0 123 6784532 32343212343456 42
4828: @item
4829: A string of the form -<digit><digit>* is treated as a single-precision
4830: (cell-sized) negative integer, and is represented using 2's-complement
4831: arithmetic. Examples are -45 -5681 -0
4832: @item
4833: A string of the form <digit><digit>*.<digit>* is treated as a double-precision
4834: (double-cell-sized) positive integer. Examples are 3465. 3.465 34.65
4835: (all three of these represent the same number).
4836: @item
4837: A string of the form -<digit><digit>*.<digit>* is treated as a
4838: double-precision (double-cell-sized) negative integer, and is
4839: represented using 2's-complement arithmetic. Examples are -3465. -3.465
4840: -34.65 (all three of these represent the same number).
4841: @item
4842: A string of the form @{+ -@}<decimal digit>@{.@}<decimal digit>*@{e
4843: E@}@{+ -@}<decimal digit><decimal digit>* is treated as a floating-point
4844: number. Examples are 1e 1e0 1.e 1.e0 +1e+0 (which all represent the same
4845: number) +12.E-4
4846: @end itemize
4847:
4848: By default, the number base used for integer number conversion is given
4849: by the contents of the variable @code{base}. Note that a lot of
4850: confusion can result from unexpected values of @code{base}. If you
4851: change @code{base} anywhere, make sure to save the old value and restore
4852: it afterwards. In general I recommend keeping @code{base} decimal, and
4853: using the prefixes described below for the popular non-decimal bases.
4854:
4855: doc-dpl
4856: doc-base
4857: doc-hex
4858: doc-decimal
4859:
4860: @cindex '-prefix for character strings
4861: @cindex &-prefix for decimal numbers
4862: @cindex %-prefix for binary numbers
4863: @cindex $-prefix for hexadecimal numbers
4864: Gforth allows you to override the value of @code{base} by using a
4865: prefix@footnote{Some Forth implementations provide a similar scheme by
4866: implementing @code{$} etc. as parsing words that process the subsequent
4867: number in the input stream and push it onto the stack. For example, see
4868: @cite{Number Conversion and Literals}, by Wil Baden; Forth Dimensions
4869: 20(3) pages 26--27. In such implementations, unlike in Gforth, a space
4870: is required between the prefix and the number.} before the first digit
4871: of an (integer) number. Four prefixes are supported:
4872:
4873: @itemize @bullet
4874: @item
4875: @code{&} -- decimal
4876: @item
4877: @code{%} -- binary
4878: @item
4879: @code{$} -- hexadecimal
4880: @item
4881: @code{'} -- base @code{max-char+1}
4882: @end itemize
4883:
4884: Here are some examples, with the equivalent decimal number shown after
4885: in braces:
4886:
4887: -$41 (-65), %1001101 (205), %1001.0001 (145 - a double-precision number),
4888: 'AB (16706; ascii A is 65, ascii B is 66, number is 65*256 + 66),
4889: 'ab (24930; ascii a is 97, ascii B is 98, number is 97*256 + 98),
4890: &905 (905), $abc (2478), $ABC (2478).
4891:
4892: @cindex number conversion - traps for the unwary
4893: @noindent
4894: Number conversion has a number of traps for the unwary:
4895:
4896: @itemize @bullet
4897: @item
4898: You cannot determine the current number base using the code sequence
4899: @code{base @@ .} -- the number base is always 10 in the current number
4900: base. Instead, use something like @code{base @@ dec.}
4901: @item
4902: If the number base is set to a value greater than 14 (for example,
4903: hexadecimal), the number 123E4 is ambiguous; the conversion rules allow
4904: it to be intepreted as either a single-precision integer or a
4905: floating-point number (Gforth treats it as an integer). The ambiguity
4906: can be resolved by explicitly stating the sign of the mantissa and/or
4907: exponent: 123E+4 or +123E4 -- if the number base is decimal, no
4908: ambiguity arises; either representation will be treated as a
4909: floating-point number.
4910: @item
4911: There is a word @code{bin} but it does @i{not} set the number base!
4912: It is used to specify file types.
4913: @item
4914: ANS Forth requires the @code{.} of a double-precision number to
4915: be the final character in the string. Allowing the @code{.} to be
4916: anywhere after the first digit is a Gforth extension.
4917: @item
4918: The number conversion process does not check for overflow.
4919: @item
4920: In Gforth, number conversion to floating-point numbers always use base
4921: 10, irrespective of the value of @code{base}. In ANS Forth,
4922: conversion to floating-point numbers whilst the value of
4923: @code{base} is not 10 is an ambiguous condition.
4924: @end itemize
4925:
4926: @ref{Input} describes words that you can use to read numbers into your
4927: programs.
4928:
4929: @node Interpret/Compile states, Literals, Number Conversion, The Text Interpreter
4930: @subsection Interpret/Compile states
4931: @cindex Interpret/Compile states
4932:
4933: A standard program is not permitted to change @code{state}
4934: explicitly. However, it can change @code{state} implicitly, using the
4935: words @code{[} and @code{]}. When @code{[} is executed it switches
4936: @code{state} to interpret state, and therefore the text interpreter
4937: starts interpreting. When @code{]} is executed it switches @code{state}
4938: to compile state and therefore the text interpreter starts
4939: compiling. The most common usage for these words is to compile literals,
4940: as shown in @ref{Literals}. However, they give you the freedom to switch
4941: modes at will.
4942:
4943: @c This is a bad example: It's non-standard, and it's not necessary.
4944: @c However, I can't think of a good example for switching into compile
4945: @c state when there is no current word (@code{state}-smart words are not a
4946: @c good reason). So maybe we should use an example for switching into
4947: @c interpret @code{state} in a colon def. - anton
4948:
4949: Here is an example of building a jump-table of execution
4950: tokens:
4951:
4952: @example
4953: : AA ." this is A" ;
4954: : BB ." this is B" ;
4955: : CC ." this is C" ;
4956:
4957: create table ' aa COMPILE, ' bb COMPILE, ' cc COMPILE,
4958: : go ( n -- ) \ n is offset into table.. 0 for 1st entry
4959: cells table + @ execute ;
4960: @end example
4961:
4962: @noindent
4963: Now @code{0 go} will display ``@code{this is A}''. The table can be
4964: built far more neatly@footnote{The source code is neater.. what is
4965: compiled in memory in each case is identical.} like this:
4966:
4967: @example
4968: create table ] aa bb cc [
4969: @end example
4970:
4971: The problem with this code is that it is not portable; it will only work
4972: on systems where code space and data space co-incide. The reason is that
4973: both tables @i{compile} execution tokens -- into code space. The
4974: Standard only allows data space to be assigned for a @code{CREATE}d
4975: word. In addition, the Standard only allows @code{@@} to access data
4976: space, whilst this example is using it to access code space. The only
4977: portable, Standard way to build this table is to build it in data space,
4978: like this:
4979:
4980: @example
4981: create table ' aa , ' bb , ' cc ,
4982: @end example
4983:
4984: @noindent
4985: A similar technique can be used to build a table of constants:
4986:
4987: @example
4988: create primes 1 , 3 , 5 , 7 , 11 ,
4989: @end example
4990:
4991: doc-state
4992: doc-[
4993: doc-]
4994:
4995: @node Literals, Interpreter Directives, Interpret/Compile states, The Text Interpreter
4996: @subsection Literals
4997: @cindex Literals
4998:
4999: Often, you want to use a number within a colon definition. When you do
5000: this, the text interpreter automatically compiles the number as a
5001: @i{literal}. A literal is a number whose run-time effect is to be pushed
5002: onto the stack. If you had to do some maths to generate the number, you
5003: might write it like this:
5004:
5005: @example
5006: : HOUR-TO-SEC ( n1 -- n2 )
5007: 60 * \ to minutes
5008: 60 * ; \ to seconds
5009: @end example
5010:
5011: It is very clear what this definition is doing, but it's inefficient
5012: since it is performing 2 multiples at run-time. An alternative would be
5013: to write:
5014:
5015: @example
5016: : HOUR-TO-SEC ( n1 -- n2 )
5017: 3600 * ; \ to seconds
5018: @end example
5019:
5020: Which does the same thing, and has the advantage of using a single
5021: multiply. Ideally, we'd like the efficiency of the second with the
5022: readability of the first.
5023:
5024: @code{Literal} allows us to achieve that. It takes a number from the
5025: stack and lays it down in the current definition just as though the
5026: number had been typed directly into the definition. Our first attempt
5027: might look like this:
5028:
5029: @example
5030: 60 \ mins per hour
5031: 60 * \ seconds per minute
5032: : HOUR-TO-SEC ( n1 -- n2 )
5033: Literal * ; \ to seconds
5034: @end example
5035:
5036: But this produces the error message @code{unstructured}. What happened?
5037: The stack notation for @code{:} is (@i{ -- colon-sys}) and the size of
5038: @i{colon-sys} is implementation-defined. In other words, once we start a
5039: colon definition we can't portably access anything that was on the stack
5040: before the definition began@footnote{@cite{Two Problems in ANS Forth},
5041: by Thomas Worthington; Forth Dimensions 20(2) pages 32--34 describes
5042: some situations where you might want to access stack items above
5043: colon-sys, and provides a solution to the problem.}. The correct way of
5044: solving this problem in this instance is to use @code{[ ]} like this:
5045:
5046: @example
5047: : HOUR-TO-SEC ( n1 -- n2 )
5048: [ 60 \ minutes per hour
5049: 60 * ] \ seconds per minute
5050: LITERAL * ; \ to seconds
5051: @end example
5052:
5053: doc-literal
5054: doc-]L
5055: doc-2literal
5056: doc-fliteral
5057:
5058: @node Interpreter Directives, , Literals, The Text Interpreter
5059: @subsection Interpreter Directives
5060: @cindex interpreter directives
5061:
5062: These words are usually used in interpret state; typically to control
5063: which parts of a source file are processed by the text
5064: interpreter. There are only a few ANS Forth Standard words, but Gforth
5065: supplements these with a rich set of immediate control structure words
5066: to compensate for the fact that the non-immediate versions can only be
5067: used in compile state (@pxref{Control Structures}). Typical usages:
5068:
5069: @example
5070: FALSE Constant ASSEMBLER
5071: .
5072: .
5073: ASSEMBLER [IF]
5074: : ASSEMBLER-FEATURE
5075: ...
5076: ;
5077: [ENDIF]
5078: .
5079: .
5080: : SEE
5081: ... \ general-purpose SEE code
5082: [ ASSEMBLER [IF] ]
5083: ... \ assembler-specific SEE code
5084: [ [ENDIF] ]
5085: ;
5086: @end example
5087:
5088: doc-[IF]
5089: doc-[ELSE]
5090: doc-[THEN]
5091: doc-[ENDIF]
5092:
5093: doc-[IFDEF]
5094: doc-[IFUNDEF]
5095:
5096: doc-[?DO]
5097: doc-[DO]
5098: doc-[FOR]
5099: doc-[LOOP]
5100: doc-[+LOOP]
5101: doc-[NEXT]
5102:
5103: doc-[BEGIN]
5104: doc-[UNTIL]
5105: doc-[AGAIN]
5106: doc-[WHILE]
5107: doc-[REPEAT]
5108:
5109:
5110:
5111: @c -------------------------------------------------------------
5112: @node Tokens for Words, Word Lists, The Text Interpreter, Words
5113: @section Tokens for Words
5114: @cindex tokens for words
5115:
5116: This section describes the creation and use of tokens that represent
5117: words.
5118:
5119: Named words have information stored in their header space entries to
5120: indicate any non-default semantics (@pxref{Interpretation and
5121: Compilation Semantics}). The semantics can be modified, using
5122: @code{immediate} and/or @code{compile-only}, at the time that the words
5123: are defined. Unnamed words have (by definition) no header space
5124: entry, and therefore must have default semantics.
5125:
5126: Named words have interpretation and compilation semantics. Unnamed words
5127: just have execution semantics.
5128:
5129: @cindex xt
5130: @cindex execution token
5131: The execution semantics of an unnamed word are represented by an
5132: @dfn{execution token} (@i{xt}). As explained in @ref{Supplying names},
5133: the execution token of the last word defined can be produced with
5134: @code{lastxt}.
5135:
5136: The interpretation semantics of a named word are also represented by an
5137: execution token. You can produce the execution token using @code{'} or
5138: @code{[']}. A simple example shows the difference between the two:
5139:
5140: @example
5141: : greet ( -- ) ." Hello" ;
5142: : foo ( -- xt ) ['] greet execute ; \ ['] parses greet at compile-time
5143: : bar ( -- ) ' execute ; \ ' parses at run-time
5144:
5145: \ the next four lines all do the same thing
5146: foo
5147: bar greet
5148: greet
5149: ' greet EXECUTE
5150: @end example
5151:
5152: An execution token occupies one cell.
5153: @cindex code field address
5154: @cindex CFA
5155: In Gforth, the abstract data type @i{execution token} is implemented
5156: as a code field address (CFA).
5157: @comment TODO note that the standard does not say what it represents..
5158: @comment and you cannot necessarily compile it in all Forths (eg native
5159: @comment compilers?).
5160:
5161: For literals, use @code{'} in interpreted code and @code{[']} in
5162: compiled code. Gforth's @code{'} and @code{[']} behave somewhat
5163: unusually by complaining about compile-only words. To get the execution
5164: token for a compile-only word @i{name}, use @code{COMP' @i{name} DROP}
5165: or @code{[COMP'] @i{name} DROP}.
5166:
5167: @cindex compilation token
5168: The compilation semantics of a named word are represented by a
5169: @dfn{compilation token} consisting of two cells: @i{w xt}. The top cell
5170: @i{xt} is an execution token. The compilation semantics represented by
5171: the compilation token can be performed with @code{execute}, which
5172: consumes the whole compilation token, with an additional stack effect
5173: determined by the represented compilation semantics.
5174:
5175: At present, the @i{w} part of a compilation token is an execution token,
5176: and the @i{xt} part represents either @code{execute} or
5177: @code{compile,}@footnote{Depending upon the compilation semantics of the
5178: word. If the word has default compilation semantics, the @i{xt} will
5179: represent @code{compile,}. Otherwise (e.g., for immediate words), the
5180: @i{xt} will represent @code{execute}.}. However, don't rely on that
5181: knowledge, unless necessary; future versions of Gforth may introduce
5182: unusual compilation tokens (e.g., a compilation token that represents
5183: the compilation semantics of a literal).
5184:
5185: You can compile the compilation semantics with @code{postpone,}. I.e.,
5186: @code{COMP' @i{word} postpone,} is equivalent to @code{postpone
5187: @i{word}}.
5188:
5189: @cindex name token
5190: @cindex name field address
5191: @cindex NFA
5192: Named words are also represented by the @dfn{name token}, (@i{nt}). In
5193: Gforth, the abstract data type @emph{name token} is implemented as a
5194: name field address (NFA).
5195:
5196: doc-execute
5197: doc-compile,
5198: doc-[']
5199: doc-'
5200: doc-[comp']
5201: doc-comp'
5202: doc-postpone,
5203:
5204: doc-find-name
5205: doc-name>int
5206: doc-name?int
5207: doc-name>comp
5208: doc-name>string
5209:
5210: @c -------------------------------------------------------------
5211: @node Word Lists, Environmental Queries, Tokens for Words, Words
5212: @section Word Lists
5213: @cindex word lists
5214: @cindex header space
5215:
5216: A wordlist is a list of named words; you can add new words and look up
5217: words by name (and you can remove words in a restricted way with
5218: markers). Every named (and @code{reveal}ed) word is in one wordlist.
5219:
5220: @cindex search order stack
5221: The text interpreter searches the wordlists present in the search order
5222: (a stack of wordlists), from the top to the bottom. Within each
5223: wordlist, the search starts conceptually at the newest word; i.e., if
5224: two words in a wordlist have the same name, the newer word is found.
5225:
5226: @cindex compilation word list
5227: New words are added to the @dfn{compilation wordlist} (aka current
5228: wordlist).
5229:
5230: @cindex wid
5231: A word list is identified by a cell-sized word list identifier (@i{wid})
5232: in much the same way as a file is identified by a file handle. The
5233: numerical value of the wid has no (portable) meaning, and might change
5234: from session to session.
5235:
5236: The ANS Forth ``Search order'' word set is intended to provide a set of
5237: low-level tools that allow various different schemes to be
5238: implemented. Gforth provides @code{vocabulary}, a traditional Forth
5239: word. @file{compat/vocabulary.fs} provides an implementation in ANS
5240: Standard Forth.
5241:
5242: @comment TODO: locals section refers to here, saying that every word list (aka
5243: @comment vocabulary) has its own methods for searching etc. Need to document that.
5244:
5245: @comment the thisone- prefix is used to pick out the true definition of a
5246: @comment word from the source files, rather than some alias.
5247: doc-forth-wordlist
5248: doc-definitions
5249: doc-get-current
5250: doc-set-current
5251: doc-get-order
5252: doc---thisone-set-order
5253: doc-wordlist
5254: doc-table
5255: doc-push-order
5256: doc-previous
5257: doc-also
5258: doc---thisone-forth
5259: doc-only
5260: doc---thisone-order
5261:
5262: doc-find
5263: doc-search-wordlist
5264:
5265: doc-words
5266: doc-vlist
5267:
5268: doc-mappedwordlist
5269: doc-root
5270: doc-vocabulary
5271: doc-seal
5272: doc-vocs
5273: doc-current
5274: doc-context
5275:
5276: @menu
5277: * Why use word lists?::
5278: * Word list examples::
5279: @end menu
5280:
5281: @node Why use word lists?, Word list examples, Word Lists, Word Lists
5282: @subsection Why use word lists?
5283: @cindex word lists - why use them?
5284:
5285: Here are some reasons for using multiple word lists:
5286:
5287: @itemize @bullet
5288: @item
5289: To improve compilation speed by reducing the number of header space
5290: entries that must be searched. This is achieved by creating a new
5291: word list that contains all of the definitions that are used in the
5292: definition of a Forth system but which would not usually be used by
5293: programs running on that system. That word list would be on the search
5294: list when the Forth system was compiled but would be removed from the
5295: search list for normal operation. This can be a useful technique for
5296: low-performance systems (for example, 8-bit processors in embedded
5297: systems) but is unlikely to be necessary in high-performance desktop
5298: systems.
5299: @item
5300: To prevent a set of words from being used outside the context in which
5301: they are valid. Two classic examples of this are an integrated editor
5302: (all of the edit commands are defined in a separate word list; the
5303: search order is set to the editor word list when the editor is invoked;
5304: the old search order is restored when the editor is terminated) and an
5305: integrated assembler (the op-codes for the machine are defined in a
5306: separate word list which is used when a @code{CODE} word is defined).
5307: @item
5308: To prevent a name-space clash between multiple definitions with the same
5309: name. For example, when building a cross-compiler you might have a word
5310: @code{IF} that generates conditional code for your target system. By
5311: placing this definition in a different word list you can control whether
5312: the host system's @code{IF} or the target system's @code{IF} get used in
5313: any particular context by controlling the order of the word lists on the
5314: search order stack.
5315: @end itemize
5316:
5317: @node Word list examples, ,Why use word lists?, Word Lists
5318: @subsection Word list examples
5319: @cindex word lists - examples
5320:
5321: Here is an example of creating and using a new wordlist using ANS
5322: Forth Standard words:
5323:
5324: @example
5325: wordlist constant my-new-words-wordlist
5326: : my-new-words get-order nip my-new-words-wordlist swap set-order ;
5327:
5328: \ add it to the search order
5329: also my-new-words
5330:
5331: \ alternatively, add it to the search order and make it
5332: \ the compilation word list
5333: also my-new-words definitions
5334: \ type "order" to see the problem
5335: @end example
5336:
5337: The problem with this example is that @code{order} has no way to
5338: associate the name @code{my-new-words} with the wid of the word list (in
5339: Gforth, @code{order} and @code{vocs} will display @code{???} for a wid
5340: that has no associated name). There is no Standard way of associating a
5341: name with a wid.
5342:
5343: In Gforth, this example can be re-coded using @code{vocabulary}, which
5344: associates a name with a wid:
5345:
5346: @example
5347: vocabulary my-new-words
5348:
5349: \ add it to the search order
5350: my-new-words
5351:
5352: \ alternatively, add it to the search order and make it
5353: \ the compilation word list
5354: my-new-words definitions
5355: \ type "order" to see that the problem is solved
5356: @end example
5357:
5358: @c -------------------------------------------------------------
5359: @node Environmental Queries, Files, Word Lists, Words
5360: @section Environmental Queries
5361: @cindex environmental queries
5362:
5363: ANS Forth introduced the idea of ``environmental queries'' as a way
5364: for a program running on a system to determine certain characteristics of the system.
5365: The Standard specifies a number of strings that might be recognised by a system.
5366:
5367: The Standard requires that the header space used for environmental queries
5368: be distinct from the header space used for definitions.
5369:
5370: Typically, environmental queries are supported by creating a set of
5371: definitions in a word list that is @i{only} used during environmental
5372: queries; that is what Gforth does. There is no Standard way of adding
5373: definitions to the set of recognised environmental queries, but any
5374: implementation that supports the loading of optional word sets must have
5375: some mechanism for doing this (after loading the word set, the
5376: associated environmental query string must return @code{true}). In
5377: Gforth, the word list used to honour environmental queries can be
5378: manipulated just like any other word list.
5379:
5380: doc-environment?
5381: doc-environment-wordlist
5382:
5383: doc-gforth
5384: doc-os-class
5385:
5386: Note that, whilst the documentation for (e.g.) @code{gforth} shows it
5387: returning two items on the stack, querying it using @code{environment?}
5388: will return an additional item; the @code{true} flag that shows that the
5389: string was recognised.
5390:
5391: @comment TODO Document the standard strings or note where they are documented herein
5392:
5393: Here are some examples of using environmental queries:
5394:
5395: @example
5396: s" address-unit-bits" environment? 0=
5397: [IF]
5398: cr .( environmental attribute address-units-bits unknown... ) cr
5399: [THEN]
5400:
5401: s" block" environment? [IF] DROP include block.fs [THEN]
5402:
5403: s" gforth" environment? [IF] 2DROP include compat/vocabulary.fs [THEN]
5404:
5405: s" gforth" environment? [IF] .( Gforth version ) TYPE
5406: [ELSE] .( Not Gforth..) [THEN]
5407: @end example
5408:
5409:
5410: Here is an example of adding a definition to the environment word list:
5411:
5412: @example
5413: get-current environment-wordlist set-current
5414: true constant block
5415: true constant block-ext
5416: set-current
5417: @end example
5418:
5419: You can see what definitions are in the environment word list like this:
5420:
5421: @example
5422: get-order 1+ environment-wordlist swap set-order words previous
5423: @end example
5424:
5425:
5426: @c -------------------------------------------------------------
5427: @node Files, Blocks, Environmental Queries, Words
5428: @section Files
5429: @cindex files
5430: @cindex I/O - file-handling
5431:
5432: Gforth provides facilities for accessing files that are stored in the
5433: host operating system's file-system. Files that are processed by Gforth
5434: can be divided into two categories:
5435:
5436: @itemize @bullet
5437: @item
5438: Files that are processed by the Text Interpreter (@dfn{Forth source files}).
5439: @item
5440: Files that are processed by some other program (@dfn{general files}).
5441: @end itemize
5442:
5443: @menu
5444: * Forth source files::
5445: * General files::
5446: * Search Paths::
5447: * Forth Search Paths::
5448: * General Search Paths::
5449: @end menu
5450:
5451:
5452: @c -------------------------------------------------------------
5453: @node Forth source files, General files, Files, Files
5454: @subsection Forth source files
5455: @cindex including files
5456: @cindex Forth source files
5457:
5458: The simplest way to interpret the contents of a file is to use one of
5459: these two formats:
5460:
5461: @example
5462: include mysource.fs
5463: s" mysource.fs" included
5464: @end example
5465:
5466: Sometimes you want to include a file only if it is not included already
5467: (by, say, another source file). In that case, you can use one of these
5468: fomats:
5469:
5470: @example
5471: require mysource.fs
5472: needs mysource.fs
5473: s" mysource.fs" required
5474: @end example
5475:
5476: @cindex stack effect of included files
5477: @cindex including files, stack effect
5478: I recommend that you write your source files such that interpreting them
5479: does not change the stack. This allows using these files with
5480: @code{required} and friends without complications. For example:
5481:
5482: @example
5483: 1 require foo.fs drop
5484: @end example
5485:
5486: doc-include-file
5487: doc-included
5488: doc-included?
5489: doc-include
5490: doc-required
5491: doc-require
5492: doc-needs
5493: doc-init-included-files
5494:
5495: A definition in ANS Forth for @code{required} is provided in
5496: @file{compat/required.fs}.
5497:
5498: @c -------------------------------------------------------------
5499: @node General files, Search Paths, Forth source files, Files
5500: @subsection General files
5501: @cindex general files
5502: @cindex file-handling
5503:
5504: Files are opened/created by name and type. The following types are
5505: recognised:
5506:
5507: doc-r/o
5508: doc-r/w
5509: doc-w/o
5510: doc-bin
5511:
5512: When a file is opened/created, it returns a file identifier,
5513: @i{wfileid} that is used for all other file commands. All file
5514: commands also return a status value, @i{wior}, that is 0 for a
5515: successful operation and an implementation-defined non-zero value in the
5516: case of an error.
5517:
5518: doc-open-file
5519: doc-create-file
5520:
5521: doc-close-file
5522: doc-delete-file
5523: doc-rename-file
5524: doc-read-file
5525: doc-read-line
5526: doc-write-file
5527: doc-write-line
5528: doc-emit-file
5529: doc-flush-file
5530:
5531: doc-file-status
5532: doc-file-position
5533: doc-reposition-file
5534: doc-file-size
5535: doc-resize-file
5536:
5537: @c ---------------------------------------------------------
5538: @node Search Paths, Forth Search Paths, General files, Files
5539: @subsection Search Paths
5540: @cindex path for @code{included}
5541: @cindex file search path
5542: @cindex @code{include} search path
5543: @cindex search path for files
5544:
5545: If you specify an absolute filename (i.e., a filename starting with
5546: @file{/} or @file{~}, or with @file{:} in the second position (as in
5547: @samp{C:...})) for @code{included} and friends, that file is included
5548: just as you would expect.
5549:
5550: For relative filenames, Gforth uses a search path similar to Forth's
5551: search order (@pxref{Word Lists}). It tries to find the given filename
5552: in the directories present in the path, and includes the first one it
5553: finds. There are separate search paths for Forth source files and
5554: general files.
5555:
5556: If the search path contains the directory @file{.} (as it should), this
5557: refers to the directory that the present file was @code{included}
5558: from. This allows files to include other files relative to their own
5559: position (irrespective of the current working directory or the absolute
5560: position). This feature is essential for libraries consisting of
5561: several files, where a file may include other files from the library.
5562: It corresponds to @code{#include "..."} in C. If the current input
5563: source is not a file, @file{.} refers to the directory of the innermost
5564: file being included, or, if there is no file being included, to the
5565: current working directory.
5566:
5567: Use @file{~+} to refer to the current working directory (as in the
5568: @code{bash}).
5569:
5570: If the filename starts with @file{./}, the search path is not searched
5571: (just as with absolute filenames), and the @file{.} has the same meaning
5572: as described above.
5573:
5574: @c ---------------------------------------------------------
5575: @node Forth Search Paths, General Search Paths, Search Paths, Files
5576: @subsubsection Forth Search Paths
5577: @cindex search path control - Forth
5578:
5579: The search path is initialized when you start Gforth (@pxref{Invoking
5580: Gforth}). You can display it and change it using these words:
5581:
5582: doc-.fpath
5583: doc-fpath+
5584: doc-fpath=
5585: doc-open-fpath-file
5586:
5587: Here is an example of using @code{fpath} and @code{require}:
5588:
5589: @example
5590: fpath= /usr/lib/forth/|./
5591: require timer.fs
5592: @end example
5593:
5594: @c ---------------------------------------------------------
5595: @node General Search Paths, , Forth Search Paths, Files
5596: @subsubsection General Search Paths
5597: @cindex search path control - for user applications
5598:
5599: Your application may need to search files in several directories, like
5600: @code{included} does. To facilitate this, Gforth allows you to define
5601: and use your own search paths, by providing generic equivalents of the
5602: Forth search path words:
5603:
5604: doc-.path
5605: doc-path+
5606: doc-path=
5607: doc-open-path-file
5608:
5609: Here's an example of creating a search path:
5610:
5611: @example
5612: \ Make a buffer for the path:
5613: create mypath 100 chars , \ maximum length (is checked)
5614: 0 , \ real len
5615: 100 chars allot \ space for path
5616: @end example
5617:
5618: @c -------------------------------------------------------------
5619: @node Blocks, Other I/O, Files, Words
5620: @section Blocks
5621: @cindex I/O - blocks
5622: @cindex blocks
5623:
5624: When you run Gforth on a modern desk-top computer, it runs under the
5625: control of an operating system which provides certain services. One of
5626: these services is @var{file services}, which allows Forth source code
5627: and data to be stored in files and read into Gforth (@pxref{Files}).
5628:
5629: Traditionally, Forth has been an important programming language on
5630: systems where it has interfaced directly to the underlying hardware with
5631: no intervening operating system. Forth provides a mechanism, called
5632: @dfn{blocks}, for accessing mass storage on such systems.
5633:
5634: A block is a 1024-byte data area, which can be used to hold data or
5635: Forth source code. No structure is imposed on the contents of the
5636: block. A block is identified by its number; blocks are numbered
5637: contiguously from 1 to an implementation-defined maximum.
5638:
5639: A typical system that used blocks but no operating system might use a
5640: single floppy-disk drive for mass storage, with the disks formatted to
5641: provide 256-byte sectors. Blocks would be implemented by assigning the
5642: first four sectors of the disk to block 1, the second four sectors to
5643: block 2 and so on, up to the limit of the capacity of the disk. The disk
5644: would not contain any file system information, just the set of blocks.
5645:
5646: @cindex blocks file
5647: On systems that do provide file services, blocks are typically
5648: implemented by storing a sequence of blocks within a single @dfn{blocks
5649: file}. The size of the blocks file will be an exact multiple of 1024
5650: bytes, corresponding to the number of blocks it contains. This is the
5651: mechanism that Gforth uses.
5652:
5653: @cindex @file{blocks.fb}
5654: Only 1 blocks file can be open at a time. If you use block words without
5655: having specified a blocks file, Gforth defaults to the blocks file
5656: @file{blocks.fb}. Gforth uses the Forth search path when attempting to
5657: locate a blocks file (@pxref{Forth Search Paths}).
5658:
5659: @cindex block buffers
5660: When you read and write blocks under program control, Gforth uses a
5661: number of @dfn{block buffers} as intermediate storage. These buffers are
5662: not used when you use @code{load} to interpret the contents of a block.
5663:
5664: The behaviour of the block buffers is directly analagous to that of a
5665: cache. Each block buffer has three states:
5666:
5667: @itemize @bullet
5668: @item
5669: Unassigned
5670: @item
5671: Assigned-clean
5672: @item
5673: Assigned-dirty
5674: @end itemize
5675:
5676: Initially, all block buffers are @i{unassigned}. In order to access a
5677: block, the block (specified by its block number) must be assigned to a
5678: block buffer.
5679:
5680: The assignment of a block to a block buffer is performed by @code{block}
5681: or @code{buffer}. Use @code{block} when you wish to modify the existing
5682: contents of a block. Use @code{buffer} when you don't care about the
5683: existing contents of the block@footnote{The ANS Forth definition of
5684: @code{buffer} is intended not to cause disk I/O; if the data associated
5685: with the particular block is already stored in a block buffer due to an
5686: earlier @code{block} command, @code{buffer} will return that block
5687: buffer and the existing contents of the block will be
5688: available. Otherwise, @code{buffer} will simply assign a new, empty
5689: block buffer for the block.}.
5690:
5691: Once a block has been assigned to a block buffer, the block buffer state
5692: becomes @i{assigned-clean}. Data can now be manipulated within the
5693: block buffer.
5694:
5695: When the contents of a block buffer is changed it is necessary,
5696: @i{before calling} @code{block} @i{or} @code{buffer} @i{again}, to
5697: either abandon the changes (by doing nothing) or commit the changes,
5698: using @code{update}. Using @code{update} does not change the blocks
5699: file; it simply changes a block buffer's state to @i{assigned-dirty}.
5700:
5701: The word @code{flush} causes all @i{assigned-dirty} blocks to be
5702: written back to the blocks file on disk. Leaving Gforth using @code{bye}
5703: also causes a @code{flush} to be performed.
5704:
5705: In Gforth, @code{block} and @code{buffer} use a @i{direct-mapped}
5706: algorithm to assign a block buffer to a block. That means that any
5707: particular block can only be assigned to one specific block buffer,
5708: called (for the particular operation) the @i{victim buffer}. If the
5709: victim buffer is @i{unassigned} or @i{assigned-clean} it can be
5710: allocated to the new block immediately. If it is @i{assigned-dirty}
5711: its current contents must be written out to disk before it can be
5712: allocated to the new block.
5713:
5714: Although no structure is imposed on the contents of a block, it is
5715: traditional to display the contents as 16 lines each of 64 characters. A
5716: block provides a single, continuous stream of input (for example, it
5717: acts as a single parse area) -- there are no end-of-line characters
5718: within a block, and no end-of-file character at the end of a
5719: block. There are two consequences of this:
5720:
5721: @itemize @bullet
5722: @item
5723: The last character of one line wraps straight into the first character
5724: of the following line
5725: @item
5726: The word @code{\} -- comment to end of line -- requires special
5727: treatment; in the context of a block it causes all characters until the
5728: end of the current 64-character ``line'' to be ignored.
5729: @end itemize
5730:
5731: In Gforth, when you use @code{block} with a non-existent block number,
5732: the current block file will be extended to the appropriate size and the
5733: block buffer will be initialised with spaces.
5734:
5735: Gforth doesn't encourage the use of blocks; the mechanism is only
5736: provided for backward compatibility -- ANS Forth requires blocks to be
5737: available when files are.
5738:
5739: Common techniques that are used when working with blocks include:
5740:
5741: @itemize @bullet
5742: @item
5743: A screen editor that allows you to edit blocks without leaving the Forth
5744: environment.
5745: @item
5746: Shadow screens; where every code block has an associated block
5747: containing comments (for example: code in odd block numbers, comments in
5748: even block numbers). Typically, the block editor provides a convenient
5749: mechanism to toggle between code and comments.
5750: @item
5751: Load blocks; a single block (typically block 1) contains a number of
5752: @code{thru} commands which @code{load} the whole of the application.
5753: @end itemize
5754:
5755: See Frank Sergeant's Pygmy Forth to see just how well blocks can be
5756: integrated into a Forth programming environment.
5757:
5758: @comment TODO what about errors on open-blocks?
5759: doc-open-blocks
5760: doc-use
5761: doc-get-block-fid
5762: doc-block-position
5763:
5764: doc-scr
5765: doc-list
5766:
5767: doc---block-block
5768: doc-buffer
5769:
5770: doc-update
5771: doc-updated?
5772: doc-save-buffers
5773: doc-empty-buffers
5774: doc-empty-buffer
5775: doc-flush
5776:
5777: doc-load
5778: doc-thru
5779: doc-+load
5780: doc-+thru
5781: xdoc--gforth--->
5782: doc-block-included
5783:
5784: @c -------------------------------------------------------------
5785: @node Other I/O, Programming Tools, Blocks, Words
5786: @section Other I/O
5787: @cindex I/O - keyboard and display
5788:
5789: @menu
5790: * Simple numeric output:: Predefined formats
5791: * Formatted numeric output:: Formatted (pictured) output
5792: * String Formats:: How Forth stores strings in memory
5793: * Displaying characters and strings:: Other stuff
5794: * Input:: Input
5795: @end menu
5796:
5797: @node Simple numeric output, Formatted numeric output, Other I/O, Other I/O
5798: @subsection Simple numeric output
5799: @cindex numeric output - simple/free-format
5800:
5801: The simplest output functions are those that display numbers from the
5802: data or floating-point stacks. Floating-point output is always displayed
5803: using base 10. Numbers displayed from the data stack use the value stored
5804: in @code{base}.
5805:
5806: doc-.
5807: doc-dec.
5808: doc-hex.
5809: doc-u.
5810: doc-.r
5811: doc-u.r
5812: doc-d.
5813: doc-ud.
5814: doc-d.r
5815: doc-ud.r
5816: doc-f.
5817: doc-fe.
5818: doc-fs.
5819:
5820: Examples of printing the number 1234.5678E23 in the different floating-point output
5821: formats are shown below:
5822:
5823: @example
5824: f. 123456779999999000000000000.
5825: fe. 123.456779999999E24
5826: fs. 1.23456779999999E26
5827: @end example
5828:
5829:
5830: @node Formatted numeric output, String Formats, Simple numeric output, Other I/O
5831: @subsection Formatted numeric output
5832: @cindex formatted numeric output
5833: @cindex pictured numeric output
5834: @cindex numeric output - formatted
5835:
5836: Forth traditionally uses a technique called @dfn{pictured numeric
5837: output} for formatted printing of integers. In this technique, digits
5838: are extracted from the number (using the current output radix defined by
5839: @code{base}), converted to ASCII codes and appended to a string that is
5840: built in a scratch-pad area of memory (@pxref{core-idef,
5841: Implementation-defined options, Implementation-defined
5842: options}). Arbitrary characters can be appended to the string during the
5843: extraction process. The completed string is specified by an address
5844: and length and can be manipulated (@code{TYPE}ed, copied, modified)
5845: under program control.
5846:
5847: All of the words described in the previous section for simple numeric
5848: output are implemented in Gforth using pictured numeric output.
5849:
5850: Three important things to remember about Pictured Numeric Output:
5851:
5852: @itemize @bullet
5853: @item
5854: It always operates on double-precision numbers; to display a
5855: single-precision number, convert it first (@pxref{Double precision} for
5856: ways of doing this).
5857: @item
5858: It always treats the double-precision number as though it were
5859: unsigned. The examples below show ways of printing signed numbers.
5860: @item
5861: The string is built up from right to left; least significant digit first.
5862: @end itemize
5863:
5864: doc-<#
5865: doc-#
5866: doc-#s
5867: doc-hold
5868: doc-sign
5869: doc-#>
5870:
5871: doc-represent
5872:
5873: Here are some examples of using pictured numeric output:
5874:
5875: @example
5876: : my-u. ( u -- )
5877: \ Simplest use of pns.. behaves like Standard u.
5878: 0 \ convert to unsigned double
5879: <# \ start conversion
5880: #s \ convert all digits
5881: #> \ complete conversion
5882: TYPE SPACE ; \ display, with trailing space
5883:
5884: : cents-only ( u -- )
5885: 0 \ convert to unsigned double
5886: <# \ start conversion
5887: # # \ convert two least-significant digits
5888: #> \ complete conversion, discard other digits
5889: TYPE SPACE ; \ display, with trailing space
5890:
5891: : dollars-and-cents ( u -- )
5892: 0 \ convert to unsigned double
5893: <# \ start conversion
5894: # # \ convert two least-significant digits
5895: [char] . hold \ insert decimal point
5896: #s \ convert remaining digits
5897: [char] $ hold \ append currency symbol
5898: #> \ complete conversion
5899: TYPE SPACE ; \ display, with trailing space
5900:
5901: : my-. ( n -- )
5902: \ handling negatives.. behaves like Standard .
5903: s>d \ convert to signed double
5904: swap over dabs \ leave sign byte followed by unsigned double
5905: <# \ start conversion
5906: #s \ convert all digits
5907: rot sign \ get at sign byte, append "-" if needed
5908: #> \ complete conversion
5909: TYPE SPACE ; \ display, with trailing space
5910:
5911: : account. ( n -- )
5912: \ accountants don't like minus signs, they use braces
5913: \ for negative numbers
5914: s>d \ convert to signed double
5915: swap over dabs \ leave sign byte followed by unsigned double
5916: <# \ start conversion
5917: 2 pick \ get copy of sign byte
5918: 0< IF [char] ) hold THEN \ right-most character of output
5919: #s \ convert all digits
5920: rot \ get at sign byte
5921: 0< IF [char] ( hold THEN
5922: #> \ complete conversion
5923: TYPE SPACE ; \ display, with trailing space
5924: @end example
5925:
5926: Here are some examples of using these words:
5927:
5928: @example
5929: 1 my-u. 1
5930: hex -1 my-u. decimal FFFFFFFF
5931: 1 cents-only 01
5932: 1234 cents-only 34
5933: 2 dollars-and-cents $0.02
5934: 1234 dollars-and-cents $12.34
5935: 123 my-. 123
5936: -123 my. -123
5937: 123 account. 123
5938: -456 account. (456)
5939: @end example
5940:
5941:
5942: @node String Formats, Displaying characters and strings, Formatted numeric output, Other I/O
5943: @subsection String Formats
5944: @cindex strings - see character strings
5945: @cindex character strings - formats
5946: @cindex I/O - see character strings
5947:
5948: Forth commonly uses two different methods for representing character
5949: strings:
5950:
5951: @itemize @bullet
5952: @item
5953: @cindex address of counted string
5954: As a @dfn{counted string}, represented by a @i{c-addr}. The char
5955: addressed by @i{c-addr} contains a character-count, @i{n}, of the
5956: string and the string occupies the subsequent @i{n} char addresses in
5957: memory.
5958: @item
5959: As cell pair on the stack; @i{c-addr u}, where @i{u} is the length
5960: of the string in characters, and @i{c-addr} is the address of the
5961: first byte of the string.
5962: @end itemize
5963:
5964: ANS Forth encourages the use of the second format when representing
5965: strings on the stack, whilst conceeding that the counted string format
5966: remains useful as a way of storing strings in memory.
5967:
5968: doc-count
5969:
5970: @xref{Memory Blocks} for words that move, copy and search
5971: for strings. @xref{Displaying characters and strings,} for words that
5972: display characters and strings.
5973:
5974:
5975: @node Displaying characters and strings, Input, String Formats, Other I/O
5976: @subsection Displaying characters and strings
5977: @cindex characters - compiling and displaying
5978: @cindex character strings - compiling and displaying
5979:
5980: This section starts with a glossary of Forth words and ends with a set
5981: of examples.
5982:
5983: doc-bl
5984: doc-space
5985: doc-spaces
5986: doc-emit
5987: doc-toupper
5988: doc-."
5989: doc-.(
5990: doc-type
5991: doc-cr
5992: @cindex cursor control
5993: doc-at-xy
5994: doc-page
5995: doc-s"
5996: doc-c"
5997: doc-char
5998: doc-[char]
5999: doc-sliteral
6000:
6001: As an example, consider the following text, stored in a file @file{test.fs}:
6002:
6003: @example
6004: .( text-1)
6005: : my-word
6006: ." text-2" cr
6007: .( text-3)
6008: ;
6009:
6010: ." text-4"
6011:
6012: : my-char
6013: [char] ALPHABET emit
6014: char emit
6015: ;
6016: @end example
6017:
6018: When you load this code into Gforth, the following output is generated:
6019:
6020: @example
6021: @kbd{include test.fs @key{RET}} text-1text-3text-4 ok
6022: @end example
6023:
6024: @itemize @bullet
6025: @item
6026: Messages @code{text-1} and @code{text-3} are displayed because @code{.(}
6027: is an immediate word; it behaves in the same way whether it is used inside
6028: or outside a colon definition.
6029: @item
6030: Message @code{text-4} is displayed because of Gforth's added interpretation
6031: semantics for @code{."}.
6032: @item
6033: Message @code{text-2} is @i{not} displayed, because the text interpreter
6034: performs the compilation semantics for @code{."} within the definition of
6035: @code{my-word}.
6036: @end itemize
6037:
6038: Here are some examples of executing @code{my-word} and @code{my-char}:
6039:
6040: @example
6041: @kbd{my-word @key{RET}} text-2
6042: ok
6043: @kbd{my-char fred @key{RET}} Af ok
6044: @kbd{my-char jim @key{RET}} Aj ok
6045: @end example
6046:
6047: @itemize @bullet
6048: @item
6049: Message @code{text-2} is displayed because of the run-time behaviour of
6050: @code{."}.
6051: @item
6052: @code{[char]} compiles the ``A'' from ``ALPHABET'' and puts its display code
6053: on the stack at run-time. @code{emit} always displays the character
6054: when @code{my-char} is executed.
6055: @item
6056: @code{char} parses a string at run-time and the second @code{emit} displays
6057: the first character of the string.
6058: @item
6059: If you type @code{see my-char} you can see that @code{[char]} discarded
6060: the text ``LPHABET'' and only compiled the display code for ``A'' into the
6061: definition of @code{my-char}.
6062: @end itemize
6063:
6064:
6065:
6066: @node Input, , Displaying characters and strings, Other I/O
6067: @subsection Input
6068: @cindex input
6069: @cindex I/O - see input
6070: @cindex parsing a string
6071:
6072: @xref{String Formats} for ways of storing character strings in memory.
6073:
6074: @comment TODO examples for >number >float accept key key? pad parse word refill
6075: @comment then index them
6076:
6077: doc-key
6078: doc-key?
6079: doc->number
6080: doc->float
6081: doc-accept
6082: doc-pad
6083: doc-parse
6084: doc-word
6085: doc-sword
6086: doc-refill
6087: @comment obsolescent words..
6088: doc-convert
6089: doc-query
6090: doc-expect
6091: doc-span
6092:
6093:
6094: @c -------------------------------------------------------------
6095: @node Programming Tools, Assembler and Code Words, Other I/O, Words
6096: @section Programming Tools
6097: @cindex programming tools
6098:
6099: @menu
6100: * Debugging:: Simple and quick.
6101: * Assertions:: Making your programs self-checking.
6102: * Singlestep Debugger:: Executing your program word by word.
6103: @end menu
6104:
6105: @node Debugging, Assertions, Programming Tools, Programming Tools
6106: @subsection Debugging
6107: @cindex debugging
6108:
6109: Languages with a slow edit/compile/link/test development loop tend to
6110: require sophisticated tracing/stepping debuggers to facilate
6111: productive debugging.
6112:
6113: A much better (faster) way in fast-compiling languages is to add
6114: printing code at well-selected places, let the program run, look at
6115: the output, see where things went wrong, add more printing code, etc.,
6116: until the bug is found.
6117:
6118: The simple debugging aids provided in @file{debugs.fs}
6119: are meant to support this style of debugging. In addition, there are
6120: words for non-destructively inspecting the stack and memory:
6121:
6122: doc-.s
6123: doc-f.s
6124:
6125: There is a word @code{.r} but it does @i{not} display the return
6126: stack! It is used for formatted numeric output.
6127:
6128: doc-depth
6129: doc-fdepth
6130: doc-clearstack
6131: doc-?
6132: doc-dump
6133:
6134: The word @code{~~} prints debugging information (by default the source
6135: location and the stack contents). It is easy to insert. If you use Emacs
6136: it is also easy to remove (@kbd{C-x ~} in the Emacs Forth mode to
6137: query-replace them with nothing). The deferred words
6138: @code{printdebugdata} and @code{printdebugline} control the output of
6139: @code{~~}. The default source location output format works well with
6140: Emacs' compilation mode, so you can step through the program at the
6141: source level using @kbd{C-x `} (the advantage over a stepping debugger
6142: is that you can step in any direction and you know where the crash has
6143: happened or where the strange data has occurred).
6144:
6145: The default actions of @code{~~} clobber the contents of the pictured
6146: numeric output string, so you should not use @code{~~}, e.g., between
6147: @code{<#} and @code{#>}.
6148:
6149: doc-~~
6150: doc-printdebugdata
6151: doc-printdebugline
6152:
6153: doc-see
6154: doc-marker
6155:
6156: Here's an example of using @code{marker} at the start of a source file
6157: that you are debugging; it ensures that you only ever have one copy of
6158: the file's definitions compiled at any time:
6159:
6160: @example
6161: [IFDEF] my-code
6162: my-code
6163: [ENDIF]
6164:
6165: marker my-code
6166: init-included-files
6167:
6168: \ .. definitions start here
6169: \ .
6170: \ .
6171: \ end
6172: @end example
6173:
6174:
6175:
6176: @node Assertions, Singlestep Debugger, Debugging, Programming Tools
6177: @subsection Assertions
6178: @cindex assertions
6179:
6180: It is a good idea to make your programs self-checking, especially if you
6181: make an assumption that may become invalid during maintenance (for
6182: example, that a certain field of a data structure is never zero). Gforth
6183: supports @dfn{assertions} for this purpose. They are used like this:
6184:
6185: @example
6186: assert( @i{flag} )
6187: @end example
6188:
6189: The code between @code{assert(} and @code{)} should compute a flag, that
6190: should be true if everything is alright and false otherwise. It should
6191: not change anything else on the stack. The overall stack effect of the
6192: assertion is @code{( -- )}. E.g.
6193:
6194: @example
6195: assert( 1 1 + 2 = ) \ what we learn in school
6196: assert( dup 0<> ) \ assert that the top of stack is not zero
6197: assert( false ) \ this code should not be reached
6198: @end example
6199:
6200: The need for assertions is different at different times. During
6201: debugging, we want more checking, in production we sometimes care more
6202: for speed. Therefore, assertions can be turned off, i.e., the assertion
6203: becomes a comment. Depending on the importance of an assertion and the
6204: time it takes to check it, you may want to turn off some assertions and
6205: keep others turned on. Gforth provides several levels of assertions for
6206: this purpose:
6207:
6208: doc-assert0(
6209: doc-assert1(
6210: doc-assert2(
6211: doc-assert3(
6212: doc-assert(
6213: doc-)
6214:
6215: The variable @code{assert-level} specifies the highest assertions that
6216: are turned on. I.e., at the default @code{assert-level} of one,
6217: @code{assert0(} and @code{assert1(} assertions perform checking, while
6218: @code{assert2(} and @code{assert3(} assertions are treated as comments.
6219:
6220: The value of @code{assert-level} is evaluated at compile-time, not at
6221: run-time. Therefore you cannot turn assertions on or off at run-time;
6222: you have to set the @code{assert-level} appropriately before compiling a
6223: piece of code. You can compile different pieces of code at different
6224: @code{assert-level}s (e.g., a trusted library at level 1 and
6225: newly-written code at level 3).
6226:
6227: doc-assert-level
6228:
6229: If an assertion fails, a message compatible with Emacs' compilation mode
6230: is produced and the execution is aborted (currently with @code{ABORT"}.
6231: If there is interest, we will introduce a special throw code. But if you
6232: intend to @code{catch} a specific condition, using @code{throw} is
6233: probably more appropriate than an assertion).
6234:
6235: Definitions in ANS Forth for these assertion words are provided
6236: in @file{compat/assert.fs}.
6237:
6238:
6239: @node Singlestep Debugger, , Assertions, Programming Tools
6240: @subsection Singlestep Debugger
6241: @cindex singlestep Debugger
6242: @cindex debugging Singlestep
6243: @cindex @code{dbg}
6244: @cindex @code{BREAK:}
6245: @cindex @code{BREAK"}
6246:
6247: When you create a new word there's often the need to check whether it
6248: behaves correctly or not. You can do this by typing @code{dbg
6249: badword}. A debug session might look like this:
6250:
6251: @example
6252: : badword 0 DO i . LOOP ; ok
6253: 2 dbg badword
6254: : badword
6255: Scanning code...
6256:
6257: Nesting debugger ready!
6258:
6259: 400D4738 8049BC4 0 -> [ 2 ] 00002 00000
6260: 400D4740 8049F68 DO -> [ 0 ]
6261: 400D4744 804A0C8 i -> [ 1 ] 00000
6262: 400D4748 400C5E60 . -> 0 [ 0 ]
6263: 400D474C 8049D0C LOOP -> [ 0 ]
6264: 400D4744 804A0C8 i -> [ 1 ] 00001
6265: 400D4748 400C5E60 . -> 1 [ 0 ]
6266: 400D474C 8049D0C LOOP -> [ 0 ]
6267: 400D4758 804B384 ; -> ok
6268: @end example
6269:
6270: Each line displayed is one step. You always have to hit return to
6271: execute the next word that is displayed. If you don't want to execute
6272: the next word in a whole, you have to type @kbd{n} for @code{nest}. Here is
6273: an overview what keys are available:
6274:
6275: @table @i
6276:
6277: @item @key{RET}
6278: Next; Execute the next word.
6279:
6280: @item n
6281: Nest; Single step through next word.
6282:
6283: @item u
6284: Unnest; Stop debugging and execute rest of word. If we got to this word
6285: with nest, continue debugging with the calling word.
6286:
6287: @item d
6288: Done; Stop debugging and execute rest.
6289:
6290: @item s
6291: Stop; Abort immediately.
6292:
6293: @end table
6294:
6295: Debugging large application with this mechanism is very difficult, because
6296: you have to nest very deeply into the program before the interesting part
6297: begins. This takes a lot of time.
6298:
6299: To do it more directly put a @code{BREAK:} command into your source code.
6300: When program execution reaches @code{BREAK:} the single step debugger is
6301: invoked and you have all the features described above.
6302:
6303: If you have more than one part to debug it is useful to know where the
6304: program has stopped at the moment. You can do this by the
6305: @code{BREAK" string"} command. This behaves like @code{BREAK:} except that
6306: string is typed out when the ``breakpoint'' is reached.
6307:
6308: doc-dbg
6309: doc-BREAK:
6310: doc-BREAK"
6311:
6312:
6313: @c -------------------------------------------------------------
6314: @node Assembler and Code Words, Threading Words, Programming Tools, Words
6315: @section Assembler and Code Words
6316: @cindex assembler
6317: @cindex code words
6318:
6319: Gforth provides some words for defining primitives (words written in
6320: machine code), and for defining the machine-code equivalent of
6321: @code{DOES>}-based defining words. However, the machine-independent
6322: nature of Gforth poses a few problems: First of all, Gforth runs on
6323: several architectures, so it can provide no standard assembler. What's
6324: worse is that the register allocation not only depends on the processor,
6325: but also on the @code{gcc} version and options used.
6326:
6327: The words that Gforth offers encapsulate some system dependences (e.g.,
6328: the header structure), so a system-independent assembler may be used in
6329: Gforth. If you do not have an assembler, you can compile machine code
6330: directly with @code{,} and @code{c,}@footnote{This isn't portable,
6331: because these words emit stuff in @i{data} space; it works because
6332: Gforth has unified code/data spaces. Assembler isn't likely to be
6333: portable anyway.}.
6334:
6335: doc-assembler
6336: doc-code
6337: doc-end-code
6338: doc-;code
6339: doc-flush-icache
6340:
6341: If @code{flush-icache} does not work correctly, @code{code} words
6342: etc. will not work (reliably), either.
6343:
6344: The typical usage of these @code{code} words can be shown most easily by
6345: analogy to the equivalent high-level defining words:
6346:
6347: @example
6348: : foo code foo
6349: <high-level Forth words> <assembler>
6350: ; end-code
6351:
6352: : bar : bar
6353: <high-level Forth words> <high-level Forth words>
6354: CREATE CREATE
6355: <high-level Forth words> <high-level Forth words>
6356: DOES> ;code
6357: <high-level Forth words> <assembler>
6358: ; end-code
6359: @end example
6360:
6361: @code{flush-icache} is always present. The other words are rarely used
6362: and reside in @code{code.fs}, which is usually not loaded. You can load
6363: it with @code{require code.fs}.
6364:
6365: @cindex registers of the inner interpreter
6366: In the assembly code you will want to refer to the inner interpreter's
6367: registers (e.g., the data stack pointer) and you may want to use other
6368: registers for temporary storage. Unfortunately, the register allocation
6369: is installation-dependent.
6370:
6371: The easiest solution is to use explicit register declarations
6372: (@pxref{Explicit Reg Vars, , Variables in Specified Registers, gcc.info,
6373: GNU C Manual}) for all of the inner interpreter's registers: You have to
6374: compile Gforth with @code{-DFORCE_REG} (configure option
6375: @code{--enable-force-reg}) and the appropriate declarations must be
6376: present in the @code{machine.h} file (see @code{mips.h} for an example;
6377: you can find a full list of all declarable register symbols with
6378: @code{grep register engine.c}). If you give explicit registers to all
6379: variables that are declared at the beginning of @code{engine()}, you
6380: should be able to use the other caller-saved registers for temporary
6381: storage. Alternatively, you can use the @code{gcc} option
6382: @code{-ffixed-REG} (@pxref{Code Gen Options, , Options for Code
6383: Generation Conventions, gcc.info, GNU C Manual}) to reserve a register
6384: (however, this restriction on register allocation may slow Gforth
6385: significantly).
6386:
6387: If this solution is not viable (e.g., because @code{gcc} does not allow
6388: you to explicitly declare all the registers you need), you have to find
6389: out by looking at the code where the inner interpreter's registers
6390: reside and which registers can be used for temporary storage. You can
6391: get an assembly listing of the engine's code with @code{make engine.s}.
6392:
6393: In any case, it is good practice to abstract your assembly code from the
6394: actual register allocation. E.g., if the data stack pointer resides in
6395: register @code{$17}, create an alias for this register called @code{sp},
6396: and use that in your assembly code.
6397:
6398: @cindex code words, portable
6399: Another option for implementing normal and defining words efficiently
6400: is to add the desired functionality to the source of Gforth. For normal
6401: words you just have to edit @file{primitives} (@pxref{Automatic
6402: Generation}). Defining words (equivalent to @code{;CODE} words, for fast
6403: defined words) may require changes in @file{engine.c}, @file{kernel.fs},
6404: @file{prims2x.fs}, and possibly @file{cross.fs}.
6405:
6406:
6407: @c -------------------------------------------------------------
6408: @node Threading Words, Locals, Assembler and Code Words, Words
6409: @section Threading Words
6410: @cindex threading words
6411:
6412: @cindex code address
6413: These words provide access to code addresses and other threading stuff
6414: in Gforth (and, possibly, other interpretive Forths). It more or less
6415: abstracts away the differences between direct and indirect threading
6416: (and, for direct threading, the machine dependences). However, at
6417: present this wordset is still incomplete. It is also pretty low-level;
6418: some day it will hopefully be made unnecessary by an internals wordset
6419: that abstracts implementation details away completely.
6420:
6421: doc-threading-method
6422: doc->code-address
6423: doc->does-code
6424: doc-code-address!
6425: doc-does-code!
6426: doc-does-handler!
6427: doc-/does-handler
6428:
6429: The code addresses produced by various defining words are produced by
6430: the following words:
6431:
6432: doc-docol:
6433: doc-docon:
6434: doc-dovar:
6435: doc-douser:
6436: doc-dodefer:
6437: doc-dofield:
6438:
6439: You can recognize words defined by a @code{CREATE}...@code{DOES>} word
6440: with @code{>does-code}. If the word was defined in that way, the value
6441: returned is non-zero and identifies the @code{DOES>} used by the
6442: defining word.
6443: @comment TODO should that be ``identifies the xt of the DOES> ??''
6444:
6445: @c -------------------------------------------------------------
6446: @node Locals, Structures, Threading Words, Words
6447: @section Locals
6448: @cindex locals
6449:
6450: Local variables can make Forth programming more enjoyable and Forth
6451: programs easier to read. Unfortunately, the locals of ANS Forth are
6452: laden with restrictions. Therefore, we provide not only the ANS Forth
6453: locals wordset, but also our own, more powerful locals wordset (we
6454: implemented the ANS Forth locals wordset through our locals wordset).
6455:
6456: The ideas in this section have also been published in the paper
6457: @cite{Automatic Scoping of Local Variables} by M. Anton Ertl, presented
6458: at EuroForth '94; it is available at
6459: @*@url{http://www.complang.tuwien.ac.at/papers/ertl94l.ps.gz}.
6460:
6461: @menu
6462: * Gforth locals::
6463: * ANS Forth locals::
6464: @end menu
6465:
6466: @node Gforth locals, ANS Forth locals, Locals, Locals
6467: @subsection Gforth locals
6468: @cindex Gforth locals
6469: @cindex locals, Gforth style
6470:
6471: Locals can be defined with
6472:
6473: @example
6474: @{ local1 local2 ... -- comment @}
6475: @end example
6476: or
6477: @example
6478: @{ local1 local2 ... @}
6479: @end example
6480:
6481: E.g.,
6482: @example
6483: : max @{ n1 n2 -- n3 @}
6484: n1 n2 > if
6485: n1
6486: else
6487: n2
6488: endif ;
6489: @end example
6490:
6491: The similarity of locals definitions with stack comments is intended. A
6492: locals definition often replaces the stack comment of a word. The order
6493: of the locals corresponds to the order in a stack comment and everything
6494: after the @code{--} is really a comment.
6495:
6496: This similarity has one disadvantage: It is too easy to confuse locals
6497: declarations with stack comments, causing bugs and making them hard to
6498: find. However, this problem can be avoided by appropriate coding
6499: conventions: Do not use both notations in the same program. If you do,
6500: they should be distinguished using additional means, e.g. by position.
6501:
6502: @cindex types of locals
6503: @cindex locals types
6504: The name of the local may be preceded by a type specifier, e.g.,
6505: @code{F:} for a floating point value:
6506:
6507: @example
6508: : CX* @{ F: Ar F: Ai F: Br F: Bi -- Cr Ci @}
6509: \ complex multiplication
6510: Ar Br f* Ai Bi f* f-
6511: Ar Bi f* Ai Br f* f+ ;
6512: @end example
6513:
6514: @cindex flavours of locals
6515: @cindex locals flavours
6516: @cindex value-flavoured locals
6517: @cindex variable-flavoured locals
6518: Gforth currently supports cells (@code{W:}, @code{W^}), doubles
6519: (@code{D:}, @code{D^}), floats (@code{F:}, @code{F^}) and characters
6520: (@code{C:}, @code{C^}) in two flavours: a value-flavoured local (defined
6521: with @code{W:}, @code{D:} etc.) produces its value and can be changed
6522: with @code{TO}. A variable-flavoured local (defined with @code{W^} etc.)
6523: produces its address (which becomes invalid when the variable's scope is
6524: left). E.g., the standard word @code{emit} can be defined in terms of
6525: @code{type} like this:
6526:
6527: @example
6528: : emit @{ C^ char* -- @}
6529: char* 1 type ;
6530: @end example
6531:
6532: @cindex default type of locals
6533: @cindex locals, default type
6534: A local without type specifier is a @code{W:} local. Both flavours of
6535: locals are initialized with values from the data or FP stack.
6536:
6537: Currently there is no way to define locals with user-defined data
6538: structures, but we are working on it.
6539:
6540: Gforth allows defining locals everywhere in a colon definition. This
6541: poses the following questions:
6542:
6543: @menu
6544: * Where are locals visible by name?::
6545: * How long do locals live?::
6546: * Programming Style::
6547: * Implementation::
6548: @end menu
6549:
6550: @node Where are locals visible by name?, How long do locals live?, Gforth locals, Gforth locals
6551: @subsubsection Where are locals visible by name?
6552: @cindex locals visibility
6553: @cindex visibility of locals
6554: @cindex scope of locals
6555:
6556: Basically, the answer is that locals are visible where you would expect
6557: it in block-structured languages, and sometimes a little longer. If you
6558: want to restrict the scope of a local, enclose its definition in
6559: @code{SCOPE}...@code{ENDSCOPE}.
6560:
6561: doc-scope
6562: doc-endscope
6563:
6564: These words behave like control structure words, so you can use them
6565: with @code{CS-PICK} and @code{CS-ROLL} to restrict the scope in
6566: arbitrary ways.
6567:
6568: If you want a more exact answer to the visibility question, here's the
6569: basic principle: A local is visible in all places that can only be
6570: reached through the definition of the local@footnote{In compiler
6571: construction terminology, all places dominated by the definition of the
6572: local.}. In other words, it is not visible in places that can be reached
6573: without going through the definition of the local. E.g., locals defined
6574: in @code{IF}...@code{ENDIF} are visible until the @code{ENDIF}, locals
6575: defined in @code{BEGIN}...@code{UNTIL} are visible after the
6576: @code{UNTIL} (until, e.g., a subsequent @code{ENDSCOPE}).
6577:
6578: The reasoning behind this solution is: We want to have the locals
6579: visible as long as it is meaningful. The user can always make the
6580: visibility shorter by using explicit scoping. In a place that can
6581: only be reached through the definition of a local, the meaning of a
6582: local name is clear. In other places it is not: How is the local
6583: initialized at the control flow path that does not contain the
6584: definition? Which local is meant, if the same name is defined twice in
6585: two independent control flow paths?
6586:
6587: This should be enough detail for nearly all users, so you can skip the
6588: rest of this section. If you really must know all the gory details and
6589: options, read on.
6590:
6591: In order to implement this rule, the compiler has to know which places
6592: are unreachable. It knows this automatically after @code{AHEAD},
6593: @code{AGAIN}, @code{EXIT} and @code{LEAVE}; in other cases (e.g., after
6594: most @code{THROW}s), you can use the word @code{UNREACHABLE} to tell the
6595: compiler that the control flow never reaches that place. If
6596: @code{UNREACHABLE} is not used where it could, the only consequence is
6597: that the visibility of some locals is more limited than the rule above
6598: says. If @code{UNREACHABLE} is used where it should not (i.e., if you
6599: lie to the compiler), buggy code will be produced.
6600:
6601: doc-unreachable
6602:
6603: Another problem with this rule is that at @code{BEGIN}, the compiler
6604: does not know which locals will be visible on the incoming
6605: back-edge. All problems discussed in the following are due to this
6606: ignorance of the compiler (we discuss the problems using @code{BEGIN}
6607: loops as examples; the discussion also applies to @code{?DO} and other
6608: loops). Perhaps the most insidious example is:
6609: @example
6610: AHEAD
6611: BEGIN
6612: x
6613: [ 1 CS-ROLL ] THEN
6614: @{ x @}
6615: ...
6616: UNTIL
6617: @end example
6618:
6619: This should be legal according to the visibility rule. The use of
6620: @code{x} can only be reached through the definition; but that appears
6621: textually below the use.
6622:
6623: From this example it is clear that the visibility rules cannot be fully
6624: implemented without major headaches. Our implementation treats common
6625: cases as advertised and the exceptions are treated in a safe way: The
6626: compiler makes a reasonable guess about the locals visible after a
6627: @code{BEGIN}; if it is too pessimistic, the
6628: user will get a spurious error about the local not being defined; if the
6629: compiler is too optimistic, it will notice this later and issue a
6630: warning. In the case above the compiler would complain about @code{x}
6631: being undefined at its use. You can see from the obscure examples in
6632: this section that it takes quite unusual control structures to get the
6633: compiler into trouble, and even then it will often do fine.
6634:
6635: If the @code{BEGIN} is reachable from above, the most optimistic guess
6636: is that all locals visible before the @code{BEGIN} will also be
6637: visible after the @code{BEGIN}. This guess is valid for all loops that
6638: are entered only through the @code{BEGIN}, in particular, for normal
6639: @code{BEGIN}...@code{WHILE}...@code{REPEAT} and
6640: @code{BEGIN}...@code{UNTIL} loops and it is implemented in our
6641: compiler. When the branch to the @code{BEGIN} is finally generated by
6642: @code{AGAIN} or @code{UNTIL}, the compiler checks the guess and
6643: warns the user if it was too optimistic:
6644: @example
6645: IF
6646: @{ x @}
6647: BEGIN
6648: \ x ?
6649: [ 1 cs-roll ] THEN
6650: ...
6651: UNTIL
6652: @end example
6653:
6654: Here, @code{x} lives only until the @code{BEGIN}, but the compiler
6655: optimistically assumes that it lives until the @code{THEN}. It notices
6656: this difference when it compiles the @code{UNTIL} and issues a
6657: warning. The user can avoid the warning, and make sure that @code{x}
6658: is not used in the wrong area by using explicit scoping:
6659: @example
6660: IF
6661: SCOPE
6662: @{ x @}
6663: ENDSCOPE
6664: BEGIN
6665: [ 1 cs-roll ] THEN
6666: ...
6667: UNTIL
6668: @end example
6669:
6670: Since the guess is optimistic, there will be no spurious error messages
6671: about undefined locals.
6672:
6673: If the @code{BEGIN} is not reachable from above (e.g., after
6674: @code{AHEAD} or @code{EXIT}), the compiler cannot even make an
6675: optimistic guess, as the locals visible after the @code{BEGIN} may be
6676: defined later. Therefore, the compiler assumes that no locals are
6677: visible after the @code{BEGIN}. However, the user can use
6678: @code{ASSUME-LIVE} to make the compiler assume that the same locals are
6679: visible at the BEGIN as at the point where the top control-flow stack
6680: item was created.
6681:
6682: doc-assume-live
6683:
6684: E.g.,
6685: @example
6686: @{ x @}
6687: AHEAD
6688: ASSUME-LIVE
6689: BEGIN
6690: x
6691: [ 1 CS-ROLL ] THEN
6692: ...
6693: UNTIL
6694: @end example
6695:
6696: Other cases where the locals are defined before the @code{BEGIN} can be
6697: handled by inserting an appropriate @code{CS-ROLL} before the
6698: @code{ASSUME-LIVE} (and changing the control-flow stack manipulation
6699: behind the @code{ASSUME-LIVE}).
6700:
6701: Cases where locals are defined after the @code{BEGIN} (but should be
6702: visible immediately after the @code{BEGIN}) can only be handled by
6703: rearranging the loop. E.g., the ``most insidious'' example above can be
6704: arranged into:
6705: @example
6706: BEGIN
6707: @{ x @}
6708: ... 0=
6709: WHILE
6710: x
6711: REPEAT
6712: @end example
6713:
6714: @node How long do locals live?, Programming Style, Where are locals visible by name?, Gforth locals
6715: @subsubsection How long do locals live?
6716: @cindex locals lifetime
6717: @cindex lifetime of locals
6718:
6719: The right answer for the lifetime question would be: A local lives at
6720: least as long as it can be accessed. For a value-flavoured local this
6721: means: until the end of its visibility. However, a variable-flavoured
6722: local could be accessed through its address far beyond its visibility
6723: scope. Ultimately, this would mean that such locals would have to be
6724: garbage collected. Since this entails un-Forth-like implementation
6725: complexities, I adopted the same cowardly solution as some other
6726: languages (e.g., C): The local lives only as long as it is visible;
6727: afterwards its address is invalid (and programs that access it
6728: afterwards are erroneous).
6729:
6730: @node Programming Style, Implementation, How long do locals live?, Gforth locals
6731: @subsubsection Programming Style
6732: @cindex locals programming style
6733: @cindex programming style, locals
6734:
6735: The freedom to define locals anywhere has the potential to change
6736: programming styles dramatically. In particular, the need to use the
6737: return stack for intermediate storage vanishes. Moreover, all stack
6738: manipulations (except @code{PICK}s and @code{ROLL}s with run-time
6739: determined arguments) can be eliminated: If the stack items are in the
6740: wrong order, just write a locals definition for all of them; then
6741: write the items in the order you want.
6742:
6743: This seems a little far-fetched and eliminating stack manipulations is
6744: unlikely to become a conscious programming objective. Still, the number
6745: of stack manipulations will be reduced dramatically if local variables
6746: are used liberally (e.g., compare @code{max} in @ref{Gforth locals} with
6747: a traditional implementation of @code{max}).
6748:
6749: This shows one potential benefit of locals: making Forth programs more
6750: readable. Of course, this benefit will only be realized if the
6751: programmers continue to honour the principle of factoring instead of
6752: using the added latitude to make the words longer.
6753:
6754: @cindex single-assignment style for locals
6755: Using @code{TO} can and should be avoided. Without @code{TO},
6756: every value-flavoured local has only a single assignment and many
6757: advantages of functional languages apply to Forth. I.e., programs are
6758: easier to analyse, to optimize and to read: It is clear from the
6759: definition what the local stands for, it does not turn into something
6760: different later.
6761:
6762: E.g., a definition using @code{TO} might look like this:
6763: @example
6764: : strcmp @{ addr1 u1 addr2 u2 -- n @}
6765: u1 u2 min 0
6766: ?do
6767: addr1 c@@ addr2 c@@ -
6768: ?dup-if
6769: unloop exit
6770: then
6771: addr1 char+ TO addr1
6772: addr2 char+ TO addr2
6773: loop
6774: u1 u2 - ;
6775: @end example
6776: Here, @code{TO} is used to update @code{addr1} and @code{addr2} at
6777: every loop iteration. @code{strcmp} is a typical example of the
6778: readability problems of using @code{TO}. When you start reading
6779: @code{strcmp}, you think that @code{addr1} refers to the start of the
6780: string. Only near the end of the loop you realize that it is something
6781: else.
6782:
6783: This can be avoided by defining two locals at the start of the loop that
6784: are initialized with the right value for the current iteration.
6785: @example
6786: : strcmp @{ addr1 u1 addr2 u2 -- n @}
6787: addr1 addr2
6788: u1 u2 min 0
6789: ?do @{ s1 s2 @}
6790: s1 c@@ s2 c@@ -
6791: ?dup-if
6792: unloop exit
6793: then
6794: s1 char+ s2 char+
6795: loop
6796: 2drop
6797: u1 u2 - ;
6798: @end example
6799: Here it is clear from the start that @code{s1} has a different value
6800: in every loop iteration.
6801:
6802: @node Implementation, , Programming Style, Gforth locals
6803: @subsubsection Implementation
6804: @cindex locals implementation
6805: @cindex implementation of locals
6806:
6807: @cindex locals stack
6808: Gforth uses an extra locals stack. The most compelling reason for
6809: this is that the return stack is not float-aligned; using an extra stack
6810: also eliminates the problems and restrictions of using the return stack
6811: as locals stack. Like the other stacks, the locals stack grows toward
6812: lower addresses. A few primitives allow an efficient implementation:
6813:
6814: doc-@local#
6815: doc-f@local#
6816: doc-laddr#
6817: doc-lp+!#
6818: doc-lp!
6819: doc->l
6820: doc-f>l
6821:
6822: In addition to these primitives, some specializations of these
6823: primitives for commonly occurring inline arguments are provided for
6824: efficiency reasons, e.g., @code{@@local0} as specialization of
6825: @code{@@local#} for the inline argument 0. The following compiling words
6826: compile the right specialized version, or the general version, as
6827: appropriate:
6828:
6829: doc-compile-@local
6830: doc-compile-f@local
6831: doc-compile-lp+!
6832:
6833: Combinations of conditional branches and @code{lp+!#} like
6834: @code{?branch-lp+!#} (the locals pointer is only changed if the branch
6835: is taken) are provided for efficiency and correctness in loops.
6836:
6837: A special area in the dictionary space is reserved for keeping the
6838: local variable names. @code{@{} switches the dictionary pointer to this
6839: area and @code{@}} switches it back and generates the locals
6840: initializing code. @code{W:} etc.@ are normal defining words. This
6841: special area is cleared at the start of every colon definition.
6842:
6843: @cindex word list for defining locals
6844: A special feature of Gforth's dictionary is used to implement the
6845: definition of locals without type specifiers: every word list (aka
6846: vocabulary) has its own methods for searching
6847: etc. (@pxref{Word Lists}). For the present purpose we defined a word list
6848: with a special search method: When it is searched for a word, it
6849: actually creates that word using @code{W:}. @code{@{} changes the search
6850: order to first search the word list containing @code{@}}, @code{W:} etc.,
6851: and then the word list for defining locals without type specifiers.
6852:
6853: The lifetime rules support a stack discipline within a colon
6854: definition: The lifetime of a local is either nested with other locals
6855: lifetimes or it does not overlap them.
6856:
6857: At @code{BEGIN}, @code{IF}, and @code{AHEAD} no code for locals stack
6858: pointer manipulation is generated. Between control structure words
6859: locals definitions can push locals onto the locals stack. @code{AGAIN}
6860: is the simplest of the other three control flow words. It has to
6861: restore the locals stack depth of the corresponding @code{BEGIN}
6862: before branching. The code looks like this:
6863: @format
6864: @code{lp+!#} current-locals-size @minus{} dest-locals-size
6865: @code{branch} <begin>
6866: @end format
6867:
6868: @code{UNTIL} is a little more complicated: If it branches back, it
6869: must adjust the stack just like @code{AGAIN}. But if it falls through,
6870: the locals stack must not be changed. The compiler generates the
6871: following code:
6872: @format
6873: @code{?branch-lp+!#} <begin> current-locals-size @minus{} dest-locals-size
6874: @end format
6875: The locals stack pointer is only adjusted if the branch is taken.
6876:
6877: @code{THEN} can produce somewhat inefficient code:
6878: @format
6879: @code{lp+!#} current-locals-size @minus{} orig-locals-size
6880: <orig target>:
6881: @code{lp+!#} orig-locals-size @minus{} new-locals-size
6882: @end format
6883: The second @code{lp+!#} adjusts the locals stack pointer from the
6884: level at the @i{orig} point to the level after the @code{THEN}. The
6885: first @code{lp+!#} adjusts the locals stack pointer from the current
6886: level to the level at the orig point, so the complete effect is an
6887: adjustment from the current level to the right level after the
6888: @code{THEN}.
6889:
6890: @cindex locals information on the control-flow stack
6891: @cindex control-flow stack items, locals information
6892: In a conventional Forth implementation a dest control-flow stack entry
6893: is just the target address and an orig entry is just the address to be
6894: patched. Our locals implementation adds a word list to every orig or dest
6895: item. It is the list of locals visible (or assumed visible) at the point
6896: described by the entry. Our implementation also adds a tag to identify
6897: the kind of entry, in particular to differentiate between live and dead
6898: (reachable and unreachable) orig entries.
6899:
6900: A few unusual operations have to be performed on locals word lists:
6901:
6902: doc-common-list
6903: doc-sub-list?
6904: doc-list-size
6905:
6906: Several features of our locals word list implementation make these
6907: operations easy to implement: The locals word lists are organised as
6908: linked lists; the tails of these lists are shared, if the lists
6909: contain some of the same locals; and the address of a name is greater
6910: than the address of the names behind it in the list.
6911:
6912: Another important implementation detail is the variable
6913: @code{dead-code}. It is used by @code{BEGIN} and @code{THEN} to
6914: determine if they can be reached directly or only through the branch
6915: that they resolve. @code{dead-code} is set by @code{UNREACHABLE},
6916: @code{AHEAD}, @code{EXIT} etc., and cleared at the start of a colon
6917: definition, by @code{BEGIN} and usually by @code{THEN}.
6918:
6919: Counted loops are similar to other loops in most respects, but
6920: @code{LEAVE} requires special attention: It performs basically the same
6921: service as @code{AHEAD}, but it does not create a control-flow stack
6922: entry. Therefore the information has to be stored elsewhere;
6923: traditionally, the information was stored in the target fields of the
6924: branches created by the @code{LEAVE}s, by organizing these fields into a
6925: linked list. Unfortunately, this clever trick does not provide enough
6926: space for storing our extended control flow information. Therefore, we
6927: introduce another stack, the leave stack. It contains the control-flow
6928: stack entries for all unresolved @code{LEAVE}s.
6929:
6930: Local names are kept until the end of the colon definition, even if
6931: they are no longer visible in any control-flow path. In a few cases
6932: this may lead to increased space needs for the locals name area, but
6933: usually less than reclaiming this space would cost in code size.
6934:
6935:
6936: @node ANS Forth locals, , Gforth locals, Locals
6937: @subsection ANS Forth locals
6938: @cindex locals, ANS Forth style
6939:
6940: The ANS Forth locals wordset does not define a syntax for locals, but
6941: words that make it possible to define various syntaxes. One of the
6942: possible syntaxes is a subset of the syntax we used in the Gforth locals
6943: wordset, i.e.:
6944:
6945: @example
6946: @{ local1 local2 ... -- comment @}
6947: @end example
6948: @noindent
6949: or
6950: @example
6951: @{ local1 local2 ... @}
6952: @end example
6953:
6954: The order of the locals corresponds to the order in a stack comment. The
6955: restrictions are:
6956:
6957: @itemize @bullet
6958: @item
6959: Locals can only be cell-sized values (no type specifiers are allowed).
6960: @item
6961: Locals can be defined only outside control structures.
6962: @item
6963: Locals can interfere with explicit usage of the return stack. For the
6964: exact (and long) rules, see the standard. If you don't use return stack
6965: accessing words in a definition using locals, you will be all right. The
6966: purpose of this rule is to make locals implementation on the return
6967: stack easier.
6968: @item
6969: The whole definition must be in one line.
6970: @end itemize
6971:
6972: Locals defined in this way behave like @code{VALUE}s (@xref{Simple
6973: Defining Words}). I.e., they are initialized from the stack. Using their
6974: name produces their value. Their value can be changed using @code{TO}.
6975:
6976: Since this syntax is supported by Gforth directly, you need not do
6977: anything to use it. If you want to port a program using this syntax to
6978: another ANS Forth system, use @file{compat/anslocal.fs} to implement the
6979: syntax on the other system.
6980:
6981: Note that a syntax shown in the standard, section A.13 looks
6982: similar, but is quite different in having the order of locals
6983: reversed. Beware!
6984:
6985: The ANS Forth locals wordset itself consists of a word:
6986:
6987: doc-(local)
6988:
6989: The ANS Forth locals extension wordset defines a syntax using @code{locals|}, but it is so
6990: awful that we strongly recommend not to use it. We have implemented this
6991: syntax to make porting to Gforth easy, but do not document it here. The
6992: problem with this syntax is that the locals are defined in an order
6993: reversed with respect to the standard stack comment notation, making
6994: programs harder to read, and easier to misread and miswrite. The only
6995: merit of this syntax is that it is easy to implement using the ANS Forth
6996: locals wordset.
6997:
6998:
6999: @c ----------------------------------------------------------
7000: @node Structures, Object-oriented Forth, Locals, Words
7001: @section Structures
7002: @cindex structures
7003: @cindex records
7004:
7005: This section presents the structure package that comes with Gforth. A
7006: version of the package implemented in ANS Forth is available in
7007: @file{compat/struct.fs}. This package was inspired by a posting on
7008: comp.lang.forth in 1989 (unfortunately I don't remember, by whom;
7009: possibly John Hayes). A version of this section has been published in
7010: ???. Marcel Hendrix provided helpful comments.
7011:
7012: @menu
7013: * Why explicit structure support?::
7014: * Structure Usage::
7015: * Structure Naming Convention::
7016: * Structure Implementation::
7017: * Structure Glossary::
7018: @end menu
7019:
7020: @node Why explicit structure support?, Structure Usage, Structures, Structures
7021: @subsection Why explicit structure support?
7022:
7023: @cindex address arithmetic for structures
7024: @cindex structures using address arithmetic
7025: If we want to use a structure containing several fields, we could simply
7026: reserve memory for it, and access the fields using address arithmetic
7027: (@pxref{Address arithmetic}). As an example, consider a structure with
7028: the following fields
7029:
7030: @table @code
7031: @item a
7032: is a float
7033: @item b
7034: is a cell
7035: @item c
7036: is a float
7037: @end table
7038:
7039: Given the (float-aligned) base address of the structure we get the
7040: address of the field
7041:
7042: @table @code
7043: @item a
7044: without doing anything further.
7045: @item b
7046: with @code{float+}
7047: @item c
7048: with @code{float+ cell+ faligned}
7049: @end table
7050:
7051: It is easy to see that this can become quite tiring.
7052:
7053: Moreover, it is not very readable, because seeing a
7054: @code{cell+} tells us neither which kind of structure is
7055: accessed nor what field is accessed; we have to somehow infer the kind
7056: of structure, and then look up in the documentation, which field of
7057: that structure corresponds to that offset.
7058:
7059: Finally, this kind of address arithmetic also causes maintenance
7060: troubles: If you add or delete a field somewhere in the middle of the
7061: structure, you have to find and change all computations for the fields
7062: afterwards.
7063:
7064: So, instead of using @code{cell+} and friends directly, how
7065: about storing the offsets in constants:
7066:
7067: @example
7068: 0 constant a-offset
7069: 0 float+ constant b-offset
7070: 0 float+ cell+ faligned c-offset
7071: @end example
7072:
7073: Now we can get the address of field @code{x} with @code{x-offset
7074: +}. This is much better in all respects. Of course, you still
7075: have to change all later offset definitions if you add a field. You can
7076: fix this by declaring the offsets in the following way:
7077:
7078: @example
7079: 0 constant a-offset
7080: a-offset float+ constant b-offset
7081: b-offset cell+ faligned constant c-offset
7082: @end example
7083:
7084: Since we always use the offsets with @code{+}, we could use a defining
7085: word @code{cfield} that includes the @code{+} in the action of the
7086: defined word:
7087:
7088: @example
7089: : cfield ( n "name" -- )
7090: create ,
7091: does> ( name execution: addr1 -- addr2 )
7092: @@ + ;
7093:
7094: 0 cfield a
7095: 0 a float+ cfield b
7096: 0 b cell+ faligned cfield c
7097: @end example
7098:
7099: Instead of @code{x-offset +}, we now simply write @code{x}.
7100:
7101: The structure field words now can be used quite nicely. However,
7102: their definition is still a bit cumbersome: We have to repeat the
7103: name, the information about size and alignment is distributed before
7104: and after the field definitions etc. The structure package presented
7105: here addresses these problems.
7106:
7107: @node Structure Usage, Structure Naming Convention, Why explicit structure support?, Structures
7108: @subsection Structure Usage
7109: @cindex structure usage
7110:
7111: @cindex @code{field} usage
7112: @cindex @code{struct} usage
7113: @cindex @code{end-struct} usage
7114: You can define a structure for a (data-less) linked list with:
7115: @example
7116: struct
7117: cell% field list-next
7118: end-struct list%
7119: @end example
7120:
7121: With the address of the list node on the stack, you can compute the
7122: address of the field that contains the address of the next node with
7123: @code{list-next}. E.g., you can determine the length of a list
7124: with:
7125:
7126: @example
7127: : list-length ( list -- n )
7128: \ "list" is a pointer to the first element of a linked list
7129: \ "n" is the length of the list
7130: 0 BEGIN ( list1 n1 )
7131: over
7132: WHILE ( list1 n1 )
7133: 1+ swap list-next @@ swap
7134: REPEAT
7135: nip ;
7136: @end example
7137:
7138: You can reserve memory for a list node in the dictionary with
7139: @code{list% %allot}, which leaves the address of the list node on the
7140: stack. For the equivalent allocation on the heap you can use @code{list%
7141: %alloc} (or, for an @code{allocate}-like stack effect (i.e., with ior),
7142: use @code{list% %allocate}). You can get the the size of a list
7143: node with @code{list% %size} and its alignment with @code{list%
7144: %alignment}.
7145:
7146: Note that in ANS Forth the body of a @code{create}d word is
7147: @code{aligned} but not necessarily @code{faligned};
7148: therefore, if you do a:
7149: @example
7150: create @emph{name} foo% %allot
7151: @end example
7152:
7153: @noindent
7154: then the memory alloted for @code{foo%} is
7155: guaranteed to start at the body of @code{@emph{name}} only if
7156: @code{foo%} contains only character, cell and double fields.
7157:
7158: @cindex strcutures containing structures
7159: You can include a structure @code{foo%} as a field of
7160: another structure, like this:
7161: @example
7162: struct
7163: ...
7164: foo% field ...
7165: ...
7166: end-struct ...
7167: @end example
7168:
7169: @cindex structure extension
7170: @cindex extended records
7171: Instead of starting with an empty structure, you can extend an
7172: existing structure. E.g., a plain linked list without data, as defined
7173: above, is hardly useful; You can extend it to a linked list of integers,
7174: like this:@footnote{This feature is also known as @emph{extended
7175: records}. It is the main innovation in the Oberon language; in other
7176: words, adding this feature to Modula-2 led Wirth to create a new
7177: language, write a new compiler etc. Adding this feature to Forth just
7178: required a few lines of code.}
7179:
7180: @example
7181: list%
7182: cell% field intlist-int
7183: end-struct intlist%
7184: @end example
7185:
7186: @code{intlist%} is a structure with two fields:
7187: @code{list-next} and @code{intlist-int}.
7188:
7189: @cindex structures containing arrays
7190: You can specify an array type containing @emph{n} elements of
7191: type @code{foo%} like this:
7192:
7193: @example
7194: foo% @emph{n} *
7195: @end example
7196:
7197: You can use this array type in any place where you can use a normal
7198: type, e.g., when defining a @code{field}, or with
7199: @code{%allot}.
7200:
7201: @cindex first field optimization
7202: The first field is at the base address of a structure and the word
7203: for this field (e.g., @code{list-next}) actually does not change
7204: the address on the stack. You may be tempted to leave it away in the
7205: interest of run-time and space efficiency. This is not necessary,
7206: because the structure package optimizes this case and compiling such
7207: words does not generate any code. So, in the interest of readability
7208: and maintainability you should include the word for the field when
7209: accessing the field.
7210:
7211: @node Structure Naming Convention, Structure Implementation, Structure Usage, Structures
7212: @subsection Structure Naming Convention
7213: @cindex structure naming convention
7214:
7215: The field names that come to (my) mind are often quite generic, and,
7216: if used, would cause frequent name clashes. E.g., many structures
7217: probably contain a @code{counter} field. The structure names
7218: that come to (my) mind are often also the logical choice for the names
7219: of words that create such a structure.
7220:
7221: Therefore, I have adopted the following naming conventions:
7222:
7223: @itemize @bullet
7224: @cindex field naming convention
7225: @item
7226: The names of fields are of the form
7227: @code{@emph{struct}-@emph{field}}, where
7228: @code{@emph{struct}} is the basic name of the structure, and
7229: @code{@emph{field}} is the basic name of the field. You can
7230: think of field words as converting the (address of the)
7231: structure into the (address of the) field.
7232:
7233: @cindex structure naming convention
7234: @item
7235: The names of structures are of the form
7236: @code{@emph{struct}%}, where
7237: @code{@emph{struct}} is the basic name of the structure.
7238: @end itemize
7239:
7240: This naming convention does not work that well for fields of extended
7241: structures; e.g., the integer list structure has a field
7242: @code{intlist-int}, but has @code{list-next}, not
7243: @code{intlist-next}.
7244:
7245: @node Structure Implementation, Structure Glossary, Structure Naming Convention, Structures
7246: @subsection Structure Implementation
7247: @cindex structure implementation
7248: @cindex implementation of structures
7249:
7250: The central idea in the implementation is to pass the data about the
7251: structure being built on the stack, not in some global
7252: variable. Everything else falls into place naturally once this design
7253: decision is made.
7254:
7255: The type description on the stack is of the form @emph{align
7256: size}. Keeping the size on the top-of-stack makes dealing with arrays
7257: very simple.
7258:
7259: @code{field} is a defining word that uses @code{Create}
7260: and @code{DOES>}. The body of the field contains the offset
7261: of the field, and the normal @code{DOES>} action is simply:
7262:
7263: @example
7264: @ +
7265: @end example
7266:
7267: @noindent
7268: i.e., add the offset to the address, giving the stack effect
7269: @i{addr1 -- addr2} for a field.
7270:
7271: @cindex first field optimization, implementation
7272: This simple structure is slightly complicated by the optimization
7273: for fields with offset 0, which requires a different
7274: @code{DOES>}-part (because we cannot rely on there being
7275: something on the stack if such a field is invoked during
7276: compilation). Therefore, we put the different @code{DOES>}-parts
7277: in separate words, and decide which one to invoke based on the
7278: offset. For a zero offset, the field is basically a noop; it is
7279: immediate, and therefore no code is generated when it is compiled.
7280:
7281: @node Structure Glossary, , Structure Implementation, Structures
7282: @subsection Structure Glossary
7283: @cindex structure glossary
7284:
7285: doc-%align
7286: doc-%alignment
7287: doc-%alloc
7288: doc-%allocate
7289: doc-%allot
7290: doc-cell%
7291: doc-char%
7292: doc-dfloat%
7293: doc-double%
7294: doc-end-struct
7295: doc-field
7296: doc-float%
7297: doc-naligned
7298: doc-sfloat%
7299: doc-%size
7300: doc-struct
7301:
7302: @c -------------------------------------------------------------
7303: @node Object-oriented Forth, Passing Commands to the OS, Structures, Words
7304: @section Object-oriented Forth
7305:
7306: Gforth comes with three packages for object-oriented programming:
7307: @file{objects.fs}, @file{oof.fs}, and @file{mini-oof.fs}; none of them
7308: is preloaded, so you have to @code{include} them before use. The most
7309: important differences between these packages (and others) are discussed
7310: in @ref{Comparison with other object models}. All packages are written
7311: in ANS Forth and can be used with any other ANS Forth.
7312:
7313: @menu
7314: * Why object-oriented programming?::
7315: * Object-Oriented Terminology::
7316: * Objects::
7317: * OOF::
7318: * Mini-OOF::
7319: * Comparison with other object models::
7320: @end menu
7321:
7322:
7323: @node Why object-oriented programming?, Object-Oriented Terminology, , Object-oriented Forth
7324: @subsubsection Why object-oriented programming?
7325: @cindex object-oriented programming motivation
7326: @cindex motivation for object-oriented programming
7327:
7328: Often we have to deal with several data structures (@emph{objects}),
7329: that have to be treated similarly in some respects, but differently in
7330: others. Graphical objects are the textbook example: circles, triangles,
7331: dinosaurs, icons, and others, and we may want to add more during program
7332: development. We want to apply some operations to any graphical object,
7333: e.g., @code{draw} for displaying it on the screen. However, @code{draw}
7334: has to do something different for every kind of object.
7335: @comment TODO add some other operations eg perimeter, area
7336: @comment and tie in to concrete examples later..
7337:
7338: We could implement @code{draw} as a big @code{CASE}
7339: control structure that executes the appropriate code depending on the
7340: kind of object to be drawn. This would be not be very elegant, and,
7341: moreover, we would have to change @code{draw} every time we add
7342: a new kind of graphical object (say, a spaceship).
7343:
7344: What we would rather do is: When defining spaceships, we would tell
7345: the system: ``Here's how you @code{draw} a spaceship; you figure
7346: out the rest''.
7347:
7348: This is the problem that all systems solve that (rightfully) call
7349: themselves object-oriented; the object-oriented packages presented here
7350: solve this problem (and not much else).
7351: @comment TODO ?list properties of oo systems.. oo vs o-based?
7352:
7353: @node Object-Oriented Terminology, Objects, Why object-oriented programming?, Object-oriented Forth
7354: @subsubsection Object-Oriented Terminology
7355: @cindex object-oriented terminology
7356: @cindex terminology for object-oriented programming
7357:
7358: This section is mainly for reference, so you don't have to understand
7359: all of it right away. The terminology is mainly Smalltalk-inspired. In
7360: short:
7361:
7362: @table @emph
7363: @cindex class
7364: @item class
7365: a data structure definition with some extras.
7366:
7367: @cindex object
7368: @item object
7369: an instance of the data structure described by the class definition.
7370:
7371: @cindex instance variables
7372: @item instance variables
7373: fields of the data structure.
7374:
7375: @cindex selector
7376: @cindex method selector
7377: @cindex virtual function
7378: @item selector
7379: (or @emph{method selector}) a word (e.g.,
7380: @code{draw}) that performs an operation on a variety of data
7381: structures (classes). A selector describes @emph{what} operation to
7382: perform. In C++ terminology: a (pure) virtual function.
7383:
7384: @cindex method
7385: @item method
7386: the concrete definition that performs the operation
7387: described by the selector for a specific class. A method specifies
7388: @emph{how} the operation is performed for a specific class.
7389:
7390: @cindex selector invocation
7391: @cindex message send
7392: @cindex invoking a selector
7393: @item selector invocation
7394: a call of a selector. One argument of the call (the TOS (top-of-stack))
7395: is used for determining which method is used. In Smalltalk terminology:
7396: a message (consisting of the selector and the other arguments) is sent
7397: to the object.
7398:
7399: @cindex receiving object
7400: @item receiving object
7401: the object used for determining the method executed by a selector
7402: invocation. In the @file{objects.fs} model, it is the object that is on
7403: the TOS when the selector is invoked. (@emph{Receiving} comes from
7404: the Smalltalk @emph{message} terminology.)
7405:
7406: @cindex child class
7407: @cindex parent class
7408: @cindex inheritance
7409: @item child class
7410: a class that has (@emph{inherits}) all properties (instance variables,
7411: selectors, methods) from a @emph{parent class}. In Smalltalk
7412: terminology: The subclass inherits from the superclass. In C++
7413: terminology: The derived class inherits from the base class.
7414:
7415: @end table
7416:
7417: @c If you wonder about the message sending terminology, it comes from
7418: @c a time when each object had it's own task and objects communicated via
7419: @c message passing; eventually the Smalltalk developers realized that
7420: @c they can do most things through simple (indirect) calls. They kept the
7421: @c terminology.
7422:
7423:
7424: @node Objects, OOF, Object-Oriented Terminology, Object-oriented Forth
7425: @subsection The @file{objects.fs} model
7426: @cindex objects
7427: @cindex object-oriented programming
7428:
7429: @cindex @file{objects.fs}
7430: @cindex @file{oof.fs}
7431:
7432: This section describes the @file{objects.fs} package. This material also
7433: has been published in @cite{Yet Another Forth Objects Package} by Anton
7434: Ertl and appeared in Forth Dimensions 19(2), pages 37--43
7435: (@url{http://www.complang.tuwien.ac.at/forth/objects/objects.html}).
7436: @c McKewan's and Zsoter's packages
7437:
7438: This section assumes that you have read @ref{Structures}.
7439:
7440: The techniques on which this model is based have been used to implement
7441: the parser generator, Gray, and have also been used in Gforth for
7442: implementing the various flavours of word lists (hashed or not,
7443: case-sensitive or not, special-purpose word lists for locals etc.).
7444:
7445:
7446: @menu
7447: * Properties of the Objects model::
7448: * Basic Objects Usage::
7449: * The Objects base class::
7450: * Creating objects::
7451: * Object-Oriented Programming Style::
7452: * Class Binding::
7453: * Method conveniences::
7454: * Classes and Scoping::
7455: * Dividing classes::
7456: * Object Interfaces::
7457: * Objects Implementation::
7458: * Objects Glossary::
7459: @end menu
7460:
7461: Marcel Hendrix provided helpful comments on this section. Andras Zsoter
7462: and Bernd Paysan helped me with the related works section.
7463:
7464: @node Properties of the Objects model, Basic Objects Usage, Objects, Objects
7465: @subsubsection Properties of the @file{objects.fs} model
7466: @cindex @file{objects.fs} properties
7467:
7468: @itemize @bullet
7469: @item
7470: It is straightforward to pass objects on the stack. Passing
7471: selectors on the stack is a little less convenient, but possible.
7472:
7473: @item
7474: Objects are just data structures in memory, and are referenced by their
7475: address. You can create words for objects with normal defining words
7476: like @code{constant}. Likewise, there is no difference between instance
7477: variables that contain objects and those that contain other data.
7478:
7479: @item
7480: Late binding is efficient and easy to use.
7481:
7482: @item
7483: It avoids parsing, and thus avoids problems with state-smartness
7484: and reduced extensibility; for convenience there are a few parsing
7485: words, but they have non-parsing counterparts. There are also a few
7486: defining words that parse. This is hard to avoid, because all standard
7487: defining words parse (except @code{:noname}); however, such
7488: words are not as bad as many other parsing words, because they are not
7489: state-smart.
7490:
7491: @item
7492: It does not try to incorporate everything. It does a few things and does
7493: them well (IMO). In particular, this model was not designed to support
7494: information hiding (although it has features that may help); you can use
7495: a separate package for achieving this.
7496:
7497: @item
7498: It is layered; you don't have to learn and use all features to use this
7499: model. Only a few features are necessary (@xref{Basic Objects Usage},
7500: @xref{The Objects base class}, @xref{Creating objects}.), the others
7501: are optional and independent of each other.
7502:
7503: @item
7504: An implementation in ANS Forth is available.
7505:
7506: @end itemize
7507:
7508:
7509: @node Basic Objects Usage, The Objects base class, Properties of the Objects model, Objects
7510: @subsubsection Basic @file{objects.fs} Usage
7511: @cindex basic objects usage
7512: @cindex objects, basic usage
7513:
7514: You can define a class for graphical objects like this:
7515:
7516: @cindex @code{class} usage
7517: @cindex @code{end-class} usage
7518: @cindex @code{selector} usage
7519: @example
7520: object class \ "object" is the parent class
7521: selector draw ( x y graphical -- )
7522: end-class graphical
7523: @end example
7524:
7525: This code defines a class @code{graphical} with an
7526: operation @code{draw}. We can perform the operation
7527: @code{draw} on any @code{graphical} object, e.g.:
7528:
7529: @example
7530: 100 100 t-rex draw
7531: @end example
7532:
7533: @noindent
7534: where @code{t-rex} is a word (say, a constant) that produces a
7535: graphical object.
7536:
7537: @comment TODO add a 2nd operation eg perimeter.. and use for
7538: @comment a concrete example
7539:
7540: @cindex abstract class
7541: How do we create a graphical object? With the present definitions,
7542: we cannot create a useful graphical object. The class
7543: @code{graphical} describes graphical objects in general, but not
7544: any concrete graphical object type (C++ users would call it an
7545: @emph{abstract class}); e.g., there is no method for the selector
7546: @code{draw} in the class @code{graphical}.
7547:
7548: For concrete graphical objects, we define child classes of the
7549: class @code{graphical}, e.g.:
7550:
7551: @cindex @code{overrides} usage
7552: @cindex @code{field} usage in class definition
7553: @example
7554: graphical class \ "graphical" is the parent class
7555: cell% field circle-radius
7556:
7557: :noname ( x y circle -- )
7558: circle-radius @@ draw-circle ;
7559: overrides draw
7560:
7561: :noname ( n-radius circle -- )
7562: circle-radius ! ;
7563: overrides construct
7564:
7565: end-class circle
7566: @end example
7567:
7568: Here we define a class @code{circle} as a child of @code{graphical},
7569: with field @code{circle-radius} (which behaves just like a field
7570: (@pxref{Structures}); it defines (using @code{overrides}) new methods
7571: for the selectors @code{draw} and @code{construct} (@code{construct} is
7572: defined in @code{object}, the parent class of @code{graphical}).
7573:
7574: Now we can create a circle on the heap (i.e.,
7575: @code{allocate}d memory) with:
7576:
7577: @cindex @code{heap-new} usage
7578: @example
7579: 50 circle heap-new constant my-circle
7580: @end example
7581:
7582: @noindent
7583: @code{heap-new} invokes @code{construct}, thus
7584: initializing the field @code{circle-radius} with 50. We can draw
7585: this new circle at (100,100) with:
7586:
7587: @example
7588: 100 100 my-circle draw
7589: @end example
7590:
7591: @cindex selector invocation, restrictions
7592: @cindex class definition, restrictions
7593: Note: You can only invoke a selector if the object on the TOS
7594: (the receiving object) belongs to the class where the selector was
7595: defined or one of its descendents; e.g., you can invoke
7596: @code{draw} only for objects belonging to @code{graphical}
7597: or its descendents (e.g., @code{circle}). Immediately before
7598: @code{end-class}, the search order has to be the same as
7599: immediately after @code{class}.
7600:
7601: @node The Objects base class, Creating objects, Basic Objects Usage, Objects
7602: @subsubsection The @file{object.fs} base class
7603: @cindex @code{object} class
7604:
7605: When you define a class, you have to specify a parent class. So how do
7606: you start defining classes? There is one class available from the start:
7607: @code{object}. It is ancestor for all classes and so is the
7608: only class that has no parent. It has two selectors: @code{construct}
7609: and @code{print}.
7610:
7611: @node Creating objects, Object-Oriented Programming Style, The Objects base class, Objects
7612: @subsubsection Creating objects
7613: @cindex creating objects
7614: @cindex object creation
7615: @cindex object allocation options
7616:
7617: @cindex @code{heap-new} discussion
7618: @cindex @code{dict-new} discussion
7619: @cindex @code{construct} discussion
7620: You can create and initialize an object of a class on the heap with
7621: @code{heap-new} ( ... class -- object ) and in the dictionary
7622: (allocation with @code{allot}) with @code{dict-new} (
7623: ... class -- object ). Both words invoke @code{construct}, which
7624: consumes the stack items indicated by "..." above.
7625:
7626: @cindex @code{init-object} discussion
7627: @cindex @code{class-inst-size} discussion
7628: If you want to allocate memory for an object yourself, you can get its
7629: alignment and size with @code{class-inst-size 2@@} ( class --
7630: align size ). Once you have memory for an object, you can initialize
7631: it with @code{init-object} ( ... class object -- );
7632: @code{construct} does only a part of the necessary work.
7633:
7634: @node Object-Oriented Programming Style, Class Binding, Creating objects, Objects
7635: @subsubsection Object-Oriented Programming Style
7636: @cindex object-oriented programming style
7637:
7638: This section is not exhaustive.
7639:
7640: @cindex stack effects of selectors
7641: @cindex selectors and stack effects
7642: In general, it is a good idea to ensure that all methods for the
7643: same selector have the same stack effect: when you invoke a selector,
7644: you often have no idea which method will be invoked, so, unless all
7645: methods have the same stack effect, you will not know the stack effect
7646: of the selector invocation.
7647:
7648: One exception to this rule is methods for the selector
7649: @code{construct}. We know which method is invoked, because we
7650: specify the class to be constructed at the same place. Actually, I
7651: defined @code{construct} as a selector only to give the users a
7652: convenient way to specify initialization. The way it is used, a
7653: mechanism different from selector invocation would be more natural
7654: (but probably would take more code and more space to explain).
7655:
7656: @node Class Binding, Method conveniences, Object-Oriented Programming Style, Objects
7657: @subsubsection Class Binding
7658: @cindex class binding
7659: @cindex early binding
7660:
7661: @cindex late binding
7662: Normal selector invocations determine the method at run-time depending
7663: on the class of the receiving object. This run-time selection is called
7664: @i{late binding}.
7665:
7666: Sometimes it's preferable to invoke a different method. For example,
7667: you might want to use the simple method for @code{print}ing
7668: @code{object}s instead of the possibly long-winded @code{print} method
7669: of the receiver class. You can achieve this by replacing the invocation
7670: of @code{print} with:
7671:
7672: @cindex @code{[bind]} usage
7673: @example
7674: [bind] object print
7675: @end example
7676:
7677: @noindent
7678: in compiled code or:
7679:
7680: @cindex @code{bind} usage
7681: @example
7682: bind object print
7683: @end example
7684:
7685: @cindex class binding, alternative to
7686: @noindent
7687: in interpreted code. Alternatively, you can define the method with a
7688: name (e.g., @code{print-object}), and then invoke it through the
7689: name. Class binding is just a (often more convenient) way to achieve
7690: the same effect; it avoids name clutter and allows you to invoke
7691: methods directly without naming them first.
7692:
7693: @cindex superclass binding
7694: @cindex parent class binding
7695: A frequent use of class binding is this: When we define a method
7696: for a selector, we often want the method to do what the selector does
7697: in the parent class, and a little more. There is a special word for
7698: this purpose: @code{[parent]}; @code{[parent]
7699: @emph{selector}} is equivalent to @code{[bind] @emph{parent
7700: selector}}, where @code{@emph{parent}} is the parent
7701: class of the current class. E.g., a method definition might look like:
7702:
7703: @cindex @code{[parent]} usage
7704: @example
7705: :noname
7706: dup [parent] foo \ do parent's foo on the receiving object
7707: ... \ do some more
7708: ; overrides foo
7709: @end example
7710:
7711: @cindex class binding as optimization
7712: In @cite{Object-oriented programming in ANS Forth} (Forth Dimensions,
7713: March 1997), Andrew McKewan presents class binding as an optimization
7714: technique. I recommend not using it for this purpose unless you are in
7715: an emergency. Late binding is pretty fast with this model anyway, so the
7716: benefit of using class binding is small; the cost of using class binding
7717: where it is not appropriate is reduced maintainability.
7718:
7719: While we are at programming style questions: You should bind
7720: selectors only to ancestor classes of the receiving object. E.g., say,
7721: you know that the receiving object is of class @code{foo} or its
7722: descendents; then you should bind only to @code{foo} and its
7723: ancestors.
7724:
7725: @node Method conveniences, Classes and Scoping, Class Binding, Objects
7726: @subsubsection Method conveniences
7727: @cindex method conveniences
7728:
7729: In a method you usually access the receiving object pretty often. If
7730: you define the method as a plain colon definition (e.g., with
7731: @code{:noname}), you may have to do a lot of stack
7732: gymnastics. To avoid this, you can define the method with @code{m:
7733: ... ;m}. E.g., you could define the method for
7734: @code{draw}ing a @code{circle} with
7735:
7736: @cindex @code{this} usage
7737: @cindex @code{m:} usage
7738: @cindex @code{;m} usage
7739: @example
7740: m: ( x y circle -- )
7741: ( x y ) this circle-radius @@ draw-circle ;m
7742: @end example
7743:
7744: @cindex @code{exit} in @code{m: ... ;m}
7745: @cindex @code{exitm} discussion
7746: @cindex @code{catch} in @code{m: ... ;m}
7747: When this method is executed, the receiver object is removed from the
7748: stack; you can access it with @code{this} (admittedly, in this
7749: example the use of @code{m: ... ;m} offers no advantage). Note
7750: that I specify the stack effect for the whole method (i.e. including
7751: the receiver object), not just for the code between @code{m:}
7752: and @code{;m}. You cannot use @code{exit} in
7753: @code{m:...;m}; instead, use
7754: @code{exitm}.@footnote{Moreover, for any word that calls
7755: @code{catch} and was defined before loading
7756: @code{objects.fs}, you have to redefine it like I redefined
7757: @code{catch}: @code{: catch this >r catch r> to-this ;}}
7758:
7759: @cindex @code{inst-var} usage
7760: You will frequently use sequences of the form @code{this
7761: @emph{field}} (in the example above: @code{this
7762: circle-radius}). If you use the field only in this way, you can
7763: define it with @code{inst-var} and eliminate the
7764: @code{this} before the field name. E.g., the @code{circle}
7765: class above could also be defined with:
7766:
7767: @example
7768: graphical class
7769: cell% inst-var radius
7770:
7771: m: ( x y circle -- )
7772: radius @@ draw-circle ;m
7773: overrides draw
7774:
7775: m: ( n-radius circle -- )
7776: radius ! ;m
7777: overrides construct
7778:
7779: end-class circle
7780: @end example
7781:
7782: @code{radius} can only be used in @code{circle} and its
7783: descendent classes and inside @code{m:...;m}.
7784:
7785: @cindex @code{inst-value} usage
7786: You can also define fields with @code{inst-value}, which is
7787: to @code{inst-var} what @code{value} is to
7788: @code{variable}. You can change the value of such a field with
7789: @code{[to-inst]}. E.g., we could also define the class
7790: @code{circle} like this:
7791:
7792: @example
7793: graphical class
7794: inst-value radius
7795:
7796: m: ( x y circle -- )
7797: radius draw-circle ;m
7798: overrides draw
7799:
7800: m: ( n-radius circle -- )
7801: [to-inst] radius ;m
7802: overrides construct
7803:
7804: end-class circle
7805: @end example
7806:
7807: Finally, you can define named methods with @code{:m}. One use of this
7808: feature is the definition of words that occur only in one class and are
7809: not intended to be overridden, but which still need method context
7810: (e.g., for accessing @code{inst-var}s). Another use is for methods that
7811: would be bound frequently, if defined anonymously.
7812:
7813:
7814: @node Classes and Scoping, Dividing classes, Method conveniences, Objects
7815: @subsubsection Classes and Scoping
7816: @cindex classes and scoping
7817: @cindex scoping and classes
7818:
7819: Inheritance is frequent, unlike structure extension. This exacerbates
7820: the problem with the field name convention (@pxref{Structure Naming
7821: Convention}): One always has to remember in which class the field was
7822: originally defined; changing a part of the class structure would require
7823: changes for renaming in otherwise unaffected code.
7824:
7825: @cindex @code{inst-var} visibility
7826: @cindex @code{inst-value} visibility
7827: To solve this problem, I added a scoping mechanism (which was not in my
7828: original charter): A field defined with @code{inst-var} (or
7829: @code{inst-value}) is visible only in the class where it is defined and in
7830: the descendent classes of this class. Using such fields only makes
7831: sense in @code{m:}-defined methods in these classes anyway.
7832:
7833: This scoping mechanism allows us to use the unadorned field name,
7834: because name clashes with unrelated words become much less likely.
7835:
7836: @cindex @code{protected} discussion
7837: @cindex @code{private} discussion
7838: Once we have this mechanism, we can also use it for controlling the
7839: visibility of other words: All words defined after
7840: @code{protected} are visible only in the current class and its
7841: descendents. @code{public} restores the compilation
7842: (i.e. @code{current}) word list that was in effect before. If you
7843: have several @code{protected}s without an intervening
7844: @code{public} or @code{set-current}, @code{public}
7845: will restore the compilation word list in effect before the first of
7846: these @code{protected}s.
7847:
7848: @node Dividing classes, Object Interfaces, Classes and Scoping, Objects
7849: @subsubsection Dividing classes
7850: @cindex Dividing classes
7851: @cindex @code{methods}...@code{end-methods}
7852:
7853: You may want to do the definition of methods separate from the
7854: definition of the class, its selectors, fields, and instance variables,
7855: i.e., separate the implementation from the definition. You can do this
7856: in the following way:
7857:
7858: @example
7859: graphical class
7860: inst-value radius
7861: end-class circle
7862:
7863: ... \ do some other stuff
7864:
7865: circle methods \ now we are ready
7866:
7867: m: ( x y circle -- )
7868: radius draw-circle ;m
7869: overrides draw
7870:
7871: m: ( n-radius circle -- )
7872: [to-inst] radius ;m
7873: overrides construct
7874:
7875: end-methods
7876: @end example
7877:
7878: You can use several @code{methods}...@code{end-methods} sections. The
7879: only things you can do to the class in these sections are: defining
7880: methods, and overriding the class's selectors. You must not define new
7881: selectors or fields.
7882:
7883: Note that you often have to override a selector before using it. In
7884: particular, you usually have to override @code{construct} with a new
7885: method before you can invoke @code{heap-new} and friends. E.g., you
7886: must not create a circle before the @code{overrides construct} sequence
7887: in the example above.
7888:
7889: @node Object Interfaces, Objects Implementation, Dividing classes, Objects
7890: @subsubsection Object Interfaces
7891: @cindex object interfaces
7892: @cindex interfaces for objects
7893:
7894: In this model you can only call selectors defined in the class of the
7895: receiving objects or in one of its ancestors. If you call a selector
7896: with a receiving object that is not in one of these classes, the
7897: result is undefined; if you are lucky, the program crashes
7898: immediately.
7899:
7900: @cindex selectors common to hardly-related classes
7901: Now consider the case when you want to have a selector (or several)
7902: available in two classes: You would have to add the selector to a
7903: common ancestor class, in the worst case to @code{object}. You
7904: may not want to do this, e.g., because someone else is responsible for
7905: this ancestor class.
7906:
7907: The solution for this problem is interfaces. An interface is a
7908: collection of selectors. If a class implements an interface, the
7909: selectors become available to the class and its descendents. A class
7910: can implement an unlimited number of interfaces. For the problem
7911: discussed above, we would define an interface for the selector(s), and
7912: both classes would implement the interface.
7913:
7914: As an example, consider an interface @code{storage} for
7915: writing objects to disk and getting them back, and a class
7916: @code{foo} that implements it. The code would look like this:
7917:
7918: @cindex @code{interface} usage
7919: @cindex @code{end-interface} usage
7920: @cindex @code{implementation} usage
7921: @example
7922: interface
7923: selector write ( file object -- )
7924: selector read1 ( file object -- )
7925: end-interface storage
7926:
7927: bar class
7928: storage implementation
7929:
7930: ... overrides write
7931: ... overrides read1
7932: ...
7933: end-class foo
7934: @end example
7935:
7936: @noindent
7937: (I would add a word @code{read} @i{( file -- object )} that uses
7938: @code{read1} internally, but that's beyond the point illustrated
7939: here.)
7940:
7941: Note that you cannot use @code{protected} in an interface; and
7942: of course you cannot define fields.
7943:
7944: In the Neon model, all selectors are available for all classes;
7945: therefore it does not need interfaces. The price you pay in this model
7946: is slower late binding, and therefore, added complexity to avoid late
7947: binding.
7948:
7949: @node Objects Implementation, Objects Glossary, Object Interfaces, Objects
7950: @subsubsection @file{objects.fs} Implementation
7951: @cindex @file{objects.fs} implementation
7952:
7953: @cindex @code{object-map} discussion
7954: An object is a piece of memory, like one of the data structures
7955: described with @code{struct...end-struct}. It has a field
7956: @code{object-map} that points to the method map for the object's
7957: class.
7958:
7959: @cindex method map
7960: @cindex virtual function table
7961: The @emph{method map}@footnote{This is Self terminology; in C++
7962: terminology: virtual function table.} is an array that contains the
7963: execution tokens (@i{xt}s) of the methods for the object's class. Each
7964: selector contains an offset into a method map.
7965:
7966: @cindex @code{selector} implementation, class
7967: @code{selector} is a defining word that uses
7968: @code{CREATE} and @code{DOES>}. The body of the
7969: selector contains the offset; the @code{does>} action for a
7970: class selector is, basically:
7971:
7972: @example
7973: ( object addr ) @@ over object-map @@ + @@ execute
7974: @end example
7975:
7976: Since @code{object-map} is the first field of the object, it
7977: does not generate any code. As you can see, calling a selector has a
7978: small, constant cost.
7979:
7980: @cindex @code{current-interface} discussion
7981: @cindex class implementation and representation
7982: A class is basically a @code{struct} combined with a method
7983: map. During the class definition the alignment and size of the class
7984: are passed on the stack, just as with @code{struct}s, so
7985: @code{field} can also be used for defining class
7986: fields. However, passing more items on the stack would be
7987: inconvenient, so @code{class} builds a data structure in memory,
7988: which is accessed through the variable
7989: @code{current-interface}. After its definition is complete, the
7990: class is represented on the stack by a pointer (e.g., as parameter for
7991: a child class definition).
7992:
7993: A new class starts off with the alignment and size of its parent,
7994: and a copy of the parent's method map. Defining new fields extends the
7995: size and alignment; likewise, defining new selectors extends the
7996: method map. @code{overrides} just stores a new @i{xt} in the method
7997: map at the offset given by the selector.
7998:
7999: @cindex class binding, implementation
8000: Class binding just gets the @i{xt} at the offset given by the selector
8001: from the class's method map and @code{compile,}s (in the case of
8002: @code{[bind]}) it.
8003:
8004: @cindex @code{this} implementation
8005: @cindex @code{catch} and @code{this}
8006: @cindex @code{this} and @code{catch}
8007: I implemented @code{this} as a @code{value}. At the
8008: start of an @code{m:...;m} method the old @code{this} is
8009: stored to the return stack and restored at the end; and the object on
8010: the TOS is stored @code{TO this}. This technique has one
8011: disadvantage: If the user does not leave the method via
8012: @code{;m}, but via @code{throw} or @code{exit},
8013: @code{this} is not restored (and @code{exit} may
8014: crash). To deal with the @code{throw} problem, I have redefined
8015: @code{catch} to save and restore @code{this}; the same
8016: should be done with any word that can catch an exception. As for
8017: @code{exit}, I simply forbid it (as a replacement, there is
8018: @code{exitm}).
8019:
8020: @cindex @code{inst-var} implementation
8021: @code{inst-var} is just the same as @code{field}, with
8022: a different @code{DOES>} action:
8023: @example
8024: @@ this +
8025: @end example
8026: Similar for @code{inst-value}.
8027:
8028: @cindex class scoping implementation
8029: Each class also has a word list that contains the words defined with
8030: @code{inst-var} and @code{inst-value}, and its protected
8031: words. It also has a pointer to its parent. @code{class} pushes
8032: the word lists of the class and all its ancestors onto the search order stack,
8033: and @code{end-class} drops them.
8034:
8035: @cindex interface implementation
8036: An interface is like a class without fields, parent and protected
8037: words; i.e., it just has a method map. If a class implements an
8038: interface, its method map contains a pointer to the method map of the
8039: interface. The positive offsets in the map are reserved for class
8040: methods, therefore interface map pointers have negative
8041: offsets. Interfaces have offsets that are unique throughout the
8042: system, unlike class selectors, whose offsets are only unique for the
8043: classes where the selector is available (invokable).
8044:
8045: This structure means that interface selectors have to perform one
8046: indirection more than class selectors to find their method. Their body
8047: contains the interface map pointer offset in the class method map, and
8048: the method offset in the interface method map. The
8049: @code{does>} action for an interface selector is, basically:
8050:
8051: @example
8052: ( object selector-body )
8053: 2dup selector-interface @@ ( object selector-body object interface-offset )
8054: swap object-map @@ + @@ ( object selector-body map )
8055: swap selector-offset @@ + @@ execute
8056: @end example
8057:
8058: where @code{object-map} and @code{selector-offset} are
8059: first fields and generate no code.
8060:
8061: As a concrete example, consider the following code:
8062:
8063: @example
8064: interface
8065: selector if1sel1
8066: selector if1sel2
8067: end-interface if1
8068:
8069: object class
8070: if1 implementation
8071: selector cl1sel1
8072: cell% inst-var cl1iv1
8073:
8074: ' m1 overrides construct
8075: ' m2 overrides if1sel1
8076: ' m3 overrides if1sel2
8077: ' m4 overrides cl1sel2
8078: end-class cl1
8079:
8080: create obj1 object dict-new drop
8081: create obj2 cl1 dict-new drop
8082: @end example
8083:
8084: The data structure created by this code (including the data structure
8085: for @code{object}) is shown in the <a
8086: href="objects-implementation.eps">figure</a>, assuming a cell size of 4.
8087: @comment TODO add this diagram..
8088:
8089: @node Objects Glossary, , Objects Implementation, Objects
8090: @subsubsection @file{objects.fs} Glossary
8091: @cindex @file{objects.fs} Glossary
8092:
8093: doc---objects-bind
8094: doc---objects-<bind>
8095: doc---objects-bind'
8096: doc---objects-[bind]
8097: doc---objects-class
8098: doc---objects-class->map
8099: doc---objects-class-inst-size
8100: doc---objects-class-override!
8101: doc---objects-construct
8102: doc---objects-current'
8103: doc---objects-[current]
8104: doc---objects-current-interface
8105: doc---objects-dict-new
8106: doc---objects-drop-order
8107: doc---objects-end-class
8108: doc---objects-end-class-noname
8109: doc---objects-end-interface
8110: doc---objects-end-interface-noname
8111: doc---objects-end-methods
8112: doc---objects-exitm
8113: doc---objects-heap-new
8114: doc---objects-implementation
8115: doc---objects-init-object
8116: doc---objects-inst-value
8117: doc---objects-inst-var
8118: doc---objects-interface
8119: doc---objects-m:
8120: doc---objects-:m
8121: doc---objects-;m
8122: doc---objects-method
8123: doc---objects-methods
8124: doc---objects-object
8125: doc---objects-overrides
8126: doc---objects-[parent]
8127: doc---objects-print
8128: doc---objects-protected
8129: doc---objects-public
8130: doc---objects-push-order
8131: doc---objects-selector
8132: doc---objects-this
8133: doc---objects-<to-inst>
8134: doc---objects-[to-inst]
8135: doc---objects-to-this
8136: doc---objects-xt-new
8137:
8138: @c -------------------------------------------------------------
8139: @node OOF, Mini-OOF, Objects, Object-oriented Forth
8140: @subsection The @file{oof.fs} model
8141: @cindex oof
8142: @cindex object-oriented programming
8143:
8144: @cindex @file{objects.fs}
8145: @cindex @file{oof.fs}
8146:
8147: This section describes the @file{oof.fs} package.
8148:
8149: The package described in this section has been used in bigFORTH since 1991, and
8150: used for two large applications: a chromatographic system used to
8151: create new medicaments, and a graphic user interface library (MINOS).
8152:
8153: You can find a description (in German) of @file{oof.fs} in @cite{Object
8154: oriented bigFORTH} by Bernd Paysan, published in @cite{Vierte Dimension}
8155: 10(2), 1994.
8156:
8157: @menu
8158: * Properties of the OOF model::
8159: * Basic OOF Usage::
8160: * The OOF base class::
8161: * Class Declaration::
8162: * Class Implementation::
8163: @end menu
8164:
8165: @node Properties of the OOF model, Basic OOF Usage, OOF, OOF
8166: @subsubsection Properties of the @file{oof.fs} model
8167: @cindex @file{oof.fs} properties
8168:
8169: @itemize @bullet
8170: @item
8171: This model combines object oriented programming with information
8172: hiding. It helps you writing large application, where scoping is
8173: necessary, because it provides class-oriented scoping.
8174:
8175: @item
8176: Named objects, object pointers, and object arrays can be created,
8177: selector invocation uses the ``object selector'' syntax. Selector invocation
8178: to objects and/or selectors on the stack is a bit less convenient, but
8179: possible.
8180:
8181: @item
8182: Selector invocation and instance variable usage of the active object is
8183: straightforward, since both make use of the active object.
8184:
8185: @item
8186: Late binding is efficient and easy to use.
8187:
8188: @item
8189: State-smart objects parse selectors. However, extensibility is provided
8190: using a (parsing) selector @code{postpone} and a selector @code{'}.
8191:
8192: @item
8193: An implementation in ANS Forth is available.
8194:
8195: @end itemize
8196:
8197:
8198: @node Basic OOF Usage, The OOF base class, Properties of the OOF model, OOF
8199: @subsubsection Basic @file{oof.fs} Usage
8200: @cindex @file{oof.fs} usage
8201:
8202: This section uses the same example as for @code{objects} (@pxref{Basic Objects Usage}).
8203:
8204: You can define a class for graphical objects like this:
8205:
8206: @cindex @code{class} usage
8207: @cindex @code{class;} usage
8208: @cindex @code{method} usage
8209: @example
8210: object class graphical \ "object" is the parent class
8211: method draw ( x y graphical -- )
8212: class;
8213: @end example
8214:
8215: This code defines a class @code{graphical} with an
8216: operation @code{draw}. We can perform the operation
8217: @code{draw} on any @code{graphical} object, e.g.:
8218:
8219: @example
8220: 100 100 t-rex draw
8221: @end example
8222:
8223: @noindent
8224: where @code{t-rex} is an object or object pointer, created with e.g.
8225: @code{graphical : t-rex}.
8226:
8227: @cindex abstract class
8228: How do we create a graphical object? With the present definitions,
8229: we cannot create a useful graphical object. The class
8230: @code{graphical} describes graphical objects in general, but not
8231: any concrete graphical object type (C++ users would call it an
8232: @emph{abstract class}); e.g., there is no method for the selector
8233: @code{draw} in the class @code{graphical}.
8234:
8235: For concrete graphical objects, we define child classes of the
8236: class @code{graphical}, e.g.:
8237:
8238: @example
8239: graphical class circle \ "graphical" is the parent class
8240: cell var circle-radius
8241: how:
8242: : draw ( x y -- )
8243: circle-radius @@ draw-circle ;
8244:
8245: : init ( n-radius -- (
8246: circle-radius ! ;
8247: class;
8248: @end example
8249:
8250: Here we define a class @code{circle} as a child of @code{graphical},
8251: with a field @code{circle-radius}; it defines new methods for the
8252: selectors @code{draw} and @code{init} (@code{init} is defined in
8253: @code{object}, the parent class of @code{graphical}).
8254:
8255: Now we can create a circle in the dictionary with:
8256:
8257: @example
8258: 50 circle : my-circle
8259: @end example
8260:
8261: @noindent
8262: @code{:} invokes @code{init}, thus initializing the field
8263: @code{circle-radius} with 50. We can draw this new circle at (100,100)
8264: with:
8265:
8266: @example
8267: 100 100 my-circle draw
8268: @end example
8269:
8270: @cindex selector invocation, restrictions
8271: @cindex class definition, restrictions
8272: Note: You can only invoke a selector if the receiving object belongs to
8273: the class where the selector was defined or one of its descendents;
8274: e.g., you can invoke @code{draw} only for objects belonging to
8275: @code{graphical} or its descendents (e.g., @code{circle}). The scoping
8276: mechanism will check if you try to invoke a selector that is not
8277: defined in this class hierarchy, so you'll get an error at compilation
8278: time.
8279:
8280:
8281: @node The OOF base class, Class Declaration, Basic OOF Usage, OOF
8282: @subsubsection The @file{oof.fs} base class
8283: @cindex @file{oof.fs} base class
8284:
8285: When you define a class, you have to specify a parent class. So how do
8286: you start defining classes? There is one class available from the start:
8287: @code{object}. You have to use it as ancestor for all classes. It is the
8288: only class that has no parent. Classes are also objects, except that
8289: they don't have instance variables; class manipulation such as
8290: inheritance or changing definitions of a class is handled through
8291: selectors of the class @code{object}.
8292:
8293: @code{object} provides a number of selectors:
8294:
8295: @itemize @bullet
8296: @item
8297: @code{class} for subclassing, @code{definitions} to add definitions
8298: later on, and @code{class?} to get type informations (is the class a
8299: subclass of the class passed on the stack?).
8300: doc---object-class
8301: doc---object-definitions
8302: doc---object-class?
8303:
8304: @item
8305: @code{init} and @code{dispose} as constructor and destructor of the
8306: object. @code{init} is invocated after the object's memory is allocated,
8307: while @code{dispose} also handles deallocation. Thus if you redefine
8308: @code{dispose}, you have to call the parent's dispose with @code{super
8309: dispose}, too.
8310: doc---object-init
8311: doc---object-dispose
8312:
8313: @item
8314: @code{new}, @code{new[]}, @code{:}, @code{ptr}, @code{asptr}, and
8315: @code{[]} to create named and unnamed objects and object arrays or
8316: object pointers.
8317: doc---object-new
8318: doc---object-new[]
8319: doc---object-:
8320: doc---object-ptr
8321: doc---object-asptr
8322: doc---object-[]
8323:
8324: @item
8325: @code{::} and @code{super} for explicit scoping. You should use explicit
8326: scoping only for super classes or classes with the same set of instance
8327: variables. Explicitly-scoped selectors use early binding.
8328: doc---object-::
8329: doc---object-super
8330:
8331: @item
8332: @code{self} to get the address of the object
8333: doc---object-self
8334:
8335: @item
8336: @code{bind}, @code{bound}, @code{link}, and @code{is} to assign object
8337: pointers and instance defers.
8338: doc---object-bind
8339: doc---object-bound
8340: doc---object-link
8341: doc---object-is
8342:
8343: @item
8344: @code{'} to obtain selector tokens, @code{send} to invocate selectors
8345: form the stack, and @code{postpone} to generate selector invocation code.
8346: doc---object-'
8347: doc---object-postpone
8348:
8349: @item
8350: @code{with} and @code{endwith} to select the active object from the
8351: stack, and enable its scope. Using @code{with} and @code{endwith}
8352: also allows you to create code using selector @code{postpone} without being
8353: trapped by the state-smart objects.
8354: doc---object-with
8355: doc---object-endwith
8356:
8357: @end itemize
8358:
8359: @node Class Declaration, Class Implementation, The OOF base class, OOF
8360: @subsubsection Class Declaration
8361: @cindex class declaration
8362:
8363: @itemize @bullet
8364: @item
8365: Instance variables
8366: doc---oof-var
8367:
8368: @item
8369: Object pointers
8370: doc---oof-ptr
8371: doc---oof-asptr
8372:
8373: @item
8374: Instance defers
8375: doc---oof-defer
8376:
8377: @item
8378: Method selectors
8379: doc---oof-early
8380: doc---oof-method
8381:
8382: @item
8383: Class-wide variables
8384: doc---oof-static
8385:
8386: @item
8387: End declaration
8388: doc---oof-how:
8389: doc---oof-class;
8390:
8391: @end itemize
8392:
8393: @c -------------------------------------------------------------
8394: @node Class Implementation, , Class Declaration, OOF
8395: @subsubsection Class Implementation
8396: @cindex class implementation
8397:
8398: @c -------------------------------------------------------------
8399: @node Mini-OOF, Comparison with other object models, OOF, Object-oriented Forth
8400: @subsection The @file{mini-oof.fs} model
8401: @cindex mini-oof
8402:
8403: Gforth's third object oriented Forth package is a 12-liner. It uses a
8404: mixture of the @file{object.fs} and the @file{oof.fs} syntax,
8405: and reduces to the bare minimum of features. This is based on a posting
8406: of Bernd Paysan in comp.arch.
8407:
8408: @menu
8409: * Basic Mini-OOF Usage::
8410: * Mini-OOF Example::
8411: * Mini-OOF Implementation::
8412: @end menu
8413:
8414: @c -------------------------------------------------------------
8415: @node Basic Mini-OOF Usage, Mini-OOF Example, , Mini-OOF
8416: @subsubsection Basic @file{mini-oof.fs} Usage
8417: @cindex mini-oof usage
8418:
8419: There is a base class (@code{class}, which allocates one cell for the
8420: object pointer) plus seven other words: to define a method, a variable,
8421: a class; to end a class, to resolve binding, to allocate an object and
8422: to compile a class method.
8423: @comment TODO better description of the last one
8424:
8425: doc-object
8426: doc-method
8427: doc-var
8428: doc-class
8429: doc-end-class
8430: doc-defines
8431: doc-new
8432: doc-::
8433:
8434:
8435: @c -------------------------------------------------------------
8436: @node Mini-OOF Example, Mini-OOF Implementation, Basic Mini-OOF Usage, Mini-OOF
8437: @subsubsection Mini-OOF Example
8438: @cindex mini-oof example
8439:
8440: A short example shows how to use this package. This example, in slightly
8441: extended form, is supplied as @file{moof-exm.fs}
8442: @comment TODO could flesh this out with some comments from the Forthwrite article
8443:
8444: @example
8445: object class
8446: method init
8447: method draw
8448: end-class graphical
8449: @end example
8450:
8451: This code defines a class @code{graphical} with an
8452: operation @code{draw}. We can perform the operation
8453: @code{draw} on any @code{graphical} object, e.g.:
8454:
8455: @example
8456: 100 100 t-rex draw
8457: @end example
8458:
8459: where @code{t-rex} is an object or object pointer, created with e.g.
8460: @code{graphical new Constant t-rex}.
8461:
8462: For concrete graphical objects, we define child classes of the
8463: class @code{graphical}, e.g.:
8464:
8465: @example
8466: graphical class
8467: cell var circle-radius
8468: end-class circle \ "graphical" is the parent class
8469:
8470: :noname ( x y -- )
8471: circle-radius @@ draw-circle ; circle defines draw
8472: :noname ( r -- )
8473: circle-radius ! ; circle defines init
8474: @end example
8475:
8476: There is no implicit init method, so we have to define one. The creation
8477: code of the object now has to call init explicitely.
8478:
8479: @example
8480: circle new Constant my-circle
8481: 50 my-circle init
8482: @end example
8483:
8484: It is also possible to add a function to create named objects with
8485: automatic call of @code{init}, given that all objects have @code{init}
8486: on the same place:
8487:
8488: @example
8489: : new: ( .. o "name" -- )
8490: new dup Constant init ;
8491: 80 circle new: large-circle
8492: @end example
8493:
8494: We can draw this new circle at (100,100) with:
8495:
8496: @example
8497: 100 100 my-circle draw
8498: @end example
8499:
8500: @node Mini-OOF Implementation, , Mini-OOF Example, Mini-OOF
8501: @subsubsection @file{mini-oof.fs} Implementation
8502:
8503: Object-oriented systems with late binding typically use a
8504: ``vtable''-approach: the first variable in each object is a pointer to a
8505: table, which contains the methods as function pointers. The vtable
8506: may also contain other information.
8507:
8508: So first, let's declare methods:
8509:
8510: @example
8511: : method ( m v -- m' v ) Create over , swap cell+ swap
8512: DOES> ( ... o -- ... ) @ over @ + @ execute ;
8513: @end example
8514:
8515: During method declaration, the number of methods and instance
8516: variables is on the stack (in address units). @code{method} creates
8517: one method and increments the method number. To execute a method, it
8518: takes the object, fetches the vtable pointer, adds the offset, and
8519: executes the @i{xt} stored there. Each method takes the object it is
8520: invoked from as top of stack parameter. The method itself should
8521: consume that object.
8522:
8523: Now, we also have to declare instance variables
8524:
8525: @example
8526: : var ( m v size -- m v' ) Create over , +
8527: DOES> ( o -- addr ) @ + ;
8528: @end example
8529:
8530: As before, a word is created with the current offset. Instance
8531: variables can have different sizes (cells, floats, doubles, chars), so
8532: all we do is take the size and add it to the offset. If your machine
8533: has alignment restrictions, put the proper @code{aligned} or
8534: @code{faligned} before the variable, to adjust the variable
8535: offset. That's why it is on the top of stack.
8536:
8537: We need a starting point (the base object) and some syntactic sugar:
8538:
8539: @example
8540: Create object 1 cells , 2 cells ,
8541: : class ( class -- class methods vars ) dup 2@ ;
8542: @end example
8543:
8544: For inheritance, the vtable of the parent object has to be
8545: copied when a new, derived class is declared. This gives all the
8546: methods of the parent class, which can be overridden, though.
8547:
8548: @example
8549: : end-class ( class methods vars -- )
8550: Create here >r , dup , 2 cells ?DO ['] noop , 1 cells +LOOP
8551: cell+ dup cell+ r> rot @ 2 cells /string move ;
8552: @end example
8553:
8554: The first line creates the vtable, initialized with
8555: @code{noop}s. The second line is the inheritance mechanism, it
8556: copies the xts from the parent vtable.
8557:
8558: We still have no way to define new methods, let's do that now:
8559:
8560: @example
8561: : defines ( xt class -- ) ' >body @ + ! ;
8562: @end example
8563:
8564: To allocate a new object, we need a word, too:
8565:
8566: @example
8567: : new ( class -- o ) here over @ allot swap over ! ;
8568: @end example
8569:
8570: Sometimes derived classes want to access the method of the
8571: parent object. There are two ways to achieve this with Mini-OOF:
8572: first, you could use named words, and second, you could look up the
8573: vtable of the parent object.
8574:
8575: @example
8576: : :: ( class "name" -- ) ' >body @ + @ compile, ;
8577: @end example
8578:
8579:
8580: Nothing can be more confusing than a good example, so here is
8581: one. First let's declare a text object (called
8582: @code{button}), that stores text and position:
8583:
8584: @example
8585: object class
8586: cell var text
8587: cell var len
8588: cell var x
8589: cell var y
8590: method init
8591: method draw
8592: end-class button
8593: @end example
8594:
8595: @noindent
8596: Now, implement the two methods, @code{draw} and @code{init}:
8597:
8598: @example
8599: :noname ( o -- )
8600: >r r@ x @ r@ y @ at-xy r@ text @ r> len @ type ;
8601: button defines draw
8602: :noname ( addr u o -- )
8603: >r 0 r@ x ! 0 r@ y ! r@ len ! r> text ! ;
8604: button defines init
8605: @end example
8606:
8607: @noindent
8608: To demonstrate inheritance, we define a class @code{bold-button}, with no
8609: new data and no new methods:
8610:
8611: @example
8612: button class
8613: end-class bold-button
8614:
8615: : bold 27 emit ." [1m" ;
8616: : normal 27 emit ." [0m" ;
8617: @end example
8618:
8619: @noindent
8620: The class @code{bold-button} has a different draw method to
8621: @code{button}, but the new method is defined in terms of the draw method
8622: for @code{button}:
8623:
8624: @example
8625: :noname bold [ button :: draw ] normal ; bold-button defines draw
8626: @end example
8627:
8628: @noindent
8629: Finally, create two objects and apply methods:
8630:
8631: @example
8632: button new Constant foo
8633: s" thin foo" foo init
8634: page
8635: foo draw
8636: bold-button new Constant bar
8637: s" fat bar" bar init
8638: 1 bar y !
8639: bar draw
8640: @end example
8641:
8642:
8643: @node Comparison with other object models, , Mini-OOF, Object-oriented Forth
8644: @subsubsection Comparison with other object models
8645: @cindex comparison of object models
8646: @cindex object models, comparison
8647:
8648: Many object-oriented Forth extensions have been proposed (@cite{A survey
8649: of object-oriented Forths} (SIGPLAN Notices, April 1996) by Bradford
8650: J. Rodriguez and W. F. S. Poehlman lists 17). This section discusses the
8651: relation of the object models described here to two well-known and two
8652: closely-related (by the use of method maps) models.
8653:
8654: @cindex Neon model
8655: The most popular model currently seems to be the Neon model (see
8656: @cite{Object-oriented programming in ANS Forth} (Forth Dimensions, March
8657: 1997) by Andrew McKewan) but this model has a number of limitations
8658: @footnote{A longer version of this critique can be
8659: found in @cite{On Standardizing Object-Oriented Forth Extensions} (Forth
8660: Dimensions, May 1997) by Anton Ertl.}:
8661:
8662: @itemize @bullet
8663: @item
8664: It uses a @code{@emph{selector
8665: object}} syntax, which makes it unnatural to pass objects on the
8666: stack.
8667:
8668: @item
8669: It requires that the selector parses the input stream (at
8670: compile time); this leads to reduced extensibility and to bugs that are+
8671: hard to find.
8672:
8673: @item
8674: It allows using every selector to every object;
8675: this eliminates the need for classes, but makes it harder to create
8676: efficient implementations.
8677: @end itemize
8678:
8679: @cindex Pountain's object-oriented model
8680: Another well-known publication is @cite{Object-Oriented Forth} (Academic
8681: Press, London, 1987) by Dick Pountain. However, it is not really about
8682: object-oriented programming, because it hardly deals with late
8683: binding. Instead, it focuses on features like information hiding and
8684: overloading that are characteristic of modular languages like Ada (83).
8685:
8686: @cindex Zsoter's object-oriented model
8687: In @cite{Does late binding have to be slow?} (Forth Dimensions 18(1) 1996, pages 31-35)
8688: Andras Zsoter describes a model that makes heavy use of an active object
8689: (like @code{this} in @file{objects.fs}): The active object is not only
8690: used for accessing all fields, but also specifies the receiving object
8691: of every selector invocation; you have to change the active object
8692: explicitly with @code{@{ ... @}}, whereas in @file{objects.fs} it
8693: changes more or less implicitly at @code{m: ... ;m}. Such a change at
8694: the method entry point is unnecessary with the Zsoter's model, because
8695: the receiving object is the active object already. On the other hand, the explicit
8696: change is absolutely necessary in that model, because otherwise no one
8697: could ever change the active object. An ANS Forth implementation of this
8698: model is available at @url{http://www.forth.org/fig/oopf.html}.
8699:
8700: @cindex @file{oof.fs}, differences to other models
8701: The @file{oof.fs} model combines information hiding and overloading
8702: resolution (by keeping names in various word lists) with object-oriented
8703: programming. It sets the active object implicitly on method entry, but
8704: also allows explicit changing (with @code{>o...o>} or with
8705: @code{with...endwith}). It uses parsing and state-smart objects and
8706: classes for resolving overloading and for early binding: the object or
8707: class parses the selector and determines the method from this. If the
8708: selector is not parsed by an object or class, it performs a call to the
8709: selector for the active object (late binding), like Zsoter's model.
8710: Fields are always accessed through the active object. The big
8711: disadvantage of this model is the parsing and the state-smartness, which
8712: reduces extensibility and increases the opportunities for subtle bugs;
8713: essentially, you are only safe if you never tick or @code{postpone} an
8714: object or class (Bernd disagrees, but I (Anton) am not convinced).
8715:
8716: @cindex @file{mini-oof.fs}, differences to other models
8717: The @file{mini-oof.fs} model is quite similar to a very stripped-down version of
8718: the @file{objects.fs} model, but syntactically it is a mixture of the @file{objects.fs} and
8719: @file{oof.fs} models.
8720:
8721: @c -------------------------------------------------------------
8722: @node Passing Commands to the OS, Miscellaneous Words, Object-oriented Forth, Words
8723: @section Passing Commands to the Operating System
8724: @cindex operating system - passing commands
8725: @cindex shell commands
8726:
8727: Gforth allows you to pass an arbitrary string to the host operating
8728: system shell (if such a thing exists) for execution.
8729:
8730: doc-sh
8731: doc-system
8732: doc-$?
8733: doc-getenv
8734:
8735: @c -------------------------------------------------------------
8736: @node Miscellaneous Words, , Passing Commands to the OS, Words
8737: @section Miscellaneous Words
8738: @cindex miscellaneous words
8739:
8740: @comment TODO find homes for these
8741:
8742: These section lists the ANS Forth words that are not documented
8743: elsewhere in this manual. Ultimately, they all need proper homes.
8744:
8745: doc-ms
8746: doc-time&date
8747:
8748: doc-[compile]
8749:
8750: The following ANS Forth words are not currently supported by Gforth
8751: (@pxref{ANS conformance}):
8752:
8753: @code{EDITOR}
8754: @code{EMIT?}
8755: @code{FORGET}
8756:
8757: @c ******************************************************************
8758: @node Error messages, Tools, Words, Top
8759: @chapter Error messages
8760: @cindex error messages
8761: @cindex backtrace
8762:
8763: A typical Gforth error message looks like this:
8764:
8765: @example
8766: in file included from :-1
8767: in file included from ./yyy.fs:1
8768: ./xxx.fs:4: Invalid memory address
8769: bar
8770: ^^^
8771: $400E664C @@
8772: $400E6664 foo
8773: @end example
8774:
8775: The message identifying the error is @code{Invalid memory address}. The
8776: error happened when text-interpreting line 4 of the file
8777: @file{./xxx.fs}. This line is given (it contains @code{bar}), and the
8778: word on the line where the error happened, is pointed out (with
8779: @code{^^^}).
8780:
8781: The file containing the error was included in line 1 of @file{./yyy.fs},
8782: and @file{yyy.fs} was included from a non-file (in this case, by giving
8783: @file{yyy.fs} as command-line parameter to Gforth).
8784:
8785: At the end of the error message you find a return stack dump that can be
8786: interpreted as a backtrace (possibly empty). On top you find the top of
8787: the return stack when the @code{throw} happened, and at the bottom you
8788: find the return stack entry just above the return stack of the topmost
8789: text interpreter.
8790:
8791: To the right of most return stack entries you see a guess for the word
8792: that pushed that return stack entry as its return address. This gives a
8793: backtrace. In our case we see that @code{bar} called @code{foo}, and
8794: @code{foo} called @code{@@} (and @code{@@} had an @emph{Invalid memory
8795: address} exception).
8796:
8797: Note that the backtrace is not perfect: We don't know which return stack
8798: entries are return addresses (so we may get false positives); and in
8799: some cases (e.g., for @code{abort"}) we cannot determine from the return
8800: address the word that pushed the return address, so for some return
8801: addresses you see no names in the return stack dump.
8802:
8803: @cindex @code{catch} and backtraces
8804: The return stack dump represents the return stack at the time when a
8805: specific @code{throw} was executed. In programs that make use of
8806: @code{catch}, it is not necessarily clear which @code{throw} should be
8807: used for the return stack dump (e.g., consider one @code{throw} that
8808: indicates an error, which is caught, and during recovery another error
8809: happens; which @code{throw} should be used for the stack dump). Gforth
8810: presents the return stack dump for the first @code{throw} after the last
8811: executed (not returned-to) @code{catch}; this works well in the usual
8812: case.
8813:
8814: @cindex @code{gforth-fast} and backtraces
8815: @cindex @code{gforth-fast}, difference from @code{gforth}
8816: @cindex backtraces with @code{gforth-fast}
8817: @cindex return stack dump with @code{gforth-fast}
8818: @code{gforth} is able to do a return stack dump for throws generated
8819: from primitives (e.g., invalid memory address, stack empty etc.);
8820: @code{gforth-fast} is only able to do a return stack dump from a
8821: directly called @code{throw} (including @code{abort} etc.). This is the
8822: only difference (apart from a speed factor of between 1.15 (K6-2) and
8823: 1.6 (21164A)) between @code{gforth} and @code{gforth-fast}. Given an
8824: exception caused by a primitive in @code{gforth-fast}, you will
8825: typically see no return stack dump at all; however, if the exception is
8826: caught by @code{catch} (e.g., for restoring some state), and then
8827: @code{throw}n again, the return stack dump will be for the first such
8828: @code{throw}.
8829:
8830: @c ******************************************************************
8831: @node Tools, ANS conformance, Error messages, Top
8832: @chapter Tools
8833:
8834: @menu
8835: * ANS Report:: Report the words used, sorted by wordset.
8836: @end menu
8837:
8838: See also @ref{Emacs and Gforth}.
8839:
8840: @node ANS Report, , Tools, Tools
8841: @section @file{ans-report.fs}: Report the words used, sorted by wordset
8842: @cindex @file{ans-report.fs}
8843: @cindex report the words used in your program
8844: @cindex words used in your program
8845:
8846: If you want to label a Forth program as ANS Forth Program, you must
8847: document which wordsets the program uses; for extension wordsets, it is
8848: helpful to list the words the program requires from these wordsets
8849: (because Forth systems are allowed to provide only some words of them).
8850:
8851: The @file{ans-report.fs} tool makes it easy for you to determine which
8852: words from which wordset and which non-ANS words your application
8853: uses. You simply have to include @file{ans-report.fs} before loading the
8854: program you want to check. After loading your program, you can get the
8855: report with @code{print-ans-report}. A typical use is to run this as
8856: batch job like this:
8857: @example
8858: gforth ans-report.fs myprog.fs -e "print-ans-report bye"
8859: @end example
8860:
8861: The output looks like this (for @file{compat/control.fs}):
8862: @example
8863: The program uses the following words
8864: from CORE :
8865: : POSTPONE THEN ; immediate ?dup IF 0=
8866: from BLOCK-EXT :
8867: \
8868: from FILE :
8869: (
8870: @end example
8871:
8872: @subsection Caveats
8873:
8874: Note that @file{ans-report.fs} just checks which words are used, not whether
8875: they are used in an ANS Forth conforming way!
8876:
8877: Some words are defined in several wordsets in the
8878: standard. @file{ans-report.fs} reports them for only one of the
8879: wordsets, and not necessarily the one you expect. It depends on usage
8880: which wordset is the right one to specify. E.g., if you only use the
8881: compilation semantics of @code{S"}, it is a Core word; if you also use
8882: its interpretation semantics, it is a File word.
8883:
8884: @c ******************************************************************
8885: @node ANS conformance, Model, Tools, Top
8886: @chapter ANS conformance
8887: @cindex ANS conformance of Gforth
8888:
8889: To the best of our knowledge, Gforth is an
8890:
8891: ANS Forth System
8892: @itemize @bullet
8893: @item providing the Core Extensions word set
8894: @item providing the Block word set
8895: @item providing the Block Extensions word set
8896: @item providing the Double-Number word set
8897: @item providing the Double-Number Extensions word set
8898: @item providing the Exception word set
8899: @item providing the Exception Extensions word set
8900: @item providing the Facility word set
8901: @item providing @code{EKEY}, @code{EKEY>CHAR}, @code{EKEY?}, @code{MS} and @code{TIME&DATE} from the Facility Extensions word set
8902: @item providing the File Access word set
8903: @item providing the File Access Extensions word set
8904: @item providing the Floating-Point word set
8905: @item providing the Floating-Point Extensions word set
8906: @item providing the Locals word set
8907: @item providing the Locals Extensions word set
8908: @item providing the Memory-Allocation word set
8909: @item providing the Memory-Allocation Extensions word set (that one's easy)
8910: @item providing the Programming-Tools word set
8911: @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
8912: @item providing the Search-Order word set
8913: @item providing the Search-Order Extensions word set
8914: @item providing the String word set
8915: @item providing the String Extensions word set (another easy one)
8916: @end itemize
8917:
8918: @cindex system documentation
8919: In addition, ANS Forth systems are required to document certain
8920: implementation choices. This chapter tries to meet these
8921: requirements. In many cases it gives a way to ask the system for the
8922: information instead of providing the information directly, in
8923: particular, if the information depends on the processor, the operating
8924: system or the installation options chosen, or if they are likely to
8925: change during the maintenance of Gforth.
8926:
8927: @comment The framework for the rest has been taken from pfe.
8928:
8929: @menu
8930: * The Core Words::
8931: * The optional Block word set::
8932: * The optional Double Number word set::
8933: * The optional Exception word set::
8934: * The optional Facility word set::
8935: * The optional File-Access word set::
8936: * The optional Floating-Point word set::
8937: * The optional Locals word set::
8938: * The optional Memory-Allocation word set::
8939: * The optional Programming-Tools word set::
8940: * The optional Search-Order word set::
8941: @end menu
8942:
8943:
8944: @c =====================================================================
8945: @node The Core Words, The optional Block word set, ANS conformance, ANS conformance
8946: @comment node-name, next, previous, up
8947: @section The Core Words
8948: @c =====================================================================
8949: @cindex core words, system documentation
8950: @cindex system documentation, core words
8951:
8952: @menu
8953: * core-idef:: Implementation Defined Options
8954: * core-ambcond:: Ambiguous Conditions
8955: * core-other:: Other System Documentation
8956: @end menu
8957:
8958: @c ---------------------------------------------------------------------
8959: @node core-idef, core-ambcond, The Core Words, The Core Words
8960: @subsection Implementation Defined Options
8961: @c ---------------------------------------------------------------------
8962: @cindex core words, implementation-defined options
8963: @cindex implementation-defined options, core words
8964:
8965:
8966: @table @i
8967: @item (Cell) aligned addresses:
8968: @cindex cell-aligned addresses
8969: @cindex aligned addresses
8970: processor-dependent. Gforth's alignment words perform natural alignment
8971: (e.g., an address aligned for a datum of size 8 is divisible by
8972: 8). Unaligned accesses usually result in a @code{-23 THROW}.
8973:
8974: @item @code{EMIT} and non-graphic characters:
8975: @cindex @code{EMIT} and non-graphic characters
8976: @cindex non-graphic characters and @code{EMIT}
8977: The character is output using the C library function (actually, macro)
8978: @code{putc}.
8979:
8980: @item character editing of @code{ACCEPT} and @code{EXPECT}:
8981: @cindex character editing of @code{ACCEPT} and @code{EXPECT}
8982: @cindex editing in @code{ACCEPT} and @code{EXPECT}
8983: @cindex @code{ACCEPT}, editing
8984: @cindex @code{EXPECT}, editing
8985: This is modeled on the GNU readline library (@pxref{Readline
8986: Interaction, , Command Line Editing, readline, The GNU Readline
8987: Library}) with Emacs-like key bindings. @kbd{Tab} deviates a little by
8988: producing a full word completion every time you type it (instead of
8989: producing the common prefix of all completions). @xref{Command-line editing}.
8990:
8991: @item character set:
8992: @cindex character set
8993: The character set of your computer and display device. Gforth is
8994: 8-bit-clean (but some other component in your system may make trouble).
8995:
8996: @item Character-aligned address requirements:
8997: @cindex character-aligned address requirements
8998: installation-dependent. Currently a character is represented by a C
8999: @code{unsigned char}; in the future we might switch to @code{wchar_t}
9000: (Comments on that requested).
9001:
9002: @item character-set extensions and matching of names:
9003: @cindex character-set extensions and matching of names
9004: @cindex case-sensitivity for name lookup
9005: @cindex name lookup, case-sensitivity
9006: @cindex locale and case-sensitivity
9007: Any character except the ASCII NUL character can be used in a
9008: name. Matching is case-insensitive (except in @code{TABLE}s). The
9009: matching is performed using the C function @code{strncasecmp}, whose
9010: function is probably influenced by the locale. E.g., the @code{C} locale
9011: does not know about accents and umlauts, so they are matched
9012: case-sensitively in that locale. For portability reasons it is best to
9013: write programs such that they work in the @code{C} locale. Then one can
9014: use libraries written by a Polish programmer (who might use words
9015: containing ISO Latin-2 encoded characters) and by a French programmer
9016: (ISO Latin-1) in the same program (of course, @code{WORDS} will produce
9017: funny results for some of the words (which ones, depends on the font you
9018: are using)). Also, the locale you prefer may not be available in other
9019: operating systems. Hopefully, Unicode will solve these problems one day.
9020:
9021: @item conditions under which control characters match a space delimiter:
9022: @cindex space delimiters
9023: @cindex control characters as delimiters
9024: If @code{WORD} is called with the space character as a delimiter, all
9025: white-space characters (as identified by the C macro @code{isspace()})
9026: are delimiters. @code{PARSE}, on the other hand, treats space like other
9027: delimiters. @code{PARSE-WORD} treats space like @code{WORD}, but behaves
9028: like @code{PARSE} otherwise. @code{(NAME)}, which is used by the outer
9029: interpreter (aka text interpreter) by default, treats all white-space
9030: characters as delimiters.
9031:
9032: @item format of the control-flow stack:
9033: @cindex control-flow stack, format
9034: The data stack is used as control-flow stack. The size of a control-flow
9035: stack item in cells is given by the constant @code{cs-item-size}. At the
9036: time of this writing, an item consists of a (pointer to a) locals list
9037: (third), an address in the code (second), and a tag for identifying the
9038: item (TOS). The following tags are used: @code{defstart},
9039: @code{live-orig}, @code{dead-orig}, @code{dest}, @code{do-dest},
9040: @code{scopestart}.
9041:
9042: @item conversion of digits > 35
9043: @cindex digits > 35
9044: The characters @code{[\]^_'} are the digits with the decimal value
9045: 36@minus{}41. There is no way to input many of the larger digits.
9046:
9047: @item display after input terminates in @code{ACCEPT} and @code{EXPECT}:
9048: @cindex @code{EXPECT}, display after end of input
9049: @cindex @code{ACCEPT}, display after end of input
9050: The cursor is moved to the end of the entered string. If the input is
9051: terminated using the @kbd{Return} key, a space is typed.
9052:
9053: @item exception abort sequence of @code{ABORT"}:
9054: @cindex exception abort sequence of @code{ABORT"}
9055: @cindex @code{ABORT"}, exception abort sequence
9056: The error string is stored into the variable @code{"error} and a
9057: @code{-2 throw} is performed.
9058:
9059: @item input line terminator:
9060: @cindex input line terminator
9061: @cindex line terminator on input
9062: @cindex newline character on input
9063: For interactive input, @kbd{C-m} (CR) and @kbd{C-j} (LF) terminate
9064: lines. One of these characters is typically produced when you type the
9065: @kbd{Enter} or @kbd{Return} key.
9066:
9067: @item maximum size of a counted string:
9068: @cindex maximum size of a counted string
9069: @cindex counted string, maximum size
9070: @code{s" /counted-string" environment? drop .}. Currently 255 characters
9071: on all ports, but this may change.
9072:
9073: @item maximum size of a parsed string:
9074: @cindex maximum size of a parsed string
9075: @cindex parsed string, maximum size
9076: Given by the constant @code{/line}. Currently 255 characters.
9077:
9078: @item maximum size of a definition name, in characters:
9079: @cindex maximum size of a definition name, in characters
9080: @cindex name, maximum length
9081: 31
9082:
9083: @item maximum string length for @code{ENVIRONMENT?}, in characters:
9084: @cindex maximum string length for @code{ENVIRONMENT?}, in characters
9085: @cindex @code{ENVIRONMENT?} string length, maximum
9086: 31
9087:
9088: @item method of selecting the user input device:
9089: @cindex user input device, method of selecting
9090: The user input device is the standard input. There is currently no way to
9091: change it from within Gforth. However, the input can typically be
9092: redirected in the command line that starts Gforth.
9093:
9094: @item method of selecting the user output device:
9095: @cindex user output device, method of selecting
9096: @code{EMIT} and @code{TYPE} output to the file-id stored in the value
9097: @code{outfile-id} (@code{stdout} by default). Gforth uses unbuffered
9098: output when the user output device is a terminal, otherwise the output
9099: is buffered.
9100:
9101: @item methods of dictionary compilation:
9102: What are we expected to document here?
9103:
9104: @item number of bits in one address unit:
9105: @cindex number of bits in one address unit
9106: @cindex address unit, size in bits
9107: @code{s" address-units-bits" environment? drop .}. 8 in all current
9108: ports.
9109:
9110: @item number representation and arithmetic:
9111: @cindex number representation and arithmetic
9112: Processor-dependent. Binary two's complement on all current ports.
9113:
9114: @item ranges for integer types:
9115: @cindex ranges for integer types
9116: @cindex integer types, ranges
9117: Installation-dependent. Make environmental queries for @code{MAX-N},
9118: @code{MAX-U}, @code{MAX-D} and @code{MAX-UD}. The lower bounds for
9119: unsigned (and positive) types is 0. The lower bound for signed types on
9120: two's complement and one's complement machines machines can be computed
9121: by adding 1 to the upper bound.
9122:
9123: @item read-only data space regions:
9124: @cindex read-only data space regions
9125: @cindex data-space, read-only regions
9126: The whole Forth data space is writable.
9127:
9128: @item size of buffer at @code{WORD}:
9129: @cindex size of buffer at @code{WORD}
9130: @cindex @code{WORD} buffer size
9131: @code{PAD HERE - .}. 104 characters on 32-bit machines. The buffer is
9132: shared with the pictured numeric output string. If overwriting
9133: @code{PAD} is acceptable, it is as large as the remaining dictionary
9134: space, although only as much can be sensibly used as fits in a counted
9135: string.
9136:
9137: @item size of one cell in address units:
9138: @cindex cell size
9139: @code{1 cells .}.
9140:
9141: @item size of one character in address units:
9142: @cindex char size
9143: @code{1 chars .}. 1 on all current ports.
9144:
9145: @item size of the keyboard terminal buffer:
9146: @cindex size of the keyboard terminal buffer
9147: @cindex terminal buffer, size
9148: Varies. You can determine the size at a specific time using @code{lp@@
9149: tib - .}. It is shared with the locals stack and TIBs of files that
9150: include the current file. You can change the amount of space for TIBs
9151: and locals stack at Gforth startup with the command line option
9152: @code{-l}.
9153:
9154: @item size of the pictured numeric output buffer:
9155: @cindex size of the pictured numeric output buffer
9156: @cindex pictured numeric output buffer, size
9157: @code{PAD HERE - .}. 104 characters on 32-bit machines. The buffer is
9158: shared with @code{WORD}.
9159:
9160: @item size of the scratch area returned by @code{PAD}:
9161: @cindex size of the scratch area returned by @code{PAD}
9162: @cindex @code{PAD} size
9163: The remainder of dictionary space. @code{unused pad here - - .}.
9164:
9165: @item system case-sensitivity characteristics:
9166: @cindex case-sensitivity characteristics
9167: Dictionary searches are case-insensitive (except in
9168: @code{TABLE}s). However, as explained above under @i{character-set
9169: extensions}, the matching for non-ASCII characters is determined by the
9170: locale you are using. In the default @code{C} locale all non-ASCII
9171: characters are matched case-sensitively.
9172:
9173: @item system prompt:
9174: @cindex system prompt
9175: @cindex prompt
9176: @code{ ok} in interpret state, @code{ compiled} in compile state.
9177:
9178: @item division rounding:
9179: @cindex division rounding
9180: installation dependent. @code{s" floored" environment? drop .}. We leave
9181: the choice to @code{gcc} (what to use for @code{/}) and to you (whether
9182: to use @code{fm/mod}, @code{sm/rem} or simply @code{/}).
9183:
9184: @item values of @code{STATE} when true:
9185: @cindex @code{STATE} values
9186: -1.
9187:
9188: @item values returned after arithmetic overflow:
9189: On two's complement machines, arithmetic is performed modulo
9190: 2**bits-per-cell for single arithmetic and 4**bits-per-cell for double
9191: arithmetic (with appropriate mapping for signed types). Division by zero
9192: typically results in a @code{-55 throw} (Floating-point unidentified
9193: fault), although a @code{-10 throw} (divide by zero) would be more
9194: appropriate.
9195:
9196: @item whether the current definition can be found after @t{DOES>}:
9197: @cindex @t{DOES>}, visibility of current definition
9198: No.
9199:
9200: @end table
9201:
9202: @c ---------------------------------------------------------------------
9203: @node core-ambcond, core-other, core-idef, The Core Words
9204: @subsection Ambiguous conditions
9205: @c ---------------------------------------------------------------------
9206: @cindex core words, ambiguous conditions
9207: @cindex ambiguous conditions, core words
9208:
9209: @table @i
9210:
9211: @item a name is neither a word nor a number:
9212: @cindex name not found
9213: @cindex undefined word
9214: @code{-13 throw} (Undefined word). Actually, @code{-13 bounce}, which
9215: preserves the data and FP stack, so you don't lose more work than
9216: necessary.
9217:
9218: @item a definition name exceeds the maximum length allowed:
9219: @cindex word name too long
9220: @code{-19 throw} (Word name too long)
9221:
9222: @item addressing a region not inside the various data spaces of the forth system:
9223: @cindex Invalid memory address
9224: The stacks, code space and header space are accessible. Machine code space is
9225: typically readable. Accessing other addresses gives results dependent on
9226: the operating system. On decent systems: @code{-9 throw} (Invalid memory
9227: address).
9228:
9229: @item argument type incompatible with parameter:
9230: @cindex argument type mismatch
9231: This is usually not caught. Some words perform checks, e.g., the control
9232: flow words, and issue a @code{ABORT"} or @code{-12 THROW} (Argument type
9233: mismatch).
9234:
9235: @item attempting to obtain the execution token of a word with undefined execution semantics:
9236: @cindex Interpreting a compile-only word, for @code{'} etc.
9237: @cindex execution token of words with undefined execution semantics
9238: @code{-14 throw} (Interpreting a compile-only word). In some cases, you
9239: get an execution token for @code{compile-only-error} (which performs a
9240: @code{-14 throw} when executed).
9241:
9242: @item dividing by zero:
9243: @cindex dividing by zero
9244: @cindex floating point unidentified fault, integer division
9245: On better platforms, this produces a @code{-10 throw} (Division by
9246: zero); on other systems, this typically results in a @code{-55 throw}
9247: (Floating-point unidentified fault).
9248:
9249: @item insufficient data stack or return stack space:
9250: @cindex insufficient data stack or return stack space
9251: @cindex stack overflow
9252: @cindex address alignment exception, stack overflow
9253: @cindex Invalid memory address, stack overflow
9254: Depending on the operating system, the installation, and the invocation
9255: of Gforth, this is either checked by the memory management hardware, or
9256: it is not checked. If it is checked, you typically get a @code{-3 throw}
9257: (Stack overflow), @code{-5 throw} (Return stack overflow), or @code{-9
9258: throw} (Invalid memory address) (depending on the platform and how you
9259: achieved the overflow) as soon as the overflow happens. If it is not
9260: checked, overflows typically result in mysterious illegal memory
9261: accesses, producing @code{-9 throw} (Invalid memory address) or
9262: @code{-23 throw} (Address alignment exception); they might also destroy
9263: the internal data structure of @code{ALLOCATE} and friends, resulting in
9264: various errors in these words.
9265:
9266: @item insufficient space for loop control parameters:
9267: @cindex insufficient space for loop control parameters
9268: like other return stack overflows.
9269:
9270: @item insufficient space in the dictionary:
9271: @cindex insufficient space in the dictionary
9272: @cindex dictionary overflow
9273: If you try to allot (either directly with @code{allot}, or indirectly
9274: with @code{,}, @code{create} etc.) more memory than available in the
9275: dictionary, you get a @code{-8 throw} (Dictionary overflow). If you try
9276: to access memory beyond the end of the dictionary, the results are
9277: similar to stack overflows.
9278:
9279: @item interpreting a word with undefined interpretation semantics:
9280: @cindex interpreting a word with undefined interpretation semantics
9281: @cindex Interpreting a compile-only word
9282: For some words, we have defined interpretation semantics. For the
9283: others: @code{-14 throw} (Interpreting a compile-only word).
9284:
9285: @item modifying the contents of the input buffer or a string literal:
9286: @cindex modifying the contents of the input buffer or a string literal
9287: These are located in writable memory and can be modified.
9288:
9289: @item overflow of the pictured numeric output string:
9290: @cindex overflow of the pictured numeric output string
9291: @cindex pictured numeric output string, overflow
9292: @code{-17 throw} (Pictured numeric ouput string overflow).
9293:
9294: @item parsed string overflow:
9295: @cindex parsed string overflow
9296: @code{PARSE} cannot overflow. @code{WORD} does not check for overflow.
9297:
9298: @item producing a result out of range:
9299: @cindex result out of range
9300: On two's complement machines, arithmetic is performed modulo
9301: 2**bits-per-cell for single arithmetic and 4**bits-per-cell for double
9302: arithmetic (with appropriate mapping for signed types). Division by zero
9303: typically results in a @code{-10 throw} (divide by zero) or @code{-55
9304: throw} (floating point unidentified fault). @code{convert} and
9305: @code{>number} currently overflow silently.
9306:
9307: @item reading from an empty data or return stack:
9308: @cindex stack empty
9309: @cindex stack underflow
9310: @cindex return stack underflow
9311: The data stack is checked by the outer (aka text) interpreter after
9312: every word executed. If it has underflowed, a @code{-4 throw} (Stack
9313: underflow) is performed. Apart from that, stacks may be checked or not,
9314: depending on operating system, installation, and invocation. If they are
9315: caught by a check, they typically result in @code{-4 throw} (Stack
9316: underflow), @code{-6 throw} (Return stack underflow) or @code{-9 throw}
9317: (Invalid memory address), depending on the platform and which stack
9318: underflows and by how much. Note that even if the system uses checking
9319: (through the MMU), your program may have to underflow by a significant
9320: number of stack items to trigger the reaction (the reason for this is
9321: that the MMU, and therefore the checking, works with a page-size
9322: granularity). If there is no checking, the symptoms resulting from an
9323: underflow are similar to those from an overflow. Unbalanced return
9324: stack errors result in a variaty of symptoms, including @code{-9 throw}
9325: (Invalid memory address) and Illegal Instruction (typically @code{-260
9326: throw}).
9327:
9328: @item unexpected end of the input buffer, resulting in an attempt to use a zero-length string as a name:
9329: @cindex unexpected end of the input buffer
9330: @cindex zero-length string as a name
9331: @cindex Attempt to use zero-length string as a name
9332: @code{Create} and its descendants perform a @code{-16 throw} (Attempt to
9333: use zero-length string as a name). Words like @code{'} probably will not
9334: find what they search. Note that it is possible to create zero-length
9335: names with @code{nextname} (should it not?).
9336:
9337: @item @code{>IN} greater than input buffer:
9338: @cindex @code{>IN} greater than input buffer
9339: The next invocation of a parsing word returns a string with length 0.
9340:
9341: @item @code{RECURSE} appears after @code{DOES>}:
9342: @cindex @code{RECURSE} appears after @code{DOES>}
9343: Compiles a recursive call to the defining word, not to the defined word.
9344:
9345: @item argument input source different than current input source for @code{RESTORE-INPUT}:
9346: @cindex argument input source different than current input source for @code{RESTORE-INPUT}
9347: @cindex argument type mismatch, @code{RESTORE-INPUT}
9348: @cindex @code{RESTORE-INPUT}, Argument type mismatch
9349: @code{-12 THROW}. Note that, once an input file is closed (e.g., because
9350: the end of the file was reached), its source-id may be
9351: reused. Therefore, restoring an input source specification referencing a
9352: closed file may lead to unpredictable results instead of a @code{-12
9353: THROW}.
9354:
9355: In the future, Gforth may be able to restore input source specifications
9356: from other than the current input source.
9357:
9358: @item data space containing definitions gets de-allocated:
9359: @cindex data space containing definitions gets de-allocated
9360: Deallocation with @code{allot} is not checked. This typically results in
9361: memory access faults or execution of illegal instructions.
9362:
9363: @item data space read/write with incorrect alignment:
9364: @cindex data space read/write with incorrect alignment
9365: @cindex alignment faults
9366: @cindex address alignment exception
9367: Processor-dependent. Typically results in a @code{-23 throw} (Address
9368: alignment exception). Under Linux-Intel on a 486 or later processor with
9369: alignment turned on, incorrect alignment results in a @code{-9 throw}
9370: (Invalid memory address). There are reportedly some processors with
9371: alignment restrictions that do not report violations.
9372:
9373: @item data space pointer not properly aligned, @code{,}, @code{C,}:
9374: @cindex data space pointer not properly aligned, @code{,}, @code{C,}
9375: Like other alignment errors.
9376:
9377: @item less than u+2 stack items (@code{PICK} and @code{ROLL}):
9378: Like other stack underflows.
9379:
9380: @item loop control parameters not available:
9381: @cindex loop control parameters not available
9382: Not checked. The counted loop words simply assume that the top of return
9383: stack items are loop control parameters and behave accordingly.
9384:
9385: @item most recent definition does not have a name (@code{IMMEDIATE}):
9386: @cindex most recent definition does not have a name (@code{IMMEDIATE})
9387: @cindex last word was headerless
9388: @code{abort" last word was headerless"}.
9389:
9390: @item name not defined by @code{VALUE} used by @code{TO}:
9391: @cindex name not defined by @code{VALUE} used by @code{TO}
9392: @cindex @code{TO} on non-@code{VALUE}s
9393: @cindex Invalid name argument, @code{TO}
9394: @code{-32 throw} (Invalid name argument) (unless name is a local or was
9395: defined by @code{CONSTANT}; in the latter case it just changes the constant).
9396:
9397: @item name not found (@code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]}):
9398: @cindex name not found (@code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]})
9399: @cindex undefined word, @code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]}
9400: @code{-13 throw} (Undefined word)
9401:
9402: @item parameters are not of the same type (@code{DO}, @code{?DO}, @code{WITHIN}):
9403: @cindex parameters are not of the same type (@code{DO}, @code{?DO}, @code{WITHIN})
9404: Gforth behaves as if they were of the same type. I.e., you can predict
9405: the behaviour by interpreting all parameters as, e.g., signed.
9406:
9407: @item @code{POSTPONE} or @code{[COMPILE]} applied to @code{TO}:
9408: @cindex @code{POSTPONE} or @code{[COMPILE]} applied to @code{TO}
9409: Assume @code{: X POSTPONE TO ; IMMEDIATE}. @code{X} performs the
9410: compilation semantics of @code{TO}.
9411:
9412: @item String longer than a counted string returned by @code{WORD}:
9413: @cindex string longer than a counted string returned by @code{WORD}
9414: @cindex @code{WORD}, string overflow
9415: Not checked. The string will be ok, but the count will, of course,
9416: contain only the least significant bits of the length.
9417:
9418: @item u greater than or equal to the number of bits in a cell (@code{LSHIFT}, @code{RSHIFT}):
9419: @cindex @code{LSHIFT}, large shift counts
9420: @cindex @code{RSHIFT}, large shift counts
9421: Processor-dependent. Typical behaviours are returning 0 and using only
9422: the low bits of the shift count.
9423:
9424: @item word not defined via @code{CREATE}:
9425: @cindex @code{>BODY} of non-@code{CREATE}d words
9426: @code{>BODY} produces the PFA of the word no matter how it was defined.
9427:
9428: @cindex @code{DOES>} of non-@code{CREATE}d words
9429: @code{DOES>} changes the execution semantics of the last defined word no
9430: matter how it was defined. E.g., @code{CONSTANT DOES>} is equivalent to
9431: @code{CREATE , DOES>}.
9432:
9433: @item words improperly used outside @code{<#} and @code{#>}:
9434: Not checked. As usual, you can expect memory faults.
9435:
9436: @end table
9437:
9438:
9439: @c ---------------------------------------------------------------------
9440: @node core-other, , core-ambcond, The Core Words
9441: @subsection Other system documentation
9442: @c ---------------------------------------------------------------------
9443: @cindex other system documentation, core words
9444: @cindex core words, other system documentation
9445:
9446: @table @i
9447: @item nonstandard words using @code{PAD}:
9448: @cindex @code{PAD} use by nonstandard words
9449: None.
9450:
9451: @item operator's terminal facilities available:
9452: @cindex operator's terminal facilities available
9453: After processing the command line, Gforth goes into interactive mode,
9454: and you can give commands to Gforth interactively. The actual facilities
9455: available depend on how you invoke Gforth.
9456:
9457: @item program data space available:
9458: @cindex program data space available
9459: @cindex data space available
9460: @code{UNUSED .} gives the remaining dictionary space. The total
9461: dictionary space can be specified with the @code{-m} switch
9462: (@pxref{Invoking Gforth}) when Gforth starts up.
9463:
9464: @item return stack space available:
9465: @cindex return stack space available
9466: You can compute the total return stack space in cells with
9467: @code{s" RETURN-STACK-CELLS" environment? drop .}. You can specify it at
9468: startup time with the @code{-r} switch (@pxref{Invoking Gforth}).
9469:
9470: @item stack space available:
9471: @cindex stack space available
9472: You can compute the total data stack space in cells with
9473: @code{s" STACK-CELLS" environment? drop .}. You can specify it at
9474: startup time with the @code{-d} switch (@pxref{Invoking Gforth}).
9475:
9476: @item system dictionary space required, in address units:
9477: @cindex system dictionary space required, in address units
9478: Type @code{here forthstart - .} after startup. At the time of this
9479: writing, this gives 80080 (bytes) on a 32-bit system.
9480: @end table
9481:
9482:
9483: @c =====================================================================
9484: @node The optional Block word set, The optional Double Number word set, The Core Words, ANS conformance
9485: @section The optional Block word set
9486: @c =====================================================================
9487: @cindex system documentation, block words
9488: @cindex block words, system documentation
9489:
9490: @menu
9491: * block-idef:: Implementation Defined Options
9492: * block-ambcond:: Ambiguous Conditions
9493: * block-other:: Other System Documentation
9494: @end menu
9495:
9496:
9497: @c ---------------------------------------------------------------------
9498: @node block-idef, block-ambcond, The optional Block word set, The optional Block word set
9499: @subsection Implementation Defined Options
9500: @c ---------------------------------------------------------------------
9501: @cindex implementation-defined options, block words
9502: @cindex block words, implementation-defined options
9503:
9504: @table @i
9505: @item the format for display by @code{LIST}:
9506: @cindex @code{LIST} display format
9507: First the screen number is displayed, then 16 lines of 64 characters,
9508: each line preceded by the line number.
9509:
9510: @item the length of a line affected by @code{\}:
9511: @cindex length of a line affected by @code{\}
9512: @cindex @code{\}, line length in blocks
9513: 64 characters.
9514: @end table
9515:
9516:
9517: @c ---------------------------------------------------------------------
9518: @node block-ambcond, block-other, block-idef, The optional Block word set
9519: @subsection Ambiguous conditions
9520: @c ---------------------------------------------------------------------
9521: @cindex block words, ambiguous conditions
9522: @cindex ambiguous conditions, block words
9523:
9524: @table @i
9525: @item correct block read was not possible:
9526: @cindex block read not possible
9527: Typically results in a @code{throw} of some OS-derived value (between
9528: -512 and -2048). If the blocks file was just not long enough, blanks are
9529: supplied for the missing portion.
9530:
9531: @item I/O exception in block transfer:
9532: @cindex I/O exception in block transfer
9533: @cindex block transfer, I/O exception
9534: Typically results in a @code{throw} of some OS-derived value (between
9535: -512 and -2048).
9536:
9537: @item invalid block number:
9538: @cindex invalid block number
9539: @cindex block number invalid
9540: @code{-35 throw} (Invalid block number)
9541:
9542: @item a program directly alters the contents of @code{BLK}:
9543: @cindex @code{BLK}, altering @code{BLK}
9544: The input stream is switched to that other block, at the same
9545: position. If the storing to @code{BLK} happens when interpreting
9546: non-block input, the system will get quite confused when the block ends.
9547:
9548: @item no current block buffer for @code{UPDATE}:
9549: @cindex @code{UPDATE}, no current block buffer
9550: @code{UPDATE} has no effect.
9551:
9552: @end table
9553:
9554: @c ---------------------------------------------------------------------
9555: @node block-other, , block-ambcond, The optional Block word set
9556: @subsection Other system documentation
9557: @c ---------------------------------------------------------------------
9558: @cindex other system documentation, block words
9559: @cindex block words, other system documentation
9560:
9561: @table @i
9562: @item any restrictions a multiprogramming system places on the use of buffer addresses:
9563: No restrictions (yet).
9564:
9565: @item the number of blocks available for source and data:
9566: depends on your disk space.
9567:
9568: @end table
9569:
9570:
9571: @c =====================================================================
9572: @node The optional Double Number word set, The optional Exception word set, The optional Block word set, ANS conformance
9573: @section The optional Double Number word set
9574: @c =====================================================================
9575: @cindex system documentation, double words
9576: @cindex double words, system documentation
9577:
9578: @menu
9579: * double-ambcond:: Ambiguous Conditions
9580: @end menu
9581:
9582:
9583: @c ---------------------------------------------------------------------
9584: @node double-ambcond, , The optional Double Number word set, The optional Double Number word set
9585: @subsection Ambiguous conditions
9586: @c ---------------------------------------------------------------------
9587: @cindex double words, ambiguous conditions
9588: @cindex ambiguous conditions, double words
9589:
9590: @table @i
9591: @item @i{d} outside of range of @i{n} in @code{D>S}:
9592: @cindex @code{D>S}, @i{d} out of range of @i{n}
9593: The least significant cell of @i{d} is produced.
9594:
9595: @end table
9596:
9597:
9598: @c =====================================================================
9599: @node The optional Exception word set, The optional Facility word set, The optional Double Number word set, ANS conformance
9600: @section The optional Exception word set
9601: @c =====================================================================
9602: @cindex system documentation, exception words
9603: @cindex exception words, system documentation
9604:
9605: @menu
9606: * exception-idef:: Implementation Defined Options
9607: @end menu
9608:
9609:
9610: @c ---------------------------------------------------------------------
9611: @node exception-idef, , The optional Exception word set, The optional Exception word set
9612: @subsection Implementation Defined Options
9613: @c ---------------------------------------------------------------------
9614: @cindex implementation-defined options, exception words
9615: @cindex exception words, implementation-defined options
9616:
9617: @table @i
9618: @item @code{THROW}-codes used in the system:
9619: @cindex @code{THROW}-codes used in the system
9620: The codes -256@minus{}-511 are used for reporting signals. The mapping
9621: from OS signal numbers to throw codes is -256@minus{}@i{signal}. The
9622: codes -512@minus{}-2047 are used for OS errors (for file and memory
9623: allocation operations). The mapping from OS error numbers to throw codes
9624: is -512@minus{}@code{errno}. One side effect of this mapping is that
9625: undefined OS errors produce a message with a strange number; e.g.,
9626: @code{-1000 THROW} results in @code{Unknown error 488} on my system.
9627: @end table
9628:
9629: @c =====================================================================
9630: @node The optional Facility word set, The optional File-Access word set, The optional Exception word set, ANS conformance
9631: @section The optional Facility word set
9632: @c =====================================================================
9633: @cindex system documentation, facility words
9634: @cindex facility words, system documentation
9635:
9636: @menu
9637: * facility-idef:: Implementation Defined Options
9638: * facility-ambcond:: Ambiguous Conditions
9639: @end menu
9640:
9641:
9642: @c ---------------------------------------------------------------------
9643: @node facility-idef, facility-ambcond, The optional Facility word set, The optional Facility word set
9644: @subsection Implementation Defined Options
9645: @c ---------------------------------------------------------------------
9646: @cindex implementation-defined options, facility words
9647: @cindex facility words, implementation-defined options
9648:
9649: @table @i
9650: @item encoding of keyboard events (@code{EKEY}):
9651: @cindex keyboard events, encoding in @code{EKEY}
9652: @cindex @code{EKEY}, encoding of keyboard events
9653: Keys corresponding to ASCII characters are encoded as ASCII characters.
9654: Other keys are encoded with the constants \code{k-left}, \code{k-right},
9655: \code{k-up}, \code{k-down}, \code{k-home}, \code{k-end}, \code{k1},
9656: \code{k2}, \code{k3}, \code{k4}, \code{k5}, \code{k6}, \code{k7},
9657: \code{k8}, \code{k9}, \code{k10}, \code{k11}, \code{k12}.
9658:
9659:
9660: @item duration of a system clock tick:
9661: @cindex duration of a system clock tick
9662: @cindex clock tick duration
9663: System dependent. With respect to @code{MS}, the time is specified in
9664: microseconds. How well the OS and the hardware implement this, is
9665: another question.
9666:
9667: @item repeatability to be expected from the execution of @code{MS}:
9668: @cindex repeatability to be expected from the execution of @code{MS}
9669: @cindex @code{MS}, repeatability to be expected
9670: System dependent. On Unix, a lot depends on load. If the system is
9671: lightly loaded, and the delay is short enough that Gforth does not get
9672: swapped out, the performance should be acceptable. Under MS-DOS and
9673: other single-tasking systems, it should be good.
9674:
9675: @end table
9676:
9677:
9678: @c ---------------------------------------------------------------------
9679: @node facility-ambcond, , facility-idef, The optional Facility word set
9680: @subsection Ambiguous conditions
9681: @c ---------------------------------------------------------------------
9682: @cindex facility words, ambiguous conditions
9683: @cindex ambiguous conditions, facility words
9684:
9685: @table @i
9686: @item @code{AT-XY} can't be performed on user output device:
9687: @cindex @code{AT-XY} can't be performed on user output device
9688: Largely terminal dependent. No range checks are done on the arguments.
9689: No errors are reported. You may see some garbage appearing, you may see
9690: simply nothing happen.
9691:
9692: @end table
9693:
9694:
9695: @c =====================================================================
9696: @node The optional File-Access word set, The optional Floating-Point word set, The optional Facility word set, ANS conformance
9697: @section The optional File-Access word set
9698: @c =====================================================================
9699: @cindex system documentation, file words
9700: @cindex file words, system documentation
9701:
9702: @menu
9703: * file-idef:: Implementation Defined Options
9704: * file-ambcond:: Ambiguous Conditions
9705: @end menu
9706:
9707: @c ---------------------------------------------------------------------
9708: @node file-idef, file-ambcond, The optional File-Access word set, The optional File-Access word set
9709: @subsection Implementation Defined Options
9710: @c ---------------------------------------------------------------------
9711: @cindex implementation-defined options, file words
9712: @cindex file words, implementation-defined options
9713:
9714: @table @i
9715: @item file access methods used:
9716: @cindex file access methods used
9717: @code{R/O}, @code{R/W} and @code{BIN} work as you would
9718: expect. @code{W/O} translates into the C file opening mode @code{w} (or
9719: @code{wb}): The file is cleared, if it exists, and created, if it does
9720: not (with both @code{open-file} and @code{create-file}). Under Unix
9721: @code{create-file} creates a file with 666 permissions modified by your
9722: umask.
9723:
9724: @item file exceptions:
9725: @cindex file exceptions
9726: The file words do not raise exceptions (except, perhaps, memory access
9727: faults when you pass illegal addresses or file-ids).
9728:
9729: @item file line terminator:
9730: @cindex file line terminator
9731: System-dependent. Gforth uses C's newline character as line
9732: terminator. What the actual character code(s) of this are is
9733: system-dependent.
9734:
9735: @item file name format:
9736: @cindex file name format
9737: System dependent. Gforth just uses the file name format of your OS.
9738:
9739: @item information returned by @code{FILE-STATUS}:
9740: @cindex @code{FILE-STATUS}, returned information
9741: @code{FILE-STATUS} returns the most powerful file access mode allowed
9742: for the file: Either @code{R/O}, @code{W/O} or @code{R/W}. If the file
9743: cannot be accessed, @code{R/O BIN} is returned. @code{BIN} is applicable
9744: along with the returned mode.
9745:
9746: @item input file state after an exception when including source:
9747: @cindex exception when including source
9748: All files that are left via the exception are closed.
9749:
9750: @item @i{ior} values and meaning:
9751: @cindex @i{ior} values and meaning
9752: The @i{ior}s returned by the file and memory allocation words are
9753: intended as throw codes. They typically are in the range
9754: -512@minus{}-2047 of OS errors. The mapping from OS error numbers to
9755: @i{ior}s is -512@minus{}@i{errno}.
9756:
9757: @item maximum depth of file input nesting:
9758: @cindex maximum depth of file input nesting
9759: @cindex file input nesting, maximum depth
9760: limited by the amount of return stack, locals/TIB stack, and the number
9761: of open files available. This should not give you troubles.
9762:
9763: @item maximum size of input line:
9764: @cindex maximum size of input line
9765: @cindex input line size, maximum
9766: @code{/line}. Currently 255.
9767:
9768: @item methods of mapping block ranges to files:
9769: @cindex mapping block ranges to files
9770: @cindex files containing blocks
9771: @cindex blocks in files
9772: By default, blocks are accessed in the file @file{blocks.fb} in the
9773: current working directory. The file can be switched with @code{USE}.
9774:
9775: @item number of string buffers provided by @code{S"}:
9776: @cindex @code{S"}, number of string buffers
9777: 1
9778:
9779: @item size of string buffer used by @code{S"}:
9780: @cindex @code{S"}, size of string buffer
9781: @code{/line}. currently 255.
9782:
9783: @end table
9784:
9785: @c ---------------------------------------------------------------------
9786: @node file-ambcond, , file-idef, The optional File-Access word set
9787: @subsection Ambiguous conditions
9788: @c ---------------------------------------------------------------------
9789: @cindex file words, ambiguous conditions
9790: @cindex ambiguous conditions, file words
9791:
9792: @table @i
9793: @item attempting to position a file outside its boundaries:
9794: @cindex @code{REPOSITION-FILE}, outside the file's boundaries
9795: @code{REPOSITION-FILE} is performed as usual: Afterwards,
9796: @code{FILE-POSITION} returns the value given to @code{REPOSITION-FILE}.
9797:
9798: @item attempting to read from file positions not yet written:
9799: @cindex reading from file positions not yet written
9800: End-of-file, i.e., zero characters are read and no error is reported.
9801:
9802: @item @i{file-id} is invalid (@code{INCLUDE-FILE}):
9803: @cindex @code{INCLUDE-FILE}, @i{file-id} is invalid
9804: An appropriate exception may be thrown, but a memory fault or other
9805: problem is more probable.
9806:
9807: @item I/O exception reading or closing @i{file-id} (@code{INCLUDE-FILE}, @code{INCLUDED}):
9808: @cindex @code{INCLUDE-FILE}, I/O exception reading or closing @i{file-id}
9809: @cindex @code{INCLUDED}, I/O exception reading or closing @i{file-id}
9810: The @i{ior} produced by the operation, that discovered the problem, is
9811: thrown.
9812:
9813: @item named file cannot be opened (@code{INCLUDED}):
9814: @cindex @code{INCLUDED}, named file cannot be opened
9815: The @i{ior} produced by @code{open-file} is thrown.
9816:
9817: @item requesting an unmapped block number:
9818: @cindex unmapped block numbers
9819: There are no unmapped legal block numbers. On some operating systems,
9820: writing a block with a large number may overflow the file system and
9821: have an error message as consequence.
9822:
9823: @item using @code{source-id} when @code{blk} is non-zero:
9824: @cindex @code{SOURCE-ID}, behaviour when @code{BLK} is non-zero
9825: @code{source-id} performs its function. Typically it will give the id of
9826: the source which loaded the block. (Better ideas?)
9827:
9828: @end table
9829:
9830:
9831: @c =====================================================================
9832: @node The optional Floating-Point word set, The optional Locals word set, The optional File-Access word set, ANS conformance
9833: @section The optional Floating-Point word set
9834: @c =====================================================================
9835: @cindex system documentation, floating-point words
9836: @cindex floating-point words, system documentation
9837:
9838: @menu
9839: * floating-idef:: Implementation Defined Options
9840: * floating-ambcond:: Ambiguous Conditions
9841: @end menu
9842:
9843:
9844: @c ---------------------------------------------------------------------
9845: @node floating-idef, floating-ambcond, The optional Floating-Point word set, The optional Floating-Point word set
9846: @subsection Implementation Defined Options
9847: @c ---------------------------------------------------------------------
9848: @cindex implementation-defined options, floating-point words
9849: @cindex floating-point words, implementation-defined options
9850:
9851: @table @i
9852: @item format and range of floating point numbers:
9853: @cindex format and range of floating point numbers
9854: @cindex floating point numbers, format and range
9855: System-dependent; the @code{double} type of C.
9856:
9857: @item results of @code{REPRESENT} when @i{float} is out of range:
9858: @cindex @code{REPRESENT}, results when @i{float} is out of range
9859: System dependent; @code{REPRESENT} is implemented using the C library
9860: function @code{ecvt()} and inherits its behaviour in this respect.
9861:
9862: @item rounding or truncation of floating-point numbers:
9863: @cindex rounding of floating-point numbers
9864: @cindex truncation of floating-point numbers
9865: @cindex floating-point numbers, rounding or truncation
9866: System dependent; the rounding behaviour is inherited from the hosting C
9867: compiler. IEEE-FP-based (i.e., most) systems by default round to
9868: nearest, and break ties by rounding to even (i.e., such that the last
9869: bit of the mantissa is 0).
9870:
9871: @item size of floating-point stack:
9872: @cindex floating-point stack size
9873: @code{s" FLOATING-STACK" environment? drop .} gives the total size of
9874: the floating-point stack (in floats). You can specify this on startup
9875: with the command-line option @code{-f} (@pxref{Invoking Gforth}).
9876:
9877: @item width of floating-point stack:
9878: @cindex floating-point stack width
9879: @code{1 floats}.
9880:
9881: @end table
9882:
9883:
9884: @c ---------------------------------------------------------------------
9885: @node floating-ambcond, , floating-idef, The optional Floating-Point word set
9886: @subsection Ambiguous conditions
9887: @c ---------------------------------------------------------------------
9888: @cindex floating-point words, ambiguous conditions
9889: @cindex ambiguous conditions, floating-point words
9890:
9891: @table @i
9892: @item @code{df@@} or @code{df!} used with an address that is not double-float aligned:
9893: @cindex @code{df@@} or @code{df!} used with an address that is not double-float aligned
9894: System-dependent. Typically results in a @code{-23 THROW} like other
9895: alignment violations.
9896:
9897: @item @code{f@@} or @code{f!} used with an address that is not float aligned:
9898: @cindex @code{f@@} used with an address that is not float aligned
9899: @cindex @code{f!} used with an address that is not float aligned
9900: System-dependent. Typically results in a @code{-23 THROW} like other
9901: alignment violations.
9902:
9903: @item floating-point result out of range:
9904: @cindex floating-point result out of range
9905: System-dependent. Can result in a @code{-55 THROW} (Floating-point
9906: unidentified fault), or can produce a special value representing, e.g.,
9907: Infinity.
9908:
9909: @item @code{sf@@} or @code{sf!} used with an address that is not single-float aligned:
9910: @cindex @code{sf@@} or @code{sf!} used with an address that is not single-float aligned
9911: System-dependent. Typically results in an alignment fault like other
9912: alignment violations.
9913:
9914: @item @code{base} is not decimal (@code{REPRESENT}, @code{F.}, @code{FE.}, @code{FS.}):
9915: @cindex @code{base} is not decimal (@code{REPRESENT}, @code{F.}, @code{FE.}, @code{FS.})
9916: The floating-point number is converted into decimal nonetheless.
9917:
9918: @item Both arguments are equal to zero (@code{FATAN2}):
9919: @cindex @code{FATAN2}, both arguments are equal to zero
9920: System-dependent. @code{FATAN2} is implemented using the C library
9921: function @code{atan2()}.
9922:
9923: @item Using @code{FTAN} on an argument @i{r1} where cos(@i{r1}) is zero:
9924: @cindex @code{FTAN} on an argument @i{r1} where cos(@i{r1}) is zero
9925: System-dependent. Anyway, typically the cos of @i{r1} will not be zero
9926: because of small errors and the tan will be a very large (or very small)
9927: but finite number.
9928:
9929: @item @i{d} cannot be presented precisely as a float in @code{D>F}:
9930: @cindex @code{D>F}, @i{d} cannot be presented precisely as a float
9931: The result is rounded to the nearest float.
9932:
9933: @item dividing by zero:
9934: @cindex dividing by zero, floating-point
9935: @cindex floating-point dividing by zero
9936: @cindex floating-point unidentified fault, FP divide-by-zero
9937: @code{-55 throw} (Floating-point unidentified fault)
9938:
9939: @item exponent too big for conversion (@code{DF!}, @code{DF@@}, @code{SF!}, @code{SF@@}):
9940: @cindex exponent too big for conversion (@code{DF!}, @code{DF@@}, @code{SF!}, @code{SF@@})
9941: System dependent. On IEEE-FP based systems the number is converted into
9942: an infinity.
9943:
9944: @item @i{float}<1 (@code{FACOSH}):
9945: @cindex @code{FACOSH}, @i{float}<1
9946: @cindex floating-point unidentified fault, @code{FACOSH}
9947: @code{-55 throw} (Floating-point unidentified fault)
9948:
9949: @item @i{float}=<-1 (@code{FLNP1}):
9950: @cindex @code{FLNP1}, @i{float}=<-1
9951: @cindex floating-point unidentified fault, @code{FLNP1}
9952: @code{-55 throw} (Floating-point unidentified fault). On IEEE-FP systems
9953: negative infinity is typically produced for @i{float}=-1.
9954:
9955: @item @i{float}=<0 (@code{FLN}, @code{FLOG}):
9956: @cindex @code{FLN}, @i{float}=<0
9957: @cindex @code{FLOG}, @i{float}=<0
9958: @cindex floating-point unidentified fault, @code{FLN} or @code{FLOG}
9959: @code{-55 throw} (Floating-point unidentified fault). On IEEE-FP systems
9960: negative infinity is typically produced for @i{float}=0.
9961:
9962: @item @i{float}<0 (@code{FASINH}, @code{FSQRT}):
9963: @cindex @code{FASINH}, @i{float}<0
9964: @cindex @code{FSQRT}, @i{float}<0
9965: @cindex floating-point unidentified fault, @code{FASINH} or @code{FSQRT}
9966: @code{-55 throw} (Floating-point unidentified fault). @code{fasinh}
9967: produces values for these inputs on my Linux box (Bug in the C library?)
9968:
9969: @item |@i{float}|>1 (@code{FACOS}, @code{FASIN}, @code{FATANH}):
9970: @cindex @code{FACOS}, |@i{float}|>1
9971: @cindex @code{FASIN}, |@i{float}|>1
9972: @cindex @code{FATANH}, |@i{float}|>1
9973: @cindex floating-point unidentified fault, @code{FACOS}, @code{FASIN} or @code{FATANH}
9974: @code{-55 throw} (Floating-point unidentified fault).
9975:
9976: @item integer part of float cannot be represented by @i{d} in @code{F>D}:
9977: @cindex @code{F>D}, integer part of float cannot be represented by @i{d}
9978: @cindex floating-point unidentified fault, @code{F>D}
9979: @code{-55 throw} (Floating-point unidentified fault).
9980:
9981: @item string larger than pictured numeric output area (@code{f.}, @code{fe.}, @code{fs.}):
9982: @cindex string larger than pictured numeric output area (@code{f.}, @code{fe.}, @code{fs.})
9983: This does not happen.
9984: @end table
9985:
9986: @c =====================================================================
9987: @node The optional Locals word set, The optional Memory-Allocation word set, The optional Floating-Point word set, ANS conformance
9988: @section The optional Locals word set
9989: @c =====================================================================
9990: @cindex system documentation, locals words
9991: @cindex locals words, system documentation
9992:
9993: @menu
9994: * locals-idef:: Implementation Defined Options
9995: * locals-ambcond:: Ambiguous Conditions
9996: @end menu
9997:
9998:
9999: @c ---------------------------------------------------------------------
10000: @node locals-idef, locals-ambcond, The optional Locals word set, The optional Locals word set
10001: @subsection Implementation Defined Options
10002: @c ---------------------------------------------------------------------
10003: @cindex implementation-defined options, locals words
10004: @cindex locals words, implementation-defined options
10005:
10006: @table @i
10007: @item maximum number of locals in a definition:
10008: @cindex maximum number of locals in a definition
10009: @cindex locals, maximum number in a definition
10010: @code{s" #locals" environment? drop .}. Currently 15. This is a lower
10011: bound, e.g., on a 32-bit machine there can be 41 locals of up to 8
10012: characters. The number of locals in a definition is bounded by the size
10013: of locals-buffer, which contains the names of the locals.
10014:
10015: @end table
10016:
10017:
10018: @c ---------------------------------------------------------------------
10019: @node locals-ambcond, , locals-idef, The optional Locals word set
10020: @subsection Ambiguous conditions
10021: @c ---------------------------------------------------------------------
10022: @cindex locals words, ambiguous conditions
10023: @cindex ambiguous conditions, locals words
10024:
10025: @table @i
10026: @item executing a named local in interpretation state:
10027: @cindex local in interpretation state
10028: @cindex Interpreting a compile-only word, for a local
10029: Locals have no interpretation semantics. If you try to perform the
10030: interpretation semantics, you will get a @code{-14 throw} somewhere
10031: (Interpreting a compile-only word). If you perform the compilation
10032: semantics, the locals access will be compiled (irrespective of state).
10033:
10034: @item @i{name} not defined by @code{VALUE} or @code{(LOCAL)} (@code{TO}):
10035: @cindex name not defined by @code{VALUE} or @code{(LOCAL)} used by @code{TO}
10036: @cindex @code{TO} on non-@code{VALUE}s and non-locals
10037: @cindex Invalid name argument, @code{TO}
10038: @code{-32 throw} (Invalid name argument)
10039:
10040: @end table
10041:
10042:
10043: @c =====================================================================
10044: @node The optional Memory-Allocation word set, The optional Programming-Tools word set, The optional Locals word set, ANS conformance
10045: @section The optional Memory-Allocation word set
10046: @c =====================================================================
10047: @cindex system documentation, memory-allocation words
10048: @cindex memory-allocation words, system documentation
10049:
10050: @menu
10051: * memory-idef:: Implementation Defined Options
10052: @end menu
10053:
10054:
10055: @c ---------------------------------------------------------------------
10056: @node memory-idef, , The optional Memory-Allocation word set, The optional Memory-Allocation word set
10057: @subsection Implementation Defined Options
10058: @c ---------------------------------------------------------------------
10059: @cindex implementation-defined options, memory-allocation words
10060: @cindex memory-allocation words, implementation-defined options
10061:
10062: @table @i
10063: @item values and meaning of @i{ior}:
10064: @cindex @i{ior} values and meaning
10065: The @i{ior}s returned by the file and memory allocation words are
10066: intended as throw codes. They typically are in the range
10067: -512@minus{}-2047 of OS errors. The mapping from OS error numbers to
10068: @i{ior}s is -512@minus{}@i{errno}.
10069:
10070: @end table
10071:
10072: @c =====================================================================
10073: @node The optional Programming-Tools word set, The optional Search-Order word set, The optional Memory-Allocation word set, ANS conformance
10074: @section The optional Programming-Tools word set
10075: @c =====================================================================
10076: @cindex system documentation, programming-tools words
10077: @cindex programming-tools words, system documentation
10078:
10079: @menu
10080: * programming-idef:: Implementation Defined Options
10081: * programming-ambcond:: Ambiguous Conditions
10082: @end menu
10083:
10084:
10085: @c ---------------------------------------------------------------------
10086: @node programming-idef, programming-ambcond, The optional Programming-Tools word set, The optional Programming-Tools word set
10087: @subsection Implementation Defined Options
10088: @c ---------------------------------------------------------------------
10089: @cindex implementation-defined options, programming-tools words
10090: @cindex programming-tools words, implementation-defined options
10091:
10092: @table @i
10093: @item ending sequence for input following @code{;CODE} and @code{CODE}:
10094: @cindex @code{;CODE} ending sequence
10095: @cindex @code{CODE} ending sequence
10096: @code{END-CODE}
10097:
10098: @item manner of processing input following @code{;CODE} and @code{CODE}:
10099: @cindex @code{;CODE}, processing input
10100: @cindex @code{CODE}, processing input
10101: The @code{ASSEMBLER} vocabulary is pushed on the search order stack, and
10102: the input is processed by the text interpreter, (starting) in interpret
10103: state.
10104:
10105: @item search order capability for @code{EDITOR} and @code{ASSEMBLER}:
10106: @cindex @code{ASSEMBLER}, search order capability
10107: The ANS Forth search order word set.
10108:
10109: @item source and format of display by @code{SEE}:
10110: @cindex @code{SEE}, source and format of output
10111: The source for @code{see} is the intermediate code used by the inner
10112: interpreter. The current @code{see} tries to output Forth source code
10113: as well as possible.
10114:
10115: @end table
10116:
10117: @c ---------------------------------------------------------------------
10118: @node programming-ambcond, , programming-idef, The optional Programming-Tools word set
10119: @subsection Ambiguous conditions
10120: @c ---------------------------------------------------------------------
10121: @cindex programming-tools words, ambiguous conditions
10122: @cindex ambiguous conditions, programming-tools words
10123:
10124: @table @i
10125:
10126: @item deleting the compilation word list (@code{FORGET}):
10127: @cindex @code{FORGET}, deleting the compilation word list
10128: Not implemented (yet).
10129:
10130: @item fewer than @i{u}+1 items on the control-flow stack (@code{CS-PICK}, @code{CS-ROLL}):
10131: @cindex @code{CS-PICK}, fewer than @i{u}+1 items on the control flow-stack
10132: @cindex @code{CS-ROLL}, fewer than @i{u}+1 items on the control flow-stack
10133: @cindex control-flow stack underflow
10134: This typically results in an @code{abort"} with a descriptive error
10135: message (may change into a @code{-22 throw} (Control structure mismatch)
10136: in the future). You may also get a memory access error. If you are
10137: unlucky, this ambiguous condition is not caught.
10138:
10139: @item @i{name} can't be found (@code{FORGET}):
10140: @cindex @code{FORGET}, @i{name} can't be found
10141: Not implemented (yet).
10142:
10143: @item @i{name} not defined via @code{CREATE}:
10144: @cindex @code{;CODE}, @i{name} not defined via @code{CREATE}
10145: @code{;CODE} behaves like @code{DOES>} in this respect, i.e., it changes
10146: the execution semantics of the last defined word no matter how it was
10147: defined.
10148:
10149: @item @code{POSTPONE} applied to @code{[IF]}:
10150: @cindex @code{POSTPONE} applied to @code{[IF]}
10151: @cindex @code{[IF]} and @code{POSTPONE}
10152: After defining @code{: X POSTPONE [IF] ; IMMEDIATE}. @code{X} is
10153: equivalent to @code{[IF]}.
10154:
10155: @item reaching the end of the input source before matching @code{[ELSE]} or @code{[THEN]}:
10156: @cindex @code{[IF]}, end of the input source before matching @code{[ELSE]} or @code{[THEN]}
10157: Continue in the same state of conditional compilation in the next outer
10158: input source. Currently there is no warning to the user about this.
10159:
10160: @item removing a needed definition (@code{FORGET}):
10161: @cindex @code{FORGET}, removing a needed definition
10162: Not implemented (yet).
10163:
10164: @end table
10165:
10166:
10167: @c =====================================================================
10168: @node The optional Search-Order word set, , The optional Programming-Tools word set, ANS conformance
10169: @section The optional Search-Order word set
10170: @c =====================================================================
10171: @cindex system documentation, search-order words
10172: @cindex search-order words, system documentation
10173:
10174: @menu
10175: * search-idef:: Implementation Defined Options
10176: * search-ambcond:: Ambiguous Conditions
10177: @end menu
10178:
10179:
10180: @c ---------------------------------------------------------------------
10181: @node search-idef, search-ambcond, The optional Search-Order word set, The optional Search-Order word set
10182: @subsection Implementation Defined Options
10183: @c ---------------------------------------------------------------------
10184: @cindex implementation-defined options, search-order words
10185: @cindex search-order words, implementation-defined options
10186:
10187: @table @i
10188: @item maximum number of word lists in search order:
10189: @cindex maximum number of word lists in search order
10190: @cindex search order, maximum depth
10191: @code{s" wordlists" environment? drop .}. Currently 16.
10192:
10193: @item minimum search order:
10194: @cindex minimum search order
10195: @cindex search order, minimum
10196: @code{root root}.
10197:
10198: @end table
10199:
10200: @c ---------------------------------------------------------------------
10201: @node search-ambcond, , search-idef, The optional Search-Order word set
10202: @subsection Ambiguous conditions
10203: @c ---------------------------------------------------------------------
10204: @cindex search-order words, ambiguous conditions
10205: @cindex ambiguous conditions, search-order words
10206:
10207: @table @i
10208: @item changing the compilation word list (during compilation):
10209: @cindex changing the compilation word list (during compilation)
10210: @cindex compilation word list, change before definition ends
10211: The word is entered into the word list that was the compilation word list
10212: at the start of the definition. Any changes to the name field (e.g.,
10213: @code{immediate}) or the code field (e.g., when executing @code{DOES>})
10214: are applied to the latest defined word (as reported by @code{last} or
10215: @code{lastxt}), if possible, irrespective of the compilation word list.
10216:
10217: @item search order empty (@code{previous}):
10218: @cindex @code{previous}, search order empty
10219: @cindex vocstack empty, @code{previous}
10220: @code{abort" Vocstack empty"}.
10221:
10222: @item too many word lists in search order (@code{also}):
10223: @cindex @code{also}, too many word lists in search order
10224: @cindex vocstack full, @code{also}
10225: @code{abort" Vocstack full"}.
10226:
10227: @end table
10228:
10229: @c ***************************************************************
10230: @node Model, Integrating Gforth, ANS conformance, Top
10231: @chapter Model
10232:
10233: This chapter has yet to be written. It will contain information, on
10234: which internal structures you can rely.
10235:
10236: @c ***************************************************************
10237: @node Integrating Gforth, Emacs and Gforth, Model, Top
10238: @chapter Integrating Gforth into C programs
10239:
10240: This is not yet implemented.
10241:
10242: Several people like to use Forth as scripting language for applications
10243: that are otherwise written in C, C++, or some other language.
10244:
10245: The Forth system ATLAST provides facilities for embedding it into
10246: applications; unfortunately it has several disadvantages: most
10247: importantly, it is not based on ANS Forth, and it is apparently dead
10248: (i.e., not developed further and not supported). The facilities
10249: provided by Gforth in this area are inspired by ATLAST's facilities, so
10250: making the switch should not be hard.
10251:
10252: We also tried to design the interface such that it can easily be
10253: implemented by other Forth systems, so that we may one day arrive at a
10254: standardized interface. Such a standard interface would allow you to
10255: replace the Forth system without having to rewrite C code.
10256:
10257: You embed the Gforth interpreter by linking with the library
10258: @code{libgforth.a} (give the compiler the option @code{-lgforth}). All
10259: global symbols in this library that belong to the interface, have the
10260: prefix @code{forth_}. (Global symbols that are used internally have the
10261: prefix @code{gforth_}).
10262:
10263: You can include the declarations of Forth types and the functions and
10264: variables of the interface with @code{#include <forth.h>}.
10265:
10266: Types.
10267:
10268: Variables.
10269:
10270: Data and FP Stack pointer. Area sizes.
10271:
10272: functions.
10273:
10274: forth_init(imagefile)
10275: forth_evaluate(string) exceptions?
10276: forth_goto(address) (or forth_execute(xt)?)
10277: forth_continue() (a corountining mechanism)
10278:
10279: Adding primitives.
10280:
10281: No checking.
10282:
10283: Signals?
10284:
10285: Accessing the Stacks
10286:
10287: @c ******************************************************************
10288: @node Emacs and Gforth, Image Files, Integrating Gforth, Top
10289: @chapter Emacs and Gforth
10290: @cindex Emacs and Gforth
10291:
10292: @cindex @file{gforth.el}
10293: @cindex @file{forth.el}
10294: @cindex Rydqvist, Goran
10295: @cindex comment editing commands
10296: @cindex @code{\}, editing with Emacs
10297: @cindex debug tracer editing commands
10298: @cindex @code{~~}, removal with Emacs
10299: @cindex Forth mode in Emacs
10300: Gforth comes with @file{gforth.el}, an improved version of
10301: @file{forth.el} by Goran Rydqvist (included in the TILE package). The
10302: improvements are:
10303:
10304: @itemize @bullet
10305: @item
10306: A better (but still not perfect) handling of indentation.
10307: @item
10308: Comment paragraph filling (@kbd{M-q})
10309: @item
10310: Commenting (@kbd{C-x \}) and uncommenting (@kbd{C-u C-x \}) of regions
10311: @item
10312: Removal of debugging tracers (@kbd{C-x ~}, @pxref{Debugging}).
10313: @end itemize
10314:
10315: I left the stuff I do not use alone, even though some of it only makes
10316: sense for TILE. To get a description of these features, enter Forth mode
10317: and type @kbd{C-h m}.
10318:
10319: @cindex source location of error or debugging output in Emacs
10320: @cindex error output, finding the source location in Emacs
10321: @cindex debugging output, finding the source location in Emacs
10322: In addition, Gforth supports Emacs quite well: The source code locations
10323: given in error messages, debugging output (from @code{~~}) and failed
10324: assertion messages are in the right format for Emacs' compilation mode
10325: (@pxref{Compilation, , Running Compilations under Emacs, emacs, Emacs
10326: Manual}) so the source location corresponding to an error or other
10327: message is only a few keystrokes away (@kbd{C-x `} for the next error,
10328: @kbd{C-c C-c} for the error under the cursor).
10329:
10330: @cindex @file{TAGS} file
10331: @cindex @file{etags.fs}
10332: @cindex viewing the source of a word in Emacs
10333: Also, if you @code{include} @file{etags.fs}, a new @file{TAGS} file will
10334: be produced (@pxref{Tags, , Tags Tables, emacs, Emacs Manual}) that
10335: contains the definitions of all words defined afterwards. You can then
10336: find the source for a word using @kbd{M-.}. Note that emacs can use
10337: several tags files at the same time (e.g., one for the Gforth sources
10338: and one for your program, @pxref{Select Tags Table,,Selecting a Tags
10339: Table,emacs, Emacs Manual}). The TAGS file for the preloaded words is
10340: @file{$(datadir)/gforth/$(VERSION)/TAGS} (e.g.,
10341: @file{/usr/local/share/gforth/0.2.0/TAGS}).
10342:
10343: @cindex @file{.emacs}
10344: To get all these benefits, add the following lines to your @file{.emacs}
10345: file:
10346:
10347: @example
10348: (autoload 'forth-mode "gforth.el")
10349: (setq auto-mode-alist (cons '("\\.fs\\'" . forth-mode) auto-mode-alist))
10350: @end example
10351:
10352: @c ******************************************************************
10353: @node Image Files, Engine, Emacs and Gforth, Top
10354: @chapter Image Files
10355: @cindex image file
10356: @cindex @file{.fi} files
10357: @cindex precompiled Forth code
10358: @cindex dictionary in persistent form
10359: @cindex persistent form of dictionary
10360:
10361: An image file is a file containing an image of the Forth dictionary,
10362: i.e., compiled Forth code and data residing in the dictionary. By
10363: convention, we use the extension @code{.fi} for image files.
10364:
10365: @menu
10366: * Image Licensing Issues:: Distribution terms for images.
10367: * Image File Background:: Why have image files?
10368: * Non-Relocatable Image Files:: don't always work.
10369: * Data-Relocatable Image Files:: are better.
10370: * Fully Relocatable Image Files:: better yet.
10371: * Stack and Dictionary Sizes:: Setting the default sizes for an image.
10372: * Running Image Files:: @code{gforth -i @i{file}} or @i{file}.
10373: * Modifying the Startup Sequence:: and turnkey applications.
10374: @end menu
10375:
10376: @node Image Licensing Issues, Image File Background, Image Files, Image Files
10377: @section Image Licensing Issues
10378: @cindex license for images
10379: @cindex image license
10380:
10381: An image created with @code{gforthmi} (@pxref{gforthmi}) or
10382: @code{savesystem} (@pxref{Non-Relocatable Image Files}) includes the
10383: original image; i.e., according to copyright law it is a derived work of
10384: the original image.
10385:
10386: Since Gforth is distributed under the GNU GPL, the newly created image
10387: falls under the GNU GPL, too. In particular, this means that if you
10388: distribute the image, you have to make all of the sources for the image
10389: available, including those you wrote. For details see @ref{License, ,
10390: GNU General Public License (Section 3)}.
10391:
10392: If you create an image with @code{cross} (@pxref{cross.fs}), the image
10393: contains only code compiled from the sources you gave it; if none of
10394: these sources is under the GPL, the terms discussed above do not apply
10395: to the image. However, if your image needs an engine (a gforth binary)
10396: that is under the GPL, you should make sure that you distribute both in
10397: a way that is at most a @emph{mere aggregation}, if you don't want the
10398: terms of the GPL to apply to the image.
10399:
10400: @node Image File Background, Non-Relocatable Image Files, Image Licensing Issues, Image Files
10401: @section Image File Background
10402: @cindex image file background
10403:
10404: Our Forth system consists not only of primitives, but also of
10405: definitions written in Forth. Since the Forth compiler itself belongs to
10406: those definitions, it is not possible to start the system with the
10407: primitives and the Forth source alone. Therefore we provide the Forth
10408: code as an image file in nearly executable form. When Gforth starts up,
10409: a C routine loads the image file into memory, optionally relocates the
10410: addresses, then sets up the memory (stacks etc.) according to
10411: information in the image file, and (finally) starts executing Forth
10412: code.
10413:
10414: The image file variants represent different compromises between the
10415: goals of making it easy to generate image files and making them
10416: portable.
10417:
10418: @cindex relocation at run-time
10419: Win32Forth 3.4 and Mitch Bradley's @code{cforth} use relocation at
10420: run-time. This avoids many of the complications discussed below (image
10421: files are data relocatable without further ado), but costs performance
10422: (one addition per memory access).
10423:
10424: @cindex relocation at load-time
10425: By contrast, the Gforth loader performs relocation at image load time. The
10426: loader also has to replace tokens that represent primitive calls with the
10427: appropriate code-field addresses (or code addresses in the case of
10428: direct threading).
10429:
10430: There are three kinds of image files, with different degrees of
10431: relocatability: non-relocatable, data-relocatable, and fully relocatable
10432: image files.
10433:
10434: @cindex image file loader
10435: @cindex relocating loader
10436: @cindex loader for image files
10437: These image file variants have several restrictions in common; they are
10438: caused by the design of the image file loader:
10439:
10440: @itemize @bullet
10441: @item
10442: There is only one segment; in particular, this means, that an image file
10443: cannot represent @code{ALLOCATE}d memory chunks (and pointers to
10444: them). The contents of the stacks are not represented, either.
10445:
10446: @item
10447: The only kinds of relocation supported are: adding the same offset to
10448: all cells that represent data addresses; and replacing special tokens
10449: with code addresses or with pieces of machine code.
10450:
10451: If any complex computations involving addresses are performed, the
10452: results cannot be represented in the image file. Several applications that
10453: use such computations come to mind:
10454: @itemize @minus
10455: @item
10456: Hashing addresses (or data structures which contain addresses) for table
10457: lookup. If you use Gforth's @code{table}s or @code{wordlist}s for this
10458: purpose, you will have no problem, because the hash tables are
10459: recomputed automatically when the system is started. If you use your own
10460: hash tables, you will have to do something similar.
10461:
10462: @item
10463: There's a cute implementation of doubly-linked lists that uses
10464: @code{XOR}ed addresses. You could represent such lists as singly-linked
10465: in the image file, and restore the doubly-linked representation on
10466: startup.@footnote{In my opinion, though, you should think thrice before
10467: using a doubly-linked list (whatever implementation).}
10468:
10469: @item
10470: The code addresses of run-time routines like @code{docol:} cannot be
10471: represented in the image file (because their tokens would be replaced by
10472: machine code in direct threaded implementations). As a workaround,
10473: compute these addresses at run-time with @code{>code-address} from the
10474: executions tokens of appropriate words (see the definitions of
10475: @code{docol:} and friends in @file{kernel.fs}).
10476:
10477: @item
10478: On many architectures addresses are represented in machine code in some
10479: shifted or mangled form. You cannot put @code{CODE} words that contain
10480: absolute addresses in this form in a relocatable image file. Workarounds
10481: are representing the address in some relative form (e.g., relative to
10482: the CFA, which is present in some register), or loading the address from
10483: a place where it is stored in a non-mangled form.
10484: @end itemize
10485: @end itemize
10486:
10487: @node Non-Relocatable Image Files, Data-Relocatable Image Files, Image File Background, Image Files
10488: @section Non-Relocatable Image Files
10489: @cindex non-relocatable image files
10490: @cindex image file, non-relocatable
10491:
10492: These files are simple memory dumps of the dictionary. They are specific
10493: to the executable (i.e., @file{gforth} file) they were created
10494: with. What's worse, they are specific to the place on which the
10495: dictionary resided when the image was created. Now, there is no
10496: guarantee that the dictionary will reside at the same place the next
10497: time you start Gforth, so there's no guarantee that a non-relocatable
10498: image will work the next time (Gforth will complain instead of crashing,
10499: though).
10500:
10501: You can create a non-relocatable image file with
10502:
10503: doc-savesystem
10504:
10505: @node Data-Relocatable Image Files, Fully Relocatable Image Files, Non-Relocatable Image Files, Image Files
10506: @section Data-Relocatable Image Files
10507: @cindex data-relocatable image files
10508: @cindex image file, data-relocatable
10509:
10510: These files contain relocatable data addresses, but fixed code addresses
10511: (instead of tokens). They are specific to the executable (i.e.,
10512: @file{gforth} file) they were created with. For direct threading on some
10513: architectures (e.g., the i386), data-relocatable images do not work. You
10514: get a data-relocatable image, if you use @file{gforthmi} with a
10515: Gforth binary that is not doubly indirect threaded (@pxref{Fully
10516: Relocatable Image Files}).
10517:
10518: @node Fully Relocatable Image Files, Stack and Dictionary Sizes, Data-Relocatable Image Files, Image Files
10519: @section Fully Relocatable Image Files
10520: @cindex fully relocatable image files
10521: @cindex image file, fully relocatable
10522:
10523: @cindex @file{kern*.fi}, relocatability
10524: @cindex @file{gforth.fi}, relocatability
10525: These image files have relocatable data addresses, and tokens for code
10526: addresses. They can be used with different binaries (e.g., with and
10527: without debugging) on the same machine, and even across machines with
10528: the same data formats (byte order, cell size, floating point
10529: format). However, they are usually specific to the version of Gforth
10530: they were created with. The files @file{gforth.fi} and @file{kernl*.fi}
10531: are fully relocatable.
10532:
10533: There are two ways to create a fully relocatable image file:
10534:
10535: @menu
10536: * gforthmi:: The normal way
10537: * cross.fs:: The hard way
10538: @end menu
10539:
10540: @node gforthmi, cross.fs, Fully Relocatable Image Files, Fully Relocatable Image Files
10541: @subsection @file{gforthmi}
10542: @cindex @file{comp-i.fs}
10543: @cindex @file{gforthmi}
10544:
10545: You will usually use @file{gforthmi}. If you want to create an
10546: image @i{file} that contains everything you would load by invoking
10547: Gforth with @code{gforth @i{options}}, you simply say:
10548: @example
10549: gforthmi @i{file} @i{options}
10550: @end example
10551:
10552: E.g., if you want to create an image @file{asm.fi} that has the file
10553: @file{asm.fs} loaded in addition to the usual stuff, you could do it
10554: like this:
10555:
10556: @example
10557: gforthmi asm.fi asm.fs
10558: @end example
10559:
10560: @file{gforthmi} is implemented as a sh script and works like this: It
10561: produces two non-relocatable images for different addresses and then
10562: compares them. Its output reflects this: first you see the output (if
10563: any) of the two Gforth invocations that produce the nonrelocatable image
10564: files, then you see the output of the comparing program: It displays the
10565: offset used for data addresses and the offset used for code addresses;
10566: moreover, for each cell that cannot be represented correctly in the
10567: image files, it displays a line like the following one:
10568:
10569: @example
10570: 78DC BFFFFA50 BFFFFA40
10571: @end example
10572:
10573: This means that at offset $78dc from @code{forthstart}, one input image
10574: contains $bffffa50, and the other contains $bffffa40. Since these cells
10575: cannot be represented correctly in the output image, you should examine
10576: these places in the dictionary and verify that these cells are dead
10577: (i.e., not read before they are written).
10578:
10579: @cindex --application, @code{gforthmi} option
10580: If you insert the option @code{--application} in front of the image file
10581: name, you will get an image that uses the @code{--appl-image} option
10582: instead of the @code{--image-file} option (@pxref{Invoking
10583: Gforth}). When you execute such an image on Unix (by typing the image
10584: name as command), the Gforth engine will pass all options to the image
10585: instead of trying to interpret them as engine options.
10586:
10587: If you type @file{gforthmi} with no arguments, it prints some usage
10588: instructions.
10589:
10590: @cindex @code{savesystem} during @file{gforthmi}
10591: @cindex @code{bye} during @file{gforthmi}
10592: @cindex doubly indirect threaded code
10593: @cindex environment variable @code{GFORTHD}
10594: @cindex @code{GFORTHD} environment variable
10595: @cindex @code{gforth-ditc}
10596: There are a few wrinkles: After processing the passed @i{options}, the
10597: words @code{savesystem} and @code{bye} must be visible. A special doubly
10598: indirect threaded version of the @file{gforth} executable is used for
10599: creating the nonrelocatable images; you can pass the exact filename of
10600: this executable through the environment variable @code{GFORTHD}
10601: (default: @file{gforth-ditc}); if you pass a version that is not doubly
10602: indirect threaded, you will not get a fully relocatable image, but a
10603: data-relocatable image (because there is no code address offset). The
10604: normal @file{gforth} executable is used for creating the relocatable
10605: image; you can pass the exact filename of this executable through the
10606: environment variable @code{GFORTH}.
10607:
10608: @node cross.fs, , gforthmi, Fully Relocatable Image Files
10609: @subsection @file{cross.fs}
10610: @cindex @file{cross.fs}
10611: @cindex cross-compiler
10612: @cindex metacompiler
10613:
10614: You can also use @code{cross}, a batch compiler that accepts a Forth-like
10615: programming language. This @code{cross} language has to be documented
10616: yet.
10617:
10618: @cindex target compiler
10619: @code{cross} also allows you to create image files for machines with
10620: different data sizes and data formats than the one used for generating
10621: the image file. You can also use it to create an application image that
10622: does not contain a Forth compiler. These features are bought with
10623: restrictions and inconveniences in programming. E.g., addresses have to
10624: be stored in memory with special words (@code{A!}, @code{A,}, etc.) in
10625: order to make the code relocatable.
10626:
10627:
10628: @node Stack and Dictionary Sizes, Running Image Files, Fully Relocatable Image Files, Image Files
10629: @section Stack and Dictionary Sizes
10630: @cindex image file, stack and dictionary sizes
10631: @cindex dictionary size default
10632: @cindex stack size default
10633:
10634: If you invoke Gforth with a command line flag for the size
10635: (@pxref{Invoking Gforth}), the size you specify is stored in the
10636: dictionary. If you save the dictionary with @code{savesystem} or create
10637: an image with @file{gforthmi}, this size will become the default
10638: for the resulting image file. E.g., the following will create a
10639: fully relocatable version of @file{gforth.fi} with a 1MB dictionary:
10640:
10641: @example
10642: gforthmi gforth.fi -m 1M
10643: @end example
10644:
10645: In other words, if you want to set the default size for the dictionary
10646: and the stacks of an image, just invoke @file{gforthmi} with the
10647: appropriate options when creating the image.
10648:
10649: @cindex stack size, cache-friendly
10650: Note: For cache-friendly behaviour (i.e., good performance), you should
10651: make the sizes of the stacks modulo, say, 2K, somewhat different. E.g.,
10652: the default stack sizes are: data: 16k (mod 2k=0); fp: 15.5k (mod
10653: 2k=1.5k); return: 15k(mod 2k=1k); locals: 14.5k (mod 2k=0.5k).
10654:
10655: @node Running Image Files, Modifying the Startup Sequence, Stack and Dictionary Sizes, Image Files
10656: @section Running Image Files
10657: @cindex running image files
10658: @cindex invoking image files
10659: @cindex image file invocation
10660:
10661: @cindex -i, invoke image file
10662: @cindex --image file, invoke image file
10663: You can invoke Gforth with an image file @i{image} instead of the
10664: default @file{gforth.fi} with the @code{-i} flag (@pxref{Invoking Gforth}):
10665: @example
10666: gforth -i @i{image}
10667: @end example
10668:
10669: @cindex executable image file
10670: @cindex image file, executable
10671: If your operating system supports starting scripts with a line of the
10672: form @code{#! ...}, you just have to type the image file name to start
10673: Gforth with this image file (note that the file extension @code{.fi} is
10674: just a convention). I.e., to run Gforth with the image file @i{image},
10675: you can just type @i{image} instead of @code{gforth -i @i{image}}.
10676: This works because every @code{.fi} file starts with a line of this
10677: format:
10678:
10679: @example
10680: #! /usr/local/bin/gforth-0.4.0 -i
10681: @end example
10682:
10683: The file and pathname for the Gforth engine specified on this line is
10684: the specific Gforth executable that it was built against; i.e. the value
10685: of the environment variable @code{GFORTH} at the time that
10686: @file{gforthmi} was executed.
10687:
10688: You can make use of the same shell capability to make a Forth source
10689: file into an executable. For example, if you place this text in a file:
10690:
10691: @example
10692: #! /usr/local/bin/gforth
10693:
10694: ." Hello, world" CR
10695: bye
10696: @end example
10697:
10698: @noindent
10699: and then make the file executable (chmod +x in Unix), you can run it
10700: directly from the command line. The sequence @code{#!} is used in two
10701: ways; firstly, it is recognised as a ``magic sequence'' by the operating
10702: system@footnote{The Unix kernel actually recognises two types of files:
10703: executable files and files of data, where the data is processed by an
10704: interpreter that is specified on the ``interpreter line'' -- the first
10705: line of the file, starting with the sequence #!. There may be a small
10706: limit (e.g., 32) on the number of characters that may be specified on
10707: the interpreter line.} secondly it is treated as a comment character by
10708: Gforth. Because of the second usage, a space is required between
10709: @code{#!} and the path to the executable.
10710:
10711: The disadvantage of this latter technique, compared with using
10712: @file{gforthmi}, is that it is slower; the Forth source code is compiled
10713: on-the-fly, each time the program is invoked.
10714:
10715: @comment TODO describe the #! magic with reference to the Power Tools book.
10716:
10717: doc-#!
10718:
10719: @node Modifying the Startup Sequence, , Running Image Files, Image Files
10720: @section Modifying the Startup Sequence
10721: @cindex startup sequence for image file
10722: @cindex image file initialization sequence
10723: @cindex initialization sequence of image file
10724:
10725: You can add your own initialization to the startup sequence through the
10726: deferred word @code{'cold}. @code{'cold} is invoked just before the
10727: image-specific command line processing (by default, loading files and
10728: evaluating (@code{-e}) strings) starts.
10729:
10730: A sequence for adding your initialization usually looks like this:
10731:
10732: @example
10733: :noname
10734: Defers 'cold \ do other initialization stuff (e.g., rehashing wordlists)
10735: ... \ your stuff
10736: ; IS 'cold
10737: @end example
10738:
10739: @cindex turnkey image files
10740: @cindex image file, turnkey applications
10741: You can make a turnkey image by letting @code{'cold} execute a word
10742: (your turnkey application) that never returns; instead, it exits Gforth
10743: via @code{bye} or @code{throw}.
10744:
10745: @cindex command-line arguments, access
10746: @cindex arguments on the command line, access
10747: You can access the (image-specific) command-line arguments through the
10748: variables @code{argc} and @code{argv}. @code{arg} provides convenient
10749: access to @code{argv}.
10750:
10751: If @code{'cold} exits normally, Gforth processes the command-line
10752: arguments as files to be loaded and strings to be evaluated. Therefore,
10753: @code{'cold} should remove the arguments it has used in this case.
10754:
10755: doc-'cold
10756: doc-argc
10757: doc-argv
10758: doc-arg
10759:
10760:
10761: @c ******************************************************************
10762: @node Engine, Binding to System Library, Image Files, Top
10763: @chapter Engine
10764: @cindex engine
10765: @cindex virtual machine
10766:
10767: Reading this chapter is not necessary for programming with Gforth. It
10768: may be helpful for finding your way in the Gforth sources.
10769:
10770: The ideas in this section have also been published in the papers
10771: @cite{ANS fig/GNU/??? Forth} (in German) by Bernd Paysan, presented at
10772: the Forth-Tagung '93 and @cite{A Portable Forth Engine} by M. Anton
10773: Ertl, presented at EuroForth '93; the latter is available at
10774: @*@url{http://www.complang.tuwien.ac.at/papers/ertl93.ps.Z}.
10775:
10776: @menu
10777: * Portability::
10778: * Threading::
10779: * Primitives::
10780: * Performance::
10781: @end menu
10782:
10783: @node Portability, Threading, Engine, Engine
10784: @section Portability
10785: @cindex engine portability
10786:
10787: An important goal of the Gforth Project is availability across a wide
10788: range of personal machines. fig-Forth, and, to a lesser extent, F83,
10789: achieved this goal by manually coding the engine in assembly language
10790: for several then-popular processors. This approach is very
10791: labor-intensive and the results are short-lived due to progress in
10792: computer architecture.
10793:
10794: @cindex C, using C for the engine
10795: Others have avoided this problem by coding in C, e.g., Mitch Bradley
10796: (cforth), Mikael Patel (TILE) and Dirk Zoller (pfe). This approach is
10797: particularly popular for UNIX-based Forths due to the large variety of
10798: architectures of UNIX machines. Unfortunately an implementation in C
10799: does not mix well with the goals of efficiency and with using
10800: traditional techniques: Indirect or direct threading cannot be expressed
10801: in C, and switch threading, the fastest technique available in C, is
10802: significantly slower. Another problem with C is that it is very
10803: cumbersome to express double integer arithmetic.
10804:
10805: @cindex GNU C for the engine
10806: @cindex long long
10807: Fortunately, there is a portable language that does not have these
10808: limitations: GNU C, the version of C processed by the GNU C compiler
10809: (@pxref{C Extensions, , Extensions to the C Language Family, gcc.info,
10810: GNU C Manual}). Its labels as values feature (@pxref{Labels as Values, ,
10811: Labels as Values, gcc.info, GNU C Manual}) makes direct and indirect
10812: threading possible, its @code{long long} type (@pxref{Long Long, ,
10813: Double-Word Integers, gcc.info, GNU C Manual}) corresponds to Forth's
10814: double numbers@footnote{Unfortunately, long longs are not implemented
10815: properly on all machines (e.g., on alpha-osf1, long longs are only 64
10816: bits, the same size as longs (and pointers), but they should be twice as
10817: long according to @pxref{Long Long, , Double-Word Integers, gcc.info, GNU
10818: C Manual}). So, we had to implement doubles in C after all. Still, on
10819: most machines we can use long longs and achieve better performance than
10820: with the emulation package.}. GNU C is available for free on all
10821: important (and many unimportant) UNIX machines, VMS, 80386s running
10822: MS-DOS, the Amiga, and the Atari ST, so a Forth written in GNU C can run
10823: on all these machines.
10824:
10825: Writing in a portable language has the reputation of producing code that
10826: is slower than assembly. For our Forth engine we repeatedly looked at
10827: the code produced by the compiler and eliminated most compiler-induced
10828: inefficiencies by appropriate changes in the source code.
10829:
10830: @cindex explicit register declarations
10831: @cindex --enable-force-reg, configuration flag
10832: @cindex -DFORCE_REG
10833: However, register allocation cannot be portably influenced by the
10834: programmer, leading to some inefficiencies on register-starved
10835: machines. We use explicit register declarations (@pxref{Explicit Reg
10836: Vars, , Variables in Specified Registers, gcc.info, GNU C Manual}) to
10837: improve the speed on some machines. They are turned on by using the
10838: configuration flag @code{--enable-force-reg} (@code{gcc} switch
10839: @code{-DFORCE_REG}). Unfortunately, this feature not only depends on the
10840: machine, but also on the compiler version: On some machines some
10841: compiler versions produce incorrect code when certain explicit register
10842: declarations are used. So by default @code{-DFORCE_REG} is not used.
10843:
10844: @node Threading, Primitives, Portability, Engine
10845: @section Threading
10846: @cindex inner interpreter implementation
10847: @cindex threaded code implementation
10848:
10849: @cindex labels as values
10850: GNU C's labels as values extension (available since @code{gcc-2.0},
10851: @pxref{Labels as Values, , Labels as Values, gcc.info, GNU C Manual})
10852: makes it possible to take the address of @i{label} by writing
10853: @code{&&@i{label}}. This address can then be used in a statement like
10854: @code{goto *@i{address}}. I.e., @code{goto *&&x} is the same as
10855: @code{goto x}.
10856:
10857: @cindex @code{NEXT}, indirect threaded
10858: @cindex indirect threaded inner interpreter
10859: @cindex inner interpreter, indirect threaded
10860: With this feature an indirect threaded @code{NEXT} looks like:
10861: @example
10862: cfa = *ip++;
10863: ca = *cfa;
10864: goto *ca;
10865: @end example
10866: @cindex instruction pointer
10867: For those unfamiliar with the names: @code{ip} is the Forth instruction
10868: pointer; the @code{cfa} (code-field address) corresponds to ANS Forths
10869: execution token and points to the code field of the next word to be
10870: executed; The @code{ca} (code address) fetched from there points to some
10871: executable code, e.g., a primitive or the colon definition handler
10872: @code{docol}.
10873:
10874: @cindex @code{NEXT}, direct threaded
10875: @cindex direct threaded inner interpreter
10876: @cindex inner interpreter, direct threaded
10877: Direct threading is even simpler:
10878: @example
10879: ca = *ip++;
10880: goto *ca;
10881: @end example
10882:
10883: Of course we have packaged the whole thing neatly in macros called
10884: @code{NEXT} and @code{NEXT1} (the part of @code{NEXT} after fetching the cfa).
10885:
10886: @menu
10887: * Scheduling::
10888: * Direct or Indirect Threaded?::
10889: * DOES>::
10890: @end menu
10891:
10892: @node Scheduling, Direct or Indirect Threaded?, Threading, Threading
10893: @subsection Scheduling
10894: @cindex inner interpreter optimization
10895:
10896: There is a little complication: Pipelined and superscalar processors,
10897: i.e., RISC and some modern CISC machines can process independent
10898: instructions while waiting for the results of an instruction. The
10899: compiler usually reorders (schedules) the instructions in a way that
10900: achieves good usage of these delay slots. However, on our first tries
10901: the compiler did not do well on scheduling primitives. E.g., for
10902: @code{+} implemented as
10903: @example
10904: n=sp[0]+sp[1];
10905: sp++;
10906: sp[0]=n;
10907: NEXT;
10908: @end example
10909: the @code{NEXT} comes strictly after the other code, i.e., there is nearly no
10910: scheduling. After a little thought the problem becomes clear: The
10911: compiler cannot know that @code{sp} and @code{ip} point to different
10912: addresses (and the version of @code{gcc} we used would not know it even
10913: if it was possible), so it could not move the load of the cfa above the
10914: store to the TOS. Indeed the pointers could be the same, if code on or
10915: very near the top of stack were executed. In the interest of speed we
10916: chose to forbid this probably unused ``feature'' and helped the compiler
10917: in scheduling: @code{NEXT} is divided into the loading part (@code{NEXT_P1})
10918: and the goto part (@code{NEXT_P2}). @code{+} now looks like:
10919: @example
10920: n=sp[0]+sp[1];
10921: sp++;
10922: NEXT_P1;
10923: sp[0]=n;
10924: NEXT_P2;
10925: @end example
10926: This can be scheduled optimally by the compiler.
10927:
10928: This division can be turned off with the switch @code{-DCISC_NEXT}. This
10929: switch is on by default on machines that do not profit from scheduling
10930: (e.g., the 80386), in order to preserve registers.
10931:
10932: @node Direct or Indirect Threaded?, DOES>, Scheduling, Threading
10933: @subsection Direct or Indirect Threaded?
10934: @cindex threading, direct or indirect?
10935:
10936: @cindex -DDIRECT_THREADED
10937: Both! After packaging the nasty details in macro definitions we
10938: realized that we could switch between direct and indirect threading by
10939: simply setting a compilation flag (@code{-DDIRECT_THREADED}) and
10940: defining a few machine-specific macros for the direct-threading case.
10941: On the Forth level we also offer access words that hide the
10942: differences between the threading methods (@pxref{Threading Words}).
10943:
10944: Indirect threading is implemented completely machine-independently.
10945: Direct threading needs routines for creating jumps to the executable
10946: code (e.g. to @code{docol} or @code{dodoes}). These routines are inherently
10947: machine-dependent, but they do not amount to many source lines. Therefore,
10948: even porting direct threading to a new machine requires little effort.
10949:
10950: @cindex --enable-indirect-threaded, configuration flag
10951: @cindex --enable-direct-threaded, configuration flag
10952: The default threading method is machine-dependent. You can enforce a
10953: specific threading method when building Gforth with the configuration
10954: flag @code{--enable-direct-threaded} or
10955: @code{--enable-indirect-threaded}. Note that direct threading is not
10956: supported on all machines.
10957:
10958: @node DOES>, , Direct or Indirect Threaded?, Threading
10959: @subsection DOES>
10960: @cindex @code{DOES>} implementation
10961:
10962: @cindex @code{dodoes} routine
10963: @cindex @code{DOES>}-code
10964: One of the most complex parts of a Forth engine is @code{dodoes}, i.e.,
10965: the chunk of code executed by every word defined by a
10966: @code{CREATE}...@code{DOES>} pair. The main problem here is: How to find
10967: the Forth code to be executed, i.e. the code after the
10968: @code{DOES>} (the @code{DOES>}-code)? There are two solutions:
10969:
10970: In fig-Forth the code field points directly to the @code{dodoes} and the
10971: @code{DOES>}code address is stored in the cell after the code address (i.e. at
10972: @code{@i{CFA} cell+}). It may seem that this solution is illegal in
10973: the Forth-79 and all later standards, because in fig-Forth this address
10974: lies in the body (which is illegal in these standards). However, by
10975: making the code field larger for all words this solution becomes legal
10976: again. We use this approach for the indirect threaded version and for
10977: direct threading on some machines. Leaving a cell unused in most words
10978: is a bit wasteful, but on the machines we are targeting this is hardly a
10979: problem. The other reason for having a code field size of two cells is
10980: to avoid having different image files for direct and indirect threaded
10981: systems (direct threaded systems require two-cell code fields on many
10982: machines).
10983:
10984: @cindex @code{DOES>}-handler
10985: The other approach is that the code field points or jumps to the cell
10986: after @code{DOES>}. In this variant there is a jump to @code{dodoes} at
10987: this address (the @code{DOES>}-handler). @code{dodoes} can then get the
10988: @code{DOES>}-code address by computing the code address, i.e., the address of
10989: the jump to dodoes, and add the length of that jump field. A variant of
10990: this is to have a call to @code{dodoes} after the @code{DOES>}; then the
10991: return address (which can be found in the return register on RISCs) is
10992: the @code{DOES>}-code address. Since the two cells available in the code field
10993: are used up by the jump to the code address in direct threading on many
10994: architectures, we use this approach for direct threading on these
10995: architectures. We did not want to add another cell to the code field.
10996:
10997: @node Primitives, Performance, Threading, Engine
10998: @section Primitives
10999: @cindex primitives, implementation
11000: @cindex virtual machine instructions, implementation
11001:
11002: @menu
11003: * Automatic Generation::
11004: * TOS Optimization::
11005: * Produced code::
11006: @end menu
11007:
11008: @node Automatic Generation, TOS Optimization, Primitives, Primitives
11009: @subsection Automatic Generation
11010: @cindex primitives, automatic generation
11011:
11012: @cindex @file{prims2x.fs}
11013: Since the primitives are implemented in a portable language, there is no
11014: longer any need to minimize the number of primitives. On the contrary,
11015: having many primitives has an advantage: speed. In order to reduce the
11016: number of errors in primitives and to make programming them easier, we
11017: provide a tool, the primitive generator (@file{prims2x.fs}), that
11018: automatically generates most (and sometimes all) of the C code for a
11019: primitive from the stack effect notation. The source for a primitive
11020: has the following form:
11021:
11022: @cindex primitive source format
11023: @format
11024: @i{Forth-name} @i{stack-effect} @i{category} [@i{pronounc.}]
11025: [@code{""}@i{glossary entry}@code{""}]
11026: @i{C code}
11027: [@code{:}
11028: @i{Forth code}]
11029: @end format
11030:
11031: The items in brackets are optional. The category and glossary fields
11032: are there for generating the documentation, the Forth code is there
11033: for manual implementations on machines without GNU C. E.g., the source
11034: for the primitive @code{+} is:
11035: @example
11036: + n1 n2 -- n core plus
11037: n = n1+n2;
11038: @end example
11039:
11040: This looks like a specification, but in fact @code{n = n1+n2} is C
11041: code. Our primitive generation tool extracts a lot of information from
11042: the stack effect notations@footnote{We use a one-stack notation, even
11043: though we have separate data and floating-point stacks; The separate
11044: notation can be generated easily from the unified notation.}: The number
11045: of items popped from and pushed on the stack, their type, and by what
11046: name they are referred to in the C code. It then generates a C code
11047: prelude and postlude for each primitive. The final C code for @code{+}
11048: looks like this:
11049:
11050: @example
11051: I_plus: /* + ( n1 n2 -- n ) */ /* label, stack effect */
11052: /* */ /* documentation */
11053: @{
11054: DEF_CA /* definition of variable ca (indirect threading) */
11055: Cell n1; /* definitions of variables */
11056: Cell n2;
11057: Cell n;
11058: n1 = (Cell) sp[1]; /* input */
11059: n2 = (Cell) TOS;
11060: sp += 1; /* stack adjustment */
11061: NAME("+") /* debugging output (with -DDEBUG) */
11062: @{
11063: n = n1+n2; /* C code taken from the source */
11064: @}
11065: NEXT_P1; /* NEXT part 1 */
11066: TOS = (Cell)n; /* output */
11067: NEXT_P2; /* NEXT part 2 */
11068: @}
11069: @end example
11070:
11071: This looks long and inefficient, but the GNU C compiler optimizes quite
11072: well and produces optimal code for @code{+} on, e.g., the R3000 and the
11073: HP RISC machines: Defining the @code{n}s does not produce any code, and
11074: using them as intermediate storage also adds no cost.
11075:
11076: There are also other optimizations that are not illustrated by this
11077: example: assignments between simple variables are usually for free (copy
11078: propagation). If one of the stack items is not used by the primitive
11079: (e.g. in @code{drop}), the compiler eliminates the load from the stack
11080: (dead code elimination). On the other hand, there are some things that
11081: the compiler does not do, therefore they are performed by
11082: @file{prims2x.fs}: The compiler does not optimize code away that stores
11083: a stack item to the place where it just came from (e.g., @code{over}).
11084:
11085: While programming a primitive is usually easy, there are a few cases
11086: where the programmer has to take the actions of the generator into
11087: account, most notably @code{?dup}, but also words that do not (always)
11088: fall through to @code{NEXT}.
11089:
11090: @node TOS Optimization, Produced code, Automatic Generation, Primitives
11091: @subsection TOS Optimization
11092: @cindex TOS optimization for primitives
11093: @cindex primitives, keeping the TOS in a register
11094:
11095: An important optimization for stack machine emulators, e.g., Forth
11096: engines, is keeping one or more of the top stack items in
11097: registers. If a word has the stack effect @i{in1}...@i{inx} @code{--}
11098: @i{out1}...@i{outy}, keeping the top @i{n} items in registers
11099: @itemize @bullet
11100: @item
11101: is better than keeping @i{n-1} items, if @i{x>=n} and @i{y>=n},
11102: due to fewer loads from and stores to the stack.
11103: @item is slower than keeping @i{n-1} items, if @i{x<>y} and @i{x<n} and
11104: @i{y<n}, due to additional moves between registers.
11105: @end itemize
11106:
11107: @cindex -DUSE_TOS
11108: @cindex -DUSE_NO_TOS
11109: In particular, keeping one item in a register is never a disadvantage,
11110: if there are enough registers. Keeping two items in registers is a
11111: disadvantage for frequent words like @code{?branch}, constants,
11112: variables, literals and @code{i}. Therefore our generator only produces
11113: code that keeps zero or one items in registers. The generated C code
11114: covers both cases; the selection between these alternatives is made at
11115: C-compile time using the switch @code{-DUSE_TOS}. @code{TOS} in the C
11116: code for @code{+} is just a simple variable name in the one-item case,
11117: otherwise it is a macro that expands into @code{sp[0]}. Note that the
11118: GNU C compiler tries to keep simple variables like @code{TOS} in
11119: registers, and it usually succeeds, if there are enough registers.
11120:
11121: @cindex -DUSE_FTOS
11122: @cindex -DUSE_NO_FTOS
11123: The primitive generator performs the TOS optimization for the
11124: floating-point stack, too (@code{-DUSE_FTOS}). For floating-point
11125: operations the benefit of this optimization is even larger:
11126: floating-point operations take quite long on most processors, but can be
11127: performed in parallel with other operations as long as their results are
11128: not used. If the FP-TOS is kept in a register, this works. If
11129: it is kept on the stack, i.e., in memory, the store into memory has to
11130: wait for the result of the floating-point operation, lengthening the
11131: execution time of the primitive considerably.
11132:
11133: The TOS optimization makes the automatic generation of primitives a
11134: bit more complicated. Just replacing all occurrences of @code{sp[0]} by
11135: @code{TOS} is not sufficient. There are some special cases to
11136: consider:
11137: @itemize @bullet
11138: @item In the case of @code{dup ( w -- w w )} the generator must not
11139: eliminate the store to the original location of the item on the stack,
11140: if the TOS optimization is turned on.
11141: @item Primitives with stack effects of the form @code{--}
11142: @i{out1}...@i{outy} must store the TOS to the stack at the start.
11143: Likewise, primitives with the stack effect @i{in1}...@i{inx} @code{--}
11144: must load the TOS from the stack at the end. But for the null stack
11145: effect @code{--} no stores or loads should be generated.
11146: @end itemize
11147:
11148: @node Produced code, , TOS Optimization, Primitives
11149: @subsection Produced code
11150: @cindex primitives, assembly code listing
11151:
11152: @cindex @file{engine.s}
11153: To see what assembly code is produced for the primitives on your machine
11154: with your compiler and your flag settings, type @code{make engine.s} and
11155: look at the resulting file @file{engine.s}.
11156:
11157: @node Performance, , Primitives, Engine
11158: @section Performance
11159: @cindex performance of some Forth interpreters
11160: @cindex engine performance
11161: @cindex benchmarking Forth systems
11162: @cindex Gforth performance
11163:
11164: On RISCs the Gforth engine is very close to optimal; i.e., it is usually
11165: impossible to write a significantly faster engine.
11166:
11167: On register-starved machines like the 386 architecture processors
11168: improvements are possible, because @code{gcc} does not utilize the
11169: registers as well as a human, even with explicit register declarations;
11170: e.g., Bernd Beuster wrote a Forth system fragment in assembly language
11171: and hand-tuned it for the 486; this system is 1.19 times faster on the
11172: Sieve benchmark on a 486DX2/66 than Gforth compiled with
11173: @code{gcc-2.6.3} with @code{-DFORCE_REG}. The situation has improved
11174: with gcc-2.95 and gforth-0.4.9; now the most important virtual machine
11175: registers fit in real registers (and we can even afford to use the TOS
11176: optimization), resulting in a speedup of 1.14 on the sieve over the
11177: earlier results.
11178:
11179: @cindex Win32Forth performance
11180: @cindex NT Forth performance
11181: @cindex eforth performance
11182: @cindex ThisForth performance
11183: @cindex PFE performance
11184: @cindex TILE performance
11185: The potential advantage of assembly language implementations
11186: is not necessarily realized in complete Forth systems: We compared
11187: Gforth-0.4.9 (direct threaded, compiled with @code{gcc-2.95.1} and
11188: @code{-DFORCE_REG}) with Win32Forth 1.2093, LMI's NT Forth (Beta, May
11189: 1994) and Eforth (with and without peephole (aka pinhole) optimization
11190: of the threaded code); all these systems were written in assembly
11191: language. We also compared Gforth with three systems written in C:
11192: PFE-0.9.14 (compiled with @code{gcc-2.6.3} with the default
11193: configuration for Linux: @code{-O2 -fomit-frame-pointer -DUSE_REGS
11194: -DUNROLL_NEXT}), ThisForth Beta (compiled with @code{gcc-2.6.3 -O3
11195: -fomit-frame-pointer}; ThisForth employs peephole optimization of the
11196: threaded code) and TILE (compiled with @code{make opt}). We benchmarked
11197: Gforth, PFE, ThisForth and TILE on a 486DX2/66 under Linux. Kenneth
11198: O'Heskin kindly provided the results for Win32Forth and NT Forth on a
11199: 486DX2/66 with similar memory performance under Windows NT. Marcel
11200: Hendrix ported Eforth to Linux, then extended it to run the benchmarks,
11201: added the peephole optimizer, ran the benchmarks and reported the
11202: results.
11203:
11204: We used four small benchmarks: the ubiquitous Sieve; bubble-sorting and
11205: matrix multiplication come from the Stanford integer benchmarks and have
11206: been translated into Forth by Martin Fraeman; we used the versions
11207: included in the TILE Forth package, but with bigger data set sizes; and
11208: a recursive Fibonacci number computation for benchmarking calling
11209: performance. The following table shows the time taken for the benchmarks
11210: scaled by the time taken by Gforth (in other words, it shows the speedup
11211: factor that Gforth achieved over the other systems).
11212:
11213: @example
11214: relative Win32- NT eforth This-
11215: time Gforth Forth Forth eforth +opt PFE Forth TILE
11216: sieve 1.00 1.58 1.30 1.58 0.97 1.80 3.63 9.79
11217: bubble 1.00 1.55 1.67 1.75 1.04 1.78 4.59
11218: matmul 1.00 1.67 1.53 1.66 0.84 1.79 4.63
11219: fib 1.00 1.75 1.53 1.40 0.99 1.99 3.43 4.93
11220: @end example
11221:
11222: You may be quite surprised by the good performance of Gforth when
11223: compared with systems written in assembly language. One important reason
11224: for the disappointing performance of these other systems is probably
11225: that they are not written optimally for the 486 (e.g., they use the
11226: @code{lods} instruction). In addition, Win32Forth uses a comfortable,
11227: but costly method for relocating the Forth image: like @code{cforth}, it
11228: computes the actual addresses at run time, resulting in two address
11229: computations per @code{NEXT} (@pxref{Image File Background}).
11230:
11231: Only Eforth with the peephole optimizer performs comparable to
11232: Gforth. The speedups achieved with peephole optimization of threaded
11233: code are quite remarkable. Adding a peephole optimizer to Gforth should
11234: cause similar speedups.
11235:
11236: The speedup of Gforth over PFE, ThisForth and TILE can be easily
11237: explained with the self-imposed restriction of the latter systems to
11238: standard C, which makes efficient threading impossible (however, the
11239: measured implementation of PFE uses a GNU C extension: @pxref{Global Reg
11240: Vars, , Defining Global Register Variables, gcc.info, GNU C Manual}).
11241: Moreover, current C compilers have a hard time optimizing other aspects
11242: of the ThisForth and the TILE source.
11243:
11244: The performance of Gforth on 386 architecture processors varies widely
11245: with the version of @code{gcc} used. E.g., @code{gcc-2.5.8} failed to
11246: allocate any of the virtual machine registers into real machine
11247: registers by itself and would not work correctly with explicit register
11248: declarations, giving a 1.5 times slower engine (on a 486DX2/66 running
11249: the Sieve) than the one measured above.
11250:
11251: Note that there have been several releases of Win32Forth since the
11252: release presented here, so the results presented above may have little
11253: predictive value for the performance of Win32Forth today (results for
11254: the current release on an i486DX2/66 are welcome).
11255:
11256: @cindex @file{Benchres}
11257: In @cite{Translating Forth to Efficient C} by M. Anton Ertl and Martin
11258: Maierhofer (presented at EuroForth '95), an indirect threaded version of
11259: Gforth is compared with Win32Forth, NT Forth, PFE, and ThisForth; that
11260: version of Gforth is slower on a 486 than the direct threaded version
11261: used here. The paper available at
11262: @*@url{http://www.complang.tuwien.ac.at/papers/ertl&maierhofer95.ps.gz};
11263: it also contains numbers for some native code systems. You can find a
11264: newer version of these measurements at
11265: @url{http://www.complang.tuwien.ac.at/forth/performance.html}. You can
11266: find numbers for Gforth on various machines in @file{Benchres}.
11267:
11268: @c ******************************************************************
11269: @node Binding to System Library, Cross Compiler, Engine, Top
11270: @chapter Binding to System Library
11271:
11272: @node Cross Compiler, Bugs, Binding to System Library, Top
11273: @chapter Cross Compiler
11274:
11275: Cross Compiler
11276:
11277: @menu
11278: * Using the Cross Compiler::
11279: * How the Cross Compiler Works::
11280: @end menu
11281:
11282: @node Using the Cross Compiler, How the Cross Compiler Works, Cross Compiler, Cross Compiler
11283: @section Using the Cross Compiler
11284:
11285: @node How the Cross Compiler Works, , Using the Cross Compiler, Cross Compiler
11286: @section How the Cross Compiler Works
11287:
11288: @node Bugs, Origin, Cross Compiler, Top
11289: @appendix Bugs
11290: @cindex bug reporting
11291:
11292: Known bugs are described in the file @file{BUGS} in the Gforth distribution.
11293:
11294: If you find a bug, please send a bug report to
11295: @email{bug-gforth@@gnu.org}. A bug report should include this
11296: information:
11297:
11298: @itemize @bullet
11299: @item
11300: The Gforth version used (it is announced at the start of an
11301: interactive Gforth session).
11302: @item
11303: The machine and operating system (on Unix
11304: systems @code{uname -a} will report this information).
11305: @item
11306: The installation options (send the file @file{config.status}).
11307: @item
11308: A complete list of changes (if any) you (or your installer) have made to the
11309: Gforth sources.
11310: @item
11311: A program (or a sequence of keyboard commands) that reproduces the bug.
11312: @item
11313: A description of what you think constitutes the buggy behaviour.
11314: @end itemize
11315:
11316: For a thorough guide on reporting bugs read @ref{Bug Reporting, , How
11317: to Report Bugs, gcc.info, GNU C Manual}.
11318:
11319:
11320: @node Origin, Forth-related information, Bugs, Top
11321: @appendix Authors and Ancestors of Gforth
11322:
11323: @section Authors and Contributors
11324: @cindex authors of Gforth
11325: @cindex contributors to Gforth
11326:
11327: The Gforth project was started in mid-1992 by Bernd Paysan and Anton
11328: Ertl. The third major author was Jens Wilke. Lennart Benschop (who was
11329: one of Gforth's first users, in mid-1993) and Stuart Ramsden inspired us
11330: with their continuous feedback. Lennart Benshop contributed
11331: @file{glosgen.fs}, while Stuart Ramsden has been working on automatic
11332: support for calling C libraries. Helpful comments also came from Paul
11333: Kleinrubatscher, Christian Pirker, Dirk Zoller, Marcel Hendrix, John
11334: Wavrik, Barrie Stott, Marc de Groot, and Jorge Acerada. Since the
11335: release of Gforth-0.2.1 there were also helpful comments from many
11336: others; thank you all, sorry for not listing you here (but digging
11337: through my mailbox to extract your names is on my to-do list). Since the
11338: release of Gforth-0.4.0 Neal Crook worked on the manual.
11339:
11340: Gforth also owes a lot to the authors of the tools we used (GCC, CVS,
11341: and autoconf, among others), and to the creators of the Internet: Gforth
11342: was developed across the Internet, and its authors did not meet
11343: physically for the first 4 years of development.
11344:
11345: @section Pedigree
11346: @cindex pedigree of Gforth
11347:
11348: Gforth descends from bigFORTH (1993) and fig-Forth. Gforth and PFE (by
11349: Dirk Zoller) will cross-fertilize each other. Of course, a significant
11350: part of the design of Gforth was prescribed by ANS Forth.
11351:
11352: Bernd Paysan wrote bigFORTH, a descendent from TurboForth, an unreleased
11353: 32 bit native code version of VolksForth for the Atari ST, written
11354: mostly by Dietrich Weineck.
11355:
11356: VolksForth descends from F83. It was written by Klaus Schleisiek, Bernd
11357: Pennemann, Georg Rehfeld and Dietrich Weineck for the C64 (called
11358: UltraForth there) in the mid-80s and ported to the Atari ST in 1986.
11359:
11360: Henry Laxen and Mike Perry wrote F83 as a model implementation of the
11361: Forth-83 standard. !! Pedigree? When?
11362:
11363: A team led by Bill Ragsdale implemented fig-Forth on many processors in
11364: 1979. Robert Selzer and Bill Ragsdale developed the original
11365: implementation of fig-Forth for the 6502 based on microForth.
11366:
11367: The principal architect of microForth was Dean Sanderson. microForth was
11368: FORTH, Inc.'s first off-the-shelf product. It was developed in 1976 for
11369: the 1802, and subsequently implemented on the 8080, the 6800 and the
11370: Z80.
11371:
11372: All earlier Forth systems were custom-made, usually by Charles Moore,
11373: who discovered (as he puts it) Forth during the late 60s. The first full
11374: Forth existed in 1971.
11375:
11376: A part of the information in this section comes from @cite{The Evolution
11377: of Forth} by Elizabeth D. Rather, Donald R. Colburn and Charles
11378: H. Moore, presented at the HOPL-II conference and preprinted in SIGPLAN
11379: Notices 28(3), 1993. You can find more historical and genealogical
11380: information about Forth there.
11381:
11382: @node Forth-related information, Word Index, Origin, Top
11383: @appendix Other Forth-related information
11384: @cindex Forth-related information
11385:
11386: @menu
11387: * Internet resources::
11388: * Books::
11389: * The Forth Interest Group::
11390: * Conferences::
11391: @end menu
11392:
11393:
11394: @node Internet resources, Books, Forth-related information, Forth-related information
11395: @section Internet resources
11396: @cindex internet resources
11397:
11398: @cindex comp.lang.forth
11399: @cindex frequently asked questions
11400: There is an active newsgroup (comp.lang.forth) discussing Forth and
11401: Forth-related issues. A frequently-asked-questions (FAQ) list
11402: is posted to the newsgroup regulary, and archived at these sites:
11403:
11404: @itemize @bullet
11405: @item
11406: @url{ftp://rtfm.mit.edu/pub/usenet-by-group/comp.lang.forth/}
11407: @item
11408: @url{ftp://ftp.forth.org/pub/Forth/FAQ/}
11409: @end itemize
11410:
11411: The FAQ list should be considered mandatory reading before posting to
11412: the newsgroup.
11413:
11414: Here are some other web sites holding Forth-related material:
11415:
11416: @itemize @bullet
11417: @item
11418: @url{http://www.taygeta.com/forth.html} -- Skip Carter's Forth pages.
11419: @item
11420: @url{http://www.jwdt.com/~paysan/gforth.html} -- the Gforth home page.
11421: @item
11422: @url{http://www.minerva.com/uathena.htm} -- home of ANS Forth Standard.
11423: @item
11424: @url{http://dec.bournemouth.ac.uk/forth/index.html} -- the Forth
11425: Research page, including links to the Journal of Forth Application and
11426: Research (JFAR) and a searchable Forth bibliography.
11427: @end itemize
11428:
11429:
11430: @node Books, The Forth Interest Group, Internet resources, Forth-related information
11431: @section Books
11432: @cindex books on Forth
11433:
11434: As the Standard is relatively new, there are not many books out yet. It
11435: is not recommended to learn Forth by using Gforth and a book that is not
11436: written for ANS Forth, as you will not know your mistakes from the
11437: deviations of the book. However, books based on the Forth-83 standard
11438: should be ok, because ANS Forth is primarily an extension of Forth-83.
11439:
11440: @cindex standard document for ANS Forth
11441: @cindex ANS Forth document
11442: The definite reference if you want to write ANS Forth programs is, of
11443: course, the ANS Forth document. It is available in printed form from the
11444: National Standards Institute Sales Department (Tel.: USA (212) 642-4900;
11445: Fax.: USA (212) 302-1286) as document @cite{X3.215-1994} for about
11446: $200. You can also get it from Global Engineering Documents (Tel.: USA
11447: (800) 854-7179; Fax.: (303) 843-9880) for about $300.
11448:
11449: @cite{dpANS6}, the last draft of the standard, which was then submitted
11450: to ANSI for publication is available electronically and for free in some
11451: MS Word format, and it has been converted to HTML
11452: (@url{http://www.taygeta.com/forth/dpans.html}; this is my favourite
11453: format); this HTML version also includes the answers to Requests for
11454: Interpretation (RFIs). Some pointers to these versions can be found
11455: through @*@url{http://www.complang.tuwien.ac.at/projects/forth.html}.
11456:
11457: @cindex introductory book on Forth
11458: @cindex book on Forth, introductory
11459: @cindex Woehr, Jack: @cite{Forth: The New Model}
11460: @cindex @cite{Forth: The new model} (book)
11461: @cite{Forth: The New Model} by Jack Woehr (Prentice-Hall, 1993) is an
11462: introductory book based on a draft version of the standard. It does not
11463: cover the whole standard. It also contains interesting background
11464: information (Jack Woehr was in the ANS Forth Technical Committee). It is
11465: not appropriate for complete newbies, but programmers experienced in
11466: other languages should find it ok.
11467:
11468: @cindex Conklin, Edward K., and Elizabeth Rather: @cite{Forth Programmer's Handbook}
11469: @cindex Rather, Elizabeth and Edward K. Conklin: @cite{Forth Programmer's Handbook}
11470: @cindex @cite{Forth Programmer's Handbook} (book)
11471: @cite{Forth Programmer's Handbook} by Edward K. Conklin, Elizabeth
11472: D. Rather and the technical staff of Forth, Inc. (Forth, Inc., 1997;
11473: ISBN 0-9662156-0-5) contains little introductory material. The majority
11474: of the book is similar to @ref{Words}, but the book covers most of the
11475: standard words and some non-standard words (whereas this manual is
11476: quite incomplete). In addition, the book contains a chapter on
11477: programming style. The major drawback of this book is that it usually
11478: does not identify what is standard and what is specific to the Forth
11479: system described in the book (probably one of Forth, Inc.'s systems).
11480: Fortunately, many of the non-standard programming practices described in
11481: the book work in Gforth, too. Still, this drawback makes the book
11482: hardly more useful than a pre-ANS book.
11483:
11484: @node The Forth Interest Group, Conferences, Books, Forth-related information
11485: @section The Forth Interest Group
11486: @cindex Forth interest group (FIG)
11487:
11488: The Forth Interest Group (FIG) is a world-wide, non-profit,
11489: member-supported organisation. It publishes a regular magazine,
11490: @var{FORTH Dimensions}, and offers other benefits of membership. You can
11491: contact the FIG through their office email address:
11492: @email{office@@forth.org} or by visiting their web site at
11493: @url{http://www.forth.org/}. This web site also includes links to FIG
11494: chapters in other countries and American cities
11495: (@url{http://www.forth.org/chapters.html}).
11496:
11497: @node Conferences, , The Forth Interest Group, Forth-related information
11498: @section Conferences
11499: @cindex Conferences
11500:
11501: There are several regular conferences related to Forth. They are all
11502: well-publicised in @var{FORTH Dimensions} and on the comp.lang.forth
11503: news group:
11504:
11505: @itemize @bullet
11506: @item
11507: FORML -- the Forth modification laboratory convenes every year near
11508: Monterey, California.
11509: @item
11510: The Rochester Forth Conference -- an annual conference traditionally
11511: held in Rochester, New York.
11512: @item
11513: EuroForth -- this European conference takes place annually.
11514: @end itemize
11515:
11516:
11517: @node Word Index, Concept Index, Forth-related information, Top
11518: @unnumbered Word Index
11519:
11520: This index is a list of Forth words that have ``glossary'' entries
11521: within this manual. Each word is listed with its stack effect and
11522: wordset.
11523:
11524: @printindex fn
11525:
11526: @node Concept Index, , Word Index, Top
11527: @unnumbered Concept and Word Index
11528:
11529: Not all entries listed in this index are present verbatim in the
11530: text. This index also duplicates, in abbreviated form, all of the words
11531: listed in the Word Index (only the names are listed for the words here).
11532:
11533: @printindex cp
11534:
11535: @contents
11536: @bye
11537:
FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>