Annotation of gforth/doc/gforth.ds, revision 1.38
1.1 anton 1: \input texinfo @c -*-texinfo-*-
2: @comment The source is gforth.ds, from which gforth.texi is generated
1.28 crook 3:
1.21 crook 4: @comment TODO: nac29jan99 - a list of things to add in the next edit:
1.28 crook 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.
1.36 anton 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
1.29 crook 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:
1.28 crook 17:
1.1 anton 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
1.29 crook 26: @comment TODO this gets left in by HTML converter
1.12 anton 27: @macro progstyle {}
28: Programming style note:
1.3 anton 29: @end macro
1.1 anton 30: @comment %**end of header (This is for running Texinfo on a region.)
31:
1.29 crook 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:
1.10 anton 54: @include version.texi
55:
1.1 anton 56: @ifinfo
1.11 anton 57: This file documents Gforth @value{VERSION}
1.1 anton 58:
1.26 crook 59: Copyright @copyright{} 1995-1999 Free Software Foundation, Inc.
1.1 anton 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
1.11 anton 91: @center for version @value{VERSION}
1.1 anton 92: @sp 2
1.34 anton 93: @center Neal Crook
1.1 anton 94: @center Anton Ertl
1.6 pazsan 95: @center Bernd Paysan
1.5 anton 96: @center Jens Wilke
1.1 anton 97: @sp 3
1.29 crook 98: @center This manual is permanently under construction and was last updated on 04-May-1999
1.1 anton 99:
100: @comment The following two commands start the copyright page.
101: @page
102: @vskip 0pt plus 1filll
1.29 crook 103: Copyright @copyright{} 1995--1999 Free Software Foundation, Inc.
1.1 anton 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
1.11 anton 129: personal machines. This manual corresponds to version @value{VERSION}.
1.1 anton 130: @end ifinfo
131:
132: @menu
1.21 crook 133: * License:: The GPL
1.26 crook 134: * Goals:: About the Gforth Project
1.29 crook 135: * Gforth Environment:: Starting (and exiting) Gforth
1.21 crook 136: * Introduction:: An introduction to ANS Forth
1.1 anton 137: * Words:: Forth words available in Gforth
1.24 anton 138: * Error messages:: How to interpret them
1.1 anton 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
1.24 anton 146: * Binding to System Library::
1.13 pazsan 147: * Cross Compiler:: The Cross Compiler
1.1 anton 148: * Bugs:: How to report them
149: * Origin:: Authors and ancestors of Gforth
1.21 crook 150: * Forth-related information:: Books and places to look on the WWW
1.1 anton 151: * Word Index:: An item for each Forth word
152: * Concept Index:: A menu covering many topics
1.12 anton 153:
1.24 anton 154: @detailmenu --- The Detailed Node Listing ---
1.12 anton 155:
1.26 crook 156: Goals of Gforth
157:
158: * Gforth Extensions Sinful?::
159:
1.29 crook 160: Gforth Environment
161:
1.32 anton 162: * Invoking Gforth:: Getting in
163: * Leaving Gforth:: Getting out
164: * Command-line editing::
1.29 crook 165: * Upper and lower case::
1.32 anton 166: * Environment variables:: ..that affect how Gforth starts up
167: * Gforth Files:: What gets installed and where
1.29 crook 168:
1.24 anton 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::
1.29 crook 177: * Where to go next::
1.24 anton 178: * Exercises::
179:
1.12 anton 180: Forth Words
181:
182: * Notation::
1.21 crook 183: * Comments::
184: * Boolean Flags::
1.12 anton 185: * Arithmetic::
186: * Stack Manipulation::
187: * Memory::
188: * Control Structures::
189: * Defining Words::
1.21 crook 190: * The Text Interpreter::
1.12 anton 191: * Tokens for Words::
1.21 crook 192: * Word Lists::
193: * Environmental Queries::
1.12 anton 194: * Files::
195: * Blocks::
196: * Other I/O::
197: * Programming Tools::
198: * Assembler and Code Words::
199: * Threading Words::
1.26 crook 200: * Locals::
201: * Structures::
202: * Object-oriented Forth::
1.21 crook 203: * Passing Commands to the OS::
204: * Miscellaneous Words::
1.12 anton 205:
206: Arithmetic
207:
208: * Single precision::
209: * Bitwise operations::
1.21 crook 210: * Double precision:: Double-cell integer arithmetic
211: * Numeric comparison::
1.32 anton 212: * Mixed precision:: Operations with single and double-cell integers
1.12 anton 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:
1.32 anton 225: * Memory model::
226: * Dictionary allocation::
227: * Heap Allocation::
228: * Memory Access::
229: * Address arithmetic::
230: * Memory Blocks::
1.12 anton 231:
232: Control Structures
233:
1.32 anton 234: * Selection:: IF.. ELSE.. ENDIF
235: * Simple Loops:: BEGIN..
236: * Counted Loops:: DO
237: * Arbitrary control structures::
238: * Calls and returns::
1.12 anton 239: * Exception Handling::
240:
241: Defining Words
242:
1.32 anton 243: * Simple Defining Words:: Variables, values and constants
244: * Colon Definitions::
245: * User-defined Defining Words::
246: * Supplying names::
247: * Interpretation and Compilation Semantics::
1.12 anton 248:
1.21 crook 249: The Text Interpreter
250:
1.29 crook 251: * Input Sources::
1.21 crook 252: * Number Conversion::
253: * Interpret/Compile states::
254: * Literals::
255: * Interpreter Directives::
256:
1.26 crook 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:
1.32 anton 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
1.26 crook 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:
1.12 anton 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:
1.24 anton 306: * Why object-oriented programming?::
307: * Object-Oriented Terminology::
308: * Objects::
309: * OOF::
310: * Mini-OOF::
1.23 crook 311: * Comparison with other object models::
1.12 anton 312:
1.24 anton 313: The @file{objects.fs} model
1.12 anton 314:
315: * Properties of the Objects model::
316: * Basic Objects Usage::
1.23 crook 317: * The Objects base class::
1.12 anton 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:
1.24 anton 327: The @file{oof.fs} model
1.12 anton 328:
329: * Properties of the OOF model::
330: * Basic OOF Usage::
1.23 crook 331: * The OOF base class::
1.12 anton 332: * Class Declaration::
333: * Class Implementation::
334:
1.24 anton 335: The @file{mini-oof.fs} model
1.23 crook 336:
337: * Basic Mini-OOF Usage::
338: * Mini-OOF Example::
339: * Mini-OOF Implementation::
340:
1.12 anton 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:
1.24 anton 415: * Image Licensing Issues:: Distribution terms for images.
416: * Image File Background:: Why have image files?
1.32 anton 417: * Non-Relocatable Image Files:: don't always work.
1.24 anton 418: * Data-Relocatable Image Files:: are better.
1.32 anton 419: * Fully Relocatable Image Files:: better yet.
1.24 anton 420: * Stack and Dictionary Sizes:: Setting the default sizes for an image.
1.32 anton 421: * Running Image Files:: @code{gforth -i @i{file}} or @i{file}.
1.24 anton 422: * Modifying the Startup Sequence:: and turnkey applications.
1.12 anton 423:
424: Fully Relocatable Image Files
425:
1.27 crook 426: * gforthmi:: The normal way
1.12 anton 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::
1.13 pazsan 447:
448: Cross Compiler
449:
450: * Using the Cross Compiler::
451: * How the Cross Compiler Works::
452:
1.24 anton 453: Other Forth-related information
1.21 crook 454:
455: * Internet resources::
456: * Books::
457: * The Forth Interest Group::
458: * Conferences::
459:
1.24 anton 460: @end detailmenu
1.1 anton 461: @end menu
462:
1.26 crook 463: @node License, Goals, Top, Top
1.1 anton 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
1.21 crook 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.
1.1 anton 862: @end iftex
863:
1.28 crook 864: @comment TODO much more blurb here.
1.26 crook 865:
866: @c ******************************************************************
1.29 crook 867: @node Goals, Gforth Environment, License, Top
1.26 crook 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
1.29 crook 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
1.26 crook 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:
1.29 crook 958:
1.26 crook 959: @c ******************************************************************
1.29 crook 960: @node Gforth Environment, Introduction, Goals, Top
961: @chapter Gforth Environment
962: @cindex Gforth environment
1.21 crook 963:
1.29 crook 964: Note: ultimately, the gforth man page will be auto-generated from the
965: material in this chapter.
1.21 crook 966:
967: @menu
1.29 crook 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
1.21 crook 974: @end menu
975:
1.30 anton 976: @xref{Image Files} for related information about the creation of images.
1.29 crook 977:
1.21 crook 978: @comment ----------------------------------------------
1.29 crook 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
1.21 crook 986:
1.30 anton 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:
1.21 crook 992: @example
1.30 anton 993: gforth [file | -e forth-code] ...
1.21 crook 994: @end example
1.29 crook 995: @noindent
996: This interprets the contents of the files and the Forth code in the order they
997: are given.
1.21 crook 998:
1.30 anton 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:
1.29 crook 1003: In general, the command line looks like this:
1.21 crook 1004:
1005: @example
1.30 anton 1006: gforth[-fast] [engine options] [image options]
1.21 crook 1007: @end example
1008:
1.30 anton 1009: The engine options must come before the rest of the command
1.29 crook 1010: line. They are:
1.26 crook 1011:
1.29 crook 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}).
1.21 crook 1019:
1.29 crook 1020: @cindex --path, command-line option
1021: @cindex -p, command-line option
1022: @item --path @i{path}
1023: @itemx -p @i{path}
1024: Uses @i{path} for searching the image file and Forth source code files
1025: instead of the default in the environment variable @code{GFORTHPATH} or
1026: the path specified at installation time (e.g.,
1027: @file{/usr/local/share/gforth/0.2.0:.}). A path is given as a list of
1028: directories, separated by @samp{:} (on Unix) or @samp{;} (on other OSs).
1.21 crook 1029:
1.29 crook 1030: @cindex --dictionary-size, command-line option
1031: @cindex -m, command-line option
1032: @cindex @i{size} parameters for command-line options
1033: @cindex size of the dictionary and the stacks
1034: @item --dictionary-size @i{size}
1035: @itemx -m @i{size}
1036: Allocate @i{size} space for the Forth dictionary space instead of
1037: using the default specified in the image (typically 256K). The
1038: @i{size} specification for this and subsequent options consists of
1039: an integer and a unit (e.g.,
1040: @code{4M}). The unit can be one of @code{b} (bytes), @code{e} (element
1041: size, in this case Cells), @code{k} (kilobytes), @code{M} (Megabytes),
1042: @code{G} (Gigabytes), and @code{T} (Terabytes). If no unit is specified,
1043: @code{e} is used.
1.21 crook 1044:
1.29 crook 1045: @cindex --data-stack-size, command-line option
1046: @cindex -d, command-line option
1047: @item --data-stack-size @i{size}
1048: @itemx -d @i{size}
1049: Allocate @i{size} space for the data stack instead of using the
1050: default specified in the image (typically 16K).
1.21 crook 1051:
1.29 crook 1052: @cindex --return-stack-size, command-line option
1053: @cindex -r, command-line option
1054: @item --return-stack-size @i{size}
1055: @itemx -r @i{size}
1056: Allocate @i{size} space for the return stack instead of using the
1057: default specified in the image (typically 15K).
1.21 crook 1058:
1.29 crook 1059: @cindex --fp-stack-size, command-line option
1060: @cindex -f, command-line option
1061: @item --fp-stack-size @i{size}
1062: @itemx -f @i{size}
1063: Allocate @i{size} space for the floating point stack instead of
1064: using the default specified in the image (typically 15.5K). In this case
1065: the unit specifier @code{e} refers to floating point numbers.
1.21 crook 1066:
1.29 crook 1067: @cindex --locals-stack-size, command-line option
1068: @cindex -l, command-line option
1069: @item --locals-stack-size @i{size}
1070: @itemx -l @i{size}
1071: Allocate @i{size} space for the locals stack instead of using the
1072: default specified in the image (typically 14.5K).
1.21 crook 1073:
1.29 crook 1074: @cindex -h, command-line option
1075: @cindex --help, command-line option
1076: @item --help
1077: @itemx -h
1078: Print a message about the command-line options
1.21 crook 1079:
1.29 crook 1080: @cindex -v, command-line option
1081: @cindex --version, command-line option
1082: @item --version
1083: @itemx -v
1084: Print version and exit
1.21 crook 1085:
1.29 crook 1086: @cindex --debug, command-line option
1087: @item --debug
1088: Print some information useful for debugging on startup.
1.21 crook 1089:
1.29 crook 1090: @cindex --offset-image, command-line option
1091: @item --offset-image
1092: Start the dictionary at a slightly different position than would be used
1093: otherwise (useful for creating data-relocatable images,
1094: @pxref{Data-Relocatable Image Files}).
1.21 crook 1095:
1.29 crook 1096: @cindex --no-offset-im, command-line option
1097: @item --no-offset-im
1098: Start the dictionary at the normal position.
1.21 crook 1099:
1.29 crook 1100: @cindex --clear-dictionary, command-line option
1101: @item --clear-dictionary
1102: Initialize all bytes in the dictionary to 0 before loading the image
1103: (@pxref{Data-Relocatable Image Files}).
1104:
1105: @cindex --die-on-signal, command-line-option
1106: @item --die-on-signal
1107: Normally Gforth handles most signals (e.g., the user interrupt SIGINT,
1108: or the segmentation violation SIGSEGV) by translating it into a Forth
1109: @code{THROW}. With this option, Gforth exits if it receives such a
1110: signal. This option is useful when the engine and/or the image might be
1111: severely broken (such that it causes another signal before recovering
1112: from the first); this option avoids endless loops in such cases.
1113: @end table
1114:
1115: @cindex loading files at startup
1116: @cindex executing code on startup
1117: @cindex batch processing with Gforth
1118: As explained above, the image-specific command-line arguments for the
1119: default image @file{gforth.fi} consist of a sequence of filenames and
1120: @code{-e @var{forth-code}} options that are interpreted in the sequence
1121: in which they are given. The @code{-e @var{forth-code}} or
1122: @code{--evaluate @var{forth-code}} option evaluates the Forth
1123: code. This option takes only one argument; if you want to evaluate more
1124: Forth words, you have to quote them or use @code{-e} several times. To exit
1125: after processing the command line (instead of entering interactive mode)
1126: append @code{-e bye} to the command line.
1127:
1128: @cindex versions, invoking other versions of Gforth
1129: If you have several versions of Gforth installed, @code{gforth} will
1130: invoke the version that was installed last. @code{gforth-@i{version}}
1131: invokes a specific version. You may want to use the option
1132: @code{--path}, if your environment contains the variable
1133: @code{GFORTHPATH}.
1134:
1135: Not yet implemented:
1136: On startup the system first executes the system initialization file
1137: (unless the option @code{--no-init-file} is given; note that the system
1138: resulting from using this option may not be ANS Forth conformant). Then
1139: the user initialization file @file{.gforth.fs} is executed, unless the
1140: option @code{--no-rc} is given; this file is first searched in @file{.},
1141: then in @file{~}, then in the normal path (see above).
1.21 crook 1142:
1143:
1144:
1.29 crook 1145: @comment ----------------------------------------------
1146: @node Leaving Gforth, Command-line editing, Invoking Gforth, Gforth Environment
1147: @section Leaving Gforth
1148: @cindex Gforth - leaving
1149: @cindex leaving Gforth
1.21 crook 1150:
1.30 anton 1151: You can leave Gforth by typing @code{bye} or @kbd{Ctrl-d} (at the start
1152: of a line) or (if you invoked Gforth with the @code{--die-on-signal}
1153: option) @kbd{Ctrl-c}. When you leave Gforth, all of your definitions and
1154: data are discarded. @xref{Image Files} for ways of saving the state of
1155: the system before leaving Gforth.
1.21 crook 1156:
1.29 crook 1157: doc-bye
1.21 crook 1158:
1.29 crook 1159: @comment ----------------------------------------------
1160: @node Command-line editing, Upper and lower case,Leaving Gforth,Gforth Environment
1161: @section Command-line editing
1162: @cindex command-line editing
1.21 crook 1163:
1.29 crook 1164: Gforth maintains a history file that records every line that you type to
1165: the text interpreter. This file is preserved between sessions, and is
1166: used to provide a command-line recall facility; if you type ctrl-P
1167: repeatedly you can recall successively older commands from this (or
1168: previous) session(s). The full list of command-line editing facilities is:
1.21 crook 1169:
1.30 anton 1170: @comment use @table? - anton
1.21 crook 1171: @itemize @bullet
1172: @item
1.30 anton 1173: @kbd{Ctrl-p} (``previous'') (or up-arrow) to recall successively older
1.29 crook 1174: commands from the history buffer.
1175: @item
1.30 anton 1176: @kbd{Ctrl-n} (``next'') (or down-arrow) to recall successively newer commands
1.29 crook 1177: from the history buffer.
1178: @item
1.30 anton 1179: @kbd{Ctrl-f} (or right-arrow) to move the cursor right, non-destructively.
1.29 crook 1180: @item
1.30 anton 1181: @kbd{Ctrl-b} (or left-arrow) to move the cursor left, non-destructively.
1.29 crook 1182: @item
1.30 anton 1183: @kbd{Ctrl-h} (backspace) to delete the character to the left of the cursor,
1.29 crook 1184: closing up the line.
1185: @item
1.30 anton 1186: @kbd{Ctrl-k} to delete (``kill'') from the cursor to the end of the line.
1.29 crook 1187: @item
1.30 anton 1188: @kbd{Ctrl-a} to move the cursor to the start of the line.
1.21 crook 1189: @item
1.30 anton 1190: @kbd{Ctrl-e} to move the cursor to the end of the line.
1.21 crook 1191: @item
1.30 anton 1192: @key{RET} (@kbd{Ctrl-m}) or @key{LFD} (@kbd{Ctrl-j}) to submit the current
1.29 crook 1193: line.
1.21 crook 1194: @item
1.30 anton 1195: @key{TAB} to step through all possible full-word completions of the word
1.29 crook 1196: currently being typed.
1.21 crook 1197: @item
1.30 anton 1198: @kbd{Ctrl-d} at the start of the line to terminate Gforth (gracefully,
1199: using @code{bye}).
1.21 crook 1200: @end itemize
1201:
1.29 crook 1202: When editing, displayable characters are inserted to the left of the
1203: cursor position; the line is always in ``insert'' (as opposed to
1204: ``overstrike'') mode.
1205:
1206: @cindex history file
1207: @cindex @file{.gforth-history}
1208: On Unix systems, the history file is @file{~/.gforth-history} by
1209: default@footnote{i.e. it is stored in the user's home directory.}. You
1210: can find out the name and location of your history file using:
1211:
1212: @example
1213: history-file type \ Unix-class systems
1.21 crook 1214:
1.29 crook 1215: history-file type \ Other systems
1216: history-dir type
1.21 crook 1217: @end example
1218:
1.29 crook 1219: If you enter long definitions by hand, you can use a text editor to
1220: paste them out of the history file into a Forth source file for reuse at
1221: a later time.
1222:
1223: Gforth never trims the size of the history file, so you should do this
1224: periodically, if necessary.
1225:
1226: @comment this is all defined in history.fs
1227:
1228:
1229:
1230: @comment ----------------------------------------------
1231: @node Upper and lower case, Environment variables,Command-line editing,Gforth Environment
1232: @section Upper and lower case
1233: @cindex case-sensitivity
1234: @cindex upper and lower case
1235:
1236: Gforth is case-insensitive, so you can enter definitions and invoke
1237: Standard words using upper, lower or mixed case (however,
1238: @pxref{core-idef, Implementation-defined options, Implementation-defined
1239: options}).
1240:
1.30 anton 1241: ANS Forth only @i{requires} implementations to recognise Standard words
1242: when they are typed entirely in upper case. Therefore, a Standard
1243: program must use upper case for all Standard words. You can use whatever
1244: case you like for words that you define, but in a standard program you
1245: have to use the words in the same case that you defined them.
1246:
1247: Gforth supports case sensitivity through @code{table}s (case-sensitive
1248: wordlists, @pxref{Word Lists}).
1249:
1250: Two people have asked how to convert Gforth to case sensitivity; while
1251: we think this is a bad idea, you can change all wordlists into tables
1252: like this:
1.29 crook 1253:
1.30 anton 1254: @example
1255: ' table-find forth-wordlist wordlist-map @ !
1256: @end example
1257:
1258: Note that you now have to type the predefined words in the same case
1259: that we defined them, which are varying. You may want to convert them
1260: to your favourite case before doing this operation (I won't explain how,
1261: because if you are even contemplating to do this, you'd better have
1262: enough knowledge of Forth systems to know this already).
1.29 crook 1263:
1264: @comment ----------------------------------------------
1265: @node Environment variables, Gforth Files, Upper and lower case,Gforth Environment
1266: @section Environment variables
1267: @cindex environment variables
1.21 crook 1268:
1.29 crook 1269: Gforth uses these environment variables:
1.21 crook 1270:
1.29 crook 1271: @itemize @bullet
1272: @item
1273: @cindex GFORTHHIST - environment variable
1274: GFORTHHIST - (Unix systems only) specifies the directory in which to
1275: open/create the history file, @file{.gforth-history}. Default:
1276: @code{$HOME}.
1.21 crook 1277:
1.29 crook 1278: @item
1279: @cindex GFORTHPATH - environment variable
1280: GFORTHPATH - specifies the path used when searching for the gforth image file and
1281: for Forth source-code files.
1.21 crook 1282:
1.29 crook 1283: @item
1284: @cindex GFORTH - environment variable
1285: GFORTH - used by @file{gforthmi} @xref{gforthmi}.
1.26 crook 1286:
1.29 crook 1287: @item
1288: @cindex GFORTHD - environment variable
1289: GFORTHD - used by @file{gforthmi} @xref{gforthmi}.
1.21 crook 1290:
1.29 crook 1291: @item
1292: @cindex TMP, TEMP - environment variable
1293: TMP, TEMP - (non-Unix systems only) used as a potential location for the
1294: history file.
1295: @end itemize
1.21 crook 1296:
1.29 crook 1297: @comment also POSIXELY_CORRECT LINES COLUMNS HOME but no interest in
1298: @comment mentioning these.
1.21 crook 1299:
1.29 crook 1300: All the Gforth environment variables default to sensible values if they
1301: are not set.
1.21 crook 1302:
1303:
1.29 crook 1304: @comment ----------------------------------------------
1305: @node Gforth Files, ,Environment variables,Gforth Environment
1306: @section Gforth files
1307: @cindex Gforth files
1.21 crook 1308:
1.30 anton 1309: When you Gforth on a Unix system in the default places, it installs
1310: files in these locations:
1.21 crook 1311:
1.26 crook 1312: @itemize @bullet
1313: @item
1.29 crook 1314: @file{/usr/local/bin/gforth}
1315: @item
1316: @file{/usr/local/bin/gforthmi}
1317: @item
1318: @file{/usr/local/man/man1/gforth.1} - man page.
1319: @item
1320: @file{/usr/local/info} - the Info version of this manual.
1321: @item
1.30 anton 1322: @file{/usr/local/lib/gforth/<version>/...} - Gforth @file{.fi} files.
1.29 crook 1323: @item
1324: @file{/usr/local/share/gforth/<version>/TAGS} - Emacs TAGS file.
1.26 crook 1325: @item
1.30 anton 1326: @file{/usr/local/share/gforth/<version>/...} - Gforth source files.
1.26 crook 1327: @item
1.30 anton 1328: @file{.../emacs/site-lisp/gforth.el} - Emacs gforth mode.
1.26 crook 1329: @end itemize
1.21 crook 1330:
1.30 anton 1331: You can select different places for installation by using
1332: @code{configure} options (listed with @code{configure --help}).
1.21 crook 1333:
1.29 crook 1334: @c ******************************************************************
1335: @node Introduction, Words, Gforth Environment, Top
1336: @comment node-name, next, previous, up
1337: @chapter An Introduction to ANS Forth
1338: @cindex Forth - an introduction
1.21 crook 1339:
1.29 crook 1340: The primary purpose of this manual is to document Gforth. However, since
1341: Forth is not a widely-known language and there is a lack of up-to-date
1342: teaching material, it seems worthwhile to provide some introductory
1343: material. @xref{Forth-related information} for other sources of Forth-related
1344: information.
1.21 crook 1345:
1.29 crook 1346: The examples in this section should work on any ANS Forth; the
1347: output shown was produced using Gforth. Each example attempts to
1348: reproduce the exact output that Gforth produces. If you try out the
1349: examples (and you should), what you should type is shown @kbd{like this}
1350: and Gforth's response is shown @code{like this}. The single exception is
1.30 anton 1351: that, where the example shows @key{RET} it means that you should
1.29 crook 1352: press the ``carriage return'' key. Unfortunately, some output formats for
1353: this manual cannot show the difference between @kbd{this} and
1354: @code{this} which will make trying out the examples harder (but not
1355: impossible).
1.21 crook 1356:
1.29 crook 1357: Forth is an unusual language. It provides an interactive development
1358: environment which includes both an interpreter and compiler. Forth
1359: programming style encourages you to break a problem down into many
1360: @cindex factoring
1361: small fragments (@dfn{factoring}), and then to develop and test each
1362: fragment interactively. Forth advocates assert that breaking the
1363: edit-compile-test cycle used by conventional programming languages can
1364: lead to great productivity improvements.
1.21 crook 1365:
1.29 crook 1366: @menu
1367: * Introducing the Text Interpreter::
1368: * Stacks and Postfix notation::
1369: * Your first definition::
1370: * How does that work?::
1371: * Forth is written in Forth::
1372: * Review - elements of a Forth system::
1373: * Where to go next::
1374: * Exercises::
1375: @end menu
1.21 crook 1376:
1.29 crook 1377: @comment ----------------------------------------------
1378: @node Introducing the Text Interpreter, Stacks and Postfix notation, Introduction, Introduction
1379: @section Introducing the Text Interpreter
1380: @cindex text interpreter
1381: @cindex outer interpreter
1.21 crook 1382:
1.30 anton 1383: @c IMO this is too detailed and the pace is too slow for
1384: @c an introduction. If you know German, take a look at
1385: @c http://www.complang.tuwien.ac.at/anton/lvas/skriptum-stack.html
1386: @c to see how I do it - anton
1387:
1.29 crook 1388: When you invoke the Forth image, you will see a startup banner printed
1389: and nothing else (if you have Gforth installed on your system, try
1.30 anton 1390: invoking it now, by typing @kbd{gforth@key{RET}}). Forth is now running
1.29 crook 1391: its command line interpreter, which is called the @dfn{Text Interpreter}
1392: (also known as the @dfn{Outer Interpreter}). (You will learn a lot
1.30 anton 1393: about the text interpreter as you read through this chapter, but
1394: @pxref{The Text Interpreter} for more detail).
1.21 crook 1395:
1.29 crook 1396: Although it's not obvious, Forth is actually waiting for your
1.30 anton 1397: input. Type a number and press the @key{RET} key:
1.21 crook 1398:
1.26 crook 1399: @example
1.30 anton 1400: @kbd{45@key{RET}} ok
1.26 crook 1401: @end example
1.21 crook 1402:
1.29 crook 1403: Rather than give you a prompt to invite you to input something, the text
1404: interpreter prints a status message @i{after} it has processed a line
1405: of input. The status message in this case (``@code{ ok}'' followed by
1406: carriage-return) indicates that the text interpreter was able to process
1407: all of your input successfully. Now type something illegal:
1408:
1409: @example
1.30 anton 1410: @kbd{qwer341@key{RET}}
1.29 crook 1411: :1: Undefined word
1412: qwer341
1413: ^^^^^^^
1414: $400D2BA8 Bounce
1415: $400DBDA8 no.extensions
1416: @end example
1.23 crook 1417:
1.29 crook 1418: The exact text, other than the ``Undefined word'' may differ slightly on
1419: your system, but the effect is the same; when the text interpreter
1420: detects an error, it discards any remaining text on a line, resets
1.30 anton 1421: certain internal state and prints an error message. @xref{Error
1422: messages} for a detailed description of error messages.
1.23 crook 1423:
1.29 crook 1424: The text interpreter waits for you to press carriage-return, and then
1425: processes your input line. Starting at the beginning of the line, it
1426: breaks the line into groups of characters separated by spaces. For each
1427: group of characters in turn, it makes two attempts to do something:
1.23 crook 1428:
1.29 crook 1429: @itemize @bullet
1430: @item
1431: It tries to treat it as a command. It does this by searching a @dfn{name
1432: dictionary}. If the group of characters matches an entry in the name
1433: dictionary, the name dictionary provides the text interpreter with
1434: information that allows the text interpreter perform some actions. In
1435: Forth jargon, we say that the group
1436: @cindex word
1437: @cindex definition
1438: @cindex execution token
1439: @cindex xt
1440: of characters names a @dfn{word}, that the dictionary search returns an
1441: @dfn{execution token (xt)} corresponding to the @dfn{definition} of the
1442: word, and that the text interpreter executes the xt. Often, the terms
1443: @dfn{word} and @dfn{definition} are used interchangeably.
1444: @item
1445: If the text interpreter fails to find a match in the name dictionary, it
1446: tries to treat the group of characters as a number in the current number
1447: base (when you start up Forth, the current number base is base 10). If
1448: the group of characters legitimately represents a number, the text
1449: interpreter pushes the number onto a stack (we'll learn more about that
1450: in the next section).
1451: @end itemize
1.23 crook 1452:
1.29 crook 1453: If the text interpreter is unable to do either of these things with any
1454: group of characters, it discards the group of characters and the rest of
1455: the line, then prints an error message. If the text interpreter reaches
1456: the end of the line without error, it prints the status message ``@code{ ok}''
1457: followed by carriage-return.
1.21 crook 1458:
1.29 crook 1459: This is the simplest command we can give to the text interpreter:
1.23 crook 1460:
1461: @example
1.30 anton 1462: @key{RET} ok
1.23 crook 1463: @end example
1.21 crook 1464:
1.29 crook 1465: The text interpreter did everything we asked it to do (nothing) without
1466: an error, so it said that everything is ``@code{ ok}''. Try a slightly longer
1467: command:
1.21 crook 1468:
1.23 crook 1469: @example
1.30 anton 1470: @kbd{12 dup fred dup@key{RET}}
1.29 crook 1471: :1: Undefined word
1472: 12 dup fred dup
1473: ^^^^
1474: $400D2BA8 Bounce
1475: $400DBDA8 no.extensions
1.23 crook 1476: @end example
1.21 crook 1477:
1.29 crook 1478: When you press the carriage-return key, the text interpreter starts to
1479: work its way along the line:
1.21 crook 1480:
1.29 crook 1481: @itemize @bullet
1482: @item
1483: When it gets to the space after the @code{2}, it takes the group of
1484: characters @code{12} and looks them up in the name
1485: dictionary@footnote{We can't tell if it found them or not, but assume
1486: for now that it did not}. There is no match for this group of characters
1487: in the name dictionary, so it tries to treat them as a number. It is
1488: able to do this successfully, so it puts the number, 12, ``on the stack''
1489: (whatever that means).
1490: @item
1491: The text interpreter resumes scanning the line and gets the next group
1492: of characters, @code{dup}. It looks it up in the name dictionary and
1493: (you'll have to take my word for this) finds it, and executes the word
1494: @code{dup} (whatever that means).
1495: @item
1496: Once again, the text interpreter resumes scanning the line and gets the
1497: group of characters @code{fred}. It looks them up in the name
1498: dictionary, but can't find them. It tries to treat them as a number, but
1499: they don't represent any legal number.
1500: @end itemize
1.21 crook 1501:
1.29 crook 1502: At this point, the text interpreter gives up and prints an error
1503: message. The error message shows exactly how far the text interpreter
1504: got in processing the line. In particular, it shows that the text
1505: interpreter made no attempt to do anything with the final character
1506: group, @code{dup}, even though we have good reason to believe that the
1507: text interpreter would have no problem looking that word up and
1508: executing it a second time.
1.21 crook 1509:
1510:
1.29 crook 1511: @comment ----------------------------------------------
1512: @node Stacks and Postfix notation, Your first definition, Introducing the Text Interpreter, Introduction
1513: @section Stacks, postfix notation and parameter passing
1514: @cindex text interpreter
1515: @cindex outer interpreter
1.21 crook 1516:
1.29 crook 1517: In procedural programming languages (like C and Pascal), the
1518: building-block of programs is the @dfn{function} or @dfn{procedure}. These
1519: functions or procedures are called with @dfn{explicit parameters}. For
1520: example, in C we might write:
1.21 crook 1521:
1.23 crook 1522: @example
1.29 crook 1523: total = total + new_volume(length,height,depth);
1.23 crook 1524: @end example
1.21 crook 1525:
1.23 crook 1526: @noindent
1.29 crook 1527: where new_volume is a function-call to another piece of code, and total,
1528: length, height and depth are all variables. length, height and depth are
1529: parameters to the function-call.
1.21 crook 1530:
1.29 crook 1531: In Forth, the equivalent of the function or procedure is the
1532: @dfn{definition} and parameters are implicitly passed between
1533: definitions using a shared stack that is visible to the
1534: programmer. Although Forth does support variables, the existence of the
1535: stack means that they are used far less often than in most other
1536: programming languages. When the text interpreter encounters a number, it
1537: will place (@dfn{push}) it on the stack. There are several stacks (the
1.30 anton 1538: actual number is implementation-dependent ...) and the particular stack
1.29 crook 1539: used for any operation is implied unambiguously by the operation being
1540: performed. The stack used for all integer operations is called the @dfn{data
1541: stack} and, since this is the stack used most commonly, references to
1542: ``the data stack'' are often abbreviated to ``the stack''.
1.21 crook 1543:
1.29 crook 1544: The stacks have a last-in, first-out (LIFO) organisation. If you type:
1.21 crook 1545:
1.23 crook 1546: @example
1.30 anton 1547: @kbd{1 2 3@key{RET}} ok
1.23 crook 1548: @end example
1.21 crook 1549:
1.29 crook 1550: Then this instructs the text interpreter to placed three numbers on the
1551: (data) stack. An analogy for the behaviour of the stack is to take a
1552: pack of playing cards and deal out the ace (1), 2 and 3 into a pile on
1553: the table. The 3 was the last card onto the pile (``last-in'') and if
1554: you take a card off the pile then, unless you're prepared to fiddle a
1555: bit, the card that you take off will be the 3 (``first-out''). The
1556: number that will be first-out of the stack is called the @dfn{top of
1557: stack}, which
1558: @cindex TOS definition
1559: is often abbreviated to @dfn{TOS}.
1.21 crook 1560:
1.29 crook 1561: To understand how parameters are passed in Forth, consider the
1562: behaviour of the definition @code{+} (pronounced ``plus''). You will not
1563: be surprised to learn that this definition performs addition. More
1564: precisely, it adds two number together and produces a result. Where does
1565: it get the two numbers from? It takes the top two numbers off the
1566: stack. Where does it place the result? On the stack. You can act-out the
1567: behaviour of @code{+} with your playing cards like this:
1.21 crook 1568:
1569: @itemize @bullet
1570: @item
1.29 crook 1571: Pick up two cards from the stack on the table
1.21 crook 1572: @item
1.29 crook 1573: Stare at them intently and ask yourself ``what @i{is} the sum of these two
1574: numbers''
1.21 crook 1575: @item
1.29 crook 1576: Decide that the answer is 5
1.21 crook 1577: @item
1.29 crook 1578: Shuffle the two cards back into the pack and find a 5
1.21 crook 1579: @item
1.29 crook 1580: Put a 5 on the remaining ace that's on the table.
1.21 crook 1581: @end itemize
1582:
1.29 crook 1583: If you don't have a pack of cards handy but you do have Forth running,
1584: you can use the definition @code{.s} to show the current state of the stack,
1585: without affecting the stack. Type:
1.21 crook 1586:
1587: @example
1.30 anton 1588: @kbd{clearstack 1 2 3@key{RET}} ok
1589: @kbd{.s@key{RET}} <3> 1 2 3 ok
1.23 crook 1590: @end example
1591:
1.29 crook 1592: The text interpreter looks up the word @code{clearstack} and executes
1593: it; it tidies up the stack and removes any entries that may have been
1594: left on it by earlier examples. The text interpreter pushes each of the
1595: three numbers in turn onto the stack. Finally, the text interpreter
1596: looks up the word @code{.s} and executes it. The effect of executing
1597: @code{.s} is to print the ``<3>'' (the total number of items on the stack)
1598: followed by a list of all the items on the stack; the item on the far
1599: right-hand side is the TOS.
1.21 crook 1600:
1.29 crook 1601: You can now type:
1.21 crook 1602:
1.29 crook 1603: @example
1.30 anton 1604: @kbd{+ .s@key{RET}} <2> 1 5 ok
1.29 crook 1605: @end example
1.21 crook 1606:
1.29 crook 1607: @noindent
1608: which is correct; there are now 2 items on the stack and the result of
1609: the addition is 5.
1.23 crook 1610:
1.29 crook 1611: If you're playing with cards, try doing a second addition: pick up the
1612: two cards, work out that their sum is 6, shuffle them into the pack,
1613: look for a 6 and place that on the table. You now have just one item on
1614: the stack. What happens if you try to do a third addition? Pick up the
1615: first card, pick up the second card -- ah! There is no second card. This
1616: is called a @dfn{stack underflow} and consitutes an error. If you try to
1617: do the same thing with Forth it will report an error (probably a Stack
1618: Underflow or an Invalid Memory Address error).
1.23 crook 1619:
1.29 crook 1620: The opposite situation to a stack underflow is a @dfn{stack overflow},
1621: which simply accepts that there is a finite amount of storage space
1622: reserved for the stack. To stretch the playing card analogy, if you had
1623: enough packs of cards and you piled the cards up on the table, you would
1624: eventually be unable to add another card; you'd hit the ceiling. Gforth
1625: allows you to set the maximum size of the stacks. In general, the only
1626: time that you will get a stack overflow is because a definition has a
1627: bug in it and is generating data on the stack uncontrollably.
1.23 crook 1628:
1.29 crook 1629: There's one final use for the playing card analogy. If you model your
1630: stack using a pack of playing cards, the maximum number of items on
1631: your stack will be 52 (I assume you didn't use the Joker). The maximum
1632: @i{value} of any item on the stack is 13 (the King). In fact, the only
1633: possible numbers are positive integer numbers 1 through 13; you can't
1634: have (for example) 0 or 27 or 3.52 or -2. If you change the way you
1635: think about some of the cards, you can accommodate different
1636: numbers. For example, you could think of the Jack as representing 0,
1637: the Queen as representing -1 and the King as representing -2. Your
1638: *range* remains unchanged (you can still only represent a total of 13
1639: numbers) but the numbers that you can represent are -2 through 10.
1.28 crook 1640:
1.29 crook 1641: In that analogy, the limit was the amount of information that a single
1642: stack entry could hold, and Forth has a similar limit. In Forth, the
1643: size of a stack entry is called a @dfn{cell}. The actual size of a cell is
1644: implementation dependent and affects the maximum value that a stack
1645: entry can hold. A Standard Forth provides a cell size of at least
1646: 16-bits, and most desktop systems use a cell size of 32-bits.
1.21 crook 1647:
1.29 crook 1648: Forth does not do any type checking for you, so you are free to
1649: manipulate and combine stack items in any way you wish. A convenient way
1650: of treating stack items is as 2's complement signed integers, and that
1651: is what Standard words like @code{+} do. Therefore you can type:
1.21 crook 1652:
1.29 crook 1653: @example
1.30 anton 1654: @kbd{-5 12 + .s@key{RET}} <1> 7 ok
1.29 crook 1655: @end example
1.21 crook 1656:
1.29 crook 1657: If you use numbers and definitions like @code{+} in order to turn Forth
1658: into a great big pocket calculator, you will realise that it's rather
1659: different from a normal calculator. Rather than typing 2 + 3 = you had
1660: to type 2 3 + (ignore the fact that you had to use @code{.s} to see the
1661: result). The terminology used to describe this difference is to say that
1662: your calculator uses @dfn{Infix Notation} (parameters and operators are
1663: mixed) whilst Forth uses @dfn{Postfix Notation} (parameters and
1664: operators are separate), also called @dfn{Reverse Polish Notation}.
1.21 crook 1665:
1.29 crook 1666: Whilst postfix notation might look confusing to begin with, it has
1667: several important advantages:
1.21 crook 1668:
1.23 crook 1669: @itemize @bullet
1670: @item
1.29 crook 1671: it is unambiguous
1.23 crook 1672: @item
1.29 crook 1673: it is more concise
1.23 crook 1674: @item
1.29 crook 1675: it fits naturally with a stack-based system
1.23 crook 1676: @end itemize
1.21 crook 1677:
1.29 crook 1678: To examine these claims in more detail, consider these sums:
1.21 crook 1679:
1.29 crook 1680: @example
1681: 6 + 5 * 4 =
1682: 4 * 5 + 6 =
1683: @end example
1.21 crook 1684:
1.29 crook 1685: If you're just learning maths or your maths is very rusty, you will
1686: probably come up with the answer 44 for the first and 26 for the
1687: second. If you are a bit of a whizz at maths you will remember the
1688: @i{convention} that multiplication takes precendence over addition, and
1689: you'd come up with the answer 26 both times. To explain the answer 26
1690: to someone who got the answer 44, you'd probably rewrite the first sum
1691: like this:
1.21 crook 1692:
1.29 crook 1693: @example
1694: 6 + (5 * 4) =
1695: @end example
1.21 crook 1696:
1.29 crook 1697: If what you really wanted was to perform the addition before the
1698: multiplication, you would have to use parentheses to force it.
1.21 crook 1699:
1.29 crook 1700: If you did the first two sums on a pocket calculator you would probably
1701: get the right answers, unless you were very cautious and entered them using
1702: these keystroke sequences:
1.21 crook 1703:
1.29 crook 1704: 6 + 5 = * 4 =
1705: 4 * 5 = + 6 =
1.21 crook 1706:
1.29 crook 1707: Postfix notation is unambiguous because the order that the operators
1708: are applied is always explicit; that also means that parentheses are
1709: never required. The operators are @i{active} (the act of quoting the
1710: operator makes the operation occur) which removes the need for ``=''.
1.28 crook 1711:
1.29 crook 1712: The sum 6 + 5 * 4 can be written (in postfix notation) in two
1713: equivalent ways:
1.26 crook 1714:
1715: @example
1.29 crook 1716: 6 5 4 * + or:
1717: 5 4 * 6 +
1.26 crook 1718: @end example
1.23 crook 1719:
1.29 crook 1720: An important thing that you should notice about this notation is that
1721: the @i{order} of the numbers does not change; if you want to subtract
1722: 2 from 10 you type @code{10 2 -}.
1.1 anton 1723:
1.29 crook 1724: The reason that Forth uses postfix notation is very simple to explain: it
1725: makes the implementation extremely simple, and it follows naturally from
1726: using the stack as a mechanism for passing parameters. Another way of
1727: thinking about this is to realise that all Forth definitions are
1728: @i{active}; they execute as they are encountered by the text
1729: interpreter. The result of this is that the syntax of Forth is trivially
1730: simple.
1.1 anton 1731:
1732:
1733:
1.29 crook 1734: @comment ----------------------------------------------
1735: @node Your first definition, How does that work?, Stacks and Postfix notation, Introduction
1736: @section Your first Forth definition
1737: @cindex first definition
1.1 anton 1738:
1.29 crook 1739: Until now, the examples we've seen have been trivial; we've just been
1740: using Forth as a bigger-than-pocket calculator. Also, each calculation
1741: we've shown has been a ``one-off'' -- to repeat it we'd need to type it in
1742: again@footnote{That's not quite true. If you press the up-arrow key on
1743: your keyboard you should be able to scroll back to any earlier command,
1744: edit it and re-enter it.} In this section we'll see how to add new
1745: words to Forth's vocabulary.
1.1 anton 1746:
1.29 crook 1747: The easiest way to create a new word is to use a @dfn{colon
1748: definition}. We'll define a few and try them out before worrying too
1749: much about how they work. Try typing in these examples; be careful to
1750: copy the spaces accurately:
1.1 anton 1751:
1.29 crook 1752: @example
1753: : add-two 2 + . ;
1754: : greet ." Hello and welcome" ;
1755: : demo 5 add-two ;
1756: @end example
1.1 anton 1757:
1.29 crook 1758: @noindent
1759: Now try them out:
1.1 anton 1760:
1.29 crook 1761: @example
1.30 anton 1762: @kbd{greet@key{RET}} Hello and welcome ok
1763: @kbd{greet greet@key{RET}} Hello and welcomeHello and welcome ok
1764: @kbd{4 add-two@key{RET}} 6 ok
1765: @kbd{demo@key{RET}} 7 ok
1766: @kbd{9 greet demo add-two@key{RET}} Hello and welcome7 11 ok
1.29 crook 1767: @end example
1.1 anton 1768:
1.29 crook 1769: The first new thing that we've introduced here is the pair of words
1770: @code{:} and @code{;}. These are used to start and terminate a new
1771: definition, respectively. The first word after the @code{:} is the name
1772: for the new definition.
1.1 anton 1773:
1.29 crook 1774: As you can see from the examples, a definition is built up of words that
1775: have already been defined; Forth makes no distinction between
1776: definitions that existed when you started the system up, and those that
1777: you define yourself.
1.1 anton 1778:
1.29 crook 1779: The examples also introduce the words @code{.} (dot), @code{."}
1780: (dot-quote) and @code{dup} (dewp). Dot takes the value from the top of
1781: the stack and displays it. It's like @code{.s} except that it only
1782: displays the top item of the stack and it is destructive; after it has
1783: executed, the number is no longer on the stack. There is always one
1784: space printed after the number, and no spaces before it. Dot-quote
1785: defines a string (a sequence of characters) that will be printed when
1786: the word is executed. The string can contain any printable characters
1787: except @code{"}. A @code{"} has a special function; it is not a Forth
1788: word but it acts as a delimiter (the way that delimiters work is
1789: described in the next section). Finally, @code{dup} duplicates the value
1790: at the top of the stack. Try typing @code{5 dup .s} to see what it does.
1.1 anton 1791:
1.29 crook 1792: We already know that the text interpreter searches through the
1793: dictionary to locate names. If you've followed the examples earlier, you
1794: will already have a definition called @code{add-two}. Lets try modifying
1795: it by typing in a new definition:
1.1 anton 1796:
1.29 crook 1797: @example
1.30 anton 1798: @kbd{: add-two dup . ." + 2 =" 2 + . ;@key{RET}} redefined add-two ok
1.29 crook 1799: @end example
1.5 anton 1800:
1.29 crook 1801: Forth recognised that we were defining a word that already exists, and
1802: printed a message to warn us of that fact. Let's try out the new
1803: definition:
1.5 anton 1804:
1.29 crook 1805: @example
1.30 anton 1806: @kbd{9 add-two@key{RET}} 9 + 2 =11 ok
1.29 crook 1807: @end example
1.1 anton 1808:
1.29 crook 1809: @noindent
1810: All that we've actually done here, though, is to create a new
1811: definition, with a particular name. The fact that there was already a
1812: definition with the same name did not make any difference to the way
1813: that the new definition was created (except that Forth printed a warning
1814: message). The old definition of add-two still exists (try @code{demo}
1815: again to see that this is true). Any new definition will use the new
1816: definition of @code{add-two}, but old definitions continue to use the
1817: version that already existed at the time that they were @code{compiled}.
1.1 anton 1818:
1.29 crook 1819: Before you go on to the next section, try defining and redefining some
1820: words of your own.
1.1 anton 1821:
1.29 crook 1822: @comment ----------------------------------------------
1823: @node How does that work?, Forth is written in Forth, Your first definition, Introduction
1824: @section How does that work?
1825: @cindex parsing words
1.1 anton 1826:
1.30 anton 1827: @c That's pretty deep (IMO way too deep) for an introduction. - anton
1828:
1829: @c Is it a good idea to talk about the interpretation semantics of a
1830: @c number? We don't have an xt to go along with it. - anton
1831:
1832: @c Now that I have eliminated execution semantics, I wonder if it would not
1833: @c be better to keep them (or add run-time semantics), to make it easier to
1834: @c explain what compilation semantics usually does. - anton
1835:
1.29 crook 1836: Now we're going to take another look at the definition of @code{add-two}
1837: from the previous section. From our knowledge of the way that the text
1838: interpreter works, we would have expected this result when we tried to
1839: define @code{add-two}:
1.21 crook 1840:
1.29 crook 1841: @example
1.30 anton 1842: @kbd{: add-two 2 + . " ;@key{RET}}
1.29 crook 1843: ^^^^^^^
1844: Error: Undefined word
1845: @end example
1.28 crook 1846:
1.29 crook 1847: The reason that this didn't happen is bound up in the way that @code{:}
1848: works. The word @code{:} does two special things. The first special
1849: thing that it does prevents the text interpreter from ever seeing the
1850: characters @code{add-two}. The text interpreter uses a variable called
1851: @cindex modifying >IN
1852: @code{>IN} (pronounced ''to-in'') to keep track of where it is in the
1853: input line. When it encounters the word @code{:} it behaves in exactly
1854: the same way as it does for any other word; it looks it up in the name
1855: dictionary, finds its xt and executes it. When @code{:} executes, it
1856: looks at the input buffer, finds the word @code{add-two} and advances the
1857: value of @code{>IN} to point past it. It then does some other stuff
1858: associated with creating the new definition (including creating an entry
1859: for @code{add-two} in the name dictionary). When the execution of @code{:}
1860: completes, control returns to the text interpreter, which is oblivious
1861: to the fact that it has been tricked into ignoring part of the input
1862: line.
1.21 crook 1863:
1.29 crook 1864: @cindex parsing words
1865: Words like @code{:} -- words that advance the value of @code{>IN} and so
1866: prevent the text interpreter from acting on the whole of the input line
1867: -- are called @dfn{parsing words}.
1.21 crook 1868:
1.29 crook 1869: @cindex @code{state} - effect on the text interpreter
1870: @cindex text interpreter - effect of state
1871: The second special thing that @code{:} does is change the value of a
1872: variable called @code{state}, which affects the way that the text
1873: interpreter behaves. When Gforth starts up, @code{state} has the value
1874: 0, and the text interpreter is said to be @dfn{interpreting}. During a
1875: colon definition (started with @code{:}), @code{state} is set to -1 and
1876: the text interpreter is said to be @dfn{compiling}. The word @code{;}
1877: ends the definition -- one of the things that it does is to change the
1878: value of @code{state} back to 0.
1.21 crook 1879:
1.29 crook 1880: We have already seen how the text interpreter behaves when it is
1881: interpreting; it looks for each character sequence in the dictionary,
1882: finds its xt and executes it, or it converts it to a number and pushes
1883: it onto the stack, or it fails to do either and generates an error.
1.21 crook 1884:
1.29 crook 1885: When the text interpreter is compiling, its behaviour is slightly
1886: different; it still looks for each character sequence in the dictionary
1.30 anton 1887: and finds it, or converts it to a number, or fails to do either and
1888: generates an error. But instead of the execution token of a word it
1889: finds and executes the compilation token. For most words executing the
1890: compilation token results in laying down (@dfn{compiling}) the execution
1891: token, i.e., some magic to make that xt or number get executed or pushed
1892: at a later time; at the time that @code{add-two} is
1893: @dfn{executed}. Therefore, when you execute @code{add-two} its
1894: @dfn{run-time effect} is exactly the same as if you had typed @code{2 +
1895: .} outside of a definition, and pressed carriage-return.
1.28 crook 1896:
1.30 anton 1897: In Forth, every word or number can be described in terms of two
1.29 crook 1898: properties:
1.28 crook 1899:
1900: @itemize @bullet
1901: @item
1.30 anton 1902: Its @dfn{interpretation semantics}, represented by the execution token.
1.28 crook 1903: @item
1.30 anton 1904: Its @dfn{compilation semantics}, represented by the compilation token.
1.29 crook 1905: @end itemize
1906:
1.30 anton 1907: The value of @code{state} determines whether the text interpreter will
1908: use the compilation or interpretation semantics of a word or number that
1909: it encounters.
1.29 crook 1910:
1911: @itemize @bullet
1.28 crook 1912: @item
1.29 crook 1913: @cindex interpretation semantics
1914: When the text interpreter encounters a word or number in @dfn{interpret}
1915: state, it performs the @dfn{interpretation semantics} of the word or
1916: number.
1.28 crook 1917: @item
1.29 crook 1918: @cindex compilation semantics
1919: When the text interpreter encounters a word or number in @dfn{compile}
1920: state, it performs the @dfn{compilation semantics} of the word or
1921: number.
1922: @end itemize
1923:
1924: @noindent
1925: Numbers are always treated in a fixed way:
1926:
1927: @itemize @bullet
1.28 crook 1928: @item
1.30 anton 1929: When the number is @dfn{interpreted}, its behaviour is to push the number onto the stack.
1.28 crook 1930: @item
1.30 anton 1931: When the number is @dfn{compiled}, a piece of code is appended to the
1932: current definition that pushes the number when it runs. (In other words,
1933: the compilation semantics of a number are to postpone its interpretation
1934: semantics until the run-time of the definition that it is being compiled
1935: into.)
1.29 crook 1936: @end itemize
1937:
1938: The behaviour of a word is not so regular, but most have @i{default
1.30 anton 1939: compilation semantics} which means that they behave like this:
1.29 crook 1940:
1941: @itemize @bullet
1.28 crook 1942: @item
1.30 anton 1943: The @dfn{interpretation semantics} of the word are to do something useful.
1944: @item
1.29 crook 1945: The @dfn{compilation semantics} of the word are to append its
1.30 anton 1946: @dfn{interpretation semantics} to the current definition (so that its
1947: run-time behaviour is to do something useful).
1.28 crook 1948: @end itemize
1949:
1.30 anton 1950: @cindex immediate words
1.29 crook 1951: The actual behaviour of any particular word depends upon the way in
1952: which it was defined. When the text interpreter finds the word in the
1953: name dictionary, it not only retrieves the xt for the word, it also
1954: retrieves some flags: the @dfn{compile-only} flag and the @dfn{immediate
1955: flag}. The compile-only flag indicates that the word has no
1.30 anton 1956: interpretation semantics (the run-time behaviour for the default
1957: compilation semantics is not affected by this flag, however); any
1958: attempt to interpret a word that has the compile-only flag set will
1959: generate an error (for example, @code{IF} has no interpretation
1960: semantics). The immediate flag changes the compilation semantics of the
1961: word; if it is set, the compilation semantics are equal to the
1962: interpretation semantics (again ignoring the compile-only flag). it. In
1963: other words, these so-called @dfn{immediate} words behave like this:
1.29 crook 1964:
1965: @itemize @bullet
1966: @item
1.30 anton 1967: The @dfn{interpretation semantics} of the word are to do something useful.
1.29 crook 1968: @item
1.30 anton 1969: The @dfn{compilation semantics} of the word are to do something useful
1970: (and actually the same thing); i.e., it is executed during compilation.
1.29 crook 1971: @end itemize
1.28 crook 1972:
1.29 crook 1973: This example shows the difference between an immediate and a
1974: non-immediate word:
1.28 crook 1975:
1.29 crook 1976: @example
1977: : show-state state @@ . ;
1978: : show-state-now show-state ; immediate
1979: : word1 show-state ;
1980: : word2 show-state-now ;
1.28 crook 1981: @end example
1.23 crook 1982:
1.29 crook 1983: The word @code{immediate} after the definition of @code{show-state-now}
1984: makes that word an immediate word. These definitions introduce a new
1985: word: @code{@@} (pronounced ``fetch''). This word fetches the value of a
1986: variable, and leaves it on the stack. Therefore, the behaviour of
1987: @code{show-state} is to print a number that represents the current value
1988: of @code{state}.
1.28 crook 1989:
1.29 crook 1990: When you execute @code{word1}, it prints the number 0, indicating that
1991: the system is interpreting. When the text interpreter compiled the
1992: definition of @code{word1}, it encountered @code{show-state} whose
1.30 anton 1993: compilation semantics are to append its interpretation semantics to the
1.29 crook 1994: current definition. When you execute @code{word1}, it performs the
1.30 anton 1995: interpretation semantics of @code{show-state}. At the time that @code{word1}
1.29 crook 1996: (and therefore @code{show-state}) are executed, the system is
1997: interpreting.
1.28 crook 1998:
1.30 anton 1999: When you pressed @key{RET} after entering the definition of @code{word2},
1.29 crook 2000: you should have seen the number -1 printed, followed by ``@code{
2001: ok}''. When the text interpreter compiled the definition of
2002: @code{word2}, it encountered @code{show-state-now}, an immediate word,
1.30 anton 2003: whose compilation semantics are therefore to perform its interpretation
1.29 crook 2004: semantics. It is executed straight away (even before the text
2005: interpreter has moved on to process another group of characters; the
2006: @code{;} in this example). The effect of executing it are to display the
2007: value of @code{state} @i{at the time that the definition of}
2008: @code{word2} @i{is being defined}. Printing -1 demonstrates that the
2009: system is compiling at this time. If you execute @code{word2} it does
2010: nothing at all.
1.28 crook 2011:
1.29 crook 2012: @cindex @code{."}, how it works
2013: Before leaving the subject of immediate words, consider the behaviour of
2014: @code{."} in the definition of @code{greet}, in the previous
2015: section. This word is both a parsing word and an immediate word. Notice
2016: that there is a space between @code{."} and the start of the text
2017: @code{Hello and welcome}, but that there is no space between the last
2018: letter of @code{welcome} and the @code{"} character. The reason for this
2019: is that @code{."} is a Forth word; it must have a space after it so that
2020: the text interpreter can identify it. The @code{"} is not a Forth word;
2021: it is a @dfn{delimiter}. The examples earlier show that, when the string
2022: is displayed, there is neither a space before the @code{H} nor after the
2023: @code{e}. Since @code{."} is an immediate word, it executes at the time
2024: that @code{greet} is defined. When it executes, its behaviour is to
2025: search forward in the input line looking for the delimiter. When it
2026: finds the delimiter, it updates @code{>IN} to point past the
2027: delimiter. It also compiles some magic code into the definition of
2028: @code{greet}; the xt of a run-time routine that prints a text string. It
2029: compiles the string @code{Hello and welcome} into memory so that it is
2030: available to be printed later. When the text interpreter gains control,
2031: the next word it finds in the input stream is @code{;} and so it
2032: terminates the definition of @code{greet}.
1.28 crook 2033:
2034:
2035: @comment ----------------------------------------------
1.29 crook 2036: @node Forth is written in Forth, Review - elements of a Forth system, How does that work?, Introduction
2037: @section Forth is written in Forth
2038: @cindex structure of Forth programs
2039:
2040: When you start up a Forth compiler, a large number of definitions
2041: already exist. In Forth, you develop a new application using bottom-up
2042: programming techniques to create new definitions that are defined in
2043: terms of existing definitions. As you create each definition you can
2044: test and debug it interactively.
2045:
2046: If you have tried out the examples in this section, you will probably
2047: have typed them in by hand; when you leave Gforth, your definitions will
2048: be lost. You can avoid this by using a text editor to enter Forth source
2049: code into a file, and then loading code from the file using
2050: @code{include} (@xref{Forth source files}). A Forth source file is
2051: processed by the text interpreter, just as though you had typed it in by
2052: hand@footnote{Actually, there are some subtle differences -- see
2053: @ref{The Text Interpreter}.}.
2054:
2055: Gforth also supports the traditional Forth alternative to using text
2056: files for program entry (@xref{Blocks}).
1.28 crook 2057:
1.29 crook 2058: In common with many, if not most, Forth compilers, most of Gforth is
2059: actually written in Forth. All of the @file{.fs} files in the
2060: installation directory@footnote{For example,
1.30 anton 2061: @file{/usr/local/share/gforth...}} are Forth source files, which you can
1.29 crook 2062: study to see examples of Forth programming.
1.28 crook 2063:
1.29 crook 2064: Gforth maintains a history file that records every line that you type to
2065: the text interpreter. This file is preserved between sessions, and is
2066: used to provide a command-line recall facility. If you enter long
2067: definitions by hand, you can use a text editor to paste them out of the
2068: history file into a Forth source file for reuse at a later time
2069: (@pxref{Command-line editing} for more information).
1.28 crook 2070:
2071:
2072: @comment ----------------------------------------------
1.29 crook 2073: @node Review - elements of a Forth system, Where to go next, Forth is written in Forth, Introduction
2074: @section Review - elements of a Forth system
2075: @cindex elements of a Forth system
1.28 crook 2076:
1.29 crook 2077: To summarise this chapter:
1.28 crook 2078:
2079: @itemize @bullet
2080: @item
1.29 crook 2081: Forth programs use @dfn{factoring} to break a problem down into small
2082: fragments called @dfn{words} or @dfn{definitions}.
2083: @item
2084: Forth program development is an interactive process.
2085: @item
2086: The main command loop that accepts input, and controls both
2087: interpretation and compilation, is called the @dfn{text interpreter}
2088: (also known as the @dfn{outer interpreter}).
2089: @item
2090: Forth has a very simple syntax, consisting of words and numbers
2091: separated by spaces or carriage-return characters. Any additional syntax
2092: is imposed by @dfn{parsing words}.
2093: @item
2094: Forth uses a stack to pass parameters between words. As a result, it
2095: uses postfix notation.
2096: @item
2097: To use a word that has previously been defined, the text interpreter
2098: searches for the word in the @dfn{name dictionary}.
2099: @item
1.30 anton 2100: Words have @dfn{interpretation semantics} and @dfn{compilation semantics}.
1.28 crook 2101: @item
1.29 crook 2102: The text interpreter uses the value of @code{state} to select between
2103: the use of the @dfn{interpretation semantics} and the @dfn{compilation
2104: semantics} of a word that it encounters.
1.28 crook 2105: @item
1.30 anton 2106: The relationship between the @dfn{interpretation semantics} and
2107: @dfn{compilation semantics} for a word
1.29 crook 2108: depend upon the way in which the word was defined (for example, whether
2109: it is an @dfn{immediate} word).
1.28 crook 2110: @item
1.29 crook 2111: Forth definitions can be implemented in Forth (called @dfn{high-level
2112: definitions}) or in some other way (usually a lower-level language and
2113: as a result often called @dfn{low-level definitions}, @dfn{code
2114: definitions} or @dfn{primitives}).
1.28 crook 2115: @item
1.29 crook 2116: Many Forth systems are implemented mainly in Forth.
1.28 crook 2117: @end itemize
2118:
2119:
1.29 crook 2120: @comment ----------------------------------------------
2121: @node Where to go next,Exercises,Review - elements of a Forth system, Introduction
2122: @section Where To Go Next
2123: @cindex where to go next
1.28 crook 2124:
1.29 crook 2125: Amazing as it may seem, if you have read (and understood) this far, you
2126: know almost all the fundamentals about the inner workings of a Forth
2127: system. You certainly know enough to be able to read and understand the
2128: rest of this manual and the ANS Forth document, to learn more about the
2129: facilities that Forth in general and Gforth in particular provide. Even
2130: scarier, you know almost enough to implement your own Forth system.
1.30 anton 2131: However, that's not a good idea just yet... better to try writing some
1.29 crook 2132: programs in Gforth.
1.28 crook 2133:
1.29 crook 2134: Forth has such a rich vocabulary that it can be hard to know where to
2135: start in learning it. This section suggests a few sets of words that are
2136: enough to write small but useful programs. Use the word index in this
2137: document to learn more about each word, then try it out and try to write
2138: small definitions using it. Start by experimenting with these words:
1.28 crook 2139:
2140: @itemize @bullet
2141: @item
1.29 crook 2142: Arithmetic: @code{+ - * / /MOD */ ABS INVERT}
2143: @item
2144: Comparison: @code{MIN MAX =}
2145: @item
2146: Logic: @code{AND OR XOR NOT}
2147: @item
2148: Stack manipulation: @code{DUP DROP SWAP OVER}
1.28 crook 2149: @item
1.29 crook 2150: Loops and decisions: @code{IF ELSE ENDIF ?DO I LOOP}
1.28 crook 2151: @item
1.29 crook 2152: Input/Output: @code{. ." EMIT CR KEY}
1.28 crook 2153: @item
1.29 crook 2154: Defining words: @code{: ; CREATE}
1.28 crook 2155: @item
1.29 crook 2156: Memory allocation words: @code{ALLOT ,}
1.28 crook 2157: @item
1.29 crook 2158: Tools: @code{SEE WORDS .S MARKER}
2159: @end itemize
2160:
2161: When you have mastered those, go on to:
2162:
2163: @itemize @bullet
1.28 crook 2164: @item
1.29 crook 2165: More defining words: @code{VARIABLE CONSTANT VALUE TO CREATE DOES>}
1.28 crook 2166: @item
1.29 crook 2167: Memory access: @code{@@ !}
1.28 crook 2168: @end itemize
1.23 crook 2169:
1.29 crook 2170: When you have mastered these, there's nothing for it but to read through
2171: the whole of this manual and find out what you've missed.
2172:
2173: @comment ----------------------------------------------
2174: @node Exercises, ,Where to go next, Introduction
2175: @section Exercises
2176: @cindex exercises
2177:
2178: TODO: provide a set of programming excercises linked into the stuff done
2179: already and into other sections of the manual. Provide solutions to all
2180: the exercises in a .fs file in the distribution.
2181:
2182: @c Get some inspiration from Starting Forth and Kelly&Spies.
2183:
2184: @c excercises:
2185: @c 1. take inches and convert to feet and inches.
2186: @c 2. take temperature and convert from fahrenheight to celcius;
2187: @c may need to care about symmetric vs floored??
2188: @c 3. take input line and do character substitution
2189: @c to encipher or decipher
2190: @c 4. as above but work on a file for in and out
2191: @c 5. take input line and convert to pig-latin
2192: @c
2193: @c thing of sets of things to exercise then come up with
2194: @c problems that need those things.
2195:
2196:
1.26 crook 2197: @c ******************************************************************
1.29 crook 2198: @node Words, Error messages, Introduction, Top
1.1 anton 2199: @chapter Forth Words
1.26 crook 2200: @cindex words
1.1 anton 2201:
2202: @menu
2203: * Notation::
1.21 crook 2204: * Comments::
2205: * Boolean Flags::
1.1 anton 2206: * Arithmetic::
2207: * Stack Manipulation::
1.5 anton 2208: * Memory::
1.1 anton 2209: * Control Structures::
2210: * Defining Words::
1.21 crook 2211: * The Text Interpreter::
1.12 anton 2212: * Tokens for Words::
1.21 crook 2213: * Word Lists::
2214: * Environmental Queries::
1.12 anton 2215: * Files::
2216: * Blocks::
2217: * Other I/O::
2218: * Programming Tools::
2219: * Assembler and Code Words::
2220: * Threading Words::
1.26 crook 2221: * Locals::
2222: * Structures::
2223: * Object-oriented Forth::
1.21 crook 2224: * Passing Commands to the OS::
2225: * Miscellaneous Words::
1.1 anton 2226: @end menu
2227:
1.21 crook 2228: @node Notation, Comments, Words, Words
1.1 anton 2229: @section Notation
2230: @cindex notation of glossary entries
2231: @cindex format of glossary entries
2232: @cindex glossary notation format
2233: @cindex word glossary entry format
2234:
2235: The Forth words are described in this section in the glossary notation
2236: that has become a de-facto standard for Forth texts, i.e.,
2237:
2238: @format
1.29 crook 2239: @i{word} @i{Stack effect} @i{wordset} @i{pronunciation}
1.1 anton 2240: @end format
1.29 crook 2241: @i{Description}
1.1 anton 2242:
2243: @table @var
2244: @item word
1.28 crook 2245: The name of the word.
1.1 anton 2246:
2247: @item Stack effect
2248: @cindex stack effect
1.29 crook 2249: The stack effect is written in the notation @code{@i{before} --
2250: @i{after}}, where @i{before} and @i{after} describe the top of
1.1 anton 2251: stack entries before and after the execution of the word. The rest of
2252: the stack is not touched by the word. The top of stack is rightmost,
2253: i.e., a stack sequence is written as it is typed in. Note that Gforth
2254: uses a separate floating point stack, but a unified stack
1.29 crook 2255: notation. Also, return stack effects are not shown in @i{stack
2256: effect}, but in @i{Description}. The name of a stack item describes
1.1 anton 2257: the type and/or the function of the item. See below for a discussion of
2258: the types.
2259:
2260: All words have two stack effects: A compile-time stack effect and a
2261: run-time stack effect. The compile-time stack-effect of most words is
1.29 crook 2262: @i{ -- }. If the compile-time stack-effect of a word deviates from
1.1 anton 2263: this standard behaviour, or the word does other unusual things at
2264: compile time, both stack effects are shown; otherwise only the run-time
2265: stack effect is shown.
2266:
2267: @cindex pronounciation of words
2268: @item pronunciation
2269: How the word is pronounced.
2270:
2271: @cindex wordset
2272: @item wordset
1.21 crook 2273: The ANS Forth standard is divided into several word sets. A standard
2274: system need not support all of them. Therefore, in theory, the fewer
2275: word sets your program uses the more portable it will be. However, we
2276: suspect that most ANS Forth systems on personal machines will feature
1.26 crook 2277: all word sets. Words that are not defined in ANS Forth have
1.21 crook 2278: @code{gforth} or @code{gforth-internal} as word set. @code{gforth}
1.1 anton 2279: describes words that will work in future releases of Gforth;
2280: @code{gforth-internal} words are more volatile. Environmental query
2281: strings are also displayed like words; you can recognize them by the
1.21 crook 2282: @code{environment} in the word set field.
1.1 anton 2283:
2284: @item Description
2285: A description of the behaviour of the word.
2286: @end table
2287:
2288: @cindex types of stack items
2289: @cindex stack item types
2290: The type of a stack item is specified by the character(s) the name
2291: starts with:
2292:
2293: @table @code
2294: @item f
2295: @cindex @code{f}, stack item type
2296: Boolean flags, i.e. @code{false} or @code{true}.
2297: @item c
2298: @cindex @code{c}, stack item type
2299: Char
2300: @item w
2301: @cindex @code{w}, stack item type
2302: Cell, can contain an integer or an address
2303: @item n
2304: @cindex @code{n}, stack item type
2305: signed integer
2306: @item u
2307: @cindex @code{u}, stack item type
2308: unsigned integer
2309: @item d
2310: @cindex @code{d}, stack item type
2311: double sized signed integer
2312: @item ud
2313: @cindex @code{ud}, stack item type
2314: double sized unsigned integer
2315: @item r
2316: @cindex @code{r}, stack item type
2317: Float (on the FP stack)
1.21 crook 2318: @item a-
1.1 anton 2319: @cindex @code{a_}, stack item type
2320: Cell-aligned address
1.21 crook 2321: @item c-
1.1 anton 2322: @cindex @code{c_}, stack item type
2323: Char-aligned address (note that a Char may have two bytes in Windows NT)
1.21 crook 2324: @item f-
1.1 anton 2325: @cindex @code{f_}, stack item type
2326: Float-aligned address
1.21 crook 2327: @item df-
1.1 anton 2328: @cindex @code{df_}, stack item type
2329: Address aligned for IEEE double precision float
1.21 crook 2330: @item sf-
1.1 anton 2331: @cindex @code{sf_}, stack item type
2332: Address aligned for IEEE single precision float
2333: @item xt
2334: @cindex @code{xt}, stack item type
2335: Execution token, same size as Cell
2336: @item wid
2337: @cindex @code{wid}, stack item type
1.21 crook 2338: Word list ID, same size as Cell
1.1 anton 2339: @item f83name
2340: @cindex @code{f83name}, stack item type
2341: Pointer to a name structure
2342: @item "
2343: @cindex @code{"}, stack item type
1.12 anton 2344: string in the input stream (not on the stack). The terminating character
2345: is a blank by default. If it is not a blank, it is shown in @code{<>}
1.1 anton 2346: quotes.
2347: @end table
2348:
1.21 crook 2349: @node Comments, Boolean Flags, Notation, Words
2350: @section Comments
1.26 crook 2351: @cindex comments
1.21 crook 2352:
1.29 crook 2353: Forth supports two styles of comment; the traditional @i{in-line} comment,
2354: @code{(} and its modern cousin, the @i{comment to end of line}; @code{\}.
1.21 crook 2355:
1.23 crook 2356: doc-(
1.21 crook 2357: doc-\
1.23 crook 2358: doc-\G
1.21 crook 2359:
2360: @node Boolean Flags, Arithmetic, Comments, Words
2361: @section Boolean Flags
1.26 crook 2362: @cindex Boolean flags
1.21 crook 2363:
2364: A Boolean flag is cell-sized. A cell with all bits clear represents the
2365: flag @code{false} and a flag with all bits set represents the flag
1.26 crook 2366: @code{true}. Words that check a flag (for example, @code{IF}) will treat
1.29 crook 2367: a cell that has @i{any} bit set as @code{true}.
1.21 crook 2368:
2369: doc-true
2370: doc-false
1.29 crook 2371: doc-on
2372: doc-off
1.21 crook 2373:
2374: @node Arithmetic, Stack Manipulation, Boolean Flags, Words
1.1 anton 2375: @section Arithmetic
2376: @cindex arithmetic words
2377:
2378: @cindex division with potentially negative operands
2379: Forth arithmetic is not checked, i.e., you will not hear about integer
2380: overflow on addition or multiplication, you may hear about division by
2381: zero if you are lucky. The operator is written after the operands, but
2382: the operands are still in the original order. I.e., the infix @code{2-1}
2383: corresponds to @code{2 1 -}. Forth offers a variety of division
2384: operators. If you perform division with potentially negative operands,
2385: you do not want to use @code{/} or @code{/mod} with its undefined
2386: behaviour, but rather @code{fm/mod} or @code{sm/mod} (probably the
2387: former, @pxref{Mixed precision}).
1.26 crook 2388: @comment TODO discuss the different division forms and the std approach
1.1 anton 2389:
2390: @menu
2391: * Single precision::
2392: * Bitwise operations::
1.21 crook 2393: * Double precision:: Double-cell integer arithmetic
2394: * Numeric comparison::
1.29 crook 2395: * Mixed precision:: Operations with single and double-cell integers
1.1 anton 2396: * Floating Point::
2397: @end menu
2398:
2399: @node Single precision, Bitwise operations, Arithmetic, Arithmetic
2400: @subsection Single precision
2401: @cindex single precision arithmetic words
2402:
1.21 crook 2403: By default, numbers in Forth are single-precision integers that are 1
1.26 crook 2404: cell in size. They can be signed or unsigned, depending upon how you
1.21 crook 2405: treat them. @xref{Number Conversion} for the rules used by the text
2406: interpreter for recognising single-precision integers.
2407:
1.1 anton 2408: doc-+
1.21 crook 2409: doc-1+
1.1 anton 2410: doc--
1.21 crook 2411: doc-1-
1.1 anton 2412: doc-*
2413: doc-/
2414: doc-mod
2415: doc-/mod
2416: doc-negate
2417: doc-abs
2418: doc-min
2419: doc-max
1.21 crook 2420: doc-d>s
1.27 crook 2421: doc-floored
1.1 anton 2422:
1.21 crook 2423: @node Bitwise operations, Double precision, Single precision, Arithmetic
1.1 anton 2424: @subsection Bitwise operations
2425: @cindex bitwise operation words
2426:
2427: doc-and
2428: doc-or
2429: doc-xor
2430: doc-invert
1.21 crook 2431: doc-lshift
2432: doc-rshift
1.1 anton 2433: doc-2*
1.21 crook 2434: doc-d2*
1.1 anton 2435: doc-2/
1.21 crook 2436: doc-d2/
2437:
2438: @node Double precision, Numeric comparison, Bitwise operations, Arithmetic
2439: @subsection Double precision
2440: @cindex double precision arithmetic words
2441:
2442: @xref{Number Conversion} for the rules used by the text interpreter for
2443: recognising double-precision integers.
2444:
2445: A double precision number is represented by a cell pair, with the most
1.31 anton 2446: significant cell at the TOS. It is trivial to convert an unsigned
1.26 crook 2447: single to an (unsigned) double; simply push a @code{0} onto the
2448: TOS. Since numbers are represented by Gforth using 2's complement
2449: arithmetic, converting a signed single to a (signed) double requires
1.31 anton 2450: sign-extension across the most significant cell. This can be achieved
1.26 crook 2451: using @code{s>d}. The moral of the story is that you cannot convert a
2452: number without knowing whether it represents an unsigned or a
2453: signed number.
1.21 crook 2454:
2455: doc-s>d
2456: doc-d+
2457: doc-d-
2458: doc-dnegate
2459: doc-dabs
2460: doc-dmin
2461: doc-dmax
2462:
2463: @node Numeric comparison, Mixed precision, Double precision, Arithmetic
2464: @subsection Numeric comparison
2465: @cindex numeric comparison words
2466:
1.28 crook 2467: doc-<
2468: doc-<=
2469: doc-<>
2470: doc-=
2471: doc->
2472: doc->=
2473:
1.21 crook 2474: doc-0<
1.23 crook 2475: doc-0<=
1.21 crook 2476: doc-0<>
2477: doc-0=
1.23 crook 2478: doc-0>
2479: doc-0>=
1.28 crook 2480:
2481: doc-u<
2482: doc-u<=
1.31 anton 2483: @c TODO why u<> and u= ... they are the same as <> and =
2484: @c commented them out because they are unnecessary
2485: @c doc-u<>
2486: @c doc-u=
1.28 crook 2487: doc-u>
2488: doc-u>=
2489:
2490: doc-within
2491:
2492: doc-d<
2493: doc-d<=
2494: doc-d<>
2495: doc-d=
2496: doc-d>
2497: doc-d>=
1.23 crook 2498:
1.21 crook 2499: doc-d0<
1.23 crook 2500: doc-d0<=
2501: doc-d0<>
1.21 crook 2502: doc-d0=
1.23 crook 2503: doc-d0>
2504: doc-d0>=
2505:
1.21 crook 2506: doc-du<
1.28 crook 2507: doc-du<=
1.31 anton 2508: @c doc-du<>
2509: @c doc-du=
1.28 crook 2510: doc-du>
2511: doc-du>=
1.1 anton 2512:
1.21 crook 2513: @node Mixed precision, Floating Point, Numeric comparison, Arithmetic
1.1 anton 2514: @subsection Mixed precision
2515: @cindex mixed precision arithmetic words
2516:
2517: doc-m+
2518: doc-*/
2519: doc-*/mod
2520: doc-m*
2521: doc-um*
2522: doc-m*/
2523: doc-um/mod
2524: doc-fm/mod
2525: doc-sm/rem
2526:
1.21 crook 2527: @node Floating Point, , Mixed precision, Arithmetic
1.1 anton 2528: @subsection Floating Point
2529: @cindex floating point arithmetic words
2530:
1.21 crook 2531: @xref{Number Conversion} for the rules used by the text interpreter for
2532: recognising floating-point numbers.
1.1 anton 2533:
1.32 anton 2534: Gforth has a separate floating point
1.26 crook 2535: stack, but the documentation uses the unified notation.
1.1 anton 2536:
2537: @cindex floating-point arithmetic, pitfalls
2538: Floating point numbers have a number of unpleasant surprises for the
2539: unwary (e.g., floating point addition is not associative) and even a few
2540: for the wary. You should not use them unless you know what you are doing
2541: or you don't care that the results you get are totally bogus. If you
2542: want to learn about the problems of floating point numbers (and how to
2543: avoid them), you might start with @cite{David Goldberg, What Every
2544: Computer Scientist Should Know About Floating-Point Arithmetic, ACM
1.17 anton 2545: Computing Surveys 23(1):5@minus{}48, March 1991}
2546: (@url{http://www.validgh.com/goldberg/paper.ps}).
1.1 anton 2547:
1.21 crook 2548: doc-d>f
2549: doc-f>d
1.1 anton 2550: doc-f+
2551: doc-f-
2552: doc-f*
2553: doc-f/
2554: doc-fnegate
2555: doc-fabs
2556: doc-fmax
2557: doc-fmin
2558: doc-floor
2559: doc-fround
2560: doc-f**
2561: doc-fsqrt
2562: doc-fexp
2563: doc-fexpm1
2564: doc-fln
2565: doc-flnp1
2566: doc-flog
2567: doc-falog
1.32 anton 2568: doc-f2*
2569: doc-f2/
2570: doc-1/f
2571: doc-precision
2572: doc-set-precision
2573:
2574: @cindex angles in trigonometric operations
2575: @cindex trigonometric operations
2576: Angles in floating point operations are given in radians (a full circle
2577: has 2 pi radians).
2578:
1.1 anton 2579: doc-fsin
2580: doc-fcos
2581: doc-fsincos
2582: doc-ftan
2583: doc-fasin
2584: doc-facos
2585: doc-fatan
2586: doc-fatan2
2587: doc-fsinh
2588: doc-fcosh
2589: doc-ftanh
2590: doc-fasinh
2591: doc-facosh
2592: doc-fatanh
1.21 crook 2593: doc-pi
1.28 crook 2594:
1.32 anton 2595: @cindex equality of floats
2596: @cindex floating-point comparisons
1.31 anton 2597: One particular problem with floating-point arithmetic is that comparison
2598: for equality often fails when you would expect it to succeed. For this
2599: reason approximate equality is often preferred (but you still have to
2600: know what you are doing). The comparison words are:
2601:
2602: doc-f~rel
2603: doc-f~abs
2604: doc-f=
2605: doc-f~
2606: doc-f<>
2607:
2608: doc-f<
2609: doc-f<=
2610: doc-f>
2611: doc-f>=
2612:
1.21 crook 2613: doc-f0<
1.28 crook 2614: doc-f0<=
2615: doc-f0<>
1.21 crook 2616: doc-f0=
1.28 crook 2617: doc-f0>
2618: doc-f0>=
2619:
1.1 anton 2620:
2621: @node Stack Manipulation, Memory, Arithmetic, Words
2622: @section Stack Manipulation
2623: @cindex stack manipulation words
2624:
2625: @cindex floating-point stack in the standard
1.21 crook 2626: Gforth maintains a number of separate stacks:
2627:
1.29 crook 2628: @cindex data stack
2629: @cindex parameter stack
1.21 crook 2630: @itemize @bullet
2631: @item
1.29 crook 2632: A data stack (also known as the @dfn{parameter stack}) -- for
2633: characters, cells, addresses, and double cells.
1.21 crook 2634:
1.29 crook 2635: @cindex floating-point stack
1.21 crook 2636: @item
2637: A floating point stack -- for floating point numbers.
2638:
1.29 crook 2639: @cindex return stack
1.21 crook 2640: @item
2641: A return stack -- for storing the return addresses of colon
1.32 anton 2642: definitions and other (non-FP) data.
1.21 crook 2643:
1.29 crook 2644: @cindex locals stack
1.21 crook 2645: @item
2646: A locals stack for storing local variables.
2647: @end itemize
2648:
1.1 anton 2649: @menu
2650: * Data stack::
2651: * Floating point stack::
2652: * Return stack::
2653: * Locals stack::
2654: * Stack pointer manipulation::
2655: @end menu
2656:
2657: @node Data stack, Floating point stack, Stack Manipulation, Stack Manipulation
2658: @subsection Data stack
2659: @cindex data stack manipulation words
2660: @cindex stack manipulations words, data stack
2661:
2662: doc-drop
2663: doc-nip
2664: doc-dup
2665: doc-over
2666: doc-tuck
2667: doc-swap
1.21 crook 2668: doc-pick
1.1 anton 2669: doc-rot
2670: doc--rot
2671: doc-?dup
2672: doc-roll
2673: doc-2drop
2674: doc-2nip
2675: doc-2dup
2676: doc-2over
2677: doc-2tuck
2678: doc-2swap
2679: doc-2rot
2680:
2681: @node Floating point stack, Return stack, Data stack, Stack Manipulation
2682: @subsection Floating point stack
2683: @cindex floating-point stack manipulation words
2684: @cindex stack manipulation words, floating-point stack
2685:
1.32 anton 2686: Whilst every sane Forth has a separate floating-point stack, it is not
2687: strictly required; an ANS Forth system could theoretically keep
2688: floating-point numbers on the data stack. As an additional difficulty,
2689: you don't know how many cells a floating-point number takes. It is
2690: reportedly possible to write words in a way that they work also for a
2691: unified stack model, but we do not recommend trying it. Instead, just
2692: say that your program has an environmental dependency on a separate
2693: floating-point stack.
2694:
2695: doc-floating-stack
2696:
1.1 anton 2697: doc-fdrop
2698: doc-fnip
2699: doc-fdup
2700: doc-fover
2701: doc-ftuck
2702: doc-fswap
1.21 crook 2703: doc-fpick
1.1 anton 2704: doc-frot
2705:
2706: @node Return stack, Locals stack, Floating point stack, Stack Manipulation
2707: @subsection Return stack
2708: @cindex return stack manipulation words
2709: @cindex stack manipulation words, return stack
2710:
1.32 anton 2711: @cindex return stack and locals
2712: @cindex locals and return stack
2713: A Forth system is allowed to keep local variables on the
2714: return stack. This is reasonable, as local variables usually eliminate
2715: the need to use the return stack explicitly. So, if you want to produce
2716: a standard compliant program and you are using local variables in a
2717: word, forget about return stack manipulations in that word (refer to the
2718: standard document for the exact rules).
2719:
1.1 anton 2720: doc->r
2721: doc-r>
2722: doc-r@
2723: doc-rdrop
2724: doc-2>r
2725: doc-2r>
2726: doc-2r@
2727: doc-2rdrop
2728:
2729: @node Locals stack, Stack pointer manipulation, Return stack, Stack Manipulation
2730: @subsection Locals stack
2731:
1.26 crook 2732: @comment TODO
1.21 crook 2733:
1.1 anton 2734: @node Stack pointer manipulation, , Locals stack, Stack Manipulation
2735: @subsection Stack pointer manipulation
2736: @cindex stack pointer manipulation words
2737:
1.21 crook 2738: doc-sp0
1.1 anton 2739: doc-sp@
2740: doc-sp!
1.21 crook 2741: doc-fp0
1.1 anton 2742: doc-fp@
2743: doc-fp!
1.21 crook 2744: doc-rp0
1.1 anton 2745: doc-rp@
2746: doc-rp!
1.21 crook 2747: doc-lp0
1.1 anton 2748: doc-lp@
2749: doc-lp!
2750:
2751: @node Memory, Control Structures, Stack Manipulation, Words
2752: @section Memory
1.26 crook 2753: @cindex memory words
1.1 anton 2754:
1.32 anton 2755: @menu
2756: * Memory model::
2757: * Dictionary allocation::
2758: * Heap Allocation::
2759: * Memory Access::
2760: * Address arithmetic::
2761: * Memory Blocks::
2762: @end menu
2763:
2764: @node Memory model, Dictionary allocation, Memory, Memory
2765: @subsection ANS Forth and Gforth memory models
2766:
2767: @c The ANS Forth description is a mess (e.g., is the heap part of
2768: @c the dictionary?), so let's not stick to closely with it.
2769:
2770: ANS Forth considers a Forth system as consisting of several memories, of
2771: which only @dfn{data space} is managed and accessible with the memory
2772: words. Memory not necessarily in data space includes the stacks, the
2773: code (called code space) and the headers (called name space). In Gforth
2774: everything is in data space, but the code for the primitives is usually
2775: read-only.
2776:
2777: Data space is divided into a number of areas: The (data space portion of
2778: the) dictionary@footnote{Sometimes, the term @dfn{dictionary} is used to
2779: refer to the search data structure embodied in word lists and headers,
2780: because it is used for looking up names, just as you would in a
2781: conventional dictionary.}, the heap, and a number of system-allocated
2782: buffers.
2783:
2784: In ANS Forth data space is also divided into contiguous regions. You
2785: can only use address arithmetic within a contiguous region, not between
2786: them. Usually each allocation gives you one contiguous region, but the
1.33 anton 2787: dictionary allocation words have additional rules (@pxref{Dictionary
1.32 anton 2788: allocation}).
2789:
2790: Gforth provides one big address space, and address arithmetic can be
2791: performed between any addresses. However, in the dictionary headers or
2792: code are interleaved with data, so almost the only contiguous data space
2793: regions there are those described by ANS Forth as contiguous; but you
2794: can be sure that the dictionary is allocated towards increasing
2795: addresses even between contiguous regions. The memory order of
2796: allocations in the heap is platform-dependent (and possibly different
2797: from one run to the next).
2798:
2799: @subsubsection ANS Forth dictionary details
2800:
2801: @c !! I have deleted some of the stuff this section refers to - anton
1.27 crook 2802:
1.32 anton 2803: This section is just informative, you can skip it if you are in a hurry.
1.27 crook 2804:
1.29 crook 2805: When you create a colon definition, the text interpreter compiles the
1.32 anton 2806: code for the definition into the code space and compiles the name
2807: of the definition into the header space, together with other
1.27 crook 2808: information about the definition (such as its execution token).
2809:
2810: When you create a variable, the execution of @code{variable} will
1.32 anton 2811: compile some code, assign one cell in data space, and compile the name
2812: of the variable into the header space.
1.27 crook 2813:
2814: @cindex memory regions - relationship between them
2815: ANS Forth does not specify the relationship between the three memory
2816: regions, and specifies that a Standard program must not access code or
2817: data space directly -- it may only access data space directly. In
2818: addition, the Standard defines what relationships you may and may not
2819: rely on when allocating regions in data space. These constraints are
2820: simply a reflection of the many diverse techniques that are used to
2821: implement Forth systems; understanding and following the requirements of
2822: the Standard allows you to write portable programs -- programs that run
2823: in the same way on any of these diverse systems. Another way of looking
2824: at this is to say that ANS Forth was designed to permit compliant Forth
2825: systems to be implemented in many diverse ways.
2826:
2827: @cindex memory regions - how they are assigned
1.29 crook 2828: Here are some examples of ways in which name, code and data spaces
2829: might be assigned in different Forth implementations:
1.27 crook 2830:
2831: @itemize @bullet
2832: @item
2833: For a Forth system that runs from RAM under a general-purpose operating
2834: system, it can be convenient to interleave name, code and data spaces in
2835: a single contiguous memory region. This organisation can be
2836: memory-efficient (for example, because the relationship between the name
1.32 anton 2837: dictionary entry and the associated code space entry can be
1.27 crook 2838: implicit, rather than requiring an explicit memory pointer to reference
1.32 anton 2839: from the header space and the code space). This is the
1.27 crook 2840: organisation used by Gforth, as this example@footnote{The addresses
2841: in the example have been truncated to fit it onto the page, and the
2842: addresses and data shown will not match the output from your system} shows:
2843: @example
2844: hex
2845: variable fred 123456 fred !
2846: variable jim abcd jim !
2847: : foo + / - ;
2848: ' fred 10 - 50 dump
2849: ..80: 5C 46 0E 40 84 66 72 65 - 64 20 20 20 20 20 20 20 \F.@.fred
2850: ..90: D0 9B 04 08 00 00 00 00 - 56 34 12 00 80 46 0E 40 ........V4...F.@
2851: ..A0: 83 6A 69 6D 20 20 20 20 - D0 9B 04 08 00 00 00 00 .jim ........
2852: ..B0: CD AB 00 00 9C 46 0E 40 - 83 66 6F 6F 20 20 20 20 .....F.@.foo
2853: ..C0: 80 9B 04 08 00 00 00 00 - E4 2E 05 08 0C 2F 05 08 ............./..
2854: @end example
2855:
2856: @item
2857: For a high-performance system running on a modern RISC processor with a
2858: modified Harvard architecture (one that has a unified main memory but
2859: separate instruction and data caches), it is desirable to separate
2860: processor instructions from processor data. This encourages a high cache
1.32 anton 2861: density and therefore a high cache hit rate. The Forth code space
1.27 crook 2862: is not necessarily made up entirely of processor instructions; its
2863: nature is dependent upon the Forth implementation.
2864:
2865: @item
2866: A Forth compiler that runs on a segmented 8086 processor could be
2867: designed to interleave the name, code and data spaces within a single
2868: 64Kbyte segment. A more common implementation choice is to use a
2869: separate 64Kbyte segment for each region, which provides more memory
2870: overall but provides an address map in which only the data space is
2871: accessible.
2872:
2873: @item
2874: Microprocessors exist that run Forth (or many of the primitives required
2875: to implement the Forth virtual machine efficiently) directly. On these
2876: processors, the relationship between name, code and data spaces may be
1.32 anton 2877: imposed as a side-effect of the architecture of the processor.
1.27 crook 2878:
2879: @item
2880: A Forth compiler that executes from ROM on an embedded system needs its
2881: data space separated from the name and code spaces so that the data
2882: space can be mapped to a RAM area.
2883:
2884: @item
2885: A Forth compiler that runs on an embedded system may have a requirement
2886: for a small memory footprint. On such a system it can be useful to
1.32 anton 2887: separate the header space from the data and code spaces; once the
2888: application has been compiled, the header space is no longer
1.27 crook 2889: required@footnote{more strictly speaking, most applications can be
1.32 anton 2890: designed so that this is the case}. The header space can be deleted
1.29 crook 2891: entirely, or could be stored in memory on a remote @i{host} system for
1.27 crook 2892: debug and development purposes. In the latter case, the compiler running
1.29 crook 2893: on the @i{target} system could implement a protocol across a
1.32 anton 2894: communication link that would allow it to interrogate the header space.
1.27 crook 2895: @end itemize
2896:
1.1 anton 2897:
1.32 anton 2898: @node Dictionary allocation, Heap Allocation, Memory model, Memory
2899: @subsection Dictionary allocation
1.27 crook 2900: @cindex reserving data space
2901: @cindex data space - reserving some
2902:
1.32 anton 2903: Dictionary allocation is a stack-oriented allocation scheme, i.e., if
2904: you want to deallocate X, you also deallocate everything
2905: allocated after X.
2906:
2907: The allocations using the words below are contiguous and grow the region
2908: towards increasing addresses. Other words that allocate dictionary
2909: memory of any kind (i.e., defining words including @code{:noname}) end
2910: the contiguous region and start a new one.
2911:
2912: In ANS Forth only @code{create}d words are guaranteed to produce an
2913: address that is the start of the following contiguous region. In
2914: particular, the cell allocated by @code{variable} is not guaranteed to
2915: be contiguous with following @code{allot}ed memory.
2916:
2917: You can deallocate memory by using @code{allot} with a negative argument
2918: (with some restrictions, see @code{allot}). For larger deallocations use
2919: @code{marker}.
1.27 crook 2920:
1.29 crook 2921:
1.27 crook 2922: doc-here
2923: doc-unused
2924: doc-allot
2925: doc-c,
1.29 crook 2926: doc-f,
1.27 crook 2927: doc-,
2928: doc-2,
1.29 crook 2929: @cindex user space
2930: doc-udp
2931: doc-uallot
1.27 crook 2932:
1.32 anton 2933: Memory accesses have to be aligned (@pxref{Address arithmetic}). So of
2934: course you should allocate memory in an aligned way, too. I.e., before
2935: allocating allocating a cell, @code{here} must be cell-aligned, etc.
2936: The words below align @code{here} if it is not already. Basically it is
2937: only already aligned for a type, if the last allocation was a multiple
2938: of the size of this type and if @code{here} was aligned for this type
2939: before.
2940:
2941: After freshly @code{create}ing a word, @code{here} is @code{align}ed in
2942: ANS Forth (@code{maxalign}ed in Gforth).
2943:
2944: doc-align
2945: doc-falign
2946: doc-sfalign
2947: doc-dfalign
2948: doc-maxalign
2949: doc-cfalign
2950:
2951:
2952: @node Heap Allocation, Memory Access, Dictionary allocation, Memory
2953: @subsection Heap allocation
2954: @cindex heap allocation
2955: @cindex dynamic allocation of memory
2956: @cindex memory-allocation word set
2957:
2958: Heap allocation supports deallocation of allocated memory in any
2959: order. Dictionary allocation is not affected by it (i.e., it does not
2960: end a contiguous region). In Gforth, these words are implemented using
2961: the standard C library calls malloc(), free() and resize().
2962:
2963: doc-allocate
2964: doc-free
2965: doc-resize
2966:
1.27 crook 2967:
1.32 anton 2968: @node Memory Access, Address arithmetic, Heap Allocation, Memory
1.1 anton 2969: @subsection Memory Access
2970: @cindex memory access words
2971:
2972: doc-@
2973: doc-!
2974: doc-+!
2975: doc-c@
2976: doc-c!
2977: doc-2@
2978: doc-2!
2979: doc-f@
2980: doc-f!
2981: doc-sf@
2982: doc-sf!
2983: doc-df@
2984: doc-df!
2985:
1.32 anton 2986: @node Address arithmetic, Memory Blocks, Memory Access, Memory
2987: @subsection Address arithmetic
1.1 anton 2988: @cindex address arithmetic words
2989:
1.32 anton 2990: Address arithmetic is the foundation on which data structures like
2991: arrays, records (@pxref{Structures}) and objects (@pxref{Object-oriented
2992: Forth}) are built.
2993:
1.1 anton 2994: ANS Forth does not specify the sizes of the data types. Instead, it
2995: offers a number of words for computing sizes and doing address
1.29 crook 2996: arithmetic. Address arithmetic is performed in terms of address units
2997: (aus); on most systems the address unit is one byte. Note that a
2998: character may have more than one au, so @code{chars} is no noop (on
2999: systems where it is a noop, it compiles to nothing).
1.1 anton 3000:
3001: @cindex alignment of addresses for types
3002: ANS Forth also defines words for aligning addresses for specific
3003: types. Many computers require that accesses to specific data types
3004: must only occur at specific addresses; e.g., that cells may only be
3005: accessed at addresses divisible by 4. Even if a machine allows unaligned
3006: accesses, it can usually perform aligned accesses faster.
3007:
3008: For the performance-conscious: alignment operations are usually only
3009: necessary during the definition of a data structure, not during the
3010: (more frequent) accesses to it.
3011:
3012: ANS Forth defines no words for character-aligning addresses. This is not
3013: an oversight, but reflects the fact that addresses that are not
3014: char-aligned have no use in the standard and therefore will not be
3015: created.
3016:
3017: @cindex @code{CREATE} and alignment
1.29 crook 3018: ANS Forth guarantees that addresses returned by @code{CREATE}d words
1.1 anton 3019: are cell-aligned; in addition, Gforth guarantees that these addresses
3020: are aligned for all purposes.
3021:
1.26 crook 3022: Note that the ANS Forth word @code{char} has nothing to do with address
3023: arithmetic.
1.1 anton 3024:
3025: doc-chars
3026: doc-char+
3027: doc-cells
3028: doc-cell+
3029: doc-cell
3030: doc-aligned
3031: doc-floats
3032: doc-float+
3033: doc-float
3034: doc-faligned
3035: doc-sfloats
3036: doc-sfloat+
3037: doc-sfaligned
3038: doc-dfloats
3039: doc-dfloat+
3040: doc-dfaligned
3041: doc-maxaligned
3042: doc-cfaligned
3043: doc-address-unit-bits
3044:
1.32 anton 3045: @node Memory Blocks, , Address arithmetic, Memory
1.1 anton 3046: @subsection Memory Blocks
3047: @cindex memory block words
1.27 crook 3048: @cindex character strings - moving and copying
3049:
3050: Memory blocks often represent character strings; @xref{String Formats}
3051: for ways of storing character strings in memory. @xref{Displaying
3052: characters and strings} for other string-processing words.
1.1 anton 3053:
1.32 anton 3054: Some of these words work on address units. Others work on character
3055: units (increments of @code{CHAR}), and expect a @code{CHAR}-aligned
3056: address. Choose the correct operation depending upon your data type.
1.21 crook 3057:
3058: When copying characters between overlapping memory regions, choose
3059: carefully between @code{cmove} and @code{cmove>}.
3060:
1.29 crook 3061: You can only use any of these words @i{portably} to access data space.
1.21 crook 3062:
1.27 crook 3063: @comment TODO - think the naming of the arguments is wrong for move
1.29 crook 3064: @comment well, really it seems to be the Standard that's wrong; it
3065: @comment describes MOVE as a word that requires a CELL-aligned source
3066: @comment and destination address but a xtranfer count that need not
3067: @comment be a multiple of CELL.
1.1 anton 3068: doc-move
3069: doc-erase
3070: doc-cmove
3071: doc-cmove>
3072: doc-fill
3073: doc-blank
1.21 crook 3074: doc-compare
3075: doc-search
1.27 crook 3076: doc--trailing
3077: doc-/string
3078:
3079: @comment TODO examples
3080:
1.1 anton 3081:
1.26 crook 3082: @node Control Structures, Defining Words, Memory, Words
1.1 anton 3083: @section Control Structures
3084: @cindex control structures
3085:
1.33 anton 3086: Control structures in Forth cannot be used interpretively, only in a
3087: colon definition@footnote{To be precise, they have no interpretation
3088: semantics (@pxref{Interpretation and Compilation Semantics}).}. We do
3089: not like this limitation, but have not seen a satisfying way around it
3090: yet, although many schemes have been proposed.
1.1 anton 3091:
3092: @menu
1.33 anton 3093: * Selection:: IF ... ELSE ... ENDIF
3094: * Simple Loops:: BEGIN ...
1.29 crook 3095: * Counted Loops:: DO
3096: * Arbitrary control structures::
3097: * Calls and returns::
1.1 anton 3098: * Exception Handling::
3099: @end menu
3100:
3101: @node Selection, Simple Loops, Control Structures, Control Structures
3102: @subsection Selection
3103: @cindex selection control structures
3104: @cindex control structures for selection
3105:
1.33 anton 3106: @c what's the purpose of all these @i? Maybe we should define a macro
3107: @c so we can produce logical markup. - anton
3108:
1.1 anton 3109: @cindex @code{IF} control structure
3110: @example
1.29 crook 3111: @i{flag}
1.1 anton 3112: IF
1.29 crook 3113: @i{code}
1.1 anton 3114: ENDIF
3115: @end example
1.21 crook 3116: @noindent
1.33 anton 3117:
3118: @var{code} is executed if @var{flag} is non-zero (that's truth as far as
3119: @code{IF} etc. are concerned).
3120:
1.1 anton 3121: @example
1.29 crook 3122: @i{flag}
1.1 anton 3123: IF
1.29 crook 3124: @i{code1}
1.1 anton 3125: ELSE
1.29 crook 3126: @i{code2}
1.1 anton 3127: ENDIF
3128: @end example
3129:
1.33 anton 3130: If @var{flag} is true, perform @var{code1}, otherwise @var{code2}.
3131:
1.1 anton 3132: You can use @code{THEN} instead of @code{ENDIF}. Indeed, @code{THEN} is
3133: standard, and @code{ENDIF} is not, although it is quite popular. We
3134: recommend using @code{ENDIF}, because it is less confusing for people
3135: who also know other languages (and is not prone to reinforcing negative
3136: prejudices against Forth in these people). Adding @code{ENDIF} to a
3137: system that only supplies @code{THEN} is simple:
3138: @example
1.21 crook 3139: : ENDIF POSTPONE THEN ; immediate
1.1 anton 3140: @end example
3141:
3142: [According to @cite{Webster's New Encyclopedic Dictionary}, @dfn{then
3143: (adv.)} has the following meanings:
3144: @quotation
3145: ... 2b: following next after in order ... 3d: as a necessary consequence
3146: (if you were there, then you saw them).
3147: @end quotation
3148: Forth's @code{THEN} has the meaning 2b, whereas @code{THEN} in Pascal
3149: and many other programming languages has the meaning 3d.]
3150:
1.21 crook 3151: Gforth also provides the words @code{?DUP-IF} and @code{?DUP-0=-IF}, so
1.1 anton 3152: you can avoid using @code{?dup}. Using these alternatives is also more
1.26 crook 3153: efficient than using @code{?dup}. Definitions in ANS Forth
1.1 anton 3154: for @code{ENDIF}, @code{?DUP-IF} and @code{?DUP-0=-IF} are provided in
3155: @file{compat/control.fs}.
3156:
3157: @cindex @code{CASE} control structure
3158: @example
1.29 crook 3159: @i{n}
1.1 anton 3160: CASE
1.29 crook 3161: @i{n1} OF @i{code1} ENDOF
3162: @i{n2} OF @i{code2} ENDOF
1.1 anton 3163: @dots{}
3164: ENDCASE
3165: @end example
3166:
1.29 crook 3167: Executes the first @i{codei}, where the @i{ni} is equal to
3168: @i{n}. A default case can be added by simply writing the code after
3169: the last @code{ENDOF}. It may use @i{n}, which is on top of the stack,
1.1 anton 3170: but must not consume it.
3171:
3172: @node Simple Loops, Counted Loops, Selection, Control Structures
3173: @subsection Simple Loops
3174: @cindex simple loops
3175: @cindex loops without count
3176:
3177: @cindex @code{WHILE} loop
3178: @example
3179: BEGIN
1.29 crook 3180: @i{code1}
3181: @i{flag}
1.1 anton 3182: WHILE
1.29 crook 3183: @i{code2}
1.1 anton 3184: REPEAT
3185: @end example
3186:
1.29 crook 3187: @i{code1} is executed and @i{flag} is computed. If it is true,
3188: @i{code2} is executed and the loop is restarted; If @i{flag} is
1.1 anton 3189: false, execution continues after the @code{REPEAT}.
3190:
3191: @cindex @code{UNTIL} loop
3192: @example
3193: BEGIN
1.29 crook 3194: @i{code}
3195: @i{flag}
1.1 anton 3196: UNTIL
3197: @end example
3198:
1.29 crook 3199: @i{code} is executed. The loop is restarted if @code{flag} is false.
1.1 anton 3200:
3201: @cindex endless loop
3202: @cindex loops, endless
3203: @example
3204: BEGIN
1.29 crook 3205: @i{code}
1.1 anton 3206: AGAIN
3207: @end example
3208:
3209: This is an endless loop.
3210:
3211: @node Counted Loops, Arbitrary control structures, Simple Loops, Control Structures
3212: @subsection Counted Loops
3213: @cindex counted loops
3214: @cindex loops, counted
3215: @cindex @code{DO} loops
3216:
3217: The basic counted loop is:
3218: @example
1.29 crook 3219: @i{limit} @i{start}
1.1 anton 3220: ?DO
1.29 crook 3221: @i{body}
1.1 anton 3222: LOOP
3223: @end example
3224:
1.29 crook 3225: This performs one iteration for every integer, starting from @i{start}
3226: and up to, but excluding @i{limit}. The counter, or @i{index}, can be
1.21 crook 3227: accessed with @code{i}. For example, the loop:
1.1 anton 3228: @example
3229: 10 0 ?DO
3230: i .
3231: LOOP
3232: @end example
1.21 crook 3233: @noindent
3234: prints @code{0 1 2 3 4 5 6 7 8 9}
3235:
1.1 anton 3236: The index of the innermost loop can be accessed with @code{i}, the index
3237: of the next loop with @code{j}, and the index of the third loop with
3238: @code{k}.
3239:
3240: doc-i
3241: doc-j
3242: doc-k
3243:
3244: The loop control data are kept on the return stack, so there are some
1.21 crook 3245: restrictions on mixing return stack accesses and counted loop words. In
3246: particuler, if you put values on the return stack outside the loop, you
3247: cannot read them inside the loop@footnote{well, not in a way that is
3248: portable.}. If you put values on the return stack within a loop, you
3249: have to remove them before the end of the loop and before accessing the
3250: index of the loop.
1.1 anton 3251:
3252: There are several variations on the counted loop:
3253:
1.21 crook 3254: @itemize @bullet
3255: @item
3256: @code{LEAVE} leaves the innermost counted loop immediately; execution
3257: continues after the associated @code{LOOP} or @code{NEXT}. For example:
3258:
3259: @example
3260: 10 0 ?DO i DUP . 3 = IF LEAVE THEN LOOP
3261: @end example
3262: prints @code{0 1 2 3}
3263:
1.1 anton 3264:
1.21 crook 3265: @item
3266: @code{UNLOOP} prepares for an abnormal loop exit, e.g., via
3267: @code{EXIT}. @code{UNLOOP} removes the loop control parameters from the
3268: return stack so @code{EXIT} can get to its return address. For example:
3269:
3270: @example
3271: : demo 10 0 ?DO i DUP . 3 = IF UNLOOP EXIT THEN LOOP ." Done" ;
3272: @end example
3273: prints @code{0 1 2 3}
3274:
3275:
3276: @item
1.29 crook 3277: If @i{start} is greater than @i{limit}, a @code{?DO} loop is entered
1.1 anton 3278: (and @code{LOOP} iterates until they become equal by wrap-around
3279: arithmetic). This behaviour is usually not what you want. Therefore,
3280: Gforth offers @code{+DO} and @code{U+DO} (as replacements for
1.29 crook 3281: @code{?DO}), which do not enter the loop if @i{start} is greater than
3282: @i{limit}; @code{+DO} is for signed loop parameters, @code{U+DO} for
1.1 anton 3283: unsigned loop parameters.
3284:
1.21 crook 3285: @item
3286: @code{?DO} can be replaced by @code{DO}. @code{DO} always enters
3287: the loop, independent of the loop parameters. Do not use @code{DO}, even
3288: if you know that the loop is entered in any case. Such knowledge tends
3289: to become invalid during maintenance of a program, and then the
3290: @code{DO} will make trouble.
3291:
3292: @item
1.29 crook 3293: @code{LOOP} can be replaced with @code{@i{n} +LOOP}; this updates the
3294: index by @i{n} instead of by 1. The loop is terminated when the border
3295: between @i{limit-1} and @i{limit} is crossed. E.g.:
1.1 anton 3296:
1.21 crook 3297: @example
3298: 4 0 +DO i . 2 +LOOP
3299: @end example
3300: @noindent
3301: prints @code{0 2}
3302:
3303: @example
3304: 4 1 +DO i . 2 +LOOP
3305: @end example
3306: @noindent
3307: prints @code{1 3}
1.1 anton 3308:
3309:
3310: @cindex negative increment for counted loops
3311: @cindex counted loops with negative increment
1.29 crook 3312: The behaviour of @code{@i{n} +LOOP} is peculiar when @i{n} is negative:
1.1 anton 3313:
1.21 crook 3314: @example
3315: -1 0 ?DO i . -1 +LOOP
3316: @end example
3317: @noindent
3318: prints @code{0 -1}
1.1 anton 3319:
1.21 crook 3320: @example
3321: 0 0 ?DO i . -1 +LOOP
3322: @end example
3323: prints nothing.
1.1 anton 3324:
1.29 crook 3325: Therefore we recommend avoiding @code{@i{n} +LOOP} with negative
3326: @i{n}. One alternative is @code{@i{u} -LOOP}, which reduces the
3327: index by @i{u} each iteration. The loop is terminated when the border
3328: between @i{limit+1} and @i{limit} is crossed. Gforth also provides
1.1 anton 3329: @code{-DO} and @code{U-DO} for down-counting loops. E.g.:
3330:
1.21 crook 3331: @example
3332: -2 0 -DO i . 1 -LOOP
3333: @end example
3334: @noindent
3335: prints @code{0 -1}
1.1 anton 3336:
1.21 crook 3337: @example
3338: -1 0 -DO i . 1 -LOOP
3339: @end example
3340: @noindent
3341: prints @code{0}
3342:
3343: @example
3344: 0 0 -DO i . 1 -LOOP
3345: @end example
3346: @noindent
3347: prints nothing.
1.1 anton 3348:
1.21 crook 3349: @end itemize
1.1 anton 3350:
3351: Unfortunately, @code{+DO}, @code{U+DO}, @code{-DO}, @code{U-DO} and
1.26 crook 3352: @code{-LOOP} are not defined in ANS Forth. However, an implementation
3353: for these words that uses only standard words is provided in
3354: @file{compat/loops.fs}.
1.1 anton 3355:
3356:
3357: @cindex @code{FOR} loops
1.26 crook 3358: Another counted loop is:
1.1 anton 3359: @example
1.29 crook 3360: @i{n}
1.1 anton 3361: FOR
1.29 crook 3362: @i{body}
1.1 anton 3363: NEXT
3364: @end example
3365: This is the preferred loop of native code compiler writers who are too
1.26 crook 3366: lazy to optimize @code{?DO} loops properly. This loop structure is not
1.29 crook 3367: defined in ANS Forth. In Gforth, this loop iterates @i{n+1} times;
3368: @code{i} produces values starting with @i{n} and ending with 0. Other
1.26 crook 3369: Forth systems may behave differently, even if they support @code{FOR}
3370: loops. To avoid problems, don't use @code{FOR} loops.
1.1 anton 3371:
3372: @node Arbitrary control structures, Calls and returns, Counted Loops, Control Structures
3373: @subsection Arbitrary control structures
3374: @cindex control structures, user-defined
3375:
3376: @cindex control-flow stack
3377: ANS Forth permits and supports using control structures in a non-nested
3378: way. Information about incomplete control structures is stored on the
3379: control-flow stack. This stack may be implemented on the Forth data
3380: stack, and this is what we have done in Gforth.
3381:
3382: @cindex @code{orig}, control-flow stack item
3383: @cindex @code{dest}, control-flow stack item
3384: An @i{orig} entry represents an unresolved forward branch, a @i{dest}
3385: entry represents a backward branch target. A few words are the basis for
3386: building any control structure possible (except control structures that
3387: need storage, like calls, coroutines, and backtracking).
3388:
3389: doc-if
3390: doc-ahead
3391: doc-then
3392: doc-begin
3393: doc-until
3394: doc-again
3395: doc-cs-pick
3396: doc-cs-roll
3397:
1.21 crook 3398: The Standard words @code{CS-PICK} and @code{CS-ROLL} allow you to
3399: manipulate the control-flow stack in a portable way. Without them, you
3400: would need to know how many stack items are occupied by a control-flow
3401: entry (many systems use one cell. In Gforth they currently take three,
3402: but this may change in the future).
3403:
1.1 anton 3404: Some standard control structure words are built from these words:
3405:
3406: doc-else
3407: doc-while
3408: doc-repeat
3409:
3410: Gforth adds some more control-structure words:
3411:
3412: doc-endif
3413: doc-?dup-if
3414: doc-?dup-0=-if
3415:
3416: Counted loop words constitute a separate group of words:
3417:
3418: doc-?do
3419: doc-+do
3420: doc-u+do
3421: doc--do
3422: doc-u-do
3423: doc-do
3424: doc-for
3425: doc-loop
3426: doc-+loop
3427: doc--loop
3428: doc-next
3429: doc-leave
3430: doc-?leave
3431: doc-unloop
3432: doc-done
3433:
1.21 crook 3434: The standard does not allow using @code{CS-PICK} and @code{CS-ROLL} on
3435: @i{do-sys}. Gforth allows it, but it's your job to ensure that for
1.1 anton 3436: every @code{?DO} etc. there is exactly one @code{UNLOOP} on any path
3437: through the definition (@code{LOOP} etc. compile an @code{UNLOOP} on the
3438: fall-through path). Also, you have to ensure that all @code{LEAVE}s are
3439: resolved (by using one of the loop-ending words or @code{DONE}).
3440:
1.26 crook 3441: Another group of control structure words are:
1.1 anton 3442:
3443: doc-case
3444: doc-endcase
3445: doc-of
3446: doc-endof
3447:
1.21 crook 3448: @i{case-sys} and @i{of-sys} cannot be processed using @code{CS-PICK} and
3449: @code{CS-ROLL}.
1.1 anton 3450:
3451: @subsubsection Programming Style
3452:
3453: In order to ensure readability we recommend that you do not create
3454: arbitrary control structures directly, but define new control structure
3455: words for the control structure you want and use these words in your
1.26 crook 3456: program. For example, instead of writing:
1.1 anton 3457:
3458: @example
1.26 crook 3459: BEGIN
1.1 anton 3460: ...
1.26 crook 3461: IF [ 1 CS-ROLL ]
1.1 anton 3462: ...
1.26 crook 3463: AGAIN THEN
1.1 anton 3464: @end example
3465:
1.21 crook 3466: @noindent
1.1 anton 3467: we recommend defining control structure words, e.g.,
3468:
3469: @example
1.26 crook 3470: : WHILE ( DEST -- ORIG DEST )
3471: POSTPONE IF
3472: 1 CS-ROLL ; immediate
3473:
3474: : REPEAT ( orig dest -- )
3475: POSTPONE AGAIN
3476: POSTPONE THEN ; immediate
1.1 anton 3477: @end example
3478:
1.21 crook 3479: @noindent
1.1 anton 3480: and then using these to create the control structure:
3481:
3482: @example
1.26 crook 3483: BEGIN
1.1 anton 3484: ...
1.26 crook 3485: WHILE
1.1 anton 3486: ...
1.26 crook 3487: REPEAT
1.1 anton 3488: @end example
3489:
3490: That's much easier to read, isn't it? Of course, @code{REPEAT} and
3491: @code{WHILE} are predefined, so in this example it would not be
3492: necessary to define them.
3493:
3494: @node Calls and returns, Exception Handling, Arbitrary control structures, Control Structures
3495: @subsection Calls and returns
3496: @cindex calling a definition
3497: @cindex returning from a definition
3498:
1.3 anton 3499: @cindex recursive definitions
3500: A definition can be called simply be writing the name of the definition
1.26 crook 3501: to be called. Normally a definition is invisible during its own
1.3 anton 3502: definition. If you want to write a directly recursive definition, you
1.26 crook 3503: can use @code{recursive} to make the current definition visible, or
3504: @code{recurse} to call the current definition directly.
1.3 anton 3505:
3506: doc-recursive
3507: doc-recurse
3508:
1.21 crook 3509: @comment TODO add example of the two recursion methods
1.12 anton 3510: @quotation
3511: @progstyle
3512: I prefer using @code{recursive} to @code{recurse}, because calling the
3513: definition by name is more descriptive (if the name is well-chosen) than
3514: the somewhat cryptic @code{recurse}. E.g., in a quicksort
3515: implementation, it is much better to read (and think) ``now sort the
3516: partitions'' than to read ``now do a recursive call''.
3517: @end quotation
1.3 anton 3518:
1.29 crook 3519: For mutual recursion, use @code{Defer}red words, like this:
1.3 anton 3520:
3521: @example
1.28 crook 3522: Defer foo
1.3 anton 3523:
3524: : bar ( ... -- ... )
3525: ... foo ... ;
3526:
3527: :noname ( ... -- ... )
3528: ... bar ... ;
3529: IS foo
3530: @end example
3531:
1.33 anton 3532: Deferred words are discussed in more detail in @ref{Simple
3533: Defining Words}.
3534:
1.26 crook 3535: The current definition returns control to the calling definition when
1.33 anton 3536: the end of the definition is reached or @code{EXIT} is encountered.
1.1 anton 3537:
3538: doc-exit
3539: doc-;s
3540:
3541: @node Exception Handling, , Calls and returns, Control Structures
3542: @subsection Exception Handling
1.26 crook 3543: @cindex exceptions
1.1 anton 3544:
1.26 crook 3545: If your program detects a fatal error condition, the simplest action
3546: that it can take is to @code{quit}. This resets the return stack and
3547: restarts the text interpreter, but does not print any error message.
1.21 crook 3548:
1.26 crook 3549: The next stage in severity is to execute @code{abort}, which has the
3550: same effect as @code{quit}, with the addition that it resets the data
3551: stack.
1.1 anton 3552:
1.26 crook 3553: A slightly more sophisticated approach is use use @code{abort"}, which
3554: compiles a string to be used as an error message and does a conditional
3555: @code{abort} at run-time. For example:
1.1 anton 3556:
1.26 crook 3557: @example
1.30 anton 3558: @kbd{: checker abort" That flag was true" ." A false flag" ;@key{RET}} ok
3559: @kbd{0 checker@key{RET}} A false flag ok
3560: @kbd{1 checker@key{RET}}
1.26 crook 3561: :1: That flag was true
3562: 1 checker
3563: ^^^^^^^
3564: $400D1648 throw
3565: $400E4660
3566: @end example
1.1 anton 3567:
1.26 crook 3568: These simple techniques allow a program to react to a fatal error
3569: condition, but they are not exactly user-friendly. The ANS Forth
3570: Exception word set provides the pair of words @code{throw} and
3571: @code{catch}, which can be used to provide sophisticated error-handling.
1.1 anton 3572:
1.26 crook 3573: @code{catch} has a similar behaviour to @code{execute}, in that it takes
1.29 crook 3574: an @i{xt} as a parameter and starts execution of the xt. However,
1.26 crook 3575: before passing control to the xt, @code{catch} pushes an
1.29 crook 3576: @dfn{exception frame} onto the @dfn{exception stack}. This exception
1.26 crook 3577: frame is used to restore the system to a known state if a detected error
3578: occurs during the execution of the xt. A typical way to use @code{catch}
3579: would be:
1.1 anton 3580:
1.26 crook 3581: @example
3582: ... ['] foo catch IF ...
3583: @end example
1.1 anton 3584:
1.33 anton 3585: @c TOS is undefined. - anton
1.26 crook 3586: Whilst @code{foo} executes, it can call other words to any level of
3587: nesting, as usual. If @code{foo} (and all the words that it calls)
1.33 anton 3588: execute successfully, control will ultimately pass to the word following
3589: the @code{catch}, and there will be a 0 at TOS. However, if any word
3590: detects an error, it can terminate the execution of @code{foo} by
3591: pushing a non-zero error code onto the stack and then performing a
3592: @code{throw}. The execution of @code{throw} will pass control to the
3593: word following the @code{catch}, but this time the TOS will hold the
3594: error code. Therefore, the @code{IF} in the example can be used to
3595: determine whether @code{foo} executed successfully.
1.1 anton 3596:
1.26 crook 3597: This simple example shows how you can use @code{throw} and @code{catch}
3598: to ``take over'' exception handling from the system:
1.1 anton 3599: @example
1.26 crook 3600: : my-div ['] / catch if ." DIVIDE ERROR" else ." OK.. " . then ;
1.1 anton 3601: @end example
3602:
1.26 crook 3603: The next example is more sophisticated and shows a multi-level
3604: @code{throw} and @code{catch}. To understand this example, start at the
3605: definition of @code{top-level} and work backwards:
3606:
1.1 anton 3607: @example
1.26 crook 3608: : lowest-level ( -- c )
3609: key dup 27 = if
3610: 1 throw \ ESCAPE key pressed
3611: else
3612: ." lowest-level successfull" CR
3613: then
3614: ;
3615:
3616: : lower-level ( -- c )
3617: lowest-level
3618: \ at this level consider a CTRL-U to be a fatal error
3619: dup 21 = if \ CTRL-U
3620: 2 throw
3621: else
3622: ." lower-level successfull" CR
3623: then
3624: ;
3625:
3626: : low-level ( -- c )
3627: ['] lower-level catch
3628: ?dup if
3629: \ error occurred - do we recognise it?
3630: dup 1 = if
3631: \ ESCAPE key pressed.. pretend it was an E
3632: [char] E
3633: else throw \ propogate the error upwards
3634: then
3635: then
3636: ." low-level successfull" CR
3637: ;
3638:
3639: : top-level ( -- )
3640: CR ['] low-level catch \ CATCH is used like EXECUTE
3641: ?dup if \ error occurred..
3642: ." Error " . ." occurred - contact your supplier"
3643: else
3644: ." The '" emit ." ' key was pressed" CR
3645: then
3646: ;
1.1 anton 3647: @end example
3648:
1.26 crook 3649: The ANS Forth document assigns @code{throw} codes thus:
1.1 anton 3650:
1.26 crook 3651: @itemize @bullet
3652: @item
3653: codes in the range -1 -- -255 are reserved to be assigned by the
3654: Standard. Assignments for codes in the range -1 -- -58 are currently
3655: documented in the Standard. In particular, @code{-1 throw} is equivalent
3656: to @code{abort} and @code{-2 throw} is equivalent to @code{abort"}.
3657: @item
3658: codes in the range -256 -- -4095 are reserved to be assigned by the system.
3659: @item
3660: all other codes may be assigned by programs.
3661: @end itemize
1.1 anton 3662:
1.26 crook 3663: Gforth provides the word @code{exception} as a mechanism for assigning
3664: system throw codes to applications. This allows multiple applications to
3665: co-exist in memory without any clash of @code{throw} codes. A definition
3666: of @code{exception} in ANS Forth is provided in
3667: @file{compat/exception.fs}.
1.1 anton 3668:
1.26 crook 3669: doc-quit
3670: doc-abort
3671: doc-abort"
1.1 anton 3672:
1.26 crook 3673: doc-catch
1.29 crook 3674: doc-throw
3675: doc---exception-exception
3676:
3677:
3678: @c -------------------------------------------------------------
3679: @node Defining Words, The Text Interpreter, Control Structures, Words
3680: @section Defining Words
3681: @cindex defining words
3682:
3683: @menu
3684: * Simple Defining Words:: Variables, values and constants
3685: * Colon Definitions::
3686: * User-defined Defining Words::
3687: * Supplying names::
3688: * Interpretation and Compilation Semantics::
3689: @end menu
3690:
3691: @node Simple Defining Words, Colon Definitions, Defining Words, Defining Words
3692: @subsection Simple Defining Words
3693: @cindex simple defining words
3694: @cindex defining words, simple
3695:
1.33 anton 3696: @c split this section?
3697:
1.29 crook 3698: Defining words are used to create new entries in the dictionary. The
3699: simplest defining word is @code{CREATE}. @code{CREATE} is used like
3700: this:
3701:
3702: @example
3703: CREATE new-word1
3704: @end example
3705:
3706: @code{CREATE} is a parsing word that generates a dictionary entry for
3707: @code{new-word1}. When @code{new-word1} is executed, all that it does is
3708: leave an address on the stack. The address represents the value of
3709: the data space pointer (@code{HERE}) at the time that @code{new-word1}
3710: was defined. Therefore, @code{CREATE} is a way of associating a name
3711: with the address of a region of memory.
3712:
1.34 anton 3713: doc-create
3714:
1.29 crook 3715: By extending this example to reserve some memory in data space, we end
3716: up with a @i{variable}. Here are two different ways to do it:
3717:
3718: @example
3719: CREATE new-word2 1 cells allot \ reserve 1 cell - initial value undefined
3720: CREATE new-word3 4 , \ reserve 1 cell and initialise it (to 4)
3721: @end example
3722:
3723: The variable can be examined and modified using @code{@@} (``fetch'') and
3724: @code{!} (``store'') like this:
3725:
3726: @example
3727: new-word2 @@ . \ get address, fetch from it and display
3728: 1234 new-word2 ! \ new value, get address, store to it
3729: @end example
3730:
3731: As a final refinement, the whole code sequence can be wrapped up in a
3732: defining word (pre-empting the subject of the next section), making it
3733: easier to create new variables:
3734:
3735: @example
1.33 anton 3736: : myvariable ( "name" -- a-addr ) CREATE 0 , ;
1.29 crook 3737:
3738: myvariable foo
3739: myvariable joe
3740:
3741: 45 3 * foo ! \ set foo to 135
3742: 1234 joe ! \ set joe to 1234
3743: 3 joe +! \ increment joe by 3.. to 1237
3744: @end example
3745:
3746: Not surprisingly, there is no need to define @code{myvariable}, since
3747: Forth already has a definition @code{Variable}. It behaves in exactly
1.33 anton 3748: the same way as @code{myvariable}. Forth also provides @code{2Variable}
3749: and @code{fvariable} for double and floating-point variables,
3750: respectively.
1.29 crook 3751:
1.34 anton 3752: doc-variable
3753: doc-2variable
3754: doc-fvariable
3755:
1.29 crook 3756: @cindex arrays
3757: A similar mechanism can be used to create arrays. For example, an
3758: 80-character text input buffer:
3759:
3760: @example
3761: CREATE text-buf 80 chars allot
3762:
3763: text-buf 0 chars c@@ \ the 1st character (offset 0)
3764: text-buf 3 chars c@@ \ the 4th character (offset 3)
3765: @end example
3766:
3767: You can build arbitrarily complex data structures by allocating
3768: appropriate areas of memory. @xref{Structures} for further discussions
3769: of this, and to learn about some Gforth tools that make it easier.
3770:
3771: @cindex user variables
3772: @cindex user space
3773: The defining word @code{User} behaves in the same way as @code{Variable}.
3774: The difference is that it reserves space in @i{user (data) space} rather
3775: than normal data space. In a Forth system that has a multi-tasker, each
3776: task has its own set of user variables.
3777:
1.34 anton 3778: doc-user
3779:
1.29 crook 3780: @comment TODO is that stuff about user variables strictly correct? Is it
3781: @comment just terminal tasks that have user variables?
3782: @comment should document tasker.fs (with some examples) elsewhere
3783: @comment in this manual, then expand on user space and user variables.
3784:
3785: After @code{CREATE} and @code{Variable}s, the next defining word to
3786: consider is @code{Constant}. @code{Constant} allows you to declare a
3787: fixed value and refer to it by name. For example:
3788:
3789: @example
3790: 12 Constant INCHES-PER-FOOT
3791: 3E+08 fconstant SPEED-O-LIGHT
3792: @end example
3793:
3794: A @code{Variable} can be both read and written, so its run-time
3795: behaviour is to supply an address through which its current value can be
3796: manipulated. In contrast, the value of a @code{Constant} cannot be
3797: changed once it has been declared@footnote{Well, often it can be -- but
3798: not in a Standard, portable way. It's safer to use a @code{Value} (read
3799: on).} so it's not necessary to supply the address -- it is more
3800: efficient to return the value of the constant directly. That's exactly
3801: what happens; the run-time effect of a constant is to put its value on
3802: the top of the stack (@ref{User-defined Defining Words} describes one
3803: way of implementing @code{Constant}).
3804:
3805: Gforth also provides @code{2Constant} and @code{fconstant} for defining
3806: double and floating-point constants, respectively.
3807:
1.34 anton 3808: doc-constant
3809: doc-2constant
3810: doc-fconstant
3811:
3812: @c that's too deep, and it's not necessarily true for all ANS Forths. - anton
1.29 crook 3813: Constants in Forth behave differently from their equivalents in other
3814: programming languages. In other languages, a constant (such as an EQU in
3815: assembler or a #define in C) only exists at compile-time; in the
3816: executable program the constant has been translated into an absolute
3817: number and, unless you are using a symbolic debugger, it's impossible to
3818: know what abstract thing that number represents. In Forth a constant has
1.32 anton 3819: an entry in the header space and remains there after the code that
1.29 crook 3820: uses it has been defined. In fact, it must remain in the dictionary
3821: since it has run-time duties to perform. For example:
3822:
3823: @example
3824: 12 Constant INCHES-PER-FOOT
3825: : FEET-TO-INCHES ( n1 -- n2 ) INCHES-PER-FOOT * ;
3826: @end example
3827:
3828: @cindex in-lining of constants
3829: When @code{FEET-TO-INCHES} is executed, it will in turn execute the xt
3830: associated with the constant @code{INCHES-PER-FOOT}. If you use
3831: @code{see} to decompile the definition of @code{FEET-TO-INCHES}, you can
3832: see that it makes a call to @code{INCHES-PER-FOOT}. Some Forth compilers
3833: attempt to optimise constants by in-lining them where they are used. You
3834: can force Gforth to in-line a constant like this:
3835:
3836: @example
3837: : FEET-TO-INCHES ( n1 -- n2 ) [ INCHES-PER-FOOT ] LITERAL * ;
3838: @end example
3839:
3840: If you use @code{see} to decompile @i{this} version of
3841: @code{FEET-TO-INCHES}, you can see that @code{INCHES-PER-FOOT} is no
1.33 anton 3842: longer present. @xref{Interpret/Compile states} and @ref{Literals} on
3843: how this works.
1.29 crook 3844:
3845: In-lining constants in this way might improve execution time
3846: fractionally, and can ensure that a constant is now only referenced at
3847: compile-time. However, the definition of the constant still remains in
3848: the dictionary. Some Forth compilers provide a mechanism for controlling
3849: a second dictionary for holding transient words such that this second
3850: dictionary can be deleted later in order to recover memory
3851: space. However, there is no standard way of doing this.
3852:
3853: One aspect of constants and variables that can sometimes be confusing is
3854: that they have different stack effects; one returns its value whilst the
3855: other returns the address of its value. The defining word @code{Value}
3856: provides an alternative to @code{Variable}, and has the same stack
3857: effect as a constant. A @code{Value} needs an additional word, @code{TO}
3858: to allow its value to be changed. Here are some examples:
3859:
3860: @example
3861: 12 Value APPLES \ a Value is initialised when it is declared.. like a
3862: \ constant but unlike a variable
3863: 34 TO APPLES \ Change the value of APPLES. TO is a parsing word
3864: APPLES \ puts 34 on the top of the stack.
3865: @end example
3866:
1.34 anton 3867: doc-value
3868: doc-to
3869:
1.29 crook 3870: The defining word @code{Defer} allows you to define a word by name
3871: without defining its behaviour; the definition of its behaviour is
3872: deferred. Here are two situation where this can be useful:
3873:
3874: @itemize @bullet
3875: @item
3876: Where you want to allow the behaviour of a word to be altered later, and
3877: for all precompiled references to the word to change when its behaviour
3878: is changed.
3879: @item
3880: For mutual recursion; @xref{Calls and returns}.
3881: @end itemize
3882:
3883: In the following example, @code{foo} always invokes the version of
3884: @code{greet} that prints ``@code{Good morning}'' whilst @code{bar}
3885: always invokes the version that prints ``@code{Hello}''. There is no way
3886: of getting @code{foo} to use the later version without re-ordering the
3887: source code and recompilng it.
3888:
3889: @example
3890: : greet ." Good morning" ;
3891: : foo ... greet ... ;
3892: : greet ." Hello" ;
3893: : bar ... greet ... ;
3894: @end example
3895:
3896: This problem can be solved by defining @code{greet} as a @code{Defer}red
3897: word. The behaviour of a @code{Defer}red word can be defined and
3898: redefined at any time by using @code{IS} to associate the xt of a
3899: previously-defined word with it. The previous example becomes:
3900:
3901: @example
3902: Defer greet
3903: : foo ... greet ... ;
3904: : bar ... greet ... ;
3905: : greet1 ." Good morning" ;
3906: : greet2 ." Hello" ;
1.35 anton 3907: ' greet2 <IS> greet \ make greet behave like greet2
3908: @end example
3909:
3910: One thing to note is that @code{<IS>} consumes it's name when it is
3911: executed. If you want to specify the name at compile time, use
3912: @code{[IS]}:
3913:
3914: @example
3915: : set-greet ( xt -- )
3916: [IS] greet ;
3917:
3918: ' greet1 set-greet
1.29 crook 3919: @end example
3920:
3921: A deferred word can only inherit default semantics from the xt (because
3922: that is all that an xt can represent -- @pxref{Tokens for Words} for
3923: more discussion of this). However, the semantics of the deferred word
3924: itself can be modified at the time that it is defined. For example:
3925:
3926: @example
3927: : bar .... ; compile-only
3928: Defer fred immediate
3929: Defer jim
3930:
1.35 anton 3931: ' bar <IS> jim \ jim has default semantics
3932: ' bar <IS> fred \ fred is immediate
1.29 crook 3933: @end example
1.1 anton 3934:
1.34 anton 3935: doc-defer
1.35 anton 3936: doc-<is>
3937: doc-[is]
3938: @comment TODO document these: what's defers [is]
1.34 anton 3939: doc-what's
3940: doc-defers
3941:
3942: Definitions in ANS Forth for @code{defer}, @code{<is>} and
3943: @code{[is]} are provided in @file{compat/defer.fs}.
3944:
1.29 crook 3945: The defining word @code{Alias} allows you to define a word by name that
3946: has the same behaviour as some other word. Here are two situation where
3947: this can be useful:
1.1 anton 3948:
1.29 crook 3949: @itemize @bullet
3950: @item
3951: When you want access to a word's definition from a different word list
3952: (for an example of this, see the definition of the @code{Root} word list
3953: in the Gforth source).
3954: @item
3955: When you want to create a synonym; a definition that can be known by
3956: either of two names (for example, @code{THEN} and @code{ENDIF} are
3957: aliases).
3958: @end itemize
1.1 anton 3959:
1.29 crook 3960: The word whose behaviour the alias is to inherit is represented by an
1.34 anton 3961: xt. Therefore, the alias only inherits default semantics from its
1.29 crook 3962: ancestor. The semantics of the alias itself can be modified at the time
3963: that it is defined. For example:
1.1 anton 3964:
1.29 crook 3965: @example
3966: : foo ... ; immediate
1.1 anton 3967:
1.29 crook 3968: ' foo Alias bar \ bar is not an immediate word
3969: ' foo Alias fooby immediate \ fooby is an immediate word
3970: @end example
1.26 crook 3971:
1.34 anton 3972: @c "combined words" is an undefined term
3973: Words that are aliases have the same xt, different headers in the
3974: dictionary, and consequently different name tokens (@pxref{Tokens for
3975: Words}) and possibly different immediate flags. An alias can only have
3976: default or immediate compilation semantics; you can define aliases for
3977: combined words with @code{interpret/compile:}.
1.27 crook 3978:
1.33 anton 3979: @c distribute this to the appropriate paragraphs? - anton
1.29 crook 3980: doc-alias
1.1 anton 3981:
1.26 crook 3982: @node Colon Definitions, User-defined Defining Words, Simple Defining Words, Defining Words
3983: @subsection Colon Definitions
3984: @cindex colon definitions
1.1 anton 3985:
1.26 crook 3986: @example
3987: : name ( ... -- ... )
3988: word1 word2 word3 ;
3989: @end example
1.1 anton 3990:
1.29 crook 3991: @noindent
3992: Creates a word called @code{name} that, upon execution, executes
1.26 crook 3993: @code{word1 word2 word3}. @code{name} is a @dfn{(colon) definition}.
1.1 anton 3994:
1.29 crook 3995: The explanation above is somewhat superficial. @xref{Your first
3996: definition} for simple examples of colon definitions, then
3997: @xref{Interpretation and Compilation Semantics} for an in-depth
3998: discussion of some of the issues involved.
1.26 crook 3999:
4000: doc-:
4001: doc-;
1.1 anton 4002:
1.26 crook 4003: @node User-defined Defining Words, Supplying names, Colon Definitions, Defining Words
4004: @subsection User-defined Defining Words
4005: @cindex user-defined defining words
4006: @cindex defining words, user-defined
1.1 anton 4007:
1.29 crook 4008: You can create a new defining word by wrapping defining-time code around
4009: an existing defining word and putting the sequence in a colon
4010: definition. For example, suppose that you have a word @code{stats} that
4011: gathers statistics about colon definitions given the @i{xt} of the
4012: definition, and you want every colon definition in your application to
4013: make a call to @code{stats}. You can define and use a new version of
4014: @code{:} like this:
4015:
4016: @example
4017: : stats ( xt -- ) DUP ." (Gathering statistics for " . ." )"
4018: ... ; \ other code
4019:
4020: : my: : lastxt postpone literal ['] stats compile, ;
4021:
4022: my: foo + - ;
4023: @end example
4024:
4025: When @code{foo} is defined using @code{my:} these steps occur:
4026:
4027: @itemize @bullet
4028: @item
4029: @code{my:} is executed.
4030: @item
4031: The @code{:} within the definition (the one between @code{my:} and
4032: @code{lastxt}) is executed, and does just what it always does; it parses
4033: the input stream for a name, builds a dictionary header for the name
4034: @code{foo} and switches @code{state} from interpret to compile.
4035: @item
4036: The word @code{lastxt} is executed. It puts the @i{xt} for the word that is
4037: being defined -- @code{foo} -- onto the stack.
4038: @item
4039: The code that was produced by @code{postpone literal} is executed; this
4040: causes the value on the stack to be compiled as a literal in the code
4041: area of @code{foo}.
4042: @item
4043: The code @code{['] stats} compiles a literal into the definition of
4044: @code{my:}. When @code{compile,} is executed, that literal -- the
4045: execution token for @code{stats} -- is layed down in the code area of
4046: @code{foo} , following the literal@footnote{Strictly speaking, the
4047: mechanism that @code{compile,} uses to convert an @i{xt} into something
4048: in the code area is implementation-dependent. A threaded implementation
4049: might spit out the execution token directly whilst another
4050: implementation might spit out a native code sequence.}.
4051: @item
4052: At this point, the execution of @code{my:} is complete, and control
4053: returns to the text interpreter. The text interpreter is in compile
4054: state, so subsequent text @code{+ -} is compiled into the definition of
4055: @code{foo} and the @code{;} terminates the definition as always.
4056: @end itemize
4057:
4058: You can use @code{see} to decompile a word that was defined using
4059: @code{my:} and see how it is different from a normal @code{:}
4060: definition. For example:
4061:
4062: @example
4063: : bar + - ; \ like foo but using : rather than my:
4064: see bar
4065: : bar
4066: + - ;
4067: see foo
4068: : foo
4069: 107645672 stats + - ;
4070:
4071: \ use ' stats . to show that 107645672 is the xt for stats
4072: @end example
4073:
4074:
1.33 anton 4075: @c a deferred word is not neccessary for these examples. - anton
1.29 crook 4076: Rather than edit your application's source code to change every @code{:}
4077: to a @code{my:}, use a deferred word:
4078:
4079: @example
4080: : real: : ; \ retain access to the original
4081: defer : \ redefine as a deferred word
4082: ' my: IS : \ use special version of :
4083: \
4084: \ load application here
4085: \
4086: ' real: IS : \ go back to the original
4087: @end example
4088:
4089: You can use techniques like this to make new defining words in terms of
4090: @i{any} existing defining word.
1.1 anton 4091:
4092:
1.29 crook 4093: @cindex defining defining words
1.26 crook 4094: @cindex @code{CREATE} ... @code{DOES>}
4095: If you want the words defined with your defining words to behave
4096: differently from words defined with standard defining words, you can
4097: write your defining word like this:
1.1 anton 4098:
4099: @example
1.26 crook 4100: : def-word ( "name" -- )
1.29 crook 4101: CREATE @i{code1}
1.26 crook 4102: DOES> ( ... -- ... )
1.29 crook 4103: @i{code2} ;
1.26 crook 4104:
4105: def-word name
1.1 anton 4106: @end example
4107:
1.29 crook 4108: @cindex child words
4109: This fragment defines a @dfn{defining word} @code{def-word} and then
4110: executes it. When @code{def-word} executes, it @code{CREATE}s a new
4111: word, @code{name}, and executes the code @i{code1}. The code @i{code2}
4112: is not executed at this time. The word @code{name} is sometimes called a
4113: @dfn{child} of @code{def-word}.
4114:
4115: When you execute @code{name}, the address of the body of @code{name} is
4116: put on the data stack and @i{code2} is executed (the address of the body
4117: of @code{name} is the address @code{HERE} returns immediately after the
4118: @code{CREATE}).
4119:
4120: @cindex atavism in child words
1.33 anton 4121: You can use @code{def-word} to define a set of child words that behave
1.29 crook 4122: differently, though atavistically; they all have a common run-time
4123: behaviour determined by @i{code2}. Typically, the @i{code1} sequence
4124: builds a data area in the body of the child word. The structure of the
4125: data is common to all children of @code{def-word}, but the data values
4126: are specific -- and private -- to each child word. When a child word is
4127: executed, the address of its private data area is passed as a parameter
4128: on TOS to be used and manipulated@footnote{It is legitimate both to read
4129: and write to this data area.} by @i{code2}.
4130:
4131: The two fragments of code that make up the defining words act (are
4132: executed) at two completely separate times:
1.1 anton 4133:
1.29 crook 4134: @itemize @bullet
4135: @item
4136: At @i{define time}, the defining word executes @i{code1} to generate a
4137: child word
4138: @item
4139: At @i{child execution time}, when a child word is invoked, @i{code2}
4140: is executed, using parameters (data) that are private and specific to
4141: the child word.
4142: @end itemize
4143:
4144: @c NAC I think this is a really bad example, because it diminishes
4145: @c rather than emphasising the fact that some important stuff happens
4146: @c at define time, and other important stuff happens at child-invocation
4147: @c time, and that those two times are potentially very different.
1.33 anton 4148:
4149: @c Well, IMO CREATE-DOES> is usually presented with much ado, making
4150: @c people think that it's hard to understand, and making those people who
4151: @c understand it easily think that it's hyped. I prefer presenting it in a
4152: @c diminished way and only emphasize the special issues later. - anton
4153:
4154: In other words, if you make the following definitions:
4155: @example
4156: : def-word1 ( "name" -- )
4157: CREATE @i{code1} ;
4158:
4159: : action1 ( ... -- ... )
4160: @i{code2} ;
4161:
4162: def-word1 name1
4163: @end example
4164:
4165: Using @code{name1 action1} is equivalent to using @code{name}.
1.1 anton 4166:
1.29 crook 4167: The classic example is that you can define @code{CONSTANT} in this way:
1.26 crook 4168:
1.1 anton 4169: @example
1.29 crook 4170: : CONSTANT ( w "name" -- )
4171: CREATE ,
1.26 crook 4172: DOES> ( -- w )
4173: @@ ;
1.1 anton 4174: @end example
4175:
1.29 crook 4176: @comment There is a beautiful description of how this works and what
4177: @comment it does in the Forthwrite 100th edition.. as well as an elegant
4178: @comment commentary on the Counting Fruits problem.
4179:
4180: When you create a constant with @code{5 CONSTANT five}, a set of
4181: define-time actions take place; first a new word @code{five} is created,
4182: then the value 5 is laid down in the body of @code{five} with
4183: @code{,}. When @code{five} is invoked, the address of the body is put on
4184: the stack, and @code{@@} retrieves the value 5. The word @code{five} has
4185: no code of its own; it simply contains a data field and a pointer to the
4186: code that follows @code{DOES>} in its defining word. That makes words
4187: created in this way very compact.
4188:
4189: The final example in this section is intended to remind you that space
4190: reserved in @code{CREATE}d words is @i{data} space and therefore can be
4191: both read and written by a Standard program@footnote{Exercise: use this
4192: example as a starting point for your own implementation of @code{Value}
4193: and @code{TO} -- if you get stuck, investigate the behaviour of @code{'} and
4194: @code{[']}.}:
4195:
4196: @example
4197: : foo ( "name" -- )
4198: CREATE -1 ,
4199: DOES> ( -- )
1.33 anton 4200: @@ . ;
1.29 crook 4201:
4202: foo first-word
4203: foo second-word
4204:
4205: 123 ' first-word >BODY !
4206: @end example
4207:
4208: If @code{first-word} had been a @code{CREATE}d word, we could simply
4209: have executed it to get the address of its data field. However, since it
4210: was defined to have @code{DOES>} actions, its execution semantics are to
4211: perform those @code{DOES>} actions. To get the address of its data field
4212: it's necessary to use @code{'} to get its xt, then @code{>BODY} to
4213: translate the xt into the address of the data field. When you execute
4214: @code{first-word}, it will display @code{123}. When you execute
4215: @code{second-word} it will display @code{-1}.
1.26 crook 4216:
4217: @cindex stack effect of @code{DOES>}-parts
4218: @cindex @code{DOES>}-parts, stack effect
1.29 crook 4219: In the examples above the stack comment after the @code{DOES>} specifies
1.26 crook 4220: the stack effect of the defined words, not the stack effect of the
4221: following code (the following code expects the address of the body on
4222: the top of stack, which is not reflected in the stack comment). This is
4223: the convention that I use and recommend (it clashes a bit with using
4224: locals declarations for stack effect specification, though).
1.1 anton 4225:
1.26 crook 4226: @subsubsection Applications of @code{CREATE..DOES>}
4227: @cindex @code{CREATE} ... @code{DOES>}, applications
1.1 anton 4228:
1.26 crook 4229: You may wonder how to use this feature. Here are some usage patterns:
1.1 anton 4230:
1.26 crook 4231: @cindex factoring similar colon definitions
4232: When you see a sequence of code occurring several times, and you can
4233: identify a meaning, you will factor it out as a colon definition. When
4234: you see similar colon definitions, you can factor them using
4235: @code{CREATE..DOES>}. E.g., an assembler usually defines several words
4236: that look very similar:
1.1 anton 4237: @example
1.26 crook 4238: : ori, ( reg-target reg-source n -- )
4239: 0 asm-reg-reg-imm ;
4240: : andi, ( reg-target reg-source n -- )
4241: 1 asm-reg-reg-imm ;
1.1 anton 4242: @end example
4243:
1.26 crook 4244: @noindent
4245: This could be factored with:
4246: @example
4247: : reg-reg-imm ( op-code -- )
4248: CREATE ,
4249: DOES> ( reg-target reg-source n -- )
4250: @@ asm-reg-reg-imm ;
4251:
4252: 0 reg-reg-imm ori,
4253: 1 reg-reg-imm andi,
4254: @end example
1.1 anton 4255:
1.26 crook 4256: @cindex currying
4257: Another view of @code{CREATE..DOES>} is to consider it as a crude way to
4258: supply a part of the parameters for a word (known as @dfn{currying} in
4259: the functional language community). E.g., @code{+} needs two
4260: parameters. Creating versions of @code{+} with one parameter fixed can
4261: be done like this:
1.1 anton 4262: @example
1.26 crook 4263: : curry+ ( n1 -- )
4264: CREATE ,
4265: DOES> ( n2 -- n1+n2 )
4266: @@ + ;
4267:
4268: 3 curry+ 3+
4269: -2 curry+ 2-
1.1 anton 4270: @end example
4271:
1.26 crook 4272: @subsubsection The gory details of @code{CREATE..DOES>}
4273: @cindex @code{CREATE} ... @code{DOES>}, details
1.1 anton 4274:
1.26 crook 4275: doc-does>
1.1 anton 4276:
1.26 crook 4277: @cindex @code{DOES>} in a separate definition
4278: This means that you need not use @code{CREATE} and @code{DOES>} in the
4279: same definition; you can put the @code{DOES>}-part in a separate
1.29 crook 4280: definition. This allows us to, e.g., select among different @code{DOES>}-parts:
1.26 crook 4281: @example
4282: : does1
4283: DOES> ( ... -- ... )
4284: ... ;
1.1 anton 4285:
1.26 crook 4286: : does2
4287: DOES> ( ... -- ... )
4288: ... ;
1.1 anton 4289:
1.26 crook 4290: : def-word ( ... -- ... )
4291: create ...
4292: IF
4293: does1
4294: ELSE
4295: does2
4296: ENDIF ;
4297: @end example
1.1 anton 4298:
1.26 crook 4299: In this example, the selection of whether to use @code{does1} or
4300: @code{does2} is made at compile-time; at the time that the child word is
1.29 crook 4301: @code{CREATE}d.
1.1 anton 4302:
1.26 crook 4303: @cindex @code{DOES>} in interpretation state
4304: In a standard program you can apply a @code{DOES>}-part only if the last
4305: word was defined with @code{CREATE}. In Gforth, the @code{DOES>}-part
4306: will override the behaviour of the last word defined in any case. In a
4307: standard program, you can use @code{DOES>} only in a colon
4308: definition. In Gforth, you can also use it in interpretation state, in a
4309: kind of one-shot mode; for example:
1.1 anton 4310: @example
1.26 crook 4311: CREATE name ( ... -- ... )
1.29 crook 4312: @i{initialization}
1.26 crook 4313: DOES>
1.29 crook 4314: @i{code} ;
1.1 anton 4315: @end example
4316:
1.26 crook 4317: @noindent
4318: is equivalent to the standard:
1.1 anton 4319: @example
1.26 crook 4320: :noname
4321: DOES>
1.29 crook 4322: @i{code} ;
1.26 crook 4323: CREATE name EXECUTE ( ... -- ... )
1.29 crook 4324: @i{initialization}
1.1 anton 4325: @end example
4326:
1.26 crook 4327: You can get the address of the body of a word with:
4328:
4329: doc->body
1.1 anton 4330:
1.26 crook 4331: @node Supplying names, Interpretation and Compilation Semantics, User-defined Defining Words, Defining Words
1.29 crook 4332: @subsection Supplying the name of a defined word
1.26 crook 4333: @cindex names for defined words
4334: @cindex defining words, name parameter
1.1 anton 4335:
1.26 crook 4336: @cindex defining words, name given in a string
1.29 crook 4337: By default, a defining word takes the name for the defined word from the
1.26 crook 4338: input stream. Sometimes you want to supply the name from a string. You
4339: can do this with:
1.1 anton 4340:
1.26 crook 4341: doc-nextname
1.1 anton 4342:
1.26 crook 4343: For example:
1.1 anton 4344:
1.26 crook 4345: @example
4346: s" foo" nextname create
4347: @end example
4348: @noindent
4349: is equivalent to:
4350: @example
4351: create foo
4352: @end example
1.1 anton 4353:
1.26 crook 4354: @cindex defining words without name
1.29 crook 4355: Sometimes you want to define an @dfn{anonymous word}; a word without a
1.26 crook 4356: name. You can do this with:
1.1 anton 4357:
1.26 crook 4358: doc-:noname
1.1 anton 4359:
1.26 crook 4360: This leaves the execution token for the word on the stack after the
4361: closing @code{;}. Here's an example in which a deferred word is
4362: initialised with an @code{xt} from an anonymous colon definition:
4363: @example
4364: Defer deferred
4365: :noname ( ... -- ... )
4366: ... ;
4367: IS deferred
4368: @end example
1.1 anton 4369:
1.29 crook 4370: @noindent
1.26 crook 4371: Gforth provides an alternative way of doing this, using two separate
4372: words:
1.1 anton 4373:
1.26 crook 4374: doc-noname
4375: @cindex execution token of last defined word
4376: doc-lastxt
1.1 anton 4377:
1.29 crook 4378: @noindent
1.26 crook 4379: The previous example can be rewritten using @code{noname} and
4380: @code{lastxt}:
1.1 anton 4381:
1.26 crook 4382: @example
4383: Defer deferred
4384: noname : ( ... -- ... )
4385: ... ;
4386: lastxt IS deferred
4387: @end example
1.1 anton 4388:
1.29 crook 4389: @noindent
1.33 anton 4390: @code{noname} and @code{nextname} work with any defining word, not just
4391: @code{:}.
4392:
1.26 crook 4393: @code{lastxt} also works when the last word was not defined as
1.29 crook 4394: @code{noname}. It also has the useful property that is is valid as soon
4395: as the header for a definition has been build. Thus:
4396:
4397: @example
4398: lastxt . : foo [ lastxt . ] ; ' foo .
4399: @end example
4400:
4401: @noindent
4402: prints 3 numbers; the last two are the same.
1.1 anton 4403:
4404:
1.26 crook 4405: @node Interpretation and Compilation Semantics, , Supplying names, Defining Words
4406: @subsection Interpretation and Compilation Semantics
4407: @cindex semantics, interpretation and compilation
1.1 anton 4408:
1.26 crook 4409: @cindex interpretation semantics
4410: The @dfn{interpretation semantics} of a word are what the text
4411: interpreter does when it encounters the word in interpret state. It also
4412: appears in some other contexts, e.g., the execution token returned by
1.29 crook 4413: @code{' @i{word}} identifies the interpretation semantics of
4414: @i{word} (in other words, @code{' @i{word} execute} is equivalent to
4415: interpret-state text interpretation of @code{@i{word}}).
1.1 anton 4416:
1.26 crook 4417: @cindex compilation semantics
4418: The @dfn{compilation semantics} of a word are what the text interpreter
4419: does when it encounters the word in compile state. It also appears in
1.29 crook 4420: other contexts, e.g, @code{POSTPONE @i{word}} compiles@footnote{In
1.26 crook 4421: standard terminology, ``appends to the current definition''.} the
1.29 crook 4422: compilation semantics of @i{word}.
1.1 anton 4423:
1.26 crook 4424: @cindex execution semantics
4425: The standard also talks about @dfn{execution semantics}. They are used
4426: only for defining the interpretation and compilation semantics of many
4427: words. By default, the interpretation semantics of a word are to
4428: @code{execute} its execution semantics, and the compilation semantics of
4429: a word are to @code{compile,} its execution semantics.@footnote{In
4430: standard terminology: The default interpretation semantics are its
4431: execution semantics; the default compilation semantics are to append its
4432: execution semantics to the execution semantics of the current
4433: definition.}
4434:
4435: @comment TODO expand, make it co-operate with new sections on text interpreter.
4436:
4437: @cindex immediate words
4438: @cindex compile-only words
4439: You can change the semantics of the most-recently defined word:
4440:
4441: doc-immediate
4442: doc-compile-only
4443: doc-restrict
4444:
4445: Note that ticking (@code{'}) a compile-only word gives an error
4446: (``Interpreting a compile-only word'').
1.1 anton 4447:
1.26 crook 4448: Gforth also allows you to define words with arbitrary combinations of
4449: interpretation and compilation semantics.
1.1 anton 4450:
1.26 crook 4451: doc-interpret/compile:
1.1 anton 4452:
1.26 crook 4453: This feature was introduced for implementing @code{TO} and @code{S"}. I
4454: recommend that you do not define such words, as cute as they may be:
4455: they make it hard to get at both parts of the word in some contexts.
4456: E.g., assume you want to get an execution token for the compilation
4457: part. Instead, define two words, one that embodies the interpretation
4458: part, and one that embodies the compilation part. Once you have done
4459: that, you can define a combined word with @code{interpret/compile:} for
4460: the convenience of your users.
1.1 anton 4461:
1.26 crook 4462: You might try to use this feature to provide an optimizing
4463: implementation of the default compilation semantics of a word. For
4464: example, by defining:
1.1 anton 4465: @example
1.26 crook 4466: :noname
4467: foo bar ;
4468: :noname
4469: POSTPONE foo POSTPONE bar ;
1.29 crook 4470: interpret/compile: opti-foobar
1.1 anton 4471: @end example
1.26 crook 4472:
1.23 crook 4473: @noindent
1.26 crook 4474: as an optimizing version of:
4475:
1.1 anton 4476: @example
1.26 crook 4477: : foobar
4478: foo bar ;
1.1 anton 4479: @end example
4480:
1.26 crook 4481: Unfortunately, this does not work correctly with @code{[compile]},
4482: because @code{[compile]} assumes that the compilation semantics of all
4483: @code{interpret/compile:} words are non-default. I.e., @code{[compile]
1.29 crook 4484: opti-foobar} would compile compilation semantics, whereas
4485: @code{[compile] foobar} would compile interpretation semantics.
1.1 anton 4486:
1.26 crook 4487: @cindex state-smart words (are a bad idea)
1.29 crook 4488: Some people try to use @dfn{state-smart} words to emulate the feature provided
1.26 crook 4489: by @code{interpret/compile:} (words are state-smart if they check
4490: @code{STATE} during execution). E.g., they would try to code
4491: @code{foobar} like this:
1.1 anton 4492:
1.26 crook 4493: @example
4494: : foobar
4495: STATE @@
4496: IF ( compilation state )
4497: POSTPONE foo POSTPONE bar
4498: ELSE
4499: foo bar
4500: ENDIF ; immediate
4501: @end example
1.1 anton 4502:
1.26 crook 4503: Although this works if @code{foobar} is only processed by the text
4504: interpreter, it does not work in other contexts (like @code{'} or
4505: @code{POSTPONE}). E.g., @code{' foobar} will produce an execution token
4506: for a state-smart word, not for the interpretation semantics of the
4507: original @code{foobar}; when you execute this execution token (directly
4508: with @code{EXECUTE} or indirectly through @code{COMPILE,}) in compile
4509: state, the result will not be what you expected (i.e., it will not
4510: perform @code{foo bar}). State-smart words are a bad idea. Simply don't
4511: write them@footnote{For a more detailed discussion of this topic, see
4512: @cite{@code{State}-smartness -- Why it is Evil and How to Exorcise it} by Anton
4513: Ertl; presented at EuroForth '98 and available from
1.33 anton 4514: @url{http://www.complang.tuwien.ac.at/papers/ertl98.ps.gz}}!
1.1 anton 4515:
1.26 crook 4516: @cindex defining words with arbitrary semantics combinations
4517: It is also possible to write defining words that define words with
4518: arbitrary combinations of interpretation and compilation semantics. In
4519: general, they look like this:
1.1 anton 4520:
1.26 crook 4521: @example
4522: : def-word
4523: create-interpret/compile
1.29 crook 4524: @i{code1}
1.26 crook 4525: interpretation>
1.29 crook 4526: @i{code2}
1.26 crook 4527: <interpretation
4528: compilation>
1.29 crook 4529: @i{code3}
1.26 crook 4530: <compilation ;
4531: @end example
1.1 anton 4532:
1.29 crook 4533: For a @i{word} defined with @code{def-word}, the interpretation
4534: semantics are to push the address of the body of @i{word} and perform
4535: @i{code2}, and the compilation semantics are to push the address of
4536: the body of @i{word} and perform @i{code3}. E.g., @code{constant}
1.26 crook 4537: can also be defined like this (except that the defined constants don't
4538: behave correctly when @code{[compile]}d):
1.1 anton 4539:
1.26 crook 4540: @example
4541: : constant ( n "name" -- )
4542: create-interpret/compile
4543: ,
4544: interpretation> ( -- n )
4545: @@
4546: <interpretation
4547: compilation> ( compilation. -- ; run-time. -- n )
4548: @@ postpone literal
4549: <compilation ;
4550: @end example
1.1 anton 4551:
1.26 crook 4552: doc-create-interpret/compile
4553: doc-interpretation>
4554: doc-<interpretation
4555: doc-compilation>
4556: doc-<compilation
1.1 anton 4557:
1.29 crook 4558: Words defined with @code{interpret/compile:} and
1.26 crook 4559: @code{create-interpret/compile} have an extended header structure that
4560: differs from other words; however, unless you try to access them with
4561: plain address arithmetic, you should not notice this. Words for
4562: accessing the header structure usually know how to deal with this; e.g.,
1.29 crook 4563: @code{'} @i{word} @code{>body} also gives you the body of a word created
4564: with @code{create-interpret/compile}.
1.1 anton 4565:
1.27 crook 4566: doc-postpone
1.29 crook 4567: @comment TODO -- expand glossary text for POSTPONE
1.27 crook 4568:
1.26 crook 4569: @c ----------------------------------------------------------
4570: @node The Text Interpreter, Tokens for Words, Defining Words, Words
4571: @section The Text Interpreter
4572: @cindex interpreter - outer
4573: @cindex text interpreter
4574: @cindex outer interpreter
1.1 anton 4575:
1.34 anton 4576: @c Should we really describe all these ugly details? IMO the text
4577: @c interpreter should be much cleaner, but that may not be possible within
4578: @c ANS Forth. - anton
4579:
1.29 crook 4580: The text interpreter@footnote{This is an expanded version of the
4581: material in @ref{Introducing the Text Interpreter}.} is an endless loop
1.34 anton 4582: that processes input from the current input device. It is also called
4583: the outer interpreter, in contrast to the inner interpreter
4584: (@pxref{Engine}) which executes the compiled Forth code on interpretive
4585: implementations.
1.27 crook 4586:
1.29 crook 4587: @cindex interpret state
4588: @cindex compile state
4589: The text interpreter operates in one of two states: @dfn{interpret
4590: state} and @dfn{compile state}. The current state is defined by the
4591: aptly-named variable, @code{state}.
4592:
4593: This section starts by describing how the text interpreter behaves when
4594: it is in interpret state, processing input from the user input device --
4595: the keyboard. This is the mode that a Forth system is in after it starts
4596: up.
4597:
4598: @cindex input buffer
4599: @cindex terminal input buffer
4600: The text interpreter works from an area of memory called the @dfn{input
4601: buffer}@footnote{When the text interpreter is processing input from the
4602: keyboard, this area of memory is called the @dfn{terminal input buffer}
4603: (TIB) and is addressed by the (obsolescent) words @code{TIB} and
4604: @code{#TIB}.}, which stores your keyboard input when you press the
1.30 anton 4605: @key{RET} key. Starting at the beginning of the input buffer, it skips
1.29 crook 4606: leading spaces (called @dfn{delimiters}) then parses a string (a
4607: sequence of non-space characters) until it reaches either a space
4608: character or the end of the buffer. Having parsed a string, it makes two
4609: attempts to process it:
1.27 crook 4610:
1.29 crook 4611: @cindex dictionary
1.27 crook 4612: @itemize @bullet
4613: @item
1.29 crook 4614: It looks for the string in a @dfn{dictionary} of definitions. If the
4615: string is found, the string names a @dfn{definition} (also known as a
4616: @dfn{word}) and the dictionary search returns information that allows
4617: the text interpreter to perform the word's @dfn{interpretation
4618: semantics}. In most cases, this simply means that the word will be
4619: executed.
1.27 crook 4620: @item
4621: If the string is not found in the dictionary, the text interpreter
1.29 crook 4622: attempts to treat it as a number, using the rules described in
4623: @ref{Number Conversion}. If the string represents a legal number in the
4624: current radix, the number is pushed onto a parameter stack (the data
4625: stack for integers, the floating-point stack for floating-point
4626: numbers).
4627: @end itemize
4628:
4629: If both attempts fail, or if the word is found in the dictionary but has
4630: no interpretation semantics@footnote{This happens if the word was
4631: defined as @code{COMPILE-ONLY}.} the text interpreter discards the
4632: remainder of the input buffer, issues an error message and waits for
4633: more input. If one of the attempts succeeds, the text interpreter
4634: repeats the parsing process until the whole of the input buffer has been
4635: processed, at which point it prints the status message ``@code{ ok}''
4636: and waits for more input.
4637:
4638: @cindex parse area
4639: The text interpreter keeps track of its position in the input buffer by
4640: updating a variable called @code{>IN} (pronounced ``to-in''). The value
4641: of @code{>IN} starts out as 0, indicating an offset of 0 from the start
4642: of the input buffer. The region from offset @code{>IN @@} to the end of
4643: the input buffer is called the @dfn{parse area}@footnote{In other words,
4644: the text interpreter processes the contents of the input buffer by
4645: parsing strings from the parse area until the parse area is empty.}.
4646: This example shows how @code{>IN} changes as the text interpreter parses
4647: the input buffer:
4648:
4649: @example
4650: : remaining >IN @@ SOURCE 2 PICK - -ROT + SWAP
4651: CR ." ->" TYPE ." <-" ; IMMEDIATE
4652:
4653: 1 2 3 remaining + remaining .
4654:
4655: : foo 1 2 3 remaining SWAP remaining ;
4656: @end example
4657:
4658: @noindent
4659: The result is:
4660:
4661: @example
4662: ->+ remaining .<-
4663: ->.<-5 ok
4664:
4665: ->SWAP remaining ;-<
4666: ->;<- ok
4667: @end example
4668:
4669: @cindex parsing words
4670: The value of @code{>IN} can also be modified by a word in the input
4671: buffer that is executed by the text interpreter. This means that a word
4672: can ``trick'' the text interpreter into either skipping a section of the
4673: input buffer@footnote{This is how parsing words work.} or into parsing a
4674: section twice. For example:
1.27 crook 4675:
1.29 crook 4676: @example
4677: : lat ." <<lat>>" ;
4678: : flat ." <<flat>>" >IN DUP @@ 3 - SWAP ! ;
4679: @end example
4680:
4681: @noindent
4682: When @code{flat} is executed, this output is produced@footnote{Exercise
4683: for the reader: what would happen if the @code{3} were replaced with
4684: @code{4}?}:
4685:
4686: @example
4687: <<flat>><<lat>>
4688: @end example
4689:
4690: @noindent
4691: Two important notes about the behaviour of the text interpreter:
1.27 crook 4692:
4693: @itemize @bullet
4694: @item
4695: It processes each input string to completion before parsing additional
1.29 crook 4696: characters from the input buffer.
4697: @item
4698: It treats the input buffer as a read-only region (and so must your code).
4699: @end itemize
4700:
4701: @noindent
4702: When the text interpreter is in compile state, its behaviour changes in
4703: these ways:
4704:
4705: @itemize @bullet
4706: @item
4707: If a parsed string is found in the dictionary, the text interpreter will
4708: perform the word's @dfn{compilation semantics}. In most cases, this
4709: simply means that the execution semantics of the word will be appended
4710: to the current definition.
1.27 crook 4711: @item
1.29 crook 4712: When a number is encountered, it is compiled into the current definition
4713: (as a literal) rather than being pushed onto a parameter stack.
4714: @item
4715: If an error occurs, @code{state} is modified to put the text interpreter
4716: back into interpret state.
4717: @item
4718: Each time a line is entered from the keyboard, Gforth prints
4719: ``@code{ compiled}'' rather than `` @code{ok}''.
4720: @end itemize
4721:
4722: @cindex text interpreter - input sources
4723: When the text interpreter is using an input device other than the
4724: keyboard, its behaviour changes in these ways:
4725:
4726: @itemize @bullet
4727: @item
4728: When the parse area is empty, the text interpreter attempts to refill
4729: the input buffer from the input source. When the input source is
4730: exhausted, the input source is set back to the user input device.
4731: @item
4732: It doesn't print out ``@code{ ok}'' or ``@code{ compiled}'' messages each
4733: time the parse area is emptied.
4734: @item
4735: If an error occurs, the input source is set back to the user input
4736: device.
1.27 crook 4737: @end itemize
1.21 crook 4738:
1.29 crook 4739: @ref{Input Sources} describes this in more detail.
4740:
1.26 crook 4741: doc->in
1.27 crook 4742: doc-source
4743:
1.26 crook 4744: doc-tib
4745: doc-#tib
1.1 anton 4746:
1.26 crook 4747: @menu
1.29 crook 4748: * Input Sources::
1.26 crook 4749: * Number Conversion::
4750: * Interpret/Compile states::
4751: * Literals::
4752: * Interpreter Directives::
4753: @end menu
1.1 anton 4754:
1.29 crook 4755: @node Input Sources, Number Conversion, The Text Interpreter, The Text Interpreter
4756: @subsection Input Sources
4757: @cindex input sources
4758: @cindex text interpreter - input sources
4759:
4760: By default, the text interpreter accepts input from the user input
4761: device (the keyboard) when Forth starts up. The text interpreter can
4762: process input from any of these sources:
4763:
4764: @itemize @bullet
4765: @item
4766: The user input device -- the keyboard.
4767: @item
4768: A file, using the words described in @ref{Forth source files}.
4769: @item
4770: A block, using the words described in @ref{Blocks}.
4771: @item
4772: A text string, using @code{evaluate}.
4773: @end itemize
4774:
4775: A program can identify the current input device from the values of
4776: @code{source-id} and @code{blk}.
4777:
4778: doc-source-id
4779: doc-blk
4780:
4781: doc-save-input
4782: doc-restore-input
4783:
4784: doc-evaluate
1.1 anton 4785:
1.29 crook 4786:
4787: @node Number Conversion, Interpret/Compile states, Input Sources, The Text Interpreter
1.26 crook 4788: @subsection Number Conversion
4789: @cindex number conversion
4790: @cindex double-cell numbers, input format
4791: @cindex input format for double-cell numbers
4792: @cindex single-cell numbers, input format
4793: @cindex input format for single-cell numbers
4794: @cindex floating-point numbers, input format
4795: @cindex input format for floating-point numbers
1.1 anton 4796:
1.29 crook 4797: This section describes the rules that the text interpreter uses when it
4798: tries to convert a string into a number.
1.1 anton 4799:
1.26 crook 4800: Let <digit> represent any character that is a legal digit in the current
1.29 crook 4801: number base@footnote{For example, 0-9 when the number base is decimal or
4802: 0-9, A-F when the number base is hexadecimal.}.
1.1 anton 4803:
1.26 crook 4804: Let <decimal digit> represent any character in the range 0-9.
1.1 anton 4805:
1.29 crook 4806: Let @{@i{a b}@} represent the @i{optional} presence of any of the characters
4807: in the braces (@i{a} or @i{b} or neither).
1.1 anton 4808:
1.26 crook 4809: Let * represent any number of instances of the previous character
4810: (including none).
1.1 anton 4811:
1.26 crook 4812: Let any other character represent itself.
1.1 anton 4813:
1.29 crook 4814: @noindent
1.26 crook 4815: Now, the conversion rules are:
1.21 crook 4816:
1.26 crook 4817: @itemize @bullet
4818: @item
4819: A string of the form <digit><digit>* is treated as a single-precision
1.29 crook 4820: (cell-sized) positive integer. Examples are 0 123 6784532 32343212343456 42
1.26 crook 4821: @item
4822: A string of the form -<digit><digit>* is treated as a single-precision
1.29 crook 4823: (cell-sized) negative integer, and is represented using 2's-complement
1.26 crook 4824: arithmetic. Examples are -45 -5681 -0
4825: @item
4826: A string of the form <digit><digit>*.<digit>* is treated as a double-precision
1.29 crook 4827: (double-cell-sized) positive integer. Examples are 3465. 3.465 34.65
4828: (all three of these represent the same number).
1.26 crook 4829: @item
4830: A string of the form -<digit><digit>*.<digit>* is treated as a
1.29 crook 4831: double-precision (double-cell-sized) negative integer, and is
1.26 crook 4832: represented using 2's-complement arithmetic. Examples are -3465. -3.465
1.29 crook 4833: -34.65 (all three of these represent the same number).
1.26 crook 4834: @item
1.29 crook 4835: A string of the form @{+ -@}<decimal digit>@{.@}<decimal digit>*@{e
4836: E@}@{+ -@}<decimal digit><decimal digit>* is treated as a floating-point
1.35 anton 4837: number. Examples are 1e 1e0 1.e 1.e0 +1e+0 (which all represent the same
1.29 crook 4838: number) +12.E-4
1.26 crook 4839: @end itemize
1.1 anton 4840:
1.26 crook 4841: By default, the number base used for integer number conversion is given
1.35 anton 4842: by the contents of the variable @code{base}. Note that a lot of
4843: confusion can result from unexpected values of @code{base}. If you
4844: change @code{base} anywhere, make sure to save the old value and restore
4845: it afterwards. In general I recommend keeping @code{base} decimal, and
4846: using the prefixes described below for the popular non-decimal bases.
1.1 anton 4847:
1.29 crook 4848: doc-dpl
1.26 crook 4849: doc-base
4850: doc-hex
4851: doc-decimal
1.1 anton 4852:
1.26 crook 4853: @cindex '-prefix for character strings
4854: @cindex &-prefix for decimal numbers
4855: @cindex %-prefix for binary numbers
4856: @cindex $-prefix for hexadecimal numbers
1.35 anton 4857: Gforth allows you to override the value of @code{base} by using a
1.29 crook 4858: prefix@footnote{Some Forth implementations provide a similar scheme by
4859: implementing @code{$} etc. as parsing words that process the subsequent
4860: number in the input stream and push it onto the stack. For example, see
4861: @cite{Number Conversion and Literals}, by Wil Baden; Forth Dimensions
4862: 20(3) pages 26--27. In such implementations, unlike in Gforth, a space
4863: is required between the prefix and the number.} before the first digit
4864: of an (integer) number. Four prefixes are supported:
1.1 anton 4865:
1.26 crook 4866: @itemize @bullet
4867: @item
1.35 anton 4868: @code{&} -- decimal
1.26 crook 4869: @item
1.35 anton 4870: @code{%} -- binary
1.26 crook 4871: @item
1.35 anton 4872: @code{$} -- hexadecimal
1.26 crook 4873: @item
1.35 anton 4874: @code{'} -- base @code{max-char+1}
1.26 crook 4875: @end itemize
1.1 anton 4876:
1.26 crook 4877: Here are some examples, with the equivalent decimal number shown after
4878: in braces:
1.1 anton 4879:
1.26 crook 4880: -$41 (-65), %1001101 (205), %1001.0001 (145 - a double-precision number),
4881: 'AB (16706; ascii A is 65, ascii B is 66, number is 65*256 + 66),
4882: 'ab (24930; ascii a is 97, ascii B is 98, number is 97*256 + 98),
4883: &905 (905), $abc (2478), $ABC (2478).
1.1 anton 4884:
1.26 crook 4885: @cindex number conversion - traps for the unwary
1.29 crook 4886: @noindent
1.26 crook 4887: Number conversion has a number of traps for the unwary:
1.1 anton 4888:
1.26 crook 4889: @itemize @bullet
4890: @item
4891: You cannot determine the current number base using the code sequence
1.35 anton 4892: @code{base @@ .} -- the number base is always 10 in the current number
4893: base. Instead, use something like @code{base @@ dec.}
1.26 crook 4894: @item
4895: If the number base is set to a value greater than 14 (for example,
4896: hexadecimal), the number 123E4 is ambiguous; the conversion rules allow
4897: it to be intepreted as either a single-precision integer or a
4898: floating-point number (Gforth treats it as an integer). The ambiguity
4899: can be resolved by explicitly stating the sign of the mantissa and/or
4900: exponent: 123E+4 or +123E4 -- if the number base is decimal, no
4901: ambiguity arises; either representation will be treated as a
4902: floating-point number.
4903: @item
1.29 crook 4904: There is a word @code{bin} but it does @i{not} set the number base!
1.26 crook 4905: It is used to specify file types.
4906: @item
4907: ANS Forth requires the @code{.} of a double-precision number to
4908: be the final character in the string. Allowing the @code{.} to be
4909: anywhere after the first digit is a Gforth extension.
4910: @item
4911: The number conversion process does not check for overflow.
4912: @item
4913: In Gforth, number conversion to floating-point numbers always use base
1.35 anton 4914: 10, irrespective of the value of @code{base}. In ANS Forth,
1.26 crook 4915: conversion to floating-point numbers whilst the value of
1.35 anton 4916: @code{base} is not 10 is an ambiguous condition.
1.26 crook 4917: @end itemize
1.1 anton 4918:
1.29 crook 4919: @ref{Input} describes words that you can use to read numbers into your
4920: programs.
1.1 anton 4921:
1.26 crook 4922: @node Interpret/Compile states, Literals, Number Conversion, The Text Interpreter
4923: @subsection Interpret/Compile states
4924: @cindex Interpret/Compile states
1.1 anton 4925:
1.29 crook 4926: A standard program is not permitted to change @code{state}
4927: explicitly. However, it can change @code{state} implicitly, using the
4928: words @code{[} and @code{]}. When @code{[} is executed it switches
4929: @code{state} to interpret state, and therefore the text interpreter
4930: starts interpreting. When @code{]} is executed it switches @code{state}
4931: to compile state and therefore the text interpreter starts
4932: compiling. The most common usage for these words is to compile literals,
4933: as shown in @ref{Literals}. However, they give you the freedom to switch
1.35 anton 4934: modes at will.
4935:
4936: @c This is a bad example: It's non-standard, and it's not necessary.
4937: @c However, I can't think of a good example for switching into compile
4938: @c state when there is no current word (@code{state}-smart words are not a
4939: @c good reason). So maybe we should use an example for switching into
4940: @c interpret @code{state} in a colon def. - anton
4941:
4942: Here is an example of building a jump-table of execution
1.29 crook 4943: tokens:
4944:
4945: @example
4946: : AA ." this is A" ;
4947: : BB ." this is B" ;
4948: : CC ." this is C" ;
4949:
4950: create table ' aa COMPILE, ' bb COMPILE, ' cc COMPILE,
4951: : go ( n -- ) \ n is offset into table.. 0 for 1st entry
4952: cells table + @ execute ;
4953: @end example
4954:
4955: @noindent
4956: Now @code{0 go} will display ``@code{this is A}''. The table can be
4957: built far more neatly@footnote{The source code is neater.. what is
4958: compiled in memory in each case is identical.} like this:
4959:
4960: @example
4961: create table ] aa bb cc [
4962: @end example
4963:
4964: The problem with this code is that it is not portable; it will only work
4965: on systems where code space and data space co-incide. The reason is that
4966: both tables @i{compile} execution tokens -- into code space. The
4967: Standard only allows data space to be assigned for a @code{CREATE}d
4968: word. In addition, the Standard only allows @code{@@} to access data
4969: space, whilst this example is using it to access code space. The only
4970: portable, Standard way to build this table is to build it in data space,
4971: like this:
4972:
4973: @example
4974: create table ' aa , ' bb , ' cc ,
4975: @end example
4976:
4977: @noindent
4978: A similar technique can be used to build a table of constants:
4979:
4980: @example
4981: create primes 1 , 3 , 5 , 7 , 11 ,
4982: @end example
1.1 anton 4983:
1.26 crook 4984: doc-state
4985: doc-[
4986: doc-]
1.1 anton 4987:
1.26 crook 4988: @node Literals, Interpreter Directives, Interpret/Compile states, The Text Interpreter
4989: @subsection Literals
4990: @cindex Literals
1.21 crook 4991:
1.29 crook 4992: Often, you want to use a number within a colon definition. When you do
4993: this, the text interpreter automatically compiles the number as a
4994: @i{literal}. A literal is a number whose run-time effect is to be pushed
4995: onto the stack. If you had to do some maths to generate the number, you
4996: might write it like this:
4997:
4998: @example
4999: : HOUR-TO-SEC ( n1 -- n2 )
5000: 60 * \ to minutes
5001: 60 * ; \ to seconds
5002: @end example
5003:
5004: It is very clear what this definition is doing, but it's inefficient
5005: since it is performing 2 multiples at run-time. An alternative would be
5006: to write:
5007:
5008: @example
5009: : HOUR-TO-SEC ( n1 -- n2 )
5010: 3600 * ; \ to seconds
5011: @end example
5012:
5013: Which does the same thing, and has the advantage of using a single
5014: multiply. Ideally, we'd like the efficiency of the second with the
5015: readability of the first.
5016:
5017: @code{Literal} allows us to achieve that. It takes a number from the
5018: stack and lays it down in the current definition just as though the
5019: number had been typed directly into the definition. Our first attempt
5020: might look like this:
5021:
5022: @example
5023: 60 \ mins per hour
5024: 60 * \ seconds per minute
5025: : HOUR-TO-SEC ( n1 -- n2 )
5026: Literal * ; \ to seconds
5027: @end example
5028:
5029: But this produces the error message @code{unstructured}. What happened?
5030: The stack notation for @code{:} is (@i{ -- colon-sys}) and the size of
5031: @i{colon-sys} is implementation-defined. In other words, once we start a
5032: colon definition we can't portably access anything that was on the stack
5033: before the definition began@footnote{@cite{Two Problems in ANS Forth},
5034: by Thomas Worthington; Forth Dimensions 20(2) pages 32--34 describes
5035: some situations where you might want to access stack items above
5036: colon-sys, and provides a solution to the problem.}. The correct way of
5037: solving this problem in this instance is to use @code{[ ]} like this:
5038:
5039: @example
5040: : HOUR-TO-SEC ( n1 -- n2 )
5041: [ 60 \ minutes per hour
5042: 60 * ] \ seconds per minute
5043: LITERAL * ; \ to seconds
5044: @end example
1.23 crook 5045:
1.26 crook 5046: doc-literal
5047: doc-]L
5048: doc-2literal
5049: doc-fliteral
1.1 anton 5050:
1.29 crook 5051: @node Interpreter Directives, , Literals, The Text Interpreter
1.26 crook 5052: @subsection Interpreter Directives
5053: @cindex interpreter directives
1.1 anton 5054:
1.29 crook 5055: These words are usually used in interpret state; typically to control
5056: which parts of a source file are processed by the text
1.26 crook 5057: interpreter. There are only a few ANS Forth Standard words, but Gforth
5058: supplements these with a rich set of immediate control structure words
5059: to compensate for the fact that the non-immediate versions can only be
1.29 crook 5060: used in compile state (@pxref{Control Structures}). Typical usages:
5061:
5062: @example
5063: FALSE Constant ASSEMBLER
5064: .
5065: .
5066: ASSEMBLER [IF]
5067: : ASSEMBLER-FEATURE
5068: ...
5069: ;
5070: [ENDIF]
5071: .
5072: .
5073: : SEE
5074: ... \ general-purpose SEE code
5075: [ ASSEMBLER [IF] ]
5076: ... \ assembler-specific SEE code
5077: [ [ENDIF] ]
5078: ;
5079: @end example
1.1 anton 5080:
1.26 crook 5081: doc-[IF]
5082: doc-[ELSE]
5083: doc-[THEN]
5084: doc-[ENDIF]
1.1 anton 5085:
1.26 crook 5086: doc-[IFDEF]
5087: doc-[IFUNDEF]
1.1 anton 5088:
1.26 crook 5089: doc-[?DO]
5090: doc-[DO]
5091: doc-[FOR]
5092: doc-[LOOP]
5093: doc-[+LOOP]
5094: doc-[NEXT]
1.1 anton 5095:
1.26 crook 5096: doc-[BEGIN]
5097: doc-[UNTIL]
5098: doc-[AGAIN]
5099: doc-[WHILE]
5100: doc-[REPEAT]
1.1 anton 5101:
1.27 crook 5102:
5103:
1.26 crook 5104: @c -------------------------------------------------------------
5105: @node Tokens for Words, Word Lists, The Text Interpreter, Words
5106: @section Tokens for Words
5107: @cindex tokens for words
1.1 anton 5108:
1.28 crook 5109: This section describes the creation and use of tokens that represent
1.29 crook 5110: words.
5111:
1.32 anton 5112: Named words have information stored in their header space entries to
1.29 crook 5113: indicate any non-default semantics (@pxref{Interpretation and
5114: Compilation Semantics}). The semantics can be modified, using
5115: @code{immediate} and/or @code{compile-only}, at the time that the words
1.32 anton 5116: are defined. Unnamed words have (by definition) no header space
1.29 crook 5117: entry, and therefore must have default semantics.
1.21 crook 5118:
1.26 crook 5119: Named words have interpretation and compilation semantics. Unnamed words
5120: just have execution semantics.
1.21 crook 5121:
1.29 crook 5122: @cindex xt
5123: @cindex execution token
5124: The execution semantics of an unnamed word are represented by an
5125: @dfn{execution token} (@i{xt}). As explained in @ref{Supplying names},
5126: the execution token of the last word defined can be produced with
5127: @code{lastxt}.
5128:
5129: The interpretation semantics of a named word are also represented by an
5130: execution token. You can produce the execution token using @code{'} or
5131: @code{[']}. A simple example shows the difference between the two:
1.21 crook 5132:
1.29 crook 5133: @example
5134: : greet ( -- ) ." Hello" ;
1.36 anton 5135: : foo ( -- xt ) ['] greet execute ; \ ['] parses greet at compile-time
5136: : bar ( -- ) ' execute ; \ ' parses at run-time
1.1 anton 5137:
1.29 crook 5138: \ the next four lines all do the same thing
1.36 anton 5139: foo
5140: bar greet
1.29 crook 5141: greet
5142: ' greet EXECUTE
5143: @end example
1.1 anton 5144:
1.29 crook 5145: An execution token occupies one cell.
1.26 crook 5146: @cindex code field address
5147: @cindex CFA
1.29 crook 5148: In Gforth, the abstract data type @i{execution token} is implemented
1.26 crook 5149: as a code field address (CFA).
5150: @comment TODO note that the standard does not say what it represents..
5151: @comment and you cannot necessarily compile it in all Forths (eg native
5152: @comment compilers?).
1.1 anton 5153:
1.29 crook 5154: For literals, use @code{'} in interpreted code and @code{[']} in
5155: compiled code. Gforth's @code{'} and @code{[']} behave somewhat
5156: unusually by complaining about compile-only words. To get the execution
5157: token for a compile-only word @i{name}, use @code{COMP' @i{name} DROP}
5158: or @code{[COMP'] @i{name} DROP}.
1.1 anton 5159:
1.26 crook 5160: @cindex compilation token
1.29 crook 5161: The compilation semantics of a named word are represented by a
5162: @dfn{compilation token} consisting of two cells: @i{w xt}. The top cell
5163: @i{xt} is an execution token. The compilation semantics represented by
5164: the compilation token can be performed with @code{execute}, which
5165: consumes the whole compilation token, with an additional stack effect
5166: determined by the represented compilation semantics.
5167:
5168: At present, the @i{w} part of a compilation token is an execution token,
5169: and the @i{xt} part represents either @code{execute} or
5170: @code{compile,}@footnote{Depending upon the compilation semantics of the
5171: word. If the word has default compilation semantics, the @i{xt} will
1.36 anton 5172: represent @code{compile,}. Otherwise (e.g., for immediate words), the
5173: @i{xt} will represent @code{execute}.}. However, don't rely on that
5174: knowledge, unless necessary; future versions of Gforth may introduce
5175: unusual compilation tokens (e.g., a compilation token that represents
5176: the compilation semantics of a literal).
1.1 anton 5177:
1.26 crook 5178: You can compile the compilation semantics with @code{postpone,}. I.e.,
1.29 crook 5179: @code{COMP' @i{word} postpone,} is equivalent to @code{postpone
5180: @i{word}}.
1.21 crook 5181:
1.26 crook 5182: @cindex name token
5183: @cindex name field address
5184: @cindex NFA
1.29 crook 5185: Named words are also represented by the @dfn{name token}, (@i{nt}). In
5186: Gforth, the abstract data type @emph{name token} is implemented as a
5187: name field address (NFA).
5188:
5189: doc-execute
5190: doc-compile,
5191: doc-[']
5192: doc-'
5193: doc-[comp']
5194: doc-comp'
5195: doc-postpone,
1.1 anton 5196:
1.26 crook 5197: doc-find-name
5198: doc-name>int
5199: doc-name?int
5200: doc-name>comp
5201: doc-name>string
1.1 anton 5202:
1.26 crook 5203: @c -------------------------------------------------------------
5204: @node Word Lists, Environmental Queries, Tokens for Words, Words
5205: @section Word Lists
5206: @cindex word lists
1.32 anton 5207: @cindex header space
1.1 anton 5208:
1.36 anton 5209: A wordlist is a list of named words; you can add new words and look up
5210: words by name (and you can remove words in a restricted way with
5211: markers). Every named (and @code{reveal}ed) word is in one wordlist.
5212:
5213: @cindex search order stack
5214: The text interpreter searches the wordlists present in the search order
5215: (a stack of wordlists), from the top to the bottom. Within each
5216: wordlist, the search starts conceptually at the newest word; i.e., if
5217: two words in a wordlist have the same name, the newer word is found.
1.1 anton 5218:
1.26 crook 5219: @cindex compilation word list
1.36 anton 5220: New words are added to the @dfn{compilation wordlist} (aka current
5221: wordlist).
1.1 anton 5222:
1.36 anton 5223: @cindex wid
5224: A word list is identified by a cell-sized word list identifier (@i{wid})
5225: in much the same way as a file is identified by a file handle. The
5226: numerical value of the wid has no (portable) meaning, and might change
5227: from session to session.
1.1 anton 5228:
1.29 crook 5229: The ANS Forth ``Search order'' word set is intended to provide a set of
5230: low-level tools that allow various different schemes to be
1.26 crook 5231: implemented. Gforth provides @code{vocabulary}, a traditional Forth
5232: word. @file{compat/vocabulary.fs} provides an implementation in ANS
5233: Standard Forth.
1.1 anton 5234:
1.27 crook 5235: @comment TODO: locals section refers to here, saying that every word list (aka
5236: @comment vocabulary) has its own methods for searching etc. Need to document that.
1.1 anton 5237:
1.27 crook 5238: @comment the thisone- prefix is used to pick out the true definition of a
5239: @comment word from the source files, rather than some alias.
1.26 crook 5240: doc-forth-wordlist
5241: doc-definitions
5242: doc-get-current
5243: doc-set-current
5244: doc-get-order
1.27 crook 5245: doc---thisone-set-order
1.26 crook 5246: doc-wordlist
1.30 anton 5247: doc-table
1.36 anton 5248: doc-push-order
5249: doc-previous
1.26 crook 5250: doc-also
1.27 crook 5251: doc---thisone-forth
1.26 crook 5252: doc-only
1.27 crook 5253: doc---thisone-order
1.15 anton 5254:
1.26 crook 5255: doc-find
5256: doc-search-wordlist
1.15 anton 5257:
1.26 crook 5258: doc-words
5259: doc-vlist
1.1 anton 5260:
1.26 crook 5261: doc-mappedwordlist
5262: doc-root
5263: doc-vocabulary
5264: doc-seal
5265: doc-vocs
5266: doc-current
5267: doc-context
1.1 anton 5268:
1.26 crook 5269: @menu
5270: * Why use word lists?::
5271: * Word list examples::
5272: @end menu
5273:
5274: @node Why use word lists?, Word list examples, Word Lists, Word Lists
5275: @subsection Why use word lists?
5276: @cindex word lists - why use them?
5277:
1.29 crook 5278: Here are some reasons for using multiple word lists:
1.26 crook 5279:
5280: @itemize @bullet
5281: @item
1.32 anton 5282: To improve compilation speed by reducing the number of header space
1.26 crook 5283: entries that must be searched. This is achieved by creating a new
5284: word list that contains all of the definitions that are used in the
5285: definition of a Forth system but which would not usually be used by
5286: programs running on that system. That word list would be on the search
5287: list when the Forth system was compiled but would be removed from the
5288: search list for normal operation. This can be a useful technique for
5289: low-performance systems (for example, 8-bit processors in embedded
5290: systems) but is unlikely to be necessary in high-performance desktop
5291: systems.
5292: @item
5293: To prevent a set of words from being used outside the context in which
5294: they are valid. Two classic examples of this are an integrated editor
5295: (all of the edit commands are defined in a separate word list; the
5296: search order is set to the editor word list when the editor is invoked;
5297: the old search order is restored when the editor is terminated) and an
5298: integrated assembler (the op-codes for the machine are defined in a
5299: separate word list which is used when a @code{CODE} word is defined).
5300: @item
5301: To prevent a name-space clash between multiple definitions with the same
5302: name. For example, when building a cross-compiler you might have a word
5303: @code{IF} that generates conditional code for your target system. By
5304: placing this definition in a different word list you can control whether
5305: the host system's @code{IF} or the target system's @code{IF} get used in
5306: any particular context by controlling the order of the word lists on the
5307: search order stack.
5308: @end itemize
1.1 anton 5309:
1.26 crook 5310: @node Word list examples, ,Why use word lists?, Word Lists
5311: @subsection Word list examples
5312: @cindex word lists - examples
1.1 anton 5313:
1.26 crook 5314: Here is an example of creating and using a new wordlist using ANS
5315: Forth Standard words:
1.1 anton 5316:
5317: @example
1.26 crook 5318: wordlist constant my-new-words-wordlist
5319: : my-new-words get-order nip my-new-words-wordlist swap set-order ;
1.21 crook 5320:
1.26 crook 5321: \ add it to the search order
5322: also my-new-words
1.21 crook 5323:
1.26 crook 5324: \ alternatively, add it to the search order and make it
5325: \ the compilation word list
5326: also my-new-words definitions
5327: \ type "order" to see the problem
1.21 crook 5328: @end example
5329:
1.26 crook 5330: The problem with this example is that @code{order} has no way to
5331: associate the name @code{my-new-words} with the wid of the word list (in
5332: Gforth, @code{order} and @code{vocs} will display @code{???} for a wid
5333: that has no associated name). There is no Standard way of associating a
5334: name with a wid.
5335:
5336: In Gforth, this example can be re-coded using @code{vocabulary}, which
5337: associates a name with a wid:
1.21 crook 5338:
1.26 crook 5339: @example
5340: vocabulary my-new-words
1.21 crook 5341:
1.26 crook 5342: \ add it to the search order
5343: my-new-words
1.21 crook 5344:
1.26 crook 5345: \ alternatively, add it to the search order and make it
5346: \ the compilation word list
5347: my-new-words definitions
5348: \ type "order" to see that the problem is solved
5349: @end example
1.23 crook 5350:
1.26 crook 5351: @c -------------------------------------------------------------
5352: @node Environmental Queries, Files, Word Lists, Words
5353: @section Environmental Queries
5354: @cindex environmental queries
1.21 crook 5355:
1.26 crook 5356: ANS Forth introduced the idea of ``environmental queries'' as a way
5357: for a program running on a system to determine certain characteristics of the system.
5358: The Standard specifies a number of strings that might be recognised by a system.
1.21 crook 5359:
1.32 anton 5360: The Standard requires that the header space used for environmental queries
5361: be distinct from the header space used for definitions.
1.21 crook 5362:
1.26 crook 5363: Typically, environmental queries are supported by creating a set of
1.29 crook 5364: definitions in a word list that is @i{only} used during environmental
1.26 crook 5365: queries; that is what Gforth does. There is no Standard way of adding
5366: definitions to the set of recognised environmental queries, but any
5367: implementation that supports the loading of optional word sets must have
5368: some mechanism for doing this (after loading the word set, the
5369: associated environmental query string must return @code{true}). In
5370: Gforth, the word list used to honour environmental queries can be
5371: manipulated just like any other word list.
1.21 crook 5372:
1.26 crook 5373: doc-environment?
5374: doc-environment-wordlist
1.21 crook 5375:
1.26 crook 5376: doc-gforth
5377: doc-os-class
1.21 crook 5378:
1.26 crook 5379: Note that, whilst the documentation for (e.g.) @code{gforth} shows it
5380: returning two items on the stack, querying it using @code{environment?}
5381: will return an additional item; the @code{true} flag that shows that the
5382: string was recognised.
1.21 crook 5383:
1.26 crook 5384: @comment TODO Document the standard strings or note where they are documented herein
1.21 crook 5385:
1.26 crook 5386: Here are some examples of using environmental queries:
1.21 crook 5387:
1.26 crook 5388: @example
5389: s" address-unit-bits" environment? 0=
5390: [IF]
5391: cr .( environmental attribute address-units-bits unknown... ) cr
5392: [THEN]
1.21 crook 5393:
1.26 crook 5394: s" block" environment? [IF] DROP include block.fs [THEN]
1.21 crook 5395:
1.26 crook 5396: s" gforth" environment? [IF] 2DROP include compat/vocabulary.fs [THEN]
1.21 crook 5397:
1.26 crook 5398: s" gforth" environment? [IF] .( Gforth version ) TYPE
5399: [ELSE] .( Not Gforth..) [THEN]
5400: @end example
1.21 crook 5401:
5402:
1.26 crook 5403: Here is an example of adding a definition to the environment word list:
1.21 crook 5404:
1.26 crook 5405: @example
5406: get-current environment-wordlist set-current
5407: true constant block
5408: true constant block-ext
5409: set-current
5410: @end example
1.21 crook 5411:
1.26 crook 5412: You can see what definitions are in the environment word list like this:
1.21 crook 5413:
1.26 crook 5414: @example
5415: get-order 1+ environment-wordlist swap set-order words previous
5416: @end example
1.21 crook 5417:
5418:
1.26 crook 5419: @c -------------------------------------------------------------
5420: @node Files, Blocks, Environmental Queries, Words
5421: @section Files
1.28 crook 5422: @cindex files
5423: @cindex I/O - file-handling
1.21 crook 5424:
1.26 crook 5425: Gforth provides facilities for accessing files that are stored in the
5426: host operating system's file-system. Files that are processed by Gforth
5427: can be divided into two categories:
1.21 crook 5428:
1.23 crook 5429: @itemize @bullet
5430: @item
1.29 crook 5431: Files that are processed by the Text Interpreter (@dfn{Forth source files}).
1.23 crook 5432: @item
1.29 crook 5433: Files that are processed by some other program (@dfn{general files}).
1.26 crook 5434: @end itemize
5435:
5436: @menu
5437: * Forth source files::
5438: * General files::
5439: * Search Paths::
5440: * Forth Search Paths::
5441: * General Search Paths::
5442: @end menu
5443:
1.21 crook 5444:
1.26 crook 5445: @c -------------------------------------------------------------
5446: @node Forth source files, General files, Files, Files
5447: @subsection Forth source files
5448: @cindex including files
5449: @cindex Forth source files
1.21 crook 5450:
1.26 crook 5451: The simplest way to interpret the contents of a file is to use one of
5452: these two formats:
1.21 crook 5453:
1.26 crook 5454: @example
5455: include mysource.fs
5456: s" mysource.fs" included
5457: @end example
1.21 crook 5458:
1.26 crook 5459: Sometimes you want to include a file only if it is not included already
5460: (by, say, another source file). In that case, you can use one of these
5461: fomats:
1.21 crook 5462:
1.26 crook 5463: @example
5464: require mysource.fs
5465: needs mysource.fs
5466: s" mysource.fs" required
5467: @end example
1.21 crook 5468:
1.26 crook 5469: @cindex stack effect of included files
5470: @cindex including files, stack effect
5471: I recommend that you write your source files such that interpreting them
5472: does not change the stack. This allows using these files with
5473: @code{required} and friends without complications. For example:
1.21 crook 5474:
1.26 crook 5475: @example
5476: 1 require foo.fs drop
5477: @end example
1.21 crook 5478:
1.26 crook 5479: doc-include-file
5480: doc-included
1.28 crook 5481: doc-included?
1.26 crook 5482: doc-include
5483: doc-required
5484: doc-require
5485: doc-needs
1.28 crook 5486: doc-init-included-files
1.21 crook 5487:
1.26 crook 5488: A definition in ANS Forth for @code{required} is provided in
5489: @file{compat/required.fs}.
1.21 crook 5490:
1.26 crook 5491: @c -------------------------------------------------------------
5492: @node General files, Search Paths, Forth source files, Files
5493: @subsection General files
5494: @cindex general files
5495: @cindex file-handling
1.21 crook 5496:
1.26 crook 5497: Files are opened/created by name and type. The following types are
5498: recognised:
1.1 anton 5499:
1.26 crook 5500: doc-r/o
5501: doc-r/w
5502: doc-w/o
5503: doc-bin
1.1 anton 5504:
1.26 crook 5505: When a file is opened/created, it returns a file identifier,
1.29 crook 5506: @i{wfileid} that is used for all other file commands. All file
5507: commands also return a status value, @i{wior}, that is 0 for a
1.26 crook 5508: successful operation and an implementation-defined non-zero value in the
5509: case of an error.
1.21 crook 5510:
1.26 crook 5511: doc-open-file
5512: doc-create-file
1.21 crook 5513:
1.26 crook 5514: doc-close-file
5515: doc-delete-file
5516: doc-rename-file
5517: doc-read-file
5518: doc-read-line
5519: doc-write-file
5520: doc-write-line
5521: doc-emit-file
5522: doc-flush-file
1.21 crook 5523:
1.26 crook 5524: doc-file-status
5525: doc-file-position
5526: doc-reposition-file
5527: doc-file-size
5528: doc-resize-file
1.21 crook 5529:
1.26 crook 5530: @c ---------------------------------------------------------
5531: @node Search Paths, Forth Search Paths, General files, Files
5532: @subsection Search Paths
5533: @cindex path for @code{included}
5534: @cindex file search path
5535: @cindex @code{include} search path
5536: @cindex search path for files
1.21 crook 5537:
1.26 crook 5538: If you specify an absolute filename (i.e., a filename starting with
5539: @file{/} or @file{~}, or with @file{:} in the second position (as in
5540: @samp{C:...})) for @code{included} and friends, that file is included
5541: just as you would expect.
1.21 crook 5542:
1.26 crook 5543: For relative filenames, Gforth uses a search path similar to Forth's
5544: search order (@pxref{Word Lists}). It tries to find the given filename
5545: in the directories present in the path, and includes the first one it
5546: finds. There are separate search paths for Forth source files and
5547: general files.
1.21 crook 5548:
1.26 crook 5549: If the search path contains the directory @file{.} (as it should), this
5550: refers to the directory that the present file was @code{included}
5551: from. This allows files to include other files relative to their own
5552: position (irrespective of the current working directory or the absolute
5553: position). This feature is essential for libraries consisting of
5554: several files, where a file may include other files from the library.
5555: It corresponds to @code{#include "..."} in C. If the current input
5556: source is not a file, @file{.} refers to the directory of the innermost
5557: file being included, or, if there is no file being included, to the
5558: current working directory.
1.21 crook 5559:
1.26 crook 5560: Use @file{~+} to refer to the current working directory (as in the
5561: @code{bash}).
1.1 anton 5562:
1.26 crook 5563: If the filename starts with @file{./}, the search path is not searched
5564: (just as with absolute filenames), and the @file{.} has the same meaning
5565: as described above.
1.1 anton 5566:
1.26 crook 5567: @c ---------------------------------------------------------
5568: @node Forth Search Paths, General Search Paths, Search Paths, Files
5569: @subsubsection Forth Search Paths
1.28 crook 5570: @cindex search path control - Forth
1.5 anton 5571:
1.26 crook 5572: The search path is initialized when you start Gforth (@pxref{Invoking
5573: Gforth}). You can display it and change it using these words:
1.5 anton 5574:
1.26 crook 5575: doc-.fpath
5576: doc-fpath+
5577: doc-fpath=
5578: doc-open-fpath-file
1.5 anton 5579:
1.26 crook 5580: Here is an example of using @code{fpath} and @code{require}:
1.5 anton 5581:
1.26 crook 5582: @example
5583: fpath= /usr/lib/forth/|./
5584: require timer.fs
5585: @end example
1.5 anton 5586:
1.26 crook 5587: @c ---------------------------------------------------------
5588: @node General Search Paths, , Forth Search Paths, Files
5589: @subsubsection General Search Paths
5590: @cindex search path control - for user applications
1.5 anton 5591:
1.26 crook 5592: Your application may need to search files in several directories, like
5593: @code{included} does. To facilitate this, Gforth allows you to define
5594: and use your own search paths, by providing generic equivalents of the
5595: Forth search path words:
1.5 anton 5596:
1.26 crook 5597: doc-.path
5598: doc-path+
5599: doc-path=
5600: doc-open-path-file
1.5 anton 5601:
1.26 crook 5602: Here's an example of creating a search path:
1.5 anton 5603:
1.26 crook 5604: @example
5605: \ Make a buffer for the path:
5606: create mypath 100 chars , \ maximum length (is checked)
5607: 0 , \ real len
5608: 100 chars allot \ space for path
5609: @end example
1.5 anton 5610:
1.26 crook 5611: @c -------------------------------------------------------------
5612: @node Blocks, Other I/O, Files, Words
5613: @section Blocks
1.28 crook 5614: @cindex I/O - blocks
5615: @cindex blocks
5616:
5617: When you run Gforth on a modern desk-top computer, it runs under the
5618: control of an operating system which provides certain services. One of
5619: these services is @var{file services}, which allows Forth source code
5620: and data to be stored in files and read into Gforth (@pxref{Files}).
5621:
5622: Traditionally, Forth has been an important programming language on
5623: systems where it has interfaced directly to the underlying hardware with
5624: no intervening operating system. Forth provides a mechanism, called
1.29 crook 5625: @dfn{blocks}, for accessing mass storage on such systems.
1.28 crook 5626:
5627: A block is a 1024-byte data area, which can be used to hold data or
5628: Forth source code. No structure is imposed on the contents of the
5629: block. A block is identified by its number; blocks are numbered
5630: contiguously from 1 to an implementation-defined maximum.
5631:
5632: A typical system that used blocks but no operating system might use a
5633: single floppy-disk drive for mass storage, with the disks formatted to
5634: provide 256-byte sectors. Blocks would be implemented by assigning the
5635: first four sectors of the disk to block 1, the second four sectors to
5636: block 2 and so on, up to the limit of the capacity of the disk. The disk
5637: would not contain any file system information, just the set of blocks.
5638:
1.29 crook 5639: @cindex blocks file
1.28 crook 5640: On systems that do provide file services, blocks are typically
1.29 crook 5641: implemented by storing a sequence of blocks within a single @dfn{blocks
1.28 crook 5642: file}. The size of the blocks file will be an exact multiple of 1024
5643: bytes, corresponding to the number of blocks it contains. This is the
5644: mechanism that Gforth uses.
5645:
1.29 crook 5646: @cindex @file{blocks.fb}
1.28 crook 5647: Only 1 blocks file can be open at a time. If you use block words without
5648: having specified a blocks file, Gforth defaults to the blocks file
5649: @file{blocks.fb}. Gforth uses the Forth search path when attempting to
5650: locate a blocks file (@pxref{Forth Search Paths}).
5651:
1.29 crook 5652: @cindex block buffers
1.28 crook 5653: When you read and write blocks under program control, Gforth uses a
1.29 crook 5654: number of @dfn{block buffers} as intermediate storage. These buffers are
1.28 crook 5655: not used when you use @code{load} to interpret the contents of a block.
5656:
5657: The behaviour of the block buffers is directly analagous to that of a
5658: cache. Each block buffer has three states:
5659:
5660: @itemize @bullet
5661: @item
5662: Unassigned
5663: @item
5664: Assigned-clean
5665: @item
5666: Assigned-dirty
5667: @end itemize
5668:
1.29 crook 5669: Initially, all block buffers are @i{unassigned}. In order to access a
1.28 crook 5670: block, the block (specified by its block number) must be assigned to a
5671: block buffer.
5672:
5673: The assignment of a block to a block buffer is performed by @code{block}
5674: or @code{buffer}. Use @code{block} when you wish to modify the existing
5675: contents of a block. Use @code{buffer} when you don't care about the
5676: existing contents of the block@footnote{The ANS Forth definition of
1.35 anton 5677: @code{buffer} is intended not to cause disk I/O; if the data associated
1.28 crook 5678: with the particular block is already stored in a block buffer due to an
5679: earlier @code{block} command, @code{buffer} will return that block
5680: buffer and the existing contents of the block will be
5681: available. Otherwise, @code{buffer} will simply assign a new, empty
1.29 crook 5682: block buffer for the block.}.
1.28 crook 5683:
5684: Once a block has been assigned to a block buffer, the block buffer state
1.29 crook 5685: becomes @i{assigned-clean}. Data can now be manipulated within the
1.28 crook 5686: block buffer.
5687:
5688: When the contents of a block buffer is changed it is necessary,
5689: @i{before calling} @code{block} @i{or} @code{buffer} @i{again}, to
5690: either abandon the changes (by doing nothing) or commit the changes,
5691: using @code{update}. Using @code{update} does not change the blocks
1.29 crook 5692: file; it simply changes a block buffer's state to @i{assigned-dirty}.
1.28 crook 5693:
1.29 crook 5694: The word @code{flush} causes all @i{assigned-dirty} blocks to be
1.28 crook 5695: written back to the blocks file on disk. Leaving Gforth using @code{bye}
5696: also causes a @code{flush} to be performed.
5697:
1.29 crook 5698: In Gforth, @code{block} and @code{buffer} use a @i{direct-mapped}
1.28 crook 5699: algorithm to assign a block buffer to a block. That means that any
5700: particular block can only be assigned to one specific block buffer,
1.29 crook 5701: called (for the particular operation) the @i{victim buffer}. If the
5702: victim buffer is @i{unassigned} or @i{assigned-clean} it can be
5703: allocated to the new block immediately. If it is @i{assigned-dirty}
1.28 crook 5704: its current contents must be written out to disk before it can be
5705: allocated to the new block.
5706:
5707: Although no structure is imposed on the contents of a block, it is
5708: traditional to display the contents as 16 lines each of 64 characters. A
5709: block provides a single, continuous stream of input (for example, it
5710: acts as a single parse area) -- there are no end-of-line characters
5711: within a block, and no end-of-file character at the end of a
5712: block. There are two consequences of this:
1.26 crook 5713:
1.28 crook 5714: @itemize @bullet
5715: @item
5716: The last character of one line wraps straight into the first character
5717: of the following line
5718: @item
5719: The word @code{\} -- comment to end of line -- requires special
5720: treatment; in the context of a block it causes all characters until the
5721: end of the current 64-character ``line'' to be ignored.
5722: @end itemize
5723:
5724: In Gforth, when you use @code{block} with a non-existent block number,
5725: the current block file will be extended to the appropriate size and the
5726: block buffer will be initialised with spaces.
5727:
1.29 crook 5728: Gforth doesn't encourage the use of blocks; the mechanism is only
5729: provided for backward compatibility -- ANS Forth requires blocks to be
5730: available when files are.
1.28 crook 5731:
5732: Common techniques that are used when working with blocks include:
5733:
5734: @itemize @bullet
5735: @item
5736: A screen editor that allows you to edit blocks without leaving the Forth
5737: environment.
5738: @item
5739: Shadow screens; where every code block has an associated block
5740: containing comments (for example: code in odd block numbers, comments in
5741: even block numbers). Typically, the block editor provides a convenient
5742: mechanism to toggle between code and comments.
5743: @item
5744: Load blocks; a single block (typically block 1) contains a number of
5745: @code{thru} commands which @code{load} the whole of the application.
5746: @end itemize
1.26 crook 5747:
1.29 crook 5748: See Frank Sergeant's Pygmy Forth to see just how well blocks can be
5749: integrated into a Forth programming environment.
1.26 crook 5750:
5751: @comment TODO what about errors on open-blocks?
5752: doc-open-blocks
5753: doc-use
5754: doc-get-block-fid
5755: doc-block-position
1.28 crook 5756:
5757: doc-scr
5758: doc-list
5759:
5760: doc---block-block
5761: doc-buffer
5762:
1.26 crook 5763: doc-update
1.28 crook 5764: doc-updated?
1.26 crook 5765: doc-save-buffers
5766: doc-empty-buffers
5767: doc-empty-buffer
5768: doc-flush
1.28 crook 5769:
1.26 crook 5770: doc-load
5771: doc-thru
5772: doc-+load
5773: doc-+thru
1.35 anton 5774: xdoc--gforth--->
1.26 crook 5775: doc-block-included
5776:
5777: @c -------------------------------------------------------------
5778: @node Other I/O, Programming Tools, Blocks, Words
5779: @section Other I/O
1.28 crook 5780: @cindex I/O - keyboard and display
1.26 crook 5781:
5782: @menu
5783: * Simple numeric output:: Predefined formats
5784: * Formatted numeric output:: Formatted (pictured) output
5785: * String Formats:: How Forth stores strings in memory
5786: * Displaying characters and strings:: Other stuff
5787: * Input:: Input
5788: @end menu
5789:
5790: @node Simple numeric output, Formatted numeric output, Other I/O, Other I/O
5791: @subsection Simple numeric output
1.28 crook 5792: @cindex numeric output - simple/free-format
1.5 anton 5793:
1.26 crook 5794: The simplest output functions are those that display numbers from the
5795: data or floating-point stacks. Floating-point output is always displayed
5796: using base 10. Numbers displayed from the data stack use the value stored
5797: in @code{base}.
1.5 anton 5798:
1.26 crook 5799: doc-.
5800: doc-dec.
5801: doc-hex.
5802: doc-u.
5803: doc-.r
5804: doc-u.r
5805: doc-d.
5806: doc-ud.
5807: doc-d.r
5808: doc-ud.r
5809: doc-f.
5810: doc-fe.
5811: doc-fs.
1.5 anton 5812:
1.26 crook 5813: Examples of printing the number 1234.5678E23 in the different floating-point output
5814: formats are shown below:
1.5 anton 5815:
5816: @example
1.26 crook 5817: f. 123456779999999000000000000.
5818: fe. 123.456779999999E24
5819: fs. 1.23456779999999E26
1.5 anton 5820: @end example
5821:
5822:
1.26 crook 5823: @node Formatted numeric output, String Formats, Simple numeric output, Other I/O
5824: @subsection Formatted numeric output
1.28 crook 5825: @cindex formatted numeric output
1.26 crook 5826: @cindex pictured numeric output
1.28 crook 5827: @cindex numeric output - formatted
1.26 crook 5828:
1.29 crook 5829: Forth traditionally uses a technique called @dfn{pictured numeric
1.26 crook 5830: output} for formatted printing of integers. In this technique, digits
5831: are extracted from the number (using the current output radix defined by
5832: @code{base}), converted to ASCII codes and appended to a string that is
5833: built in a scratch-pad area of memory (@pxref{core-idef,
5834: Implementation-defined options, Implementation-defined
5835: options}). Arbitrary characters can be appended to the string during the
5836: extraction process. The completed string is specified by an address
5837: and length and can be manipulated (@code{TYPE}ed, copied, modified)
5838: under program control.
1.5 anton 5839:
1.26 crook 5840: All of the words described in the previous section for simple numeric
5841: output are implemented in Gforth using pictured numeric output.
1.5 anton 5842:
1.26 crook 5843: Three important things to remember about Pictured Numeric Output:
1.5 anton 5844:
1.26 crook 5845: @itemize @bullet
5846: @item
1.28 crook 5847: It always operates on double-precision numbers; to display a
5848: single-precision number, convert it first (@pxref{Double precision} for
5849: ways of doing this).
1.26 crook 5850: @item
1.28 crook 5851: It always treats the double-precision number as though it were
5852: unsigned. The examples below show ways of printing signed numbers.
1.26 crook 5853: @item
5854: The string is built up from right to left; least significant digit first.
5855: @end itemize
1.5 anton 5856:
1.26 crook 5857: doc-<#
5858: doc-#
5859: doc-#s
5860: doc-hold
5861: doc-sign
5862: doc-#>
1.5 anton 5863:
1.26 crook 5864: doc-represent
1.5 anton 5865:
1.26 crook 5866: Here are some examples of using pictured numeric output:
1.5 anton 5867:
5868: @example
1.26 crook 5869: : my-u. ( u -- )
5870: \ Simplest use of pns.. behaves like Standard u.
5871: 0 \ convert to unsigned double
5872: <# \ start conversion
5873: #s \ convert all digits
5874: #> \ complete conversion
5875: TYPE SPACE ; \ display, with trailing space
1.5 anton 5876:
1.26 crook 5877: : cents-only ( u -- )
5878: 0 \ convert to unsigned double
5879: <# \ start conversion
5880: # # \ convert two least-significant digits
5881: #> \ complete conversion, discard other digits
5882: TYPE SPACE ; \ display, with trailing space
1.5 anton 5883:
1.26 crook 5884: : dollars-and-cents ( u -- )
5885: 0 \ convert to unsigned double
5886: <# \ start conversion
5887: # # \ convert two least-significant digits
5888: [char] . hold \ insert decimal point
5889: #s \ convert remaining digits
5890: [char] $ hold \ append currency symbol
5891: #> \ complete conversion
5892: TYPE SPACE ; \ display, with trailing space
1.5 anton 5893:
1.26 crook 5894: : my-. ( n -- )
5895: \ handling negatives.. behaves like Standard .
5896: s>d \ convert to signed double
5897: swap over dabs \ leave sign byte followed by unsigned double
5898: <# \ start conversion
5899: #s \ convert all digits
5900: rot sign \ get at sign byte, append "-" if needed
5901: #> \ complete conversion
5902: TYPE SPACE ; \ display, with trailing space
1.5 anton 5903:
1.26 crook 5904: : account. ( n -- )
5905: \ accountants don't like minus signs, they use braces
5906: \ for negative numbers
5907: s>d \ convert to signed double
5908: swap over dabs \ leave sign byte followed by unsigned double
5909: <# \ start conversion
5910: 2 pick \ get copy of sign byte
5911: 0< IF [char] ) hold THEN \ right-most character of output
5912: #s \ convert all digits
5913: rot \ get at sign byte
5914: 0< IF [char] ( hold THEN
5915: #> \ complete conversion
5916: TYPE SPACE ; \ display, with trailing space
1.5 anton 5917: @end example
5918:
1.26 crook 5919: Here are some examples of using these words:
1.5 anton 5920:
5921: @example
1.26 crook 5922: 1 my-u. 1
5923: hex -1 my-u. decimal FFFFFFFF
5924: 1 cents-only 01
5925: 1234 cents-only 34
5926: 2 dollars-and-cents $0.02
5927: 1234 dollars-and-cents $12.34
5928: 123 my-. 123
5929: -123 my. -123
5930: 123 account. 123
5931: -456 account. (456)
1.5 anton 5932: @end example
5933:
5934:
1.26 crook 5935: @node String Formats, Displaying characters and strings, Formatted numeric output, Other I/O
5936: @subsection String Formats
1.27 crook 5937: @cindex strings - see character strings
5938: @cindex character strings - formats
1.28 crook 5939: @cindex I/O - see character strings
1.26 crook 5940:
1.27 crook 5941: Forth commonly uses two different methods for representing character
5942: strings:
1.26 crook 5943:
5944: @itemize @bullet
5945: @item
5946: @cindex address of counted string
1.29 crook 5947: As a @dfn{counted string}, represented by a @i{c-addr}. The char
5948: addressed by @i{c-addr} contains a character-count, @i{n}, of the
5949: string and the string occupies the subsequent @i{n} char addresses in
1.26 crook 5950: memory.
5951: @item
1.29 crook 5952: As cell pair on the stack; @i{c-addr u}, where @i{u} is the length
5953: of the string in characters, and @i{c-addr} is the address of the
1.26 crook 5954: first byte of the string.
5955: @end itemize
5956:
5957: ANS Forth encourages the use of the second format when representing
5958: strings on the stack, whilst conceeding that the counted string format
5959: remains useful as a way of storing strings in memory.
5960:
5961: doc-count
5962:
5963: @xref{Memory Blocks} for words that move, copy and search
5964: for strings. @xref{Displaying characters and strings,} for words that
5965: display characters and strings.
5966:
5967:
5968: @node Displaying characters and strings, Input, String Formats, Other I/O
5969: @subsection Displaying characters and strings
1.27 crook 5970: @cindex characters - compiling and displaying
5971: @cindex character strings - compiling and displaying
1.26 crook 5972:
5973: This section starts with a glossary of Forth words and ends with a set
5974: of examples.
5975:
5976: doc-bl
5977: doc-space
5978: doc-spaces
5979: doc-emit
5980: doc-toupper
5981: doc-."
5982: doc-.(
5983: doc-type
5984: doc-cr
1.27 crook 5985: @cindex cursor control
1.26 crook 5986: doc-at-xy
5987: doc-page
5988: doc-s"
5989: doc-c"
5990: doc-char
5991: doc-[char]
5992: doc-sliteral
5993:
5994: As an example, consider the following text, stored in a file @file{test.fs}:
1.5 anton 5995:
5996: @example
1.26 crook 5997: .( text-1)
5998: : my-word
5999: ." text-2" cr
6000: .( text-3)
6001: ;
6002:
6003: ." text-4"
6004:
6005: : my-char
6006: [char] ALPHABET emit
6007: char emit
6008: ;
1.5 anton 6009: @end example
6010:
1.26 crook 6011: When you load this code into Gforth, the following output is generated:
1.5 anton 6012:
1.26 crook 6013: @example
1.30 anton 6014: @kbd{include test.fs @key{RET}} text-1text-3text-4 ok
1.26 crook 6015: @end example
1.5 anton 6016:
1.26 crook 6017: @itemize @bullet
6018: @item
6019: Messages @code{text-1} and @code{text-3} are displayed because @code{.(}
6020: is an immediate word; it behaves in the same way whether it is used inside
6021: or outside a colon definition.
6022: @item
6023: Message @code{text-4} is displayed because of Gforth's added interpretation
6024: semantics for @code{."}.
6025: @item
1.29 crook 6026: Message @code{text-2} is @i{not} displayed, because the text interpreter
1.26 crook 6027: performs the compilation semantics for @code{."} within the definition of
6028: @code{my-word}.
6029: @end itemize
1.5 anton 6030:
1.26 crook 6031: Here are some examples of executing @code{my-word} and @code{my-char}:
1.5 anton 6032:
1.26 crook 6033: @example
1.30 anton 6034: @kbd{my-word @key{RET}} text-2
1.26 crook 6035: ok
1.30 anton 6036: @kbd{my-char fred @key{RET}} Af ok
6037: @kbd{my-char jim @key{RET}} Aj ok
1.26 crook 6038: @end example
1.5 anton 6039:
6040: @itemize @bullet
6041: @item
1.26 crook 6042: Message @code{text-2} is displayed because of the run-time behaviour of
6043: @code{."}.
6044: @item
6045: @code{[char]} compiles the ``A'' from ``ALPHABET'' and puts its display code
6046: on the stack at run-time. @code{emit} always displays the character
6047: when @code{my-char} is executed.
6048: @item
6049: @code{char} parses a string at run-time and the second @code{emit} displays
6050: the first character of the string.
1.5 anton 6051: @item
1.26 crook 6052: If you type @code{see my-char} you can see that @code{[char]} discarded
6053: the text ``LPHABET'' and only compiled the display code for ``A'' into the
6054: definition of @code{my-char}.
1.5 anton 6055: @end itemize
6056:
6057:
6058:
1.26 crook 6059: @node Input, , Displaying characters and strings, Other I/O
6060: @subsection Input
6061: @cindex input
1.28 crook 6062: @cindex I/O - see input
6063: @cindex parsing a string
1.5 anton 6064:
1.27 crook 6065: @xref{String Formats} for ways of storing character strings in memory.
1.5 anton 6066:
1.27 crook 6067: @comment TODO examples for >number >float accept key key? pad parse word refill
1.29 crook 6068: @comment then index them
1.27 crook 6069:
6070: doc-key
6071: doc-key?
1.26 crook 6072: doc->number
6073: doc->float
6074: doc-accept
1.27 crook 6075: doc-pad
6076: doc-parse
6077: doc-word
6078: doc-sword
6079: doc-refill
6080: @comment obsolescent words..
6081: doc-convert
1.26 crook 6082: doc-query
6083: doc-expect
1.27 crook 6084: doc-span
1.5 anton 6085:
6086:
6087: @c -------------------------------------------------------------
1.26 crook 6088: @node Programming Tools, Assembler and Code Words, Other I/O, Words
6089: @section Programming Tools
6090: @cindex programming tools
1.12 anton 6091:
6092: @menu
1.26 crook 6093: * Debugging:: Simple and quick.
6094: * Assertions:: Making your programs self-checking.
6095: * Singlestep Debugger:: Executing your program word by word.
1.5 anton 6096: @end menu
6097:
1.26 crook 6098: @node Debugging, Assertions, Programming Tools, Programming Tools
6099: @subsection Debugging
6100: @cindex debugging
1.5 anton 6101:
1.26 crook 6102: Languages with a slow edit/compile/link/test development loop tend to
6103: require sophisticated tracing/stepping debuggers to facilate
6104: productive debugging.
1.5 anton 6105:
1.26 crook 6106: A much better (faster) way in fast-compiling languages is to add
6107: printing code at well-selected places, let the program run, look at
6108: the output, see where things went wrong, add more printing code, etc.,
6109: until the bug is found.
1.5 anton 6110:
1.26 crook 6111: The simple debugging aids provided in @file{debugs.fs}
6112: are meant to support this style of debugging. In addition, there are
6113: words for non-destructively inspecting the stack and memory:
1.5 anton 6114:
1.26 crook 6115: doc-.s
6116: doc-f.s
1.5 anton 6117:
1.29 crook 6118: There is a word @code{.r} but it does @i{not} display the return
1.26 crook 6119: stack! It is used for formatted numeric output.
1.5 anton 6120:
1.26 crook 6121: doc-depth
6122: doc-fdepth
6123: doc-clearstack
6124: doc-?
6125: doc-dump
1.5 anton 6126:
1.26 crook 6127: The word @code{~~} prints debugging information (by default the source
6128: location and the stack contents). It is easy to insert. If you use Emacs
6129: it is also easy to remove (@kbd{C-x ~} in the Emacs Forth mode to
6130: query-replace them with nothing). The deferred words
6131: @code{printdebugdata} and @code{printdebugline} control the output of
6132: @code{~~}. The default source location output format works well with
6133: Emacs' compilation mode, so you can step through the program at the
6134: source level using @kbd{C-x `} (the advantage over a stepping debugger
6135: is that you can step in any direction and you know where the crash has
6136: happened or where the strange data has occurred).
1.5 anton 6137:
1.26 crook 6138: The default actions of @code{~~} clobber the contents of the pictured
6139: numeric output string, so you should not use @code{~~}, e.g., between
6140: @code{<#} and @code{#>}.
1.5 anton 6141:
1.26 crook 6142: doc-~~
6143: doc-printdebugdata
6144: doc-printdebugline
1.5 anton 6145:
1.26 crook 6146: doc-see
6147: doc-marker
1.5 anton 6148:
1.26 crook 6149: Here's an example of using @code{marker} at the start of a source file
6150: that you are debugging; it ensures that you only ever have one copy of
6151: the file's definitions compiled at any time:
1.5 anton 6152:
1.26 crook 6153: @example
6154: [IFDEF] my-code
6155: my-code
6156: [ENDIF]
1.5 anton 6157:
1.26 crook 6158: marker my-code
1.28 crook 6159: init-included-files
1.5 anton 6160:
1.26 crook 6161: \ .. definitions start here
6162: \ .
6163: \ .
6164: \ end
6165: @end example
1.5 anton 6166:
6167:
6168:
1.26 crook 6169: @node Assertions, Singlestep Debugger, Debugging, Programming Tools
6170: @subsection Assertions
6171: @cindex assertions
1.5 anton 6172:
1.26 crook 6173: It is a good idea to make your programs self-checking, especially if you
6174: make an assumption that may become invalid during maintenance (for
6175: example, that a certain field of a data structure is never zero). Gforth
1.29 crook 6176: supports @dfn{assertions} for this purpose. They are used like this:
1.23 crook 6177:
1.26 crook 6178: @example
1.29 crook 6179: assert( @i{flag} )
1.26 crook 6180: @end example
1.23 crook 6181:
1.26 crook 6182: The code between @code{assert(} and @code{)} should compute a flag, that
6183: should be true if everything is alright and false otherwise. It should
6184: not change anything else on the stack. The overall stack effect of the
6185: assertion is @code{( -- )}. E.g.
1.23 crook 6186:
1.26 crook 6187: @example
6188: assert( 1 1 + 2 = ) \ what we learn in school
6189: assert( dup 0<> ) \ assert that the top of stack is not zero
6190: assert( false ) \ this code should not be reached
6191: @end example
1.23 crook 6192:
1.26 crook 6193: The need for assertions is different at different times. During
6194: debugging, we want more checking, in production we sometimes care more
6195: for speed. Therefore, assertions can be turned off, i.e., the assertion
6196: becomes a comment. Depending on the importance of an assertion and the
6197: time it takes to check it, you may want to turn off some assertions and
6198: keep others turned on. Gforth provides several levels of assertions for
6199: this purpose:
1.23 crook 6200:
1.26 crook 6201: doc-assert0(
6202: doc-assert1(
6203: doc-assert2(
6204: doc-assert3(
6205: doc-assert(
6206: doc-)
1.23 crook 6207:
1.26 crook 6208: The variable @code{assert-level} specifies the highest assertions that
6209: are turned on. I.e., at the default @code{assert-level} of one,
6210: @code{assert0(} and @code{assert1(} assertions perform checking, while
6211: @code{assert2(} and @code{assert3(} assertions are treated as comments.
6212:
6213: The value of @code{assert-level} is evaluated at compile-time, not at
6214: run-time. Therefore you cannot turn assertions on or off at run-time;
6215: you have to set the @code{assert-level} appropriately before compiling a
6216: piece of code. You can compile different pieces of code at different
6217: @code{assert-level}s (e.g., a trusted library at level 1 and
6218: newly-written code at level 3).
1.23 crook 6219:
1.26 crook 6220: doc-assert-level
1.23 crook 6221:
1.26 crook 6222: If an assertion fails, a message compatible with Emacs' compilation mode
6223: is produced and the execution is aborted (currently with @code{ABORT"}.
6224: If there is interest, we will introduce a special throw code. But if you
6225: intend to @code{catch} a specific condition, using @code{throw} is
6226: probably more appropriate than an assertion).
1.23 crook 6227:
1.26 crook 6228: Definitions in ANS Forth for these assertion words are provided
6229: in @file{compat/assert.fs}.
1.23 crook 6230:
6231:
1.26 crook 6232: @node Singlestep Debugger, , Assertions, Programming Tools
6233: @subsection Singlestep Debugger
6234: @cindex singlestep Debugger
6235: @cindex debugging Singlestep
6236: @cindex @code{dbg}
6237: @cindex @code{BREAK:}
6238: @cindex @code{BREAK"}
1.23 crook 6239:
1.26 crook 6240: When you create a new word there's often the need to check whether it
6241: behaves correctly or not. You can do this by typing @code{dbg
6242: badword}. A debug session might look like this:
1.23 crook 6243:
1.26 crook 6244: @example
6245: : badword 0 DO i . LOOP ; ok
6246: 2 dbg badword
6247: : badword
6248: Scanning code...
1.23 crook 6249:
1.26 crook 6250: Nesting debugger ready!
1.23 crook 6251:
1.26 crook 6252: 400D4738 8049BC4 0 -> [ 2 ] 00002 00000
6253: 400D4740 8049F68 DO -> [ 0 ]
6254: 400D4744 804A0C8 i -> [ 1 ] 00000
6255: 400D4748 400C5E60 . -> 0 [ 0 ]
6256: 400D474C 8049D0C LOOP -> [ 0 ]
6257: 400D4744 804A0C8 i -> [ 1 ] 00001
6258: 400D4748 400C5E60 . -> 1 [ 0 ]
6259: 400D474C 8049D0C LOOP -> [ 0 ]
6260: 400D4758 804B384 ; -> ok
6261: @end example
1.23 crook 6262:
1.26 crook 6263: Each line displayed is one step. You always have to hit return to
6264: execute the next word that is displayed. If you don't want to execute
6265: the next word in a whole, you have to type @kbd{n} for @code{nest}. Here is
6266: an overview what keys are available:
1.23 crook 6267:
1.26 crook 6268: @table @i
1.23 crook 6269:
1.30 anton 6270: @item @key{RET}
1.26 crook 6271: Next; Execute the next word.
1.23 crook 6272:
1.26 crook 6273: @item n
6274: Nest; Single step through next word.
1.5 anton 6275:
1.26 crook 6276: @item u
6277: Unnest; Stop debugging and execute rest of word. If we got to this word
6278: with nest, continue debugging with the calling word.
1.5 anton 6279:
1.26 crook 6280: @item d
6281: Done; Stop debugging and execute rest.
1.5 anton 6282:
1.26 crook 6283: @item s
6284: Stop; Abort immediately.
1.5 anton 6285:
1.26 crook 6286: @end table
1.5 anton 6287:
1.26 crook 6288: Debugging large application with this mechanism is very difficult, because
6289: you have to nest very deeply into the program before the interesting part
6290: begins. This takes a lot of time.
1.5 anton 6291:
1.26 crook 6292: To do it more directly put a @code{BREAK:} command into your source code.
6293: When program execution reaches @code{BREAK:} the single step debugger is
6294: invoked and you have all the features described above.
1.23 crook 6295:
1.26 crook 6296: If you have more than one part to debug it is useful to know where the
6297: program has stopped at the moment. You can do this by the
6298: @code{BREAK" string"} command. This behaves like @code{BREAK:} except that
6299: string is typed out when the ``breakpoint'' is reached.
6300:
6301: doc-dbg
6302: doc-BREAK:
6303: doc-BREAK"
6304:
6305:
6306: @c -------------------------------------------------------------
6307: @node Assembler and Code Words, Threading Words, Programming Tools, Words
6308: @section Assembler and Code Words
6309: @cindex assembler
6310: @cindex code words
1.5 anton 6311:
1.26 crook 6312: Gforth provides some words for defining primitives (words written in
1.29 crook 6313: machine code), and for defining the machine-code equivalent of
1.26 crook 6314: @code{DOES>}-based defining words. However, the machine-independent
6315: nature of Gforth poses a few problems: First of all, Gforth runs on
6316: several architectures, so it can provide no standard assembler. What's
6317: worse is that the register allocation not only depends on the processor,
6318: but also on the @code{gcc} version and options used.
1.5 anton 6319:
1.29 crook 6320: The words that Gforth offers encapsulate some system dependences (e.g.,
6321: the header structure), so a system-independent assembler may be used in
1.26 crook 6322: Gforth. If you do not have an assembler, you can compile machine code
1.29 crook 6323: directly with @code{,} and @code{c,}@footnote{This isn't portable,
6324: because these words emit stuff in @i{data} space; it works because
6325: Gforth has unified code/data spaces. Assembler isn't likely to be
6326: portable anyway.}.
1.5 anton 6327:
1.26 crook 6328: doc-assembler
6329: doc-code
6330: doc-end-code
6331: doc-;code
6332: doc-flush-icache
1.5 anton 6333:
1.26 crook 6334: If @code{flush-icache} does not work correctly, @code{code} words
6335: etc. will not work (reliably), either.
1.5 anton 6336:
1.29 crook 6337: The typical usage of these @code{code} words can be shown most easily by
6338: analogy to the equivalent high-level defining words:
6339:
6340: @example
6341: : foo code foo
6342: <high-level Forth words> <assembler>
6343: ; end-code
6344:
6345: : bar : bar
6346: <high-level Forth words> <high-level Forth words>
6347: CREATE CREATE
6348: <high-level Forth words> <high-level Forth words>
6349: DOES> ;code
6350: <high-level Forth words> <assembler>
6351: ; end-code
6352: @end example
6353:
1.26 crook 6354: @code{flush-icache} is always present. The other words are rarely used
6355: and reside in @code{code.fs}, which is usually not loaded. You can load
6356: it with @code{require code.fs}.
1.5 anton 6357:
1.26 crook 6358: @cindex registers of the inner interpreter
6359: In the assembly code you will want to refer to the inner interpreter's
6360: registers (e.g., the data stack pointer) and you may want to use other
6361: registers for temporary storage. Unfortunately, the register allocation
6362: is installation-dependent.
1.5 anton 6363:
1.26 crook 6364: The easiest solution is to use explicit register declarations
6365: (@pxref{Explicit Reg Vars, , Variables in Specified Registers, gcc.info,
6366: GNU C Manual}) for all of the inner interpreter's registers: You have to
6367: compile Gforth with @code{-DFORCE_REG} (configure option
6368: @code{--enable-force-reg}) and the appropriate declarations must be
6369: present in the @code{machine.h} file (see @code{mips.h} for an example;
6370: you can find a full list of all declarable register symbols with
6371: @code{grep register engine.c}). If you give explicit registers to all
6372: variables that are declared at the beginning of @code{engine()}, you
6373: should be able to use the other caller-saved registers for temporary
6374: storage. Alternatively, you can use the @code{gcc} option
6375: @code{-ffixed-REG} (@pxref{Code Gen Options, , Options for Code
6376: Generation Conventions, gcc.info, GNU C Manual}) to reserve a register
6377: (however, this restriction on register allocation may slow Gforth
6378: significantly).
1.5 anton 6379:
1.26 crook 6380: If this solution is not viable (e.g., because @code{gcc} does not allow
6381: you to explicitly declare all the registers you need), you have to find
6382: out by looking at the code where the inner interpreter's registers
6383: reside and which registers can be used for temporary storage. You can
6384: get an assembly listing of the engine's code with @code{make engine.s}.
1.5 anton 6385:
1.26 crook 6386: In any case, it is good practice to abstract your assembly code from the
6387: actual register allocation. E.g., if the data stack pointer resides in
6388: register @code{$17}, create an alias for this register called @code{sp},
6389: and use that in your assembly code.
1.5 anton 6390:
1.26 crook 6391: @cindex code words, portable
6392: Another option for implementing normal and defining words efficiently
6393: is to add the desired functionality to the source of Gforth. For normal
6394: words you just have to edit @file{primitives} (@pxref{Automatic
6395: Generation}). Defining words (equivalent to @code{;CODE} words, for fast
6396: defined words) may require changes in @file{engine.c}, @file{kernel.fs},
6397: @file{prims2x.fs}, and possibly @file{cross.fs}.
1.5 anton 6398:
6399:
1.26 crook 6400: @c -------------------------------------------------------------
6401: @node Threading Words, Locals, Assembler and Code Words, Words
6402: @section Threading Words
6403: @cindex threading words
1.5 anton 6404:
1.26 crook 6405: @cindex code address
6406: These words provide access to code addresses and other threading stuff
6407: in Gforth (and, possibly, other interpretive Forths). It more or less
6408: abstracts away the differences between direct and indirect threading
6409: (and, for direct threading, the machine dependences). However, at
6410: present this wordset is still incomplete. It is also pretty low-level;
6411: some day it will hopefully be made unnecessary by an internals wordset
6412: that abstracts implementation details away completely.
1.5 anton 6413:
1.26 crook 6414: doc-threading-method
6415: doc->code-address
6416: doc->does-code
6417: doc-code-address!
6418: doc-does-code!
6419: doc-does-handler!
6420: doc-/does-handler
1.5 anton 6421:
1.26 crook 6422: The code addresses produced by various defining words are produced by
6423: the following words:
1.5 anton 6424:
1.26 crook 6425: doc-docol:
6426: doc-docon:
6427: doc-dovar:
6428: doc-douser:
6429: doc-dodefer:
6430: doc-dofield:
1.5 anton 6431:
1.26 crook 6432: You can recognize words defined by a @code{CREATE}...@code{DOES>} word
6433: with @code{>does-code}. If the word was defined in that way, the value
6434: returned is non-zero and identifies the @code{DOES>} used by the
6435: defining word.
6436: @comment TODO should that be ``identifies the xt of the DOES> ??''
1.5 anton 6437:
1.26 crook 6438: @c -------------------------------------------------------------
6439: @node Locals, Structures, Threading Words, Words
6440: @section Locals
6441: @cindex locals
1.5 anton 6442:
1.26 crook 6443: Local variables can make Forth programming more enjoyable and Forth
6444: programs easier to read. Unfortunately, the locals of ANS Forth are
6445: laden with restrictions. Therefore, we provide not only the ANS Forth
6446: locals wordset, but also our own, more powerful locals wordset (we
6447: implemented the ANS Forth locals wordset through our locals wordset).
1.5 anton 6448:
1.26 crook 6449: The ideas in this section have also been published in the paper
6450: @cite{Automatic Scoping of Local Variables} by M. Anton Ertl, presented
6451: at EuroForth '94; it is available at
6452: @*@url{http://www.complang.tuwien.ac.at/papers/ertl94l.ps.gz}.
1.5 anton 6453:
1.26 crook 6454: @menu
6455: * Gforth locals::
6456: * ANS Forth locals::
6457: @end menu
1.5 anton 6458:
1.26 crook 6459: @node Gforth locals, ANS Forth locals, Locals, Locals
6460: @subsection Gforth locals
6461: @cindex Gforth locals
6462: @cindex locals, Gforth style
1.5 anton 6463:
1.26 crook 6464: Locals can be defined with
1.5 anton 6465:
6466: @example
1.26 crook 6467: @{ local1 local2 ... -- comment @}
6468: @end example
6469: or
6470: @example
6471: @{ local1 local2 ... @}
1.5 anton 6472: @end example
6473:
1.26 crook 6474: E.g.,
1.5 anton 6475: @example
1.26 crook 6476: : max @{ n1 n2 -- n3 @}
6477: n1 n2 > if
6478: n1
6479: else
6480: n2
6481: endif ;
1.5 anton 6482: @end example
6483:
1.26 crook 6484: The similarity of locals definitions with stack comments is intended. A
6485: locals definition often replaces the stack comment of a word. The order
6486: of the locals corresponds to the order in a stack comment and everything
6487: after the @code{--} is really a comment.
1.5 anton 6488:
1.26 crook 6489: This similarity has one disadvantage: It is too easy to confuse locals
6490: declarations with stack comments, causing bugs and making them hard to
6491: find. However, this problem can be avoided by appropriate coding
6492: conventions: Do not use both notations in the same program. If you do,
6493: they should be distinguished using additional means, e.g. by position.
6494:
6495: @cindex types of locals
6496: @cindex locals types
6497: The name of the local may be preceded by a type specifier, e.g.,
6498: @code{F:} for a floating point value:
6499:
6500: @example
6501: : CX* @{ F: Ar F: Ai F: Br F: Bi -- Cr Ci @}
6502: \ complex multiplication
6503: Ar Br f* Ai Bi f* f-
6504: Ar Bi f* Ai Br f* f+ ;
6505: @end example
6506:
6507: @cindex flavours of locals
6508: @cindex locals flavours
6509: @cindex value-flavoured locals
6510: @cindex variable-flavoured locals
6511: Gforth currently supports cells (@code{W:}, @code{W^}), doubles
6512: (@code{D:}, @code{D^}), floats (@code{F:}, @code{F^}) and characters
6513: (@code{C:}, @code{C^}) in two flavours: a value-flavoured local (defined
6514: with @code{W:}, @code{D:} etc.) produces its value and can be changed
6515: with @code{TO}. A variable-flavoured local (defined with @code{W^} etc.)
6516: produces its address (which becomes invalid when the variable's scope is
6517: left). E.g., the standard word @code{emit} can be defined in terms of
6518: @code{type} like this:
1.5 anton 6519:
6520: @example
1.26 crook 6521: : emit @{ C^ char* -- @}
6522: char* 1 type ;
1.5 anton 6523: @end example
6524:
1.26 crook 6525: @cindex default type of locals
6526: @cindex locals, default type
6527: A local without type specifier is a @code{W:} local. Both flavours of
6528: locals are initialized with values from the data or FP stack.
1.5 anton 6529:
1.26 crook 6530: Currently there is no way to define locals with user-defined data
6531: structures, but we are working on it.
1.5 anton 6532:
1.26 crook 6533: Gforth allows defining locals everywhere in a colon definition. This
6534: poses the following questions:
1.5 anton 6535:
1.26 crook 6536: @menu
6537: * Where are locals visible by name?::
6538: * How long do locals live?::
6539: * Programming Style::
6540: * Implementation::
6541: @end menu
1.5 anton 6542:
1.26 crook 6543: @node Where are locals visible by name?, How long do locals live?, Gforth locals, Gforth locals
6544: @subsubsection Where are locals visible by name?
6545: @cindex locals visibility
6546: @cindex visibility of locals
6547: @cindex scope of locals
1.5 anton 6548:
1.26 crook 6549: Basically, the answer is that locals are visible where you would expect
6550: it in block-structured languages, and sometimes a little longer. If you
6551: want to restrict the scope of a local, enclose its definition in
6552: @code{SCOPE}...@code{ENDSCOPE}.
1.5 anton 6553:
1.26 crook 6554: doc-scope
6555: doc-endscope
1.5 anton 6556:
1.26 crook 6557: These words behave like control structure words, so you can use them
6558: with @code{CS-PICK} and @code{CS-ROLL} to restrict the scope in
6559: arbitrary ways.
1.5 anton 6560:
1.26 crook 6561: If you want a more exact answer to the visibility question, here's the
6562: basic principle: A local is visible in all places that can only be
6563: reached through the definition of the local@footnote{In compiler
6564: construction terminology, all places dominated by the definition of the
6565: local.}. In other words, it is not visible in places that can be reached
6566: without going through the definition of the local. E.g., locals defined
6567: in @code{IF}...@code{ENDIF} are visible until the @code{ENDIF}, locals
6568: defined in @code{BEGIN}...@code{UNTIL} are visible after the
6569: @code{UNTIL} (until, e.g., a subsequent @code{ENDSCOPE}).
1.5 anton 6570:
1.26 crook 6571: The reasoning behind this solution is: We want to have the locals
6572: visible as long as it is meaningful. The user can always make the
6573: visibility shorter by using explicit scoping. In a place that can
6574: only be reached through the definition of a local, the meaning of a
6575: local name is clear. In other places it is not: How is the local
6576: initialized at the control flow path that does not contain the
6577: definition? Which local is meant, if the same name is defined twice in
6578: two independent control flow paths?
1.5 anton 6579:
1.26 crook 6580: This should be enough detail for nearly all users, so you can skip the
6581: rest of this section. If you really must know all the gory details and
6582: options, read on.
1.5 anton 6583:
1.26 crook 6584: In order to implement this rule, the compiler has to know which places
6585: are unreachable. It knows this automatically after @code{AHEAD},
6586: @code{AGAIN}, @code{EXIT} and @code{LEAVE}; in other cases (e.g., after
6587: most @code{THROW}s), you can use the word @code{UNREACHABLE} to tell the
6588: compiler that the control flow never reaches that place. If
6589: @code{UNREACHABLE} is not used where it could, the only consequence is
6590: that the visibility of some locals is more limited than the rule above
6591: says. If @code{UNREACHABLE} is used where it should not (i.e., if you
6592: lie to the compiler), buggy code will be produced.
1.5 anton 6593:
1.26 crook 6594: doc-unreachable
1.5 anton 6595:
1.26 crook 6596: Another problem with this rule is that at @code{BEGIN}, the compiler
6597: does not know which locals will be visible on the incoming
6598: back-edge. All problems discussed in the following are due to this
6599: ignorance of the compiler (we discuss the problems using @code{BEGIN}
6600: loops as examples; the discussion also applies to @code{?DO} and other
6601: loops). Perhaps the most insidious example is:
1.5 anton 6602: @example
1.26 crook 6603: AHEAD
6604: BEGIN
6605: x
6606: [ 1 CS-ROLL ] THEN
6607: @{ x @}
6608: ...
6609: UNTIL
6610: @end example
1.5 anton 6611:
1.26 crook 6612: This should be legal according to the visibility rule. The use of
6613: @code{x} can only be reached through the definition; but that appears
6614: textually below the use.
1.5 anton 6615:
1.26 crook 6616: From this example it is clear that the visibility rules cannot be fully
6617: implemented without major headaches. Our implementation treats common
6618: cases as advertised and the exceptions are treated in a safe way: The
6619: compiler makes a reasonable guess about the locals visible after a
6620: @code{BEGIN}; if it is too pessimistic, the
6621: user will get a spurious error about the local not being defined; if the
6622: compiler is too optimistic, it will notice this later and issue a
6623: warning. In the case above the compiler would complain about @code{x}
6624: being undefined at its use. You can see from the obscure examples in
6625: this section that it takes quite unusual control structures to get the
6626: compiler into trouble, and even then it will often do fine.
1.5 anton 6627:
1.26 crook 6628: If the @code{BEGIN} is reachable from above, the most optimistic guess
6629: is that all locals visible before the @code{BEGIN} will also be
6630: visible after the @code{BEGIN}. This guess is valid for all loops that
6631: are entered only through the @code{BEGIN}, in particular, for normal
6632: @code{BEGIN}...@code{WHILE}...@code{REPEAT} and
6633: @code{BEGIN}...@code{UNTIL} loops and it is implemented in our
6634: compiler. When the branch to the @code{BEGIN} is finally generated by
6635: @code{AGAIN} or @code{UNTIL}, the compiler checks the guess and
6636: warns the user if it was too optimistic:
6637: @example
6638: IF
6639: @{ x @}
6640: BEGIN
6641: \ x ?
6642: [ 1 cs-roll ] THEN
6643: ...
6644: UNTIL
1.5 anton 6645: @end example
6646:
1.26 crook 6647: Here, @code{x} lives only until the @code{BEGIN}, but the compiler
6648: optimistically assumes that it lives until the @code{THEN}. It notices
6649: this difference when it compiles the @code{UNTIL} and issues a
6650: warning. The user can avoid the warning, and make sure that @code{x}
6651: is not used in the wrong area by using explicit scoping:
6652: @example
6653: IF
6654: SCOPE
6655: @{ x @}
6656: ENDSCOPE
6657: BEGIN
6658: [ 1 cs-roll ] THEN
6659: ...
6660: UNTIL
6661: @end example
1.5 anton 6662:
1.26 crook 6663: Since the guess is optimistic, there will be no spurious error messages
6664: about undefined locals.
1.5 anton 6665:
1.26 crook 6666: If the @code{BEGIN} is not reachable from above (e.g., after
6667: @code{AHEAD} or @code{EXIT}), the compiler cannot even make an
6668: optimistic guess, as the locals visible after the @code{BEGIN} may be
6669: defined later. Therefore, the compiler assumes that no locals are
6670: visible after the @code{BEGIN}. However, the user can use
6671: @code{ASSUME-LIVE} to make the compiler assume that the same locals are
6672: visible at the BEGIN as at the point where the top control-flow stack
6673: item was created.
1.5 anton 6674:
1.26 crook 6675: doc-assume-live
1.5 anton 6676:
1.26 crook 6677: E.g.,
1.5 anton 6678: @example
1.26 crook 6679: @{ x @}
6680: AHEAD
6681: ASSUME-LIVE
6682: BEGIN
6683: x
6684: [ 1 CS-ROLL ] THEN
6685: ...
6686: UNTIL
1.5 anton 6687: @end example
6688:
1.26 crook 6689: Other cases where the locals are defined before the @code{BEGIN} can be
6690: handled by inserting an appropriate @code{CS-ROLL} before the
6691: @code{ASSUME-LIVE} (and changing the control-flow stack manipulation
6692: behind the @code{ASSUME-LIVE}).
1.5 anton 6693:
1.26 crook 6694: Cases where locals are defined after the @code{BEGIN} (but should be
6695: visible immediately after the @code{BEGIN}) can only be handled by
6696: rearranging the loop. E.g., the ``most insidious'' example above can be
6697: arranged into:
1.5 anton 6698: @example
1.26 crook 6699: BEGIN
6700: @{ x @}
6701: ... 0=
6702: WHILE
6703: x
6704: REPEAT
1.5 anton 6705: @end example
6706:
1.26 crook 6707: @node How long do locals live?, Programming Style, Where are locals visible by name?, Gforth locals
6708: @subsubsection How long do locals live?
6709: @cindex locals lifetime
6710: @cindex lifetime of locals
1.5 anton 6711:
1.26 crook 6712: The right answer for the lifetime question would be: A local lives at
6713: least as long as it can be accessed. For a value-flavoured local this
6714: means: until the end of its visibility. However, a variable-flavoured
6715: local could be accessed through its address far beyond its visibility
6716: scope. Ultimately, this would mean that such locals would have to be
6717: garbage collected. Since this entails un-Forth-like implementation
6718: complexities, I adopted the same cowardly solution as some other
6719: languages (e.g., C): The local lives only as long as it is visible;
6720: afterwards its address is invalid (and programs that access it
6721: afterwards are erroneous).
1.5 anton 6722:
1.26 crook 6723: @node Programming Style, Implementation, How long do locals live?, Gforth locals
6724: @subsubsection Programming Style
6725: @cindex locals programming style
6726: @cindex programming style, locals
1.5 anton 6727:
1.26 crook 6728: The freedom to define locals anywhere has the potential to change
6729: programming styles dramatically. In particular, the need to use the
6730: return stack for intermediate storage vanishes. Moreover, all stack
6731: manipulations (except @code{PICK}s and @code{ROLL}s with run-time
6732: determined arguments) can be eliminated: If the stack items are in the
6733: wrong order, just write a locals definition for all of them; then
6734: write the items in the order you want.
1.5 anton 6735:
1.26 crook 6736: This seems a little far-fetched and eliminating stack manipulations is
6737: unlikely to become a conscious programming objective. Still, the number
6738: of stack manipulations will be reduced dramatically if local variables
6739: are used liberally (e.g., compare @code{max} in @ref{Gforth locals} with
6740: a traditional implementation of @code{max}).
1.5 anton 6741:
1.26 crook 6742: This shows one potential benefit of locals: making Forth programs more
6743: readable. Of course, this benefit will only be realized if the
6744: programmers continue to honour the principle of factoring instead of
6745: using the added latitude to make the words longer.
1.5 anton 6746:
1.26 crook 6747: @cindex single-assignment style for locals
6748: Using @code{TO} can and should be avoided. Without @code{TO},
6749: every value-flavoured local has only a single assignment and many
6750: advantages of functional languages apply to Forth. I.e., programs are
6751: easier to analyse, to optimize and to read: It is clear from the
6752: definition what the local stands for, it does not turn into something
6753: different later.
1.5 anton 6754:
1.26 crook 6755: E.g., a definition using @code{TO} might look like this:
1.5 anton 6756: @example
1.26 crook 6757: : strcmp @{ addr1 u1 addr2 u2 -- n @}
6758: u1 u2 min 0
6759: ?do
6760: addr1 c@@ addr2 c@@ -
6761: ?dup-if
6762: unloop exit
6763: then
6764: addr1 char+ TO addr1
6765: addr2 char+ TO addr2
6766: loop
6767: u1 u2 - ;
1.5 anton 6768: @end example
1.26 crook 6769: Here, @code{TO} is used to update @code{addr1} and @code{addr2} at
6770: every loop iteration. @code{strcmp} is a typical example of the
6771: readability problems of using @code{TO}. When you start reading
6772: @code{strcmp}, you think that @code{addr1} refers to the start of the
6773: string. Only near the end of the loop you realize that it is something
6774: else.
1.5 anton 6775:
1.26 crook 6776: This can be avoided by defining two locals at the start of the loop that
6777: are initialized with the right value for the current iteration.
1.5 anton 6778: @example
1.26 crook 6779: : strcmp @{ addr1 u1 addr2 u2 -- n @}
6780: addr1 addr2
6781: u1 u2 min 0
6782: ?do @{ s1 s2 @}
6783: s1 c@@ s2 c@@ -
6784: ?dup-if
6785: unloop exit
6786: then
6787: s1 char+ s2 char+
6788: loop
6789: 2drop
6790: u1 u2 - ;
1.5 anton 6791: @end example
1.26 crook 6792: Here it is clear from the start that @code{s1} has a different value
6793: in every loop iteration.
1.5 anton 6794:
1.26 crook 6795: @node Implementation, , Programming Style, Gforth locals
6796: @subsubsection Implementation
6797: @cindex locals implementation
6798: @cindex implementation of locals
1.5 anton 6799:
1.26 crook 6800: @cindex locals stack
6801: Gforth uses an extra locals stack. The most compelling reason for
6802: this is that the return stack is not float-aligned; using an extra stack
6803: also eliminates the problems and restrictions of using the return stack
6804: as locals stack. Like the other stacks, the locals stack grows toward
6805: lower addresses. A few primitives allow an efficient implementation:
1.5 anton 6806:
1.26 crook 6807: doc-@local#
6808: doc-f@local#
6809: doc-laddr#
6810: doc-lp+!#
6811: doc-lp!
6812: doc->l
6813: doc-f>l
1.5 anton 6814:
1.26 crook 6815: In addition to these primitives, some specializations of these
6816: primitives for commonly occurring inline arguments are provided for
6817: efficiency reasons, e.g., @code{@@local0} as specialization of
6818: @code{@@local#} for the inline argument 0. The following compiling words
6819: compile the right specialized version, or the general version, as
6820: appropriate:
1.6 pazsan 6821:
1.26 crook 6822: doc-compile-@local
6823: doc-compile-f@local
6824: doc-compile-lp+!
1.12 anton 6825:
1.26 crook 6826: Combinations of conditional branches and @code{lp+!#} like
6827: @code{?branch-lp+!#} (the locals pointer is only changed if the branch
6828: is taken) are provided for efficiency and correctness in loops.
1.6 pazsan 6829:
1.26 crook 6830: A special area in the dictionary space is reserved for keeping the
6831: local variable names. @code{@{} switches the dictionary pointer to this
6832: area and @code{@}} switches it back and generates the locals
6833: initializing code. @code{W:} etc.@ are normal defining words. This
6834: special area is cleared at the start of every colon definition.
1.6 pazsan 6835:
1.26 crook 6836: @cindex word list for defining locals
6837: A special feature of Gforth's dictionary is used to implement the
6838: definition of locals without type specifiers: every word list (aka
6839: vocabulary) has its own methods for searching
6840: etc. (@pxref{Word Lists}). For the present purpose we defined a word list
6841: with a special search method: When it is searched for a word, it
6842: actually creates that word using @code{W:}. @code{@{} changes the search
6843: order to first search the word list containing @code{@}}, @code{W:} etc.,
6844: and then the word list for defining locals without type specifiers.
1.12 anton 6845:
1.26 crook 6846: The lifetime rules support a stack discipline within a colon
6847: definition: The lifetime of a local is either nested with other locals
6848: lifetimes or it does not overlap them.
1.6 pazsan 6849:
1.26 crook 6850: At @code{BEGIN}, @code{IF}, and @code{AHEAD} no code for locals stack
6851: pointer manipulation is generated. Between control structure words
6852: locals definitions can push locals onto the locals stack. @code{AGAIN}
6853: is the simplest of the other three control flow words. It has to
6854: restore the locals stack depth of the corresponding @code{BEGIN}
6855: before branching. The code looks like this:
6856: @format
6857: @code{lp+!#} current-locals-size @minus{} dest-locals-size
6858: @code{branch} <begin>
6859: @end format
1.6 pazsan 6860:
1.26 crook 6861: @code{UNTIL} is a little more complicated: If it branches back, it
6862: must adjust the stack just like @code{AGAIN}. But if it falls through,
6863: the locals stack must not be changed. The compiler generates the
6864: following code:
6865: @format
6866: @code{?branch-lp+!#} <begin> current-locals-size @minus{} dest-locals-size
6867: @end format
6868: The locals stack pointer is only adjusted if the branch is taken.
1.6 pazsan 6869:
1.26 crook 6870: @code{THEN} can produce somewhat inefficient code:
6871: @format
6872: @code{lp+!#} current-locals-size @minus{} orig-locals-size
6873: <orig target>:
6874: @code{lp+!#} orig-locals-size @minus{} new-locals-size
6875: @end format
6876: The second @code{lp+!#} adjusts the locals stack pointer from the
1.29 crook 6877: level at the @i{orig} point to the level after the @code{THEN}. The
1.26 crook 6878: first @code{lp+!#} adjusts the locals stack pointer from the current
6879: level to the level at the orig point, so the complete effect is an
6880: adjustment from the current level to the right level after the
6881: @code{THEN}.
1.6 pazsan 6882:
1.26 crook 6883: @cindex locals information on the control-flow stack
6884: @cindex control-flow stack items, locals information
6885: In a conventional Forth implementation a dest control-flow stack entry
6886: is just the target address and an orig entry is just the address to be
6887: patched. Our locals implementation adds a word list to every orig or dest
6888: item. It is the list of locals visible (or assumed visible) at the point
6889: described by the entry. Our implementation also adds a tag to identify
6890: the kind of entry, in particular to differentiate between live and dead
6891: (reachable and unreachable) orig entries.
1.6 pazsan 6892:
1.26 crook 6893: A few unusual operations have to be performed on locals word lists:
1.6 pazsan 6894:
1.26 crook 6895: doc-common-list
6896: doc-sub-list?
6897: doc-list-size
1.6 pazsan 6898:
1.26 crook 6899: Several features of our locals word list implementation make these
6900: operations easy to implement: The locals word lists are organised as
6901: linked lists; the tails of these lists are shared, if the lists
6902: contain some of the same locals; and the address of a name is greater
6903: than the address of the names behind it in the list.
1.6 pazsan 6904:
1.26 crook 6905: Another important implementation detail is the variable
6906: @code{dead-code}. It is used by @code{BEGIN} and @code{THEN} to
6907: determine if they can be reached directly or only through the branch
6908: that they resolve. @code{dead-code} is set by @code{UNREACHABLE},
6909: @code{AHEAD}, @code{EXIT} etc., and cleared at the start of a colon
6910: definition, by @code{BEGIN} and usually by @code{THEN}.
1.6 pazsan 6911:
1.26 crook 6912: Counted loops are similar to other loops in most respects, but
6913: @code{LEAVE} requires special attention: It performs basically the same
6914: service as @code{AHEAD}, but it does not create a control-flow stack
6915: entry. Therefore the information has to be stored elsewhere;
6916: traditionally, the information was stored in the target fields of the
6917: branches created by the @code{LEAVE}s, by organizing these fields into a
6918: linked list. Unfortunately, this clever trick does not provide enough
6919: space for storing our extended control flow information. Therefore, we
6920: introduce another stack, the leave stack. It contains the control-flow
6921: stack entries for all unresolved @code{LEAVE}s.
1.6 pazsan 6922:
1.26 crook 6923: Local names are kept until the end of the colon definition, even if
6924: they are no longer visible in any control-flow path. In a few cases
6925: this may lead to increased space needs for the locals name area, but
6926: usually less than reclaiming this space would cost in code size.
1.6 pazsan 6927:
6928:
1.26 crook 6929: @node ANS Forth locals, , Gforth locals, Locals
6930: @subsection ANS Forth locals
6931: @cindex locals, ANS Forth style
1.6 pazsan 6932:
1.26 crook 6933: The ANS Forth locals wordset does not define a syntax for locals, but
6934: words that make it possible to define various syntaxes. One of the
6935: possible syntaxes is a subset of the syntax we used in the Gforth locals
6936: wordset, i.e.:
1.6 pazsan 6937:
6938: @example
1.26 crook 6939: @{ local1 local2 ... -- comment @}
1.6 pazsan 6940: @end example
1.23 crook 6941: @noindent
1.26 crook 6942: or
1.6 pazsan 6943: @example
1.26 crook 6944: @{ local1 local2 ... @}
1.6 pazsan 6945: @end example
6946:
1.26 crook 6947: The order of the locals corresponds to the order in a stack comment. The
6948: restrictions are:
1.6 pazsan 6949:
6950: @itemize @bullet
6951: @item
1.26 crook 6952: Locals can only be cell-sized values (no type specifiers are allowed).
1.6 pazsan 6953: @item
1.26 crook 6954: Locals can be defined only outside control structures.
1.6 pazsan 6955: @item
1.26 crook 6956: Locals can interfere with explicit usage of the return stack. For the
6957: exact (and long) rules, see the standard. If you don't use return stack
6958: accessing words in a definition using locals, you will be all right. The
6959: purpose of this rule is to make locals implementation on the return
6960: stack easier.
1.6 pazsan 6961: @item
1.26 crook 6962: The whole definition must be in one line.
6963: @end itemize
1.6 pazsan 6964:
1.26 crook 6965: Locals defined in this way behave like @code{VALUE}s (@xref{Simple
6966: Defining Words}). I.e., they are initialized from the stack. Using their
6967: name produces their value. Their value can be changed using @code{TO}.
1.6 pazsan 6968:
1.26 crook 6969: Since this syntax is supported by Gforth directly, you need not do
6970: anything to use it. If you want to port a program using this syntax to
6971: another ANS Forth system, use @file{compat/anslocal.fs} to implement the
6972: syntax on the other system.
1.6 pazsan 6973:
1.26 crook 6974: Note that a syntax shown in the standard, section A.13 looks
6975: similar, but is quite different in having the order of locals
6976: reversed. Beware!
1.6 pazsan 6977:
1.26 crook 6978: The ANS Forth locals wordset itself consists of a word:
1.6 pazsan 6979:
1.26 crook 6980: doc-(local)
1.6 pazsan 6981:
1.26 crook 6982: The ANS Forth locals extension wordset defines a syntax using @code{locals|}, but it is so
6983: awful that we strongly recommend not to use it. We have implemented this
6984: syntax to make porting to Gforth easy, but do not document it here. The
6985: problem with this syntax is that the locals are defined in an order
6986: reversed with respect to the standard stack comment notation, making
6987: programs harder to read, and easier to misread and miswrite. The only
6988: merit of this syntax is that it is easy to implement using the ANS Forth
6989: locals wordset.
1.7 pazsan 6990:
6991:
1.26 crook 6992: @c ----------------------------------------------------------
6993: @node Structures, Object-oriented Forth, Locals, Words
6994: @section Structures
6995: @cindex structures
6996: @cindex records
1.7 pazsan 6997:
1.26 crook 6998: This section presents the structure package that comes with Gforth. A
6999: version of the package implemented in ANS Forth is available in
7000: @file{compat/struct.fs}. This package was inspired by a posting on
7001: comp.lang.forth in 1989 (unfortunately I don't remember, by whom;
7002: possibly John Hayes). A version of this section has been published in
7003: ???. Marcel Hendrix provided helpful comments.
1.7 pazsan 7004:
1.26 crook 7005: @menu
7006: * Why explicit structure support?::
7007: * Structure Usage::
7008: * Structure Naming Convention::
7009: * Structure Implementation::
7010: * Structure Glossary::
7011: @end menu
1.7 pazsan 7012:
1.26 crook 7013: @node Why explicit structure support?, Structure Usage, Structures, Structures
7014: @subsection Why explicit structure support?
1.7 pazsan 7015:
1.26 crook 7016: @cindex address arithmetic for structures
7017: @cindex structures using address arithmetic
7018: If we want to use a structure containing several fields, we could simply
7019: reserve memory for it, and access the fields using address arithmetic
1.32 anton 7020: (@pxref{Address arithmetic}). As an example, consider a structure with
1.26 crook 7021: the following fields
1.7 pazsan 7022:
1.26 crook 7023: @table @code
7024: @item a
7025: is a float
7026: @item b
7027: is a cell
7028: @item c
7029: is a float
7030: @end table
1.7 pazsan 7031:
1.26 crook 7032: Given the (float-aligned) base address of the structure we get the
7033: address of the field
1.13 pazsan 7034:
1.26 crook 7035: @table @code
7036: @item a
7037: without doing anything further.
7038: @item b
7039: with @code{float+}
7040: @item c
7041: with @code{float+ cell+ faligned}
7042: @end table
1.13 pazsan 7043:
1.26 crook 7044: It is easy to see that this can become quite tiring.
1.13 pazsan 7045:
1.26 crook 7046: Moreover, it is not very readable, because seeing a
7047: @code{cell+} tells us neither which kind of structure is
7048: accessed nor what field is accessed; we have to somehow infer the kind
7049: of structure, and then look up in the documentation, which field of
7050: that structure corresponds to that offset.
1.13 pazsan 7051:
1.26 crook 7052: Finally, this kind of address arithmetic also causes maintenance
7053: troubles: If you add or delete a field somewhere in the middle of the
7054: structure, you have to find and change all computations for the fields
7055: afterwards.
1.13 pazsan 7056:
1.26 crook 7057: So, instead of using @code{cell+} and friends directly, how
7058: about storing the offsets in constants:
1.13 pazsan 7059:
7060: @example
1.26 crook 7061: 0 constant a-offset
7062: 0 float+ constant b-offset
7063: 0 float+ cell+ faligned c-offset
1.13 pazsan 7064: @end example
7065:
1.26 crook 7066: Now we can get the address of field @code{x} with @code{x-offset
7067: +}. This is much better in all respects. Of course, you still
7068: have to change all later offset definitions if you add a field. You can
7069: fix this by declaring the offsets in the following way:
1.13 pazsan 7070:
7071: @example
1.26 crook 7072: 0 constant a-offset
7073: a-offset float+ constant b-offset
7074: b-offset cell+ faligned constant c-offset
1.13 pazsan 7075: @end example
7076:
1.26 crook 7077: Since we always use the offsets with @code{+}, we could use a defining
7078: word @code{cfield} that includes the @code{+} in the action of the
7079: defined word:
1.8 pazsan 7080:
7081: @example
1.26 crook 7082: : cfield ( n "name" -- )
7083: create ,
7084: does> ( name execution: addr1 -- addr2 )
7085: @@ + ;
1.13 pazsan 7086:
1.26 crook 7087: 0 cfield a
7088: 0 a float+ cfield b
7089: 0 b cell+ faligned cfield c
1.13 pazsan 7090: @end example
7091:
1.26 crook 7092: Instead of @code{x-offset +}, we now simply write @code{x}.
7093:
7094: The structure field words now can be used quite nicely. However,
7095: their definition is still a bit cumbersome: We have to repeat the
7096: name, the information about size and alignment is distributed before
7097: and after the field definitions etc. The structure package presented
7098: here addresses these problems.
7099:
7100: @node Structure Usage, Structure Naming Convention, Why explicit structure support?, Structures
7101: @subsection Structure Usage
7102: @cindex structure usage
1.13 pazsan 7103:
1.26 crook 7104: @cindex @code{field} usage
7105: @cindex @code{struct} usage
7106: @cindex @code{end-struct} usage
7107: You can define a structure for a (data-less) linked list with:
1.13 pazsan 7108: @example
1.26 crook 7109: struct
7110: cell% field list-next
7111: end-struct list%
1.13 pazsan 7112: @end example
7113:
1.26 crook 7114: With the address of the list node on the stack, you can compute the
7115: address of the field that contains the address of the next node with
7116: @code{list-next}. E.g., you can determine the length of a list
7117: with:
1.13 pazsan 7118:
7119: @example
1.26 crook 7120: : list-length ( list -- n )
7121: \ "list" is a pointer to the first element of a linked list
7122: \ "n" is the length of the list
7123: 0 BEGIN ( list1 n1 )
7124: over
7125: WHILE ( list1 n1 )
7126: 1+ swap list-next @@ swap
7127: REPEAT
7128: nip ;
1.13 pazsan 7129: @end example
7130:
1.26 crook 7131: You can reserve memory for a list node in the dictionary with
7132: @code{list% %allot}, which leaves the address of the list node on the
7133: stack. For the equivalent allocation on the heap you can use @code{list%
7134: %alloc} (or, for an @code{allocate}-like stack effect (i.e., with ior),
7135: use @code{list% %allocate}). You can get the the size of a list
7136: node with @code{list% %size} and its alignment with @code{list%
7137: %alignment}.
1.13 pazsan 7138:
1.26 crook 7139: Note that in ANS Forth the body of a @code{create}d word is
7140: @code{aligned} but not necessarily @code{faligned};
7141: therefore, if you do a:
1.13 pazsan 7142: @example
1.26 crook 7143: create @emph{name} foo% %allot
1.8 pazsan 7144: @end example
7145:
1.26 crook 7146: @noindent
7147: then the memory alloted for @code{foo%} is
7148: guaranteed to start at the body of @code{@emph{name}} only if
7149: @code{foo%} contains only character, cell and double fields.
1.20 pazsan 7150:
1.26 crook 7151: @cindex strcutures containing structures
7152: You can include a structure @code{foo%} as a field of
7153: another structure, like this:
1.20 pazsan 7154: @example
1.26 crook 7155: struct
7156: ...
7157: foo% field ...
7158: ...
7159: end-struct ...
1.20 pazsan 7160: @end example
7161:
1.26 crook 7162: @cindex structure extension
7163: @cindex extended records
7164: Instead of starting with an empty structure, you can extend an
7165: existing structure. E.g., a plain linked list without data, as defined
7166: above, is hardly useful; You can extend it to a linked list of integers,
7167: like this:@footnote{This feature is also known as @emph{extended
7168: records}. It is the main innovation in the Oberon language; in other
7169: words, adding this feature to Modula-2 led Wirth to create a new
7170: language, write a new compiler etc. Adding this feature to Forth just
7171: required a few lines of code.}
1.20 pazsan 7172:
7173: @example
1.26 crook 7174: list%
7175: cell% field intlist-int
7176: end-struct intlist%
1.20 pazsan 7177: @end example
7178:
1.26 crook 7179: @code{intlist%} is a structure with two fields:
7180: @code{list-next} and @code{intlist-int}.
1.20 pazsan 7181:
1.26 crook 7182: @cindex structures containing arrays
7183: You can specify an array type containing @emph{n} elements of
7184: type @code{foo%} like this:
1.20 pazsan 7185:
7186: @example
1.26 crook 7187: foo% @emph{n} *
1.20 pazsan 7188: @end example
7189:
1.26 crook 7190: You can use this array type in any place where you can use a normal
7191: type, e.g., when defining a @code{field}, or with
7192: @code{%allot}.
1.20 pazsan 7193:
1.26 crook 7194: @cindex first field optimization
7195: The first field is at the base address of a structure and the word
7196: for this field (e.g., @code{list-next}) actually does not change
7197: the address on the stack. You may be tempted to leave it away in the
7198: interest of run-time and space efficiency. This is not necessary,
7199: because the structure package optimizes this case and compiling such
7200: words does not generate any code. So, in the interest of readability
7201: and maintainability you should include the word for the field when
7202: accessing the field.
1.20 pazsan 7203:
1.26 crook 7204: @node Structure Naming Convention, Structure Implementation, Structure Usage, Structures
7205: @subsection Structure Naming Convention
7206: @cindex structure naming convention
1.20 pazsan 7207:
1.26 crook 7208: The field names that come to (my) mind are often quite generic, and,
7209: if used, would cause frequent name clashes. E.g., many structures
7210: probably contain a @code{counter} field. The structure names
7211: that come to (my) mind are often also the logical choice for the names
7212: of words that create such a structure.
1.20 pazsan 7213:
1.26 crook 7214: Therefore, I have adopted the following naming conventions:
1.20 pazsan 7215:
1.26 crook 7216: @itemize @bullet
7217: @cindex field naming convention
7218: @item
7219: The names of fields are of the form
7220: @code{@emph{struct}-@emph{field}}, where
7221: @code{@emph{struct}} is the basic name of the structure, and
7222: @code{@emph{field}} is the basic name of the field. You can
7223: think of field words as converting the (address of the)
7224: structure into the (address of the) field.
1.20 pazsan 7225:
1.26 crook 7226: @cindex structure naming convention
7227: @item
7228: The names of structures are of the form
7229: @code{@emph{struct}%}, where
7230: @code{@emph{struct}} is the basic name of the structure.
7231: @end itemize
1.20 pazsan 7232:
1.26 crook 7233: This naming convention does not work that well for fields of extended
7234: structures; e.g., the integer list structure has a field
7235: @code{intlist-int}, but has @code{list-next}, not
7236: @code{intlist-next}.
1.20 pazsan 7237:
1.26 crook 7238: @node Structure Implementation, Structure Glossary, Structure Naming Convention, Structures
7239: @subsection Structure Implementation
7240: @cindex structure implementation
7241: @cindex implementation of structures
1.20 pazsan 7242:
1.26 crook 7243: The central idea in the implementation is to pass the data about the
7244: structure being built on the stack, not in some global
7245: variable. Everything else falls into place naturally once this design
7246: decision is made.
1.20 pazsan 7247:
1.26 crook 7248: The type description on the stack is of the form @emph{align
7249: size}. Keeping the size on the top-of-stack makes dealing with arrays
7250: very simple.
1.20 pazsan 7251:
1.26 crook 7252: @code{field} is a defining word that uses @code{Create}
7253: and @code{DOES>}. The body of the field contains the offset
7254: of the field, and the normal @code{DOES>} action is simply:
1.20 pazsan 7255:
7256: @example
1.26 crook 7257: @ +
1.20 pazsan 7258: @end example
7259:
1.23 crook 7260: @noindent
1.26 crook 7261: i.e., add the offset to the address, giving the stack effect
1.29 crook 7262: @i{addr1 -- addr2} for a field.
1.20 pazsan 7263:
1.26 crook 7264: @cindex first field optimization, implementation
7265: This simple structure is slightly complicated by the optimization
7266: for fields with offset 0, which requires a different
7267: @code{DOES>}-part (because we cannot rely on there being
7268: something on the stack if such a field is invoked during
7269: compilation). Therefore, we put the different @code{DOES>}-parts
7270: in separate words, and decide which one to invoke based on the
7271: offset. For a zero offset, the field is basically a noop; it is
7272: immediate, and therefore no code is generated when it is compiled.
1.20 pazsan 7273:
1.26 crook 7274: @node Structure Glossary, , Structure Implementation, Structures
7275: @subsection Structure Glossary
7276: @cindex structure glossary
1.20 pazsan 7277:
1.26 crook 7278: doc-%align
7279: doc-%alignment
7280: doc-%alloc
7281: doc-%allocate
7282: doc-%allot
7283: doc-cell%
7284: doc-char%
7285: doc-dfloat%
7286: doc-double%
7287: doc-end-struct
7288: doc-field
7289: doc-float%
7290: doc-naligned
7291: doc-sfloat%
7292: doc-%size
7293: doc-struct
1.23 crook 7294:
1.26 crook 7295: @c -------------------------------------------------------------
7296: @node Object-oriented Forth, Passing Commands to the OS, Structures, Words
7297: @section Object-oriented Forth
1.20 pazsan 7298:
1.26 crook 7299: Gforth comes with three packages for object-oriented programming:
7300: @file{objects.fs}, @file{oof.fs}, and @file{mini-oof.fs}; none of them
7301: is preloaded, so you have to @code{include} them before use. The most
7302: important differences between these packages (and others) are discussed
7303: in @ref{Comparison with other object models}. All packages are written
7304: in ANS Forth and can be used with any other ANS Forth.
1.20 pazsan 7305:
1.26 crook 7306: @menu
7307: * Why object-oriented programming?::
7308: * Object-Oriented Terminology::
7309: * Objects::
7310: * OOF::
7311: * Mini-OOF::
7312: * Comparison with other object models::
7313: @end menu
1.20 pazsan 7314:
1.23 crook 7315:
1.26 crook 7316: @node Why object-oriented programming?, Object-Oriented Terminology, , Object-oriented Forth
7317: @subsubsection Why object-oriented programming?
7318: @cindex object-oriented programming motivation
7319: @cindex motivation for object-oriented programming
1.23 crook 7320:
1.26 crook 7321: Often we have to deal with several data structures (@emph{objects}),
7322: that have to be treated similarly in some respects, but differently in
7323: others. Graphical objects are the textbook example: circles, triangles,
7324: dinosaurs, icons, and others, and we may want to add more during program
7325: development. We want to apply some operations to any graphical object,
7326: e.g., @code{draw} for displaying it on the screen. However, @code{draw}
7327: has to do something different for every kind of object.
7328: @comment TODO add some other operations eg perimeter, area
7329: @comment and tie in to concrete examples later..
1.23 crook 7330:
1.26 crook 7331: We could implement @code{draw} as a big @code{CASE}
7332: control structure that executes the appropriate code depending on the
7333: kind of object to be drawn. This would be not be very elegant, and,
7334: moreover, we would have to change @code{draw} every time we add
7335: a new kind of graphical object (say, a spaceship).
1.23 crook 7336:
1.26 crook 7337: What we would rather do is: When defining spaceships, we would tell
7338: the system: ``Here's how you @code{draw} a spaceship; you figure
7339: out the rest''.
1.23 crook 7340:
1.26 crook 7341: This is the problem that all systems solve that (rightfully) call
7342: themselves object-oriented; the object-oriented packages presented here
7343: solve this problem (and not much else).
7344: @comment TODO ?list properties of oo systems.. oo vs o-based?
1.23 crook 7345:
1.26 crook 7346: @node Object-Oriented Terminology, Objects, Why object-oriented programming?, Object-oriented Forth
7347: @subsubsection Object-Oriented Terminology
7348: @cindex object-oriented terminology
7349: @cindex terminology for object-oriented programming
1.23 crook 7350:
1.26 crook 7351: This section is mainly for reference, so you don't have to understand
7352: all of it right away. The terminology is mainly Smalltalk-inspired. In
7353: short:
1.23 crook 7354:
1.26 crook 7355: @table @emph
7356: @cindex class
7357: @item class
7358: a data structure definition with some extras.
1.23 crook 7359:
1.26 crook 7360: @cindex object
7361: @item object
7362: an instance of the data structure described by the class definition.
1.23 crook 7363:
1.26 crook 7364: @cindex instance variables
7365: @item instance variables
7366: fields of the data structure.
1.23 crook 7367:
1.26 crook 7368: @cindex selector
7369: @cindex method selector
7370: @cindex virtual function
7371: @item selector
7372: (or @emph{method selector}) a word (e.g.,
7373: @code{draw}) that performs an operation on a variety of data
7374: structures (classes). A selector describes @emph{what} operation to
7375: perform. In C++ terminology: a (pure) virtual function.
1.23 crook 7376:
1.26 crook 7377: @cindex method
7378: @item method
7379: the concrete definition that performs the operation
7380: described by the selector for a specific class. A method specifies
7381: @emph{how} the operation is performed for a specific class.
1.23 crook 7382:
1.26 crook 7383: @cindex selector invocation
7384: @cindex message send
7385: @cindex invoking a selector
7386: @item selector invocation
7387: a call of a selector. One argument of the call (the TOS (top-of-stack))
7388: is used for determining which method is used. In Smalltalk terminology:
7389: a message (consisting of the selector and the other arguments) is sent
7390: to the object.
1.1 anton 7391:
1.26 crook 7392: @cindex receiving object
7393: @item receiving object
7394: the object used for determining the method executed by a selector
7395: invocation. In the @file{objects.fs} model, it is the object that is on
7396: the TOS when the selector is invoked. (@emph{Receiving} comes from
7397: the Smalltalk @emph{message} terminology.)
1.1 anton 7398:
1.26 crook 7399: @cindex child class
7400: @cindex parent class
7401: @cindex inheritance
7402: @item child class
7403: a class that has (@emph{inherits}) all properties (instance variables,
7404: selectors, methods) from a @emph{parent class}. In Smalltalk
7405: terminology: The subclass inherits from the superclass. In C++
7406: terminology: The derived class inherits from the base class.
1.1 anton 7407:
1.26 crook 7408: @end table
1.21 crook 7409:
1.26 crook 7410: @c If you wonder about the message sending terminology, it comes from
7411: @c a time when each object had it's own task and objects communicated via
7412: @c message passing; eventually the Smalltalk developers realized that
7413: @c they can do most things through simple (indirect) calls. They kept the
7414: @c terminology.
1.1 anton 7415:
7416:
1.26 crook 7417: @node Objects, OOF, Object-Oriented Terminology, Object-oriented Forth
7418: @subsection The @file{objects.fs} model
7419: @cindex objects
7420: @cindex object-oriented programming
1.1 anton 7421:
1.26 crook 7422: @cindex @file{objects.fs}
7423: @cindex @file{oof.fs}
1.1 anton 7424:
1.37 anton 7425: This section describes the @file{objects.fs} package. This material also
7426: has been published in @cite{Yet Another Forth Objects Package} by Anton
7427: Ertl and appeared in Forth Dimensions 19(2), pages 37--43
7428: (@url{http://www.complang.tuwien.ac.at/forth/objects/objects.html}).
1.26 crook 7429: @c McKewan's and Zsoter's packages
1.1 anton 7430:
1.26 crook 7431: This section assumes that you have read @ref{Structures}.
1.1 anton 7432:
1.26 crook 7433: The techniques on which this model is based have been used to implement
7434: the parser generator, Gray, and have also been used in Gforth for
7435: implementing the various flavours of word lists (hashed or not,
7436: case-sensitive or not, special-purpose word lists for locals etc.).
1.1 anton 7437:
7438:
1.26 crook 7439: @menu
7440: * Properties of the Objects model::
7441: * Basic Objects Usage::
1.37 anton 7442: * The Objects base class::
1.26 crook 7443: * Creating objects::
7444: * Object-Oriented Programming Style::
7445: * Class Binding::
7446: * Method conveniences::
7447: * Classes and Scoping::
1.37 anton 7448: * Dividing classes::
1.26 crook 7449: * Object Interfaces::
7450: * Objects Implementation::
7451: * Objects Glossary::
7452: @end menu
1.1 anton 7453:
1.26 crook 7454: Marcel Hendrix provided helpful comments on this section. Andras Zsoter
7455: and Bernd Paysan helped me with the related works section.
1.1 anton 7456:
1.26 crook 7457: @node Properties of the Objects model, Basic Objects Usage, Objects, Objects
7458: @subsubsection Properties of the @file{objects.fs} model
7459: @cindex @file{objects.fs} properties
1.1 anton 7460:
1.26 crook 7461: @itemize @bullet
7462: @item
7463: It is straightforward to pass objects on the stack. Passing
7464: selectors on the stack is a little less convenient, but possible.
1.1 anton 7465:
1.26 crook 7466: @item
7467: Objects are just data structures in memory, and are referenced by their
7468: address. You can create words for objects with normal defining words
7469: like @code{constant}. Likewise, there is no difference between instance
7470: variables that contain objects and those that contain other data.
1.1 anton 7471:
1.26 crook 7472: @item
7473: Late binding is efficient and easy to use.
1.21 crook 7474:
1.26 crook 7475: @item
7476: It avoids parsing, and thus avoids problems with state-smartness
7477: and reduced extensibility; for convenience there are a few parsing
7478: words, but they have non-parsing counterparts. There are also a few
7479: defining words that parse. This is hard to avoid, because all standard
7480: defining words parse (except @code{:noname}); however, such
7481: words are not as bad as many other parsing words, because they are not
7482: state-smart.
1.21 crook 7483:
1.26 crook 7484: @item
7485: It does not try to incorporate everything. It does a few things and does
7486: them well (IMO). In particular, this model was not designed to support
7487: information hiding (although it has features that may help); you can use
7488: a separate package for achieving this.
1.21 crook 7489:
1.26 crook 7490: @item
7491: It is layered; you don't have to learn and use all features to use this
7492: model. Only a few features are necessary (@xref{Basic Objects Usage},
7493: @xref{The Objects base class}, @xref{Creating objects}.), the others
7494: are optional and independent of each other.
1.21 crook 7495:
1.26 crook 7496: @item
7497: An implementation in ANS Forth is available.
1.21 crook 7498:
1.26 crook 7499: @end itemize
1.21 crook 7500:
7501:
1.26 crook 7502: @node Basic Objects Usage, The Objects base class, Properties of the Objects model, Objects
7503: @subsubsection Basic @file{objects.fs} Usage
7504: @cindex basic objects usage
7505: @cindex objects, basic usage
1.21 crook 7506:
1.26 crook 7507: You can define a class for graphical objects like this:
1.21 crook 7508:
1.26 crook 7509: @cindex @code{class} usage
7510: @cindex @code{end-class} usage
7511: @cindex @code{selector} usage
7512: @example
7513: object class \ "object" is the parent class
7514: selector draw ( x y graphical -- )
7515: end-class graphical
7516: @end example
1.21 crook 7517:
1.26 crook 7518: This code defines a class @code{graphical} with an
7519: operation @code{draw}. We can perform the operation
7520: @code{draw} on any @code{graphical} object, e.g.:
1.21 crook 7521:
1.26 crook 7522: @example
7523: 100 100 t-rex draw
7524: @end example
1.21 crook 7525:
1.26 crook 7526: @noindent
7527: where @code{t-rex} is a word (say, a constant) that produces a
7528: graphical object.
1.21 crook 7529:
1.29 crook 7530: @comment TODO add a 2nd operation eg perimeter.. and use for
1.26 crook 7531: @comment a concrete example
1.21 crook 7532:
1.26 crook 7533: @cindex abstract class
7534: How do we create a graphical object? With the present definitions,
7535: we cannot create a useful graphical object. The class
7536: @code{graphical} describes graphical objects in general, but not
7537: any concrete graphical object type (C++ users would call it an
7538: @emph{abstract class}); e.g., there is no method for the selector
7539: @code{draw} in the class @code{graphical}.
1.21 crook 7540:
1.26 crook 7541: For concrete graphical objects, we define child classes of the
7542: class @code{graphical}, e.g.:
1.21 crook 7543:
1.26 crook 7544: @cindex @code{overrides} usage
7545: @cindex @code{field} usage in class definition
7546: @example
7547: graphical class \ "graphical" is the parent class
7548: cell% field circle-radius
1.21 crook 7549:
1.26 crook 7550: :noname ( x y circle -- )
7551: circle-radius @@ draw-circle ;
7552: overrides draw
1.21 crook 7553:
1.26 crook 7554: :noname ( n-radius circle -- )
7555: circle-radius ! ;
7556: overrides construct
1.21 crook 7557:
1.26 crook 7558: end-class circle
1.21 crook 7559: @end example
7560:
1.26 crook 7561: Here we define a class @code{circle} as a child of @code{graphical},
7562: with field @code{circle-radius} (which behaves just like a field
7563: (@pxref{Structures}); it defines (using @code{overrides}) new methods
7564: for the selectors @code{draw} and @code{construct} (@code{construct} is
7565: defined in @code{object}, the parent class of @code{graphical}).
1.21 crook 7566:
1.26 crook 7567: Now we can create a circle on the heap (i.e.,
7568: @code{allocate}d memory) with:
1.21 crook 7569:
1.26 crook 7570: @cindex @code{heap-new} usage
1.21 crook 7571: @example
1.26 crook 7572: 50 circle heap-new constant my-circle
7573: @end example
1.21 crook 7574:
1.26 crook 7575: @noindent
7576: @code{heap-new} invokes @code{construct}, thus
7577: initializing the field @code{circle-radius} with 50. We can draw
7578: this new circle at (100,100) with:
1.21 crook 7579:
1.26 crook 7580: @example
7581: 100 100 my-circle draw
1.21 crook 7582: @end example
7583:
1.26 crook 7584: @cindex selector invocation, restrictions
7585: @cindex class definition, restrictions
7586: Note: You can only invoke a selector if the object on the TOS
7587: (the receiving object) belongs to the class where the selector was
7588: defined or one of its descendents; e.g., you can invoke
7589: @code{draw} only for objects belonging to @code{graphical}
7590: or its descendents (e.g., @code{circle}). Immediately before
7591: @code{end-class}, the search order has to be the same as
7592: immediately after @code{class}.
1.21 crook 7593:
1.26 crook 7594: @node The Objects base class, Creating objects, Basic Objects Usage, Objects
7595: @subsubsection The @file{object.fs} base class
7596: @cindex @code{object} class
1.21 crook 7597:
1.26 crook 7598: When you define a class, you have to specify a parent class. So how do
7599: you start defining classes? There is one class available from the start:
7600: @code{object}. It is ancestor for all classes and so is the
7601: only class that has no parent. It has two selectors: @code{construct}
7602: and @code{print}.
1.21 crook 7603:
1.26 crook 7604: @node Creating objects, Object-Oriented Programming Style, The Objects base class, Objects
7605: @subsubsection Creating objects
7606: @cindex creating objects
7607: @cindex object creation
7608: @cindex object allocation options
1.21 crook 7609:
1.26 crook 7610: @cindex @code{heap-new} discussion
7611: @cindex @code{dict-new} discussion
7612: @cindex @code{construct} discussion
7613: You can create and initialize an object of a class on the heap with
7614: @code{heap-new} ( ... class -- object ) and in the dictionary
7615: (allocation with @code{allot}) with @code{dict-new} (
7616: ... class -- object ). Both words invoke @code{construct}, which
7617: consumes the stack items indicated by "..." above.
1.21 crook 7618:
1.26 crook 7619: @cindex @code{init-object} discussion
7620: @cindex @code{class-inst-size} discussion
7621: If you want to allocate memory for an object yourself, you can get its
7622: alignment and size with @code{class-inst-size 2@@} ( class --
7623: align size ). Once you have memory for an object, you can initialize
7624: it with @code{init-object} ( ... class object -- );
7625: @code{construct} does only a part of the necessary work.
1.21 crook 7626:
1.26 crook 7627: @node Object-Oriented Programming Style, Class Binding, Creating objects, Objects
7628: @subsubsection Object-Oriented Programming Style
7629: @cindex object-oriented programming style
1.21 crook 7630:
1.26 crook 7631: This section is not exhaustive.
1.1 anton 7632:
1.26 crook 7633: @cindex stack effects of selectors
7634: @cindex selectors and stack effects
7635: In general, it is a good idea to ensure that all methods for the
7636: same selector have the same stack effect: when you invoke a selector,
7637: you often have no idea which method will be invoked, so, unless all
7638: methods have the same stack effect, you will not know the stack effect
7639: of the selector invocation.
1.21 crook 7640:
1.26 crook 7641: One exception to this rule is methods for the selector
7642: @code{construct}. We know which method is invoked, because we
7643: specify the class to be constructed at the same place. Actually, I
7644: defined @code{construct} as a selector only to give the users a
7645: convenient way to specify initialization. The way it is used, a
7646: mechanism different from selector invocation would be more natural
7647: (but probably would take more code and more space to explain).
1.21 crook 7648:
1.26 crook 7649: @node Class Binding, Method conveniences, Object-Oriented Programming Style, Objects
7650: @subsubsection Class Binding
7651: @cindex class binding
7652: @cindex early binding
1.21 crook 7653:
1.26 crook 7654: @cindex late binding
7655: Normal selector invocations determine the method at run-time depending
7656: on the class of the receiving object. This run-time selection is called
1.29 crook 7657: @i{late binding}.
1.21 crook 7658:
1.26 crook 7659: Sometimes it's preferable to invoke a different method. For example,
7660: you might want to use the simple method for @code{print}ing
7661: @code{object}s instead of the possibly long-winded @code{print} method
7662: of the receiver class. You can achieve this by replacing the invocation
7663: of @code{print} with:
1.21 crook 7664:
1.26 crook 7665: @cindex @code{[bind]} usage
7666: @example
7667: [bind] object print
1.21 crook 7668: @end example
7669:
1.26 crook 7670: @noindent
7671: in compiled code or:
1.21 crook 7672:
1.26 crook 7673: @cindex @code{bind} usage
1.21 crook 7674: @example
1.26 crook 7675: bind object print
1.21 crook 7676: @end example
7677:
1.26 crook 7678: @cindex class binding, alternative to
7679: @noindent
7680: in interpreted code. Alternatively, you can define the method with a
7681: name (e.g., @code{print-object}), and then invoke it through the
7682: name. Class binding is just a (often more convenient) way to achieve
7683: the same effect; it avoids name clutter and allows you to invoke
7684: methods directly without naming them first.
7685:
7686: @cindex superclass binding
7687: @cindex parent class binding
7688: A frequent use of class binding is this: When we define a method
7689: for a selector, we often want the method to do what the selector does
7690: in the parent class, and a little more. There is a special word for
7691: this purpose: @code{[parent]}; @code{[parent]
7692: @emph{selector}} is equivalent to @code{[bind] @emph{parent
7693: selector}}, where @code{@emph{parent}} is the parent
7694: class of the current class. E.g., a method definition might look like:
1.21 crook 7695:
1.26 crook 7696: @cindex @code{[parent]} usage
1.21 crook 7697: @example
1.26 crook 7698: :noname
7699: dup [parent] foo \ do parent's foo on the receiving object
7700: ... \ do some more
7701: ; overrides foo
1.21 crook 7702: @end example
7703:
1.26 crook 7704: @cindex class binding as optimization
7705: In @cite{Object-oriented programming in ANS Forth} (Forth Dimensions,
7706: March 1997), Andrew McKewan presents class binding as an optimization
7707: technique. I recommend not using it for this purpose unless you are in
7708: an emergency. Late binding is pretty fast with this model anyway, so the
7709: benefit of using class binding is small; the cost of using class binding
7710: where it is not appropriate is reduced maintainability.
1.21 crook 7711:
1.26 crook 7712: While we are at programming style questions: You should bind
7713: selectors only to ancestor classes of the receiving object. E.g., say,
7714: you know that the receiving object is of class @code{foo} or its
7715: descendents; then you should bind only to @code{foo} and its
7716: ancestors.
1.21 crook 7717:
1.26 crook 7718: @node Method conveniences, Classes and Scoping, Class Binding, Objects
7719: @subsubsection Method conveniences
7720: @cindex method conveniences
1.1 anton 7721:
1.26 crook 7722: In a method you usually access the receiving object pretty often. If
7723: you define the method as a plain colon definition (e.g., with
7724: @code{:noname}), you may have to do a lot of stack
7725: gymnastics. To avoid this, you can define the method with @code{m:
7726: ... ;m}. E.g., you could define the method for
7727: @code{draw}ing a @code{circle} with
1.20 pazsan 7728:
1.26 crook 7729: @cindex @code{this} usage
7730: @cindex @code{m:} usage
7731: @cindex @code{;m} usage
7732: @example
7733: m: ( x y circle -- )
7734: ( x y ) this circle-radius @@ draw-circle ;m
7735: @end example
1.20 pazsan 7736:
1.26 crook 7737: @cindex @code{exit} in @code{m: ... ;m}
7738: @cindex @code{exitm} discussion
7739: @cindex @code{catch} in @code{m: ... ;m}
7740: When this method is executed, the receiver object is removed from the
7741: stack; you can access it with @code{this} (admittedly, in this
7742: example the use of @code{m: ... ;m} offers no advantage). Note
7743: that I specify the stack effect for the whole method (i.e. including
7744: the receiver object), not just for the code between @code{m:}
7745: and @code{;m}. You cannot use @code{exit} in
7746: @code{m:...;m}; instead, use
7747: @code{exitm}.@footnote{Moreover, for any word that calls
7748: @code{catch} and was defined before loading
7749: @code{objects.fs}, you have to redefine it like I redefined
7750: @code{catch}: @code{: catch this >r catch r> to-this ;}}
1.20 pazsan 7751:
1.26 crook 7752: @cindex @code{inst-var} usage
7753: You will frequently use sequences of the form @code{this
7754: @emph{field}} (in the example above: @code{this
7755: circle-radius}). If you use the field only in this way, you can
7756: define it with @code{inst-var} and eliminate the
7757: @code{this} before the field name. E.g., the @code{circle}
7758: class above could also be defined with:
1.20 pazsan 7759:
1.26 crook 7760: @example
7761: graphical class
7762: cell% inst-var radius
1.20 pazsan 7763:
1.26 crook 7764: m: ( x y circle -- )
7765: radius @@ draw-circle ;m
7766: overrides draw
1.20 pazsan 7767:
1.26 crook 7768: m: ( n-radius circle -- )
7769: radius ! ;m
7770: overrides construct
1.12 anton 7771:
1.26 crook 7772: end-class circle
7773: @end example
1.12 anton 7774:
1.26 crook 7775: @code{radius} can only be used in @code{circle} and its
7776: descendent classes and inside @code{m:...;m}.
1.12 anton 7777:
1.26 crook 7778: @cindex @code{inst-value} usage
7779: You can also define fields with @code{inst-value}, which is
7780: to @code{inst-var} what @code{value} is to
7781: @code{variable}. You can change the value of such a field with
7782: @code{[to-inst]}. E.g., we could also define the class
7783: @code{circle} like this:
1.12 anton 7784:
1.26 crook 7785: @example
7786: graphical class
7787: inst-value radius
1.12 anton 7788:
1.26 crook 7789: m: ( x y circle -- )
7790: radius draw-circle ;m
7791: overrides draw
1.12 anton 7792:
1.26 crook 7793: m: ( n-radius circle -- )
7794: [to-inst] radius ;m
7795: overrides construct
1.21 crook 7796:
1.26 crook 7797: end-class circle
1.12 anton 7798: @end example
7799:
1.38 ! anton 7800: Finally, you can define named methods with @code{:m}. One use of this
! 7801: feature is the definition of words that occur only in one class and are
! 7802: not intended to be overridden, but which still need method context
! 7803: (e.g., for accessing @code{inst-var}s). Another use is for methods that
! 7804: would be bound frequently, if defined anonymously.
! 7805:
1.12 anton 7806:
1.37 anton 7807: @node Classes and Scoping, Dividing classes, Method conveniences, Objects
1.26 crook 7808: @subsubsection Classes and Scoping
7809: @cindex classes and scoping
7810: @cindex scoping and classes
1.12 anton 7811:
1.26 crook 7812: Inheritance is frequent, unlike structure extension. This exacerbates
7813: the problem with the field name convention (@pxref{Structure Naming
7814: Convention}): One always has to remember in which class the field was
7815: originally defined; changing a part of the class structure would require
7816: changes for renaming in otherwise unaffected code.
1.12 anton 7817:
1.26 crook 7818: @cindex @code{inst-var} visibility
7819: @cindex @code{inst-value} visibility
7820: To solve this problem, I added a scoping mechanism (which was not in my
7821: original charter): A field defined with @code{inst-var} (or
7822: @code{inst-value}) is visible only in the class where it is defined and in
7823: the descendent classes of this class. Using such fields only makes
7824: sense in @code{m:}-defined methods in these classes anyway.
1.12 anton 7825:
1.26 crook 7826: This scoping mechanism allows us to use the unadorned field name,
7827: because name clashes with unrelated words become much less likely.
1.12 anton 7828:
1.26 crook 7829: @cindex @code{protected} discussion
7830: @cindex @code{private} discussion
7831: Once we have this mechanism, we can also use it for controlling the
7832: visibility of other words: All words defined after
7833: @code{protected} are visible only in the current class and its
7834: descendents. @code{public} restores the compilation
7835: (i.e. @code{current}) word list that was in effect before. If you
7836: have several @code{protected}s without an intervening
7837: @code{public} or @code{set-current}, @code{public}
7838: will restore the compilation word list in effect before the first of
7839: these @code{protected}s.
1.12 anton 7840:
1.37 anton 7841: @node Dividing classes, Object Interfaces, Classes and Scoping, Objects
7842: @subsubsection Dividing classes
7843: @cindex Dividing classes
7844: @cindex @code{methods}...@code{end-methods}
7845:
7846: You may want to do the definition of methods separate from the
7847: definition of the class, its selectors, fields, and instance variables,
7848: i.e., separate the implementation from the definition. You can do this
7849: in the following way:
7850:
7851: @example
7852: graphical class
7853: inst-value radius
7854: end-class circle
7855:
7856: ... \ do some other stuff
7857:
7858: circle methods \ now we are ready
7859:
7860: m: ( x y circle -- )
7861: radius draw-circle ;m
7862: overrides draw
7863:
7864: m: ( n-radius circle -- )
7865: [to-inst] radius ;m
7866: overrides construct
7867:
7868: end-methods
7869: @end example
7870:
7871: You can use several @code{methods}...@code{end-methods} sections. The
7872: only things you can do to the class in these sections are: defining
7873: methods, and overriding the class's selectors. You must not define new
7874: selectors or fields.
7875:
7876: Note that you often have to override a selector before using it. In
7877: particular, you usually have to override @code{construct} with a new
7878: method before you can invoke @code{heap-new} and friends. E.g., you
7879: must not create a circle before the @code{overrides construct} sequence
7880: in the example above.
7881:
7882: @node Object Interfaces, Objects Implementation, Dividing classes, Objects
1.26 crook 7883: @subsubsection Object Interfaces
7884: @cindex object interfaces
7885: @cindex interfaces for objects
1.12 anton 7886:
1.26 crook 7887: In this model you can only call selectors defined in the class of the
7888: receiving objects or in one of its ancestors. If you call a selector
7889: with a receiving object that is not in one of these classes, the
7890: result is undefined; if you are lucky, the program crashes
7891: immediately.
1.12 anton 7892:
1.26 crook 7893: @cindex selectors common to hardly-related classes
7894: Now consider the case when you want to have a selector (or several)
7895: available in two classes: You would have to add the selector to a
7896: common ancestor class, in the worst case to @code{object}. You
7897: may not want to do this, e.g., because someone else is responsible for
7898: this ancestor class.
1.12 anton 7899:
1.26 crook 7900: The solution for this problem is interfaces. An interface is a
7901: collection of selectors. If a class implements an interface, the
7902: selectors become available to the class and its descendents. A class
7903: can implement an unlimited number of interfaces. For the problem
7904: discussed above, we would define an interface for the selector(s), and
7905: both classes would implement the interface.
1.12 anton 7906:
1.26 crook 7907: As an example, consider an interface @code{storage} for
7908: writing objects to disk and getting them back, and a class
7909: @code{foo} that implements it. The code would look like this:
1.12 anton 7910:
1.26 crook 7911: @cindex @code{interface} usage
7912: @cindex @code{end-interface} usage
7913: @cindex @code{implementation} usage
7914: @example
7915: interface
7916: selector write ( file object -- )
7917: selector read1 ( file object -- )
7918: end-interface storage
1.12 anton 7919:
1.26 crook 7920: bar class
7921: storage implementation
1.12 anton 7922:
1.26 crook 7923: ... overrides write
1.37 anton 7924: ... overrides read1
1.26 crook 7925: ...
7926: end-class foo
1.12 anton 7927: @end example
7928:
1.26 crook 7929: @noindent
1.29 crook 7930: (I would add a word @code{read} @i{( file -- object )} that uses
1.26 crook 7931: @code{read1} internally, but that's beyond the point illustrated
7932: here.)
1.12 anton 7933:
1.26 crook 7934: Note that you cannot use @code{protected} in an interface; and
7935: of course you cannot define fields.
1.12 anton 7936:
1.26 crook 7937: In the Neon model, all selectors are available for all classes;
7938: therefore it does not need interfaces. The price you pay in this model
7939: is slower late binding, and therefore, added complexity to avoid late
7940: binding.
1.12 anton 7941:
1.26 crook 7942: @node Objects Implementation, Objects Glossary, Object Interfaces, Objects
7943: @subsubsection @file{objects.fs} Implementation
7944: @cindex @file{objects.fs} implementation
1.12 anton 7945:
1.26 crook 7946: @cindex @code{object-map} discussion
7947: An object is a piece of memory, like one of the data structures
7948: described with @code{struct...end-struct}. It has a field
7949: @code{object-map} that points to the method map for the object's
7950: class.
1.12 anton 7951:
1.26 crook 7952: @cindex method map
7953: @cindex virtual function table
7954: The @emph{method map}@footnote{This is Self terminology; in C++
7955: terminology: virtual function table.} is an array that contains the
1.29 crook 7956: execution tokens (@i{xt}s) of the methods for the object's class. Each
1.26 crook 7957: selector contains an offset into a method map.
1.12 anton 7958:
1.26 crook 7959: @cindex @code{selector} implementation, class
7960: @code{selector} is a defining word that uses
7961: @code{CREATE} and @code{DOES>}. The body of the
7962: selector contains the offset; the @code{does>} action for a
7963: class selector is, basically:
1.21 crook 7964:
1.26 crook 7965: @example
7966: ( object addr ) @@ over object-map @@ + @@ execute
7967: @end example
1.12 anton 7968:
1.26 crook 7969: Since @code{object-map} is the first field of the object, it
7970: does not generate any code. As you can see, calling a selector has a
7971: small, constant cost.
1.12 anton 7972:
1.26 crook 7973: @cindex @code{current-interface} discussion
7974: @cindex class implementation and representation
7975: A class is basically a @code{struct} combined with a method
7976: map. During the class definition the alignment and size of the class
7977: are passed on the stack, just as with @code{struct}s, so
7978: @code{field} can also be used for defining class
7979: fields. However, passing more items on the stack would be
7980: inconvenient, so @code{class} builds a data structure in memory,
7981: which is accessed through the variable
7982: @code{current-interface}. After its definition is complete, the
7983: class is represented on the stack by a pointer (e.g., as parameter for
7984: a child class definition).
1.1 anton 7985:
1.26 crook 7986: A new class starts off with the alignment and size of its parent,
7987: and a copy of the parent's method map. Defining new fields extends the
7988: size and alignment; likewise, defining new selectors extends the
1.29 crook 7989: method map. @code{overrides} just stores a new @i{xt} in the method
1.26 crook 7990: map at the offset given by the selector.
1.20 pazsan 7991:
1.26 crook 7992: @cindex class binding, implementation
1.29 crook 7993: Class binding just gets the @i{xt} at the offset given by the selector
1.26 crook 7994: from the class's method map and @code{compile,}s (in the case of
7995: @code{[bind]}) it.
1.21 crook 7996:
1.26 crook 7997: @cindex @code{this} implementation
7998: @cindex @code{catch} and @code{this}
7999: @cindex @code{this} and @code{catch}
8000: I implemented @code{this} as a @code{value}. At the
8001: start of an @code{m:...;m} method the old @code{this} is
8002: stored to the return stack and restored at the end; and the object on
8003: the TOS is stored @code{TO this}. This technique has one
8004: disadvantage: If the user does not leave the method via
8005: @code{;m}, but via @code{throw} or @code{exit},
8006: @code{this} is not restored (and @code{exit} may
8007: crash). To deal with the @code{throw} problem, I have redefined
8008: @code{catch} to save and restore @code{this}; the same
8009: should be done with any word that can catch an exception. As for
8010: @code{exit}, I simply forbid it (as a replacement, there is
8011: @code{exitm}).
1.21 crook 8012:
1.26 crook 8013: @cindex @code{inst-var} implementation
8014: @code{inst-var} is just the same as @code{field}, with
8015: a different @code{DOES>} action:
8016: @example
8017: @@ this +
8018: @end example
8019: Similar for @code{inst-value}.
1.21 crook 8020:
1.26 crook 8021: @cindex class scoping implementation
8022: Each class also has a word list that contains the words defined with
8023: @code{inst-var} and @code{inst-value}, and its protected
8024: words. It also has a pointer to its parent. @code{class} pushes
8025: the word lists of the class and all its ancestors onto the search order stack,
8026: and @code{end-class} drops them.
1.21 crook 8027:
1.26 crook 8028: @cindex interface implementation
8029: An interface is like a class without fields, parent and protected
8030: words; i.e., it just has a method map. If a class implements an
8031: interface, its method map contains a pointer to the method map of the
8032: interface. The positive offsets in the map are reserved for class
8033: methods, therefore interface map pointers have negative
8034: offsets. Interfaces have offsets that are unique throughout the
8035: system, unlike class selectors, whose offsets are only unique for the
8036: classes where the selector is available (invokable).
1.21 crook 8037:
1.26 crook 8038: This structure means that interface selectors have to perform one
8039: indirection more than class selectors to find their method. Their body
8040: contains the interface map pointer offset in the class method map, and
8041: the method offset in the interface method map. The
8042: @code{does>} action for an interface selector is, basically:
1.21 crook 8043:
8044: @example
1.26 crook 8045: ( object selector-body )
8046: 2dup selector-interface @@ ( object selector-body object interface-offset )
8047: swap object-map @@ + @@ ( object selector-body map )
8048: swap selector-offset @@ + @@ execute
1.21 crook 8049: @end example
8050:
1.26 crook 8051: where @code{object-map} and @code{selector-offset} are
8052: first fields and generate no code.
8053:
8054: As a concrete example, consider the following code:
1.21 crook 8055:
1.26 crook 8056: @example
8057: interface
8058: selector if1sel1
8059: selector if1sel2
8060: end-interface if1
1.21 crook 8061:
1.26 crook 8062: object class
8063: if1 implementation
8064: selector cl1sel1
8065: cell% inst-var cl1iv1
1.21 crook 8066:
1.26 crook 8067: ' m1 overrides construct
8068: ' m2 overrides if1sel1
8069: ' m3 overrides if1sel2
8070: ' m4 overrides cl1sel2
8071: end-class cl1
1.21 crook 8072:
1.26 crook 8073: create obj1 object dict-new drop
8074: create obj2 cl1 dict-new drop
8075: @end example
1.21 crook 8076:
1.26 crook 8077: The data structure created by this code (including the data structure
8078: for @code{object}) is shown in the <a
8079: href="objects-implementation.eps">figure</a>, assuming a cell size of 4.
1.29 crook 8080: @comment TODO add this diagram..
1.21 crook 8081:
1.26 crook 8082: @node Objects Glossary, , Objects Implementation, Objects
8083: @subsubsection @file{objects.fs} Glossary
8084: @cindex @file{objects.fs} Glossary
1.21 crook 8085:
1.26 crook 8086: doc---objects-bind
8087: doc---objects-<bind>
8088: doc---objects-bind'
8089: doc---objects-[bind]
8090: doc---objects-class
8091: doc---objects-class->map
8092: doc---objects-class-inst-size
8093: doc---objects-class-override!
8094: doc---objects-construct
8095: doc---objects-current'
8096: doc---objects-[current]
8097: doc---objects-current-interface
8098: doc---objects-dict-new
8099: doc---objects-drop-order
8100: doc---objects-end-class
8101: doc---objects-end-class-noname
8102: doc---objects-end-interface
8103: doc---objects-end-interface-noname
1.37 anton 8104: doc---objects-end-methods
1.26 crook 8105: doc---objects-exitm
8106: doc---objects-heap-new
8107: doc---objects-implementation
8108: doc---objects-init-object
8109: doc---objects-inst-value
8110: doc---objects-inst-var
8111: doc---objects-interface
1.38 ! anton 8112: doc---objects-m:
! 8113: doc---objects-:m
1.26 crook 8114: doc---objects-;m
8115: doc---objects-method
1.37 anton 8116: doc---objects-methods
1.26 crook 8117: doc---objects-object
8118: doc---objects-overrides
8119: doc---objects-[parent]
8120: doc---objects-print
8121: doc---objects-protected
8122: doc---objects-public
8123: doc---objects-push-order
8124: doc---objects-selector
8125: doc---objects-this
8126: doc---objects-<to-inst>
8127: doc---objects-[to-inst]
8128: doc---objects-to-this
8129: doc---objects-xt-new
1.21 crook 8130:
1.26 crook 8131: @c -------------------------------------------------------------
8132: @node OOF, Mini-OOF, Objects, Object-oriented Forth
8133: @subsection The @file{oof.fs} model
8134: @cindex oof
8135: @cindex object-oriented programming
1.21 crook 8136:
1.26 crook 8137: @cindex @file{objects.fs}
8138: @cindex @file{oof.fs}
1.21 crook 8139:
1.26 crook 8140: This section describes the @file{oof.fs} package.
1.21 crook 8141:
1.26 crook 8142: The package described in this section has been used in bigFORTH since 1991, and
8143: used for two large applications: a chromatographic system used to
8144: create new medicaments, and a graphic user interface library (MINOS).
1.21 crook 8145:
1.26 crook 8146: You can find a description (in German) of @file{oof.fs} in @cite{Object
8147: oriented bigFORTH} by Bernd Paysan, published in @cite{Vierte Dimension}
8148: 10(2), 1994.
1.21 crook 8149:
1.26 crook 8150: @menu
8151: * Properties of the OOF model::
8152: * Basic OOF Usage::
8153: * The OOF base class::
8154: * Class Declaration::
8155: * Class Implementation::
8156: @end menu
1.21 crook 8157:
1.26 crook 8158: @node Properties of the OOF model, Basic OOF Usage, OOF, OOF
8159: @subsubsection Properties of the @file{oof.fs} model
8160: @cindex @file{oof.fs} properties
1.21 crook 8161:
1.26 crook 8162: @itemize @bullet
8163: @item
8164: This model combines object oriented programming with information
8165: hiding. It helps you writing large application, where scoping is
8166: necessary, because it provides class-oriented scoping.
1.21 crook 8167:
1.26 crook 8168: @item
8169: Named objects, object pointers, and object arrays can be created,
8170: selector invocation uses the ``object selector'' syntax. Selector invocation
8171: to objects and/or selectors on the stack is a bit less convenient, but
8172: possible.
1.21 crook 8173:
1.26 crook 8174: @item
8175: Selector invocation and instance variable usage of the active object is
8176: straightforward, since both make use of the active object.
1.21 crook 8177:
1.26 crook 8178: @item
8179: Late binding is efficient and easy to use.
1.21 crook 8180:
1.26 crook 8181: @item
8182: State-smart objects parse selectors. However, extensibility is provided
8183: using a (parsing) selector @code{postpone} and a selector @code{'}.
1.21 crook 8184:
8185: @item
1.26 crook 8186: An implementation in ANS Forth is available.
8187:
1.21 crook 8188: @end itemize
8189:
8190:
1.26 crook 8191: @node Basic OOF Usage, The OOF base class, Properties of the OOF model, OOF
8192: @subsubsection Basic @file{oof.fs} Usage
8193: @cindex @file{oof.fs} usage
8194:
8195: This section uses the same example as for @code{objects} (@pxref{Basic Objects Usage}).
1.21 crook 8196:
1.26 crook 8197: You can define a class for graphical objects like this:
1.21 crook 8198:
1.26 crook 8199: @cindex @code{class} usage
8200: @cindex @code{class;} usage
8201: @cindex @code{method} usage
8202: @example
8203: object class graphical \ "object" is the parent class
8204: method draw ( x y graphical -- )
8205: class;
8206: @end example
1.21 crook 8207:
1.26 crook 8208: This code defines a class @code{graphical} with an
8209: operation @code{draw}. We can perform the operation
8210: @code{draw} on any @code{graphical} object, e.g.:
1.21 crook 8211:
1.26 crook 8212: @example
8213: 100 100 t-rex draw
8214: @end example
1.21 crook 8215:
1.26 crook 8216: @noindent
8217: where @code{t-rex} is an object or object pointer, created with e.g.
8218: @code{graphical : t-rex}.
1.21 crook 8219:
1.26 crook 8220: @cindex abstract class
8221: How do we create a graphical object? With the present definitions,
8222: we cannot create a useful graphical object. The class
8223: @code{graphical} describes graphical objects in general, but not
8224: any concrete graphical object type (C++ users would call it an
8225: @emph{abstract class}); e.g., there is no method for the selector
8226: @code{draw} in the class @code{graphical}.
1.21 crook 8227:
1.26 crook 8228: For concrete graphical objects, we define child classes of the
8229: class @code{graphical}, e.g.:
1.21 crook 8230:
8231: @example
1.26 crook 8232: graphical class circle \ "graphical" is the parent class
8233: cell var circle-radius
8234: how:
8235: : draw ( x y -- )
8236: circle-radius @@ draw-circle ;
8237:
8238: : init ( n-radius -- (
8239: circle-radius ! ;
8240: class;
8241: @end example
8242:
8243: Here we define a class @code{circle} as a child of @code{graphical},
8244: with a field @code{circle-radius}; it defines new methods for the
8245: selectors @code{draw} and @code{init} (@code{init} is defined in
8246: @code{object}, the parent class of @code{graphical}).
1.21 crook 8247:
1.26 crook 8248: Now we can create a circle in the dictionary with:
1.21 crook 8249:
1.26 crook 8250: @example
8251: 50 circle : my-circle
1.21 crook 8252: @end example
8253:
1.26 crook 8254: @noindent
8255: @code{:} invokes @code{init}, thus initializing the field
8256: @code{circle-radius} with 50. We can draw this new circle at (100,100)
8257: with:
1.21 crook 8258:
8259: @example
1.26 crook 8260: 100 100 my-circle draw
1.21 crook 8261: @end example
8262:
1.26 crook 8263: @cindex selector invocation, restrictions
8264: @cindex class definition, restrictions
8265: Note: You can only invoke a selector if the receiving object belongs to
8266: the class where the selector was defined or one of its descendents;
8267: e.g., you can invoke @code{draw} only for objects belonging to
8268: @code{graphical} or its descendents (e.g., @code{circle}). The scoping
8269: mechanism will check if you try to invoke a selector that is not
8270: defined in this class hierarchy, so you'll get an error at compilation
8271: time.
8272:
8273:
8274: @node The OOF base class, Class Declaration, Basic OOF Usage, OOF
8275: @subsubsection The @file{oof.fs} base class
8276: @cindex @file{oof.fs} base class
8277:
8278: When you define a class, you have to specify a parent class. So how do
8279: you start defining classes? There is one class available from the start:
8280: @code{object}. You have to use it as ancestor for all classes. It is the
8281: only class that has no parent. Classes are also objects, except that
8282: they don't have instance variables; class manipulation such as
8283: inheritance or changing definitions of a class is handled through
8284: selectors of the class @code{object}.
8285:
8286: @code{object} provides a number of selectors:
8287:
1.21 crook 8288: @itemize @bullet
8289: @item
1.26 crook 8290: @code{class} for subclassing, @code{definitions} to add definitions
8291: later on, and @code{class?} to get type informations (is the class a
8292: subclass of the class passed on the stack?).
8293: doc---object-class
8294: doc---object-definitions
8295: doc---object-class?
8296:
1.21 crook 8297: @item
1.26 crook 8298: @code{init} and @code{dispose} as constructor and destructor of the
8299: object. @code{init} is invocated after the object's memory is allocated,
8300: while @code{dispose} also handles deallocation. Thus if you redefine
8301: @code{dispose}, you have to call the parent's dispose with @code{super
8302: dispose}, too.
8303: doc---object-init
8304: doc---object-dispose
8305:
1.21 crook 8306: @item
1.26 crook 8307: @code{new}, @code{new[]}, @code{:}, @code{ptr}, @code{asptr}, and
8308: @code{[]} to create named and unnamed objects and object arrays or
8309: object pointers.
8310: doc---object-new
8311: doc---object-new[]
8312: doc---object-:
8313: doc---object-ptr
8314: doc---object-asptr
8315: doc---object-[]
1.21 crook 8316:
1.26 crook 8317: @item
8318: @code{::} and @code{super} for explicit scoping. You should use explicit
8319: scoping only for super classes or classes with the same set of instance
8320: variables. Explicitly-scoped selectors use early binding.
8321: doc---object-::
8322: doc---object-super
1.21 crook 8323:
1.26 crook 8324: @item
8325: @code{self} to get the address of the object
8326: doc---object-self
1.21 crook 8327:
8328: @item
1.26 crook 8329: @code{bind}, @code{bound}, @code{link}, and @code{is} to assign object
8330: pointers and instance defers.
8331: doc---object-bind
8332: doc---object-bound
8333: doc---object-link
8334: doc---object-is
8335:
1.21 crook 8336: @item
1.26 crook 8337: @code{'} to obtain selector tokens, @code{send} to invocate selectors
8338: form the stack, and @code{postpone} to generate selector invocation code.
8339: doc---object-'
8340: doc---object-postpone
8341:
1.21 crook 8342: @item
1.26 crook 8343: @code{with} and @code{endwith} to select the active object from the
8344: stack, and enable its scope. Using @code{with} and @code{endwith}
8345: also allows you to create code using selector @code{postpone} without being
8346: trapped by the state-smart objects.
8347: doc---object-with
8348: doc---object-endwith
8349:
1.21 crook 8350: @end itemize
8351:
1.26 crook 8352: @node Class Declaration, Class Implementation, The OOF base class, OOF
8353: @subsubsection Class Declaration
8354: @cindex class declaration
8355:
8356: @itemize @bullet
8357: @item
8358: Instance variables
8359: doc---oof-var
1.21 crook 8360:
1.26 crook 8361: @item
8362: Object pointers
8363: doc---oof-ptr
8364: doc---oof-asptr
1.21 crook 8365:
1.26 crook 8366: @item
8367: Instance defers
8368: doc---oof-defer
1.21 crook 8369:
1.26 crook 8370: @item
8371: Method selectors
8372: doc---oof-early
8373: doc---oof-method
1.21 crook 8374:
1.26 crook 8375: @item
8376: Class-wide variables
8377: doc---oof-static
1.21 crook 8378:
1.26 crook 8379: @item
8380: End declaration
8381: doc---oof-how:
8382: doc---oof-class;
1.21 crook 8383:
1.26 crook 8384: @end itemize
1.21 crook 8385:
1.26 crook 8386: @c -------------------------------------------------------------
8387: @node Class Implementation, , Class Declaration, OOF
8388: @subsubsection Class Implementation
8389: @cindex class implementation
1.21 crook 8390:
1.26 crook 8391: @c -------------------------------------------------------------
8392: @node Mini-OOF, Comparison with other object models, OOF, Object-oriented Forth
8393: @subsection The @file{mini-oof.fs} model
8394: @cindex mini-oof
1.1 anton 8395:
1.26 crook 8396: Gforth's third object oriented Forth package is a 12-liner. It uses a
8397: mixture of the @file{object.fs} and the @file{oof.fs} syntax,
8398: and reduces to the bare minimum of features. This is based on a posting
8399: of Bernd Paysan in comp.arch.
1.1 anton 8400:
8401: @menu
1.26 crook 8402: * Basic Mini-OOF Usage::
8403: * Mini-OOF Example::
8404: * Mini-OOF Implementation::
1.1 anton 8405: @end menu
8406:
1.26 crook 8407: @c -------------------------------------------------------------
8408: @node Basic Mini-OOF Usage, Mini-OOF Example, , Mini-OOF
8409: @subsubsection Basic @file{mini-oof.fs} Usage
8410: @cindex mini-oof usage
1.1 anton 8411:
1.28 crook 8412: There is a base class (@code{class}, which allocates one cell for the
8413: object pointer) plus seven other words: to define a method, a variable,
8414: a class; to end a class, to resolve binding, to allocate an object and
8415: to compile a class method.
1.26 crook 8416: @comment TODO better description of the last one
1.1 anton 8417:
1.26 crook 8418: doc-object
8419: doc-method
8420: doc-var
8421: doc-class
8422: doc-end-class
8423: doc-defines
8424: doc-new
8425: doc-::
1.1 anton 8426:
1.21 crook 8427:
1.26 crook 8428: @c -------------------------------------------------------------
8429: @node Mini-OOF Example, Mini-OOF Implementation, Basic Mini-OOF Usage, Mini-OOF
8430: @subsubsection Mini-OOF Example
8431: @cindex mini-oof example
1.21 crook 8432:
1.26 crook 8433: A short example shows how to use this package. This example, in slightly
8434: extended form, is supplied as @file{moof-exm.fs}
1.29 crook 8435: @comment TODO could flesh this out with some comments from the Forthwrite article
1.21 crook 8436:
1.26 crook 8437: @example
8438: object class
8439: method init
8440: method draw
8441: end-class graphical
8442: @end example
1.21 crook 8443:
1.26 crook 8444: This code defines a class @code{graphical} with an
8445: operation @code{draw}. We can perform the operation
8446: @code{draw} on any @code{graphical} object, e.g.:
1.1 anton 8447:
1.26 crook 8448: @example
8449: 100 100 t-rex draw
8450: @end example
1.1 anton 8451:
1.26 crook 8452: where @code{t-rex} is an object or object pointer, created with e.g.
8453: @code{graphical new Constant t-rex}.
1.1 anton 8454:
1.26 crook 8455: For concrete graphical objects, we define child classes of the
8456: class @code{graphical}, e.g.:
1.21 crook 8457:
8458: @example
1.26 crook 8459: graphical class
8460: cell var circle-radius
8461: end-class circle \ "graphical" is the parent class
1.21 crook 8462:
1.26 crook 8463: :noname ( x y -- )
8464: circle-radius @@ draw-circle ; circle defines draw
8465: :noname ( r -- )
8466: circle-radius ! ; circle defines init
1.21 crook 8467: @end example
8468:
1.26 crook 8469: There is no implicit init method, so we have to define one. The creation
8470: code of the object now has to call init explicitely.
1.21 crook 8471:
1.26 crook 8472: @example
8473: circle new Constant my-circle
8474: 50 my-circle init
8475: @end example
1.21 crook 8476:
1.26 crook 8477: It is also possible to add a function to create named objects with
8478: automatic call of @code{init}, given that all objects have @code{init}
8479: on the same place:
1.1 anton 8480:
8481: @example
1.26 crook 8482: : new: ( .. o "name" -- )
8483: new dup Constant init ;
8484: 80 circle new: large-circle
1.1 anton 8485: @end example
8486:
1.26 crook 8487: We can draw this new circle at (100,100) with:
1.1 anton 8488:
8489: @example
1.26 crook 8490: 100 100 my-circle draw
1.1 anton 8491: @end example
8492:
1.26 crook 8493: @node Mini-OOF Implementation, , Mini-OOF Example, Mini-OOF
8494: @subsubsection @file{mini-oof.fs} Implementation
1.1 anton 8495:
1.26 crook 8496: Object-oriented systems with late binding typically use a
8497: ``vtable''-approach: the first variable in each object is a pointer to a
8498: table, which contains the methods as function pointers. The vtable
8499: may also contain other information.
1.1 anton 8500:
1.26 crook 8501: So first, let's declare methods:
1.1 anton 8502:
1.26 crook 8503: @example
8504: : method ( m v -- m' v ) Create over , swap cell+ swap
8505: DOES> ( ... o -- ... ) @ over @ + @ execute ;
8506: @end example
1.1 anton 8507:
1.26 crook 8508: During method declaration, the number of methods and instance
8509: variables is on the stack (in address units). @code{method} creates
8510: one method and increments the method number. To execute a method, it
8511: takes the object, fetches the vtable pointer, adds the offset, and
1.29 crook 8512: executes the @i{xt} stored there. Each method takes the object it is
1.26 crook 8513: invoked from as top of stack parameter. The method itself should
8514: consume that object.
1.1 anton 8515:
1.26 crook 8516: Now, we also have to declare instance variables
1.21 crook 8517:
1.26 crook 8518: @example
8519: : var ( m v size -- m v' ) Create over , +
8520: DOES> ( o -- addr ) @ + ;
8521: @end example
1.21 crook 8522:
1.26 crook 8523: As before, a word is created with the current offset. Instance
8524: variables can have different sizes (cells, floats, doubles, chars), so
8525: all we do is take the size and add it to the offset. If your machine
8526: has alignment restrictions, put the proper @code{aligned} or
8527: @code{faligned} before the variable, to adjust the variable
8528: offset. That's why it is on the top of stack.
1.2 jwilke 8529:
1.26 crook 8530: We need a starting point (the base object) and some syntactic sugar:
1.21 crook 8531:
1.26 crook 8532: @example
8533: Create object 1 cells , 2 cells ,
8534: : class ( class -- class methods vars ) dup 2@ ;
8535: @end example
1.21 crook 8536:
1.26 crook 8537: For inheritance, the vtable of the parent object has to be
8538: copied when a new, derived class is declared. This gives all the
8539: methods of the parent class, which can be overridden, though.
1.21 crook 8540:
1.2 jwilke 8541: @example
1.26 crook 8542: : end-class ( class methods vars -- )
8543: Create here >r , dup , 2 cells ?DO ['] noop , 1 cells +LOOP
8544: cell+ dup cell+ r> rot @ 2 cells /string move ;
8545: @end example
8546:
8547: The first line creates the vtable, initialized with
8548: @code{noop}s. The second line is the inheritance mechanism, it
8549: copies the xts from the parent vtable.
1.2 jwilke 8550:
1.26 crook 8551: We still have no way to define new methods, let's do that now:
1.2 jwilke 8552:
1.26 crook 8553: @example
8554: : defines ( xt class -- ) ' >body @ + ! ;
1.2 jwilke 8555: @end example
8556:
1.26 crook 8557: To allocate a new object, we need a word, too:
1.2 jwilke 8558:
1.26 crook 8559: @example
8560: : new ( class -- o ) here over @ allot swap over ! ;
8561: @end example
1.2 jwilke 8562:
1.26 crook 8563: Sometimes derived classes want to access the method of the
8564: parent object. There are two ways to achieve this with Mini-OOF:
8565: first, you could use named words, and second, you could look up the
8566: vtable of the parent object.
1.2 jwilke 8567:
1.26 crook 8568: @example
8569: : :: ( class "name" -- ) ' >body @ + @ compile, ;
8570: @end example
1.2 jwilke 8571:
8572:
1.26 crook 8573: Nothing can be more confusing than a good example, so here is
8574: one. First let's declare a text object (called
8575: @code{button}), that stores text and position:
1.2 jwilke 8576:
1.26 crook 8577: @example
8578: object class
8579: cell var text
8580: cell var len
8581: cell var x
8582: cell var y
8583: method init
8584: method draw
8585: end-class button
8586: @end example
1.2 jwilke 8587:
1.26 crook 8588: @noindent
8589: Now, implement the two methods, @code{draw} and @code{init}:
1.2 jwilke 8590:
1.26 crook 8591: @example
8592: :noname ( o -- )
8593: >r r@ x @ r@ y @ at-xy r@ text @ r> len @ type ;
8594: button defines draw
8595: :noname ( addr u o -- )
8596: >r 0 r@ x ! 0 r@ y ! r@ len ! r> text ! ;
8597: button defines init
8598: @end example
1.2 jwilke 8599:
1.26 crook 8600: @noindent
8601: To demonstrate inheritance, we define a class @code{bold-button}, with no
8602: new data and no new methods:
1.2 jwilke 8603:
1.26 crook 8604: @example
8605: button class
8606: end-class bold-button
1.1 anton 8607:
1.26 crook 8608: : bold 27 emit ." [1m" ;
8609: : normal 27 emit ." [0m" ;
8610: @end example
1.1 anton 8611:
1.26 crook 8612: @noindent
8613: The class @code{bold-button} has a different draw method to
8614: @code{button}, but the new method is defined in terms of the draw method
8615: for @code{button}:
1.1 anton 8616:
1.26 crook 8617: @example
8618: :noname bold [ button :: draw ] normal ; bold-button defines draw
8619: @end example
1.1 anton 8620:
1.26 crook 8621: @noindent
8622: Finally, create two objects and apply methods:
1.1 anton 8623:
1.26 crook 8624: @example
8625: button new Constant foo
8626: s" thin foo" foo init
8627: page
8628: foo draw
8629: bold-button new Constant bar
8630: s" fat bar" bar init
8631: 1 bar y !
8632: bar draw
8633: @end example
1.1 anton 8634:
8635:
1.26 crook 8636: @node Comparison with other object models, , Mini-OOF, Object-oriented Forth
8637: @subsubsection Comparison with other object models
8638: @cindex comparison of object models
8639: @cindex object models, comparison
1.1 anton 8640:
1.26 crook 8641: Many object-oriented Forth extensions have been proposed (@cite{A survey
8642: of object-oriented Forths} (SIGPLAN Notices, April 1996) by Bradford
8643: J. Rodriguez and W. F. S. Poehlman lists 17). This section discusses the
8644: relation of the object models described here to two well-known and two
8645: closely-related (by the use of method maps) models.
1.1 anton 8646:
1.26 crook 8647: @cindex Neon model
8648: The most popular model currently seems to be the Neon model (see
8649: @cite{Object-oriented programming in ANS Forth} (Forth Dimensions, March
8650: 1997) by Andrew McKewan) but this model has a number of limitations
8651: @footnote{A longer version of this critique can be
8652: found in @cite{On Standardizing Object-Oriented Forth Extensions} (Forth
8653: Dimensions, May 1997) by Anton Ertl.}:
1.1 anton 8654:
1.26 crook 8655: @itemize @bullet
8656: @item
8657: It uses a @code{@emph{selector
8658: object}} syntax, which makes it unnatural to pass objects on the
8659: stack.
1.1 anton 8660:
1.26 crook 8661: @item
8662: It requires that the selector parses the input stream (at
8663: compile time); this leads to reduced extensibility and to bugs that are+
8664: hard to find.
1.1 anton 8665:
1.26 crook 8666: @item
8667: It allows using every selector to every object;
8668: this eliminates the need for classes, but makes it harder to create
8669: efficient implementations.
8670: @end itemize
1.1 anton 8671:
1.26 crook 8672: @cindex Pountain's object-oriented model
8673: Another well-known publication is @cite{Object-Oriented Forth} (Academic
8674: Press, London, 1987) by Dick Pountain. However, it is not really about
8675: object-oriented programming, because it hardly deals with late
8676: binding. Instead, it focuses on features like information hiding and
8677: overloading that are characteristic of modular languages like Ada (83).
1.1 anton 8678:
1.26 crook 8679: @cindex Zsoter's object-oriented model
8680: In @cite{Does late binding have to be slow?} (Forth Dimensions 18(1) 1996, pages 31-35)
8681: Andras Zsoter describes a model that makes heavy use of an active object
8682: (like @code{this} in @file{objects.fs}): The active object is not only
8683: used for accessing all fields, but also specifies the receiving object
8684: of every selector invocation; you have to change the active object
8685: explicitly with @code{@{ ... @}}, whereas in @file{objects.fs} it
8686: changes more or less implicitly at @code{m: ... ;m}. Such a change at
8687: the method entry point is unnecessary with the Zsoter's model, because
8688: the receiving object is the active object already. On the other hand, the explicit
8689: change is absolutely necessary in that model, because otherwise no one
8690: could ever change the active object. An ANS Forth implementation of this
8691: model is available at @url{http://www.forth.org/fig/oopf.html}.
1.1 anton 8692:
1.26 crook 8693: @cindex @file{oof.fs}, differences to other models
8694: The @file{oof.fs} model combines information hiding and overloading
8695: resolution (by keeping names in various word lists) with object-oriented
8696: programming. It sets the active object implicitly on method entry, but
8697: also allows explicit changing (with @code{>o...o>} or with
8698: @code{with...endwith}). It uses parsing and state-smart objects and
8699: classes for resolving overloading and for early binding: the object or
8700: class parses the selector and determines the method from this. If the
8701: selector is not parsed by an object or class, it performs a call to the
8702: selector for the active object (late binding), like Zsoter's model.
8703: Fields are always accessed through the active object. The big
8704: disadvantage of this model is the parsing and the state-smartness, which
8705: reduces extensibility and increases the opportunities for subtle bugs;
8706: essentially, you are only safe if you never tick or @code{postpone} an
8707: object or class (Bernd disagrees, but I (Anton) am not convinced).
1.1 anton 8708:
1.26 crook 8709: @cindex @file{mini-oof.fs}, differences to other models
8710: The @file{mini-oof.fs} model is quite similar to a very stripped-down version of
8711: the @file{objects.fs} model, but syntactically it is a mixture of the @file{objects.fs} and
8712: @file{oof.fs} models.
1.1 anton 8713:
1.26 crook 8714: @c -------------------------------------------------------------
8715: @node Passing Commands to the OS, Miscellaneous Words, Object-oriented Forth, Words
1.21 crook 8716: @section Passing Commands to the Operating System
8717: @cindex operating system - passing commands
8718: @cindex shell commands
8719:
8720: Gforth allows you to pass an arbitrary string to the host operating
8721: system shell (if such a thing exists) for execution.
8722:
8723: doc-sh
8724: doc-system
8725: doc-$?
1.23 crook 8726: doc-getenv
1.21 crook 8727:
1.26 crook 8728: @c -------------------------------------------------------------
1.21 crook 8729: @node Miscellaneous Words, , Passing Commands to the OS, Words
8730: @section Miscellaneous Words
8731: @cindex miscellaneous words
8732:
1.29 crook 8733: @comment TODO find homes for these
8734:
1.26 crook 8735: These section lists the ANS Forth words that are not documented
1.21 crook 8736: elsewhere in this manual. Ultimately, they all need proper homes.
8737:
8738: doc-ms
8739: doc-time&date
1.27 crook 8740:
1.21 crook 8741: doc-[compile]
8742:
1.26 crook 8743: The following ANS Forth words are not currently supported by Gforth
1.27 crook 8744: (@pxref{ANS conformance}):
1.21 crook 8745:
8746: @code{EDITOR}
8747: @code{EKEY}
8748: @code{EKEY>CHAR}
8749: @code{EKEY?}
8750: @code{EMIT?}
8751: @code{FORGET}
8752:
1.24 anton 8753: @c ******************************************************************
8754: @node Error messages, Tools, Words, Top
8755: @chapter Error messages
8756: @cindex error messages
8757: @cindex backtrace
8758:
8759: A typical Gforth error message looks like this:
8760:
8761: @example
8762: in file included from :-1
8763: in file included from ./yyy.fs:1
8764: ./xxx.fs:4: Invalid memory address
8765: bar
8766: ^^^
1.25 anton 8767: $400E664C @@
8768: $400E6664 foo
1.24 anton 8769: @end example
8770:
8771: The message identifying the error is @code{Invalid memory address}. The
8772: error happened when text-interpreting line 4 of the file
8773: @file{./xxx.fs}. This line is given (it contains @code{bar}), and the
8774: word on the line where the error happened, is pointed out (with
8775: @code{^^^}).
8776:
8777: The file containing the error was included in line 1 of @file{./yyy.fs},
8778: and @file{yyy.fs} was included from a non-file (in this case, by giving
8779: @file{yyy.fs} as command-line parameter to Gforth).
8780:
8781: At the end of the error message you find a return stack dump that can be
8782: interpreted as a backtrace (possibly empty). On top you find the top of
8783: the return stack when the @code{throw} happened, and at the bottom you
8784: find the return stack entry just above the return stack of the topmost
8785: text interpreter.
8786:
8787: To the right of most return stack entries you see a guess for the word
8788: that pushed that return stack entry as its return address. This gives a
8789: backtrace. In our case we see that @code{bar} called @code{foo}, and
8790: @code{foo} called @code{@@} (and @code{@@} had an @emph{Invalid memory
8791: address} exception).
8792:
8793: Note that the backtrace is not perfect: We don't know which return stack
8794: entries are return addresses (so we may get false positives); and in
8795: some cases (e.g., for @code{abort"}) we cannot determine from the return
8796: address the word that pushed the return address, so for some return
8797: addresses you see no names in the return stack dump.
1.25 anton 8798:
8799: @cindex @code{catch} and backtraces
8800: The return stack dump represents the return stack at the time when a
8801: specific @code{throw} was executed. In programs that make use of
8802: @code{catch}, it is not necessarily clear which @code{throw} should be
8803: used for the return stack dump (e.g., consider one @code{throw} that
8804: indicates an error, which is caught, and during recovery another error
8805: happens; which @code{throw} should be used for the stack dump). Gforth
8806: presents the return stack dump for the first @code{throw} after the last
8807: executed (not returned-to) @code{catch}; this works well in the usual
8808: case.
8809:
8810: @cindex @code{gforth-fast} and backtraces
8811: @cindex @code{gforth-fast}, difference from @code{gforth}
8812: @cindex backtraces with @code{gforth-fast}
8813: @cindex return stack dump with @code{gforth-fast}
8814: @code{gforth} is able to do a return stack dump for throws generated
8815: from primitives (e.g., invalid memory address, stack empty etc.);
8816: @code{gforth-fast} is only able to do a return stack dump from a
8817: directly called @code{throw} (including @code{abort} etc.). This is the
1.30 anton 8818: only difference (apart from a speed factor of between 1.15 (K6-2) and
8819: 1.6 (21164A)) between @code{gforth} and @code{gforth-fast}. Given an
8820: exception caused by a primitive in @code{gforth-fast}, you will
8821: typically see no return stack dump at all; however, if the exception is
8822: caught by @code{catch} (e.g., for restoring some state), and then
8823: @code{throw}n again, the return stack dump will be for the first such
8824: @code{throw}.
1.2 jwilke 8825:
1.5 anton 8826: @c ******************************************************************
1.24 anton 8827: @node Tools, ANS conformance, Error messages, Top
1.1 anton 8828: @chapter Tools
8829:
8830: @menu
8831: * ANS Report:: Report the words used, sorted by wordset.
8832: @end menu
8833:
8834: See also @ref{Emacs and Gforth}.
8835:
8836: @node ANS Report, , Tools, Tools
8837: @section @file{ans-report.fs}: Report the words used, sorted by wordset
8838: @cindex @file{ans-report.fs}
8839: @cindex report the words used in your program
8840: @cindex words used in your program
8841:
8842: If you want to label a Forth program as ANS Forth Program, you must
8843: document which wordsets the program uses; for extension wordsets, it is
8844: helpful to list the words the program requires from these wordsets
8845: (because Forth systems are allowed to provide only some words of them).
8846:
8847: The @file{ans-report.fs} tool makes it easy for you to determine which
8848: words from which wordset and which non-ANS words your application
8849: uses. You simply have to include @file{ans-report.fs} before loading the
8850: program you want to check. After loading your program, you can get the
8851: report with @code{print-ans-report}. A typical use is to run this as
8852: batch job like this:
8853: @example
8854: gforth ans-report.fs myprog.fs -e "print-ans-report bye"
8855: @end example
8856:
8857: The output looks like this (for @file{compat/control.fs}):
8858: @example
8859: The program uses the following words
8860: from CORE :
8861: : POSTPONE THEN ; immediate ?dup IF 0=
8862: from BLOCK-EXT :
8863: \
8864: from FILE :
8865: (
8866: @end example
8867:
8868: @subsection Caveats
8869:
8870: Note that @file{ans-report.fs} just checks which words are used, not whether
8871: they are used in an ANS Forth conforming way!
8872:
8873: Some words are defined in several wordsets in the
8874: standard. @file{ans-report.fs} reports them for only one of the
8875: wordsets, and not necessarily the one you expect. It depends on usage
8876: which wordset is the right one to specify. E.g., if you only use the
8877: compilation semantics of @code{S"}, it is a Core word; if you also use
8878: its interpretation semantics, it is a File word.
8879:
8880: @c ******************************************************************
8881: @node ANS conformance, Model, Tools, Top
8882: @chapter ANS conformance
8883: @cindex ANS conformance of Gforth
8884:
8885: To the best of our knowledge, Gforth is an
8886:
8887: ANS Forth System
8888: @itemize @bullet
8889: @item providing the Core Extensions word set
8890: @item providing the Block word set
8891: @item providing the Block Extensions word set
8892: @item providing the Double-Number word set
8893: @item providing the Double-Number Extensions word set
8894: @item providing the Exception word set
8895: @item providing the Exception Extensions word set
8896: @item providing the Facility word set
8897: @item providing @code{MS} and @code{TIME&DATE} from the Facility Extensions word set
8898: @item providing the File Access word set
8899: @item providing the File Access Extensions word set
8900: @item providing the Floating-Point word set
8901: @item providing the Floating-Point Extensions word set
8902: @item providing the Locals word set
8903: @item providing the Locals Extensions word set
8904: @item providing the Memory-Allocation word set
8905: @item providing the Memory-Allocation Extensions word set (that one's easy)
8906: @item providing the Programming-Tools word set
8907: @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
8908: @item providing the Search-Order word set
8909: @item providing the Search-Order Extensions word set
8910: @item providing the String word set
8911: @item providing the String Extensions word set (another easy one)
8912: @end itemize
8913:
8914: @cindex system documentation
8915: In addition, ANS Forth systems are required to document certain
8916: implementation choices. This chapter tries to meet these
8917: requirements. In many cases it gives a way to ask the system for the
8918: information instead of providing the information directly, in
8919: particular, if the information depends on the processor, the operating
8920: system or the installation options chosen, or if they are likely to
8921: change during the maintenance of Gforth.
8922:
8923: @comment The framework for the rest has been taken from pfe.
8924:
8925: @menu
8926: * The Core Words::
8927: * The optional Block word set::
8928: * The optional Double Number word set::
8929: * The optional Exception word set::
8930: * The optional Facility word set::
8931: * The optional File-Access word set::
8932: * The optional Floating-Point word set::
8933: * The optional Locals word set::
8934: * The optional Memory-Allocation word set::
8935: * The optional Programming-Tools word set::
8936: * The optional Search-Order word set::
8937: @end menu
8938:
8939:
8940: @c =====================================================================
8941: @node The Core Words, The optional Block word set, ANS conformance, ANS conformance
8942: @comment node-name, next, previous, up
8943: @section The Core Words
8944: @c =====================================================================
8945: @cindex core words, system documentation
8946: @cindex system documentation, core words
8947:
8948: @menu
8949: * core-idef:: Implementation Defined Options
8950: * core-ambcond:: Ambiguous Conditions
8951: * core-other:: Other System Documentation
8952: @end menu
8953:
8954: @c ---------------------------------------------------------------------
8955: @node core-idef, core-ambcond, The Core Words, The Core Words
8956: @subsection Implementation Defined Options
8957: @c ---------------------------------------------------------------------
8958: @cindex core words, implementation-defined options
8959: @cindex implementation-defined options, core words
8960:
8961:
8962: @table @i
8963: @item (Cell) aligned addresses:
8964: @cindex cell-aligned addresses
8965: @cindex aligned addresses
8966: processor-dependent. Gforth's alignment words perform natural alignment
8967: (e.g., an address aligned for a datum of size 8 is divisible by
8968: 8). Unaligned accesses usually result in a @code{-23 THROW}.
8969:
8970: @item @code{EMIT} and non-graphic characters:
8971: @cindex @code{EMIT} and non-graphic characters
8972: @cindex non-graphic characters and @code{EMIT}
8973: The character is output using the C library function (actually, macro)
8974: @code{putc}.
8975:
8976: @item character editing of @code{ACCEPT} and @code{EXPECT}:
8977: @cindex character editing of @code{ACCEPT} and @code{EXPECT}
8978: @cindex editing in @code{ACCEPT} and @code{EXPECT}
8979: @cindex @code{ACCEPT}, editing
8980: @cindex @code{EXPECT}, editing
8981: This is modeled on the GNU readline library (@pxref{Readline
8982: Interaction, , Command Line Editing, readline, The GNU Readline
8983: Library}) with Emacs-like key bindings. @kbd{Tab} deviates a little by
8984: producing a full word completion every time you type it (instead of
1.28 crook 8985: producing the common prefix of all completions). @xref{Command-line editing}.
1.1 anton 8986:
8987: @item character set:
8988: @cindex character set
8989: The character set of your computer and display device. Gforth is
8990: 8-bit-clean (but some other component in your system may make trouble).
8991:
8992: @item Character-aligned address requirements:
8993: @cindex character-aligned address requirements
8994: installation-dependent. Currently a character is represented by a C
8995: @code{unsigned char}; in the future we might switch to @code{wchar_t}
8996: (Comments on that requested).
8997:
8998: @item character-set extensions and matching of names:
8999: @cindex character-set extensions and matching of names
1.26 crook 9000: @cindex case-sensitivity for name lookup
9001: @cindex name lookup, case-sensitivity
9002: @cindex locale and case-sensitivity
1.21 crook 9003: Any character except the ASCII NUL character can be used in a
1.1 anton 9004: name. Matching is case-insensitive (except in @code{TABLE}s). The
9005: matching is performed using the C function @code{strncasecmp}, whose
9006: function is probably influenced by the locale. E.g., the @code{C} locale
9007: does not know about accents and umlauts, so they are matched
9008: case-sensitively in that locale. For portability reasons it is best to
9009: write programs such that they work in the @code{C} locale. Then one can
9010: use libraries written by a Polish programmer (who might use words
9011: containing ISO Latin-2 encoded characters) and by a French programmer
9012: (ISO Latin-1) in the same program (of course, @code{WORDS} will produce
9013: funny results for some of the words (which ones, depends on the font you
9014: are using)). Also, the locale you prefer may not be available in other
9015: operating systems. Hopefully, Unicode will solve these problems one day.
9016:
9017: @item conditions under which control characters match a space delimiter:
9018: @cindex space delimiters
9019: @cindex control characters as delimiters
9020: If @code{WORD} is called with the space character as a delimiter, all
9021: white-space characters (as identified by the C macro @code{isspace()})
9022: are delimiters. @code{PARSE}, on the other hand, treats space like other
9023: delimiters. @code{PARSE-WORD} treats space like @code{WORD}, but behaves
9024: like @code{PARSE} otherwise. @code{(NAME)}, which is used by the outer
9025: interpreter (aka text interpreter) by default, treats all white-space
9026: characters as delimiters.
9027:
1.26 crook 9028: @item format of the control-flow stack:
9029: @cindex control-flow stack, format
9030: The data stack is used as control-flow stack. The size of a control-flow
1.1 anton 9031: stack item in cells is given by the constant @code{cs-item-size}. At the
9032: time of this writing, an item consists of a (pointer to a) locals list
9033: (third), an address in the code (second), and a tag for identifying the
9034: item (TOS). The following tags are used: @code{defstart},
9035: @code{live-orig}, @code{dead-orig}, @code{dest}, @code{do-dest},
9036: @code{scopestart}.
9037:
9038: @item conversion of digits > 35
9039: @cindex digits > 35
9040: The characters @code{[\]^_'} are the digits with the decimal value
9041: 36@minus{}41. There is no way to input many of the larger digits.
9042:
9043: @item display after input terminates in @code{ACCEPT} and @code{EXPECT}:
9044: @cindex @code{EXPECT}, display after end of input
9045: @cindex @code{ACCEPT}, display after end of input
9046: The cursor is moved to the end of the entered string. If the input is
9047: terminated using the @kbd{Return} key, a space is typed.
9048:
9049: @item exception abort sequence of @code{ABORT"}:
9050: @cindex exception abort sequence of @code{ABORT"}
9051: @cindex @code{ABORT"}, exception abort sequence
9052: The error string is stored into the variable @code{"error} and a
9053: @code{-2 throw} is performed.
9054:
9055: @item input line terminator:
9056: @cindex input line terminator
9057: @cindex line terminator on input
1.26 crook 9058: @cindex newline character on input
1.1 anton 9059: For interactive input, @kbd{C-m} (CR) and @kbd{C-j} (LF) terminate
9060: lines. One of these characters is typically produced when you type the
9061: @kbd{Enter} or @kbd{Return} key.
9062:
9063: @item maximum size of a counted string:
9064: @cindex maximum size of a counted string
9065: @cindex counted string, maximum size
9066: @code{s" /counted-string" environment? drop .}. Currently 255 characters
9067: on all ports, but this may change.
9068:
9069: @item maximum size of a parsed string:
9070: @cindex maximum size of a parsed string
9071: @cindex parsed string, maximum size
9072: Given by the constant @code{/line}. Currently 255 characters.
9073:
9074: @item maximum size of a definition name, in characters:
9075: @cindex maximum size of a definition name, in characters
9076: @cindex name, maximum length
9077: 31
9078:
9079: @item maximum string length for @code{ENVIRONMENT?}, in characters:
9080: @cindex maximum string length for @code{ENVIRONMENT?}, in characters
9081: @cindex @code{ENVIRONMENT?} string length, maximum
9082: 31
9083:
9084: @item method of selecting the user input device:
9085: @cindex user input device, method of selecting
9086: The user input device is the standard input. There is currently no way to
9087: change it from within Gforth. However, the input can typically be
9088: redirected in the command line that starts Gforth.
9089:
9090: @item method of selecting the user output device:
9091: @cindex user output device, method of selecting
9092: @code{EMIT} and @code{TYPE} output to the file-id stored in the value
1.10 anton 9093: @code{outfile-id} (@code{stdout} by default). Gforth uses unbuffered
9094: output when the user output device is a terminal, otherwise the output
9095: is buffered.
1.1 anton 9096:
9097: @item methods of dictionary compilation:
9098: What are we expected to document here?
9099:
9100: @item number of bits in one address unit:
9101: @cindex number of bits in one address unit
9102: @cindex address unit, size in bits
9103: @code{s" address-units-bits" environment? drop .}. 8 in all current
9104: ports.
9105:
9106: @item number representation and arithmetic:
9107: @cindex number representation and arithmetic
9108: Processor-dependent. Binary two's complement on all current ports.
9109:
9110: @item ranges for integer types:
9111: @cindex ranges for integer types
9112: @cindex integer types, ranges
9113: Installation-dependent. Make environmental queries for @code{MAX-N},
9114: @code{MAX-U}, @code{MAX-D} and @code{MAX-UD}. The lower bounds for
9115: unsigned (and positive) types is 0. The lower bound for signed types on
9116: two's complement and one's complement machines machines can be computed
9117: by adding 1 to the upper bound.
9118:
9119: @item read-only data space regions:
9120: @cindex read-only data space regions
9121: @cindex data-space, read-only regions
9122: The whole Forth data space is writable.
9123:
9124: @item size of buffer at @code{WORD}:
9125: @cindex size of buffer at @code{WORD}
9126: @cindex @code{WORD} buffer size
9127: @code{PAD HERE - .}. 104 characters on 32-bit machines. The buffer is
9128: shared with the pictured numeric output string. If overwriting
9129: @code{PAD} is acceptable, it is as large as the remaining dictionary
9130: space, although only as much can be sensibly used as fits in a counted
9131: string.
9132:
9133: @item size of one cell in address units:
9134: @cindex cell size
9135: @code{1 cells .}.
9136:
9137: @item size of one character in address units:
9138: @cindex char size
9139: @code{1 chars .}. 1 on all current ports.
9140:
9141: @item size of the keyboard terminal buffer:
9142: @cindex size of the keyboard terminal buffer
9143: @cindex terminal buffer, size
9144: Varies. You can determine the size at a specific time using @code{lp@@
9145: tib - .}. It is shared with the locals stack and TIBs of files that
9146: include the current file. You can change the amount of space for TIBs
9147: and locals stack at Gforth startup with the command line option
9148: @code{-l}.
9149:
9150: @item size of the pictured numeric output buffer:
9151: @cindex size of the pictured numeric output buffer
9152: @cindex pictured numeric output buffer, size
9153: @code{PAD HERE - .}. 104 characters on 32-bit machines. The buffer is
9154: shared with @code{WORD}.
9155:
9156: @item size of the scratch area returned by @code{PAD}:
9157: @cindex size of the scratch area returned by @code{PAD}
9158: @cindex @code{PAD} size
9159: The remainder of dictionary space. @code{unused pad here - - .}.
9160:
9161: @item system case-sensitivity characteristics:
9162: @cindex case-sensitivity characteristics
1.26 crook 9163: Dictionary searches are case-insensitive (except in
1.1 anton 9164: @code{TABLE}s). However, as explained above under @i{character-set
9165: extensions}, the matching for non-ASCII characters is determined by the
9166: locale you are using. In the default @code{C} locale all non-ASCII
9167: characters are matched case-sensitively.
9168:
9169: @item system prompt:
9170: @cindex system prompt
9171: @cindex prompt
9172: @code{ ok} in interpret state, @code{ compiled} in compile state.
9173:
9174: @item division rounding:
9175: @cindex division rounding
9176: installation dependent. @code{s" floored" environment? drop .}. We leave
9177: the choice to @code{gcc} (what to use for @code{/}) and to you (whether
9178: to use @code{fm/mod}, @code{sm/rem} or simply @code{/}).
9179:
9180: @item values of @code{STATE} when true:
9181: @cindex @code{STATE} values
9182: -1.
9183:
9184: @item values returned after arithmetic overflow:
9185: On two's complement machines, arithmetic is performed modulo
9186: 2**bits-per-cell for single arithmetic and 4**bits-per-cell for double
9187: arithmetic (with appropriate mapping for signed types). Division by zero
9188: typically results in a @code{-55 throw} (Floating-point unidentified
9189: fault), although a @code{-10 throw} (divide by zero) would be more
9190: appropriate.
9191:
9192: @item whether the current definition can be found after @t{DOES>}:
9193: @cindex @t{DOES>}, visibility of current definition
9194: No.
9195:
9196: @end table
9197:
9198: @c ---------------------------------------------------------------------
9199: @node core-ambcond, core-other, core-idef, The Core Words
9200: @subsection Ambiguous conditions
9201: @c ---------------------------------------------------------------------
9202: @cindex core words, ambiguous conditions
9203: @cindex ambiguous conditions, core words
9204:
9205: @table @i
9206:
9207: @item a name is neither a word nor a number:
9208: @cindex name not found
1.26 crook 9209: @cindex undefined word
1.1 anton 9210: @code{-13 throw} (Undefined word). Actually, @code{-13 bounce}, which
9211: preserves the data and FP stack, so you don't lose more work than
9212: necessary.
9213:
9214: @item a definition name exceeds the maximum length allowed:
1.26 crook 9215: @cindex word name too long
1.1 anton 9216: @code{-19 throw} (Word name too long)
9217:
9218: @item addressing a region not inside the various data spaces of the forth system:
9219: @cindex Invalid memory address
1.32 anton 9220: The stacks, code space and header space are accessible. Machine code space is
1.1 anton 9221: typically readable. Accessing other addresses gives results dependent on
9222: the operating system. On decent systems: @code{-9 throw} (Invalid memory
9223: address).
9224:
9225: @item argument type incompatible with parameter:
1.26 crook 9226: @cindex argument type mismatch
1.1 anton 9227: This is usually not caught. Some words perform checks, e.g., the control
9228: flow words, and issue a @code{ABORT"} or @code{-12 THROW} (Argument type
9229: mismatch).
9230:
9231: @item attempting to obtain the execution token of a word with undefined execution semantics:
9232: @cindex Interpreting a compile-only word, for @code{'} etc.
9233: @cindex execution token of words with undefined execution semantics
9234: @code{-14 throw} (Interpreting a compile-only word). In some cases, you
9235: get an execution token for @code{compile-only-error} (which performs a
9236: @code{-14 throw} when executed).
9237:
9238: @item dividing by zero:
9239: @cindex dividing by zero
9240: @cindex floating point unidentified fault, integer division
1.24 anton 9241: On better platforms, this produces a @code{-10 throw} (Division by
9242: zero); on other systems, this typically results in a @code{-55 throw}
9243: (Floating-point unidentified fault).
1.1 anton 9244:
9245: @item insufficient data stack or return stack space:
9246: @cindex insufficient data stack or return stack space
9247: @cindex stack overflow
1.26 crook 9248: @cindex address alignment exception, stack overflow
1.1 anton 9249: @cindex Invalid memory address, stack overflow
9250: Depending on the operating system, the installation, and the invocation
9251: of Gforth, this is either checked by the memory management hardware, or
1.24 anton 9252: it is not checked. If it is checked, you typically get a @code{-3 throw}
9253: (Stack overflow), @code{-5 throw} (Return stack overflow), or @code{-9
9254: throw} (Invalid memory address) (depending on the platform and how you
9255: achieved the overflow) as soon as the overflow happens. If it is not
9256: checked, overflows typically result in mysterious illegal memory
9257: accesses, producing @code{-9 throw} (Invalid memory address) or
9258: @code{-23 throw} (Address alignment exception); they might also destroy
9259: the internal data structure of @code{ALLOCATE} and friends, resulting in
9260: various errors in these words.
1.1 anton 9261:
9262: @item insufficient space for loop control parameters:
9263: @cindex insufficient space for loop control parameters
9264: like other return stack overflows.
9265:
9266: @item insufficient space in the dictionary:
9267: @cindex insufficient space in the dictionary
9268: @cindex dictionary overflow
1.12 anton 9269: If you try to allot (either directly with @code{allot}, or indirectly
9270: with @code{,}, @code{create} etc.) more memory than available in the
9271: dictionary, you get a @code{-8 throw} (Dictionary overflow). If you try
9272: to access memory beyond the end of the dictionary, the results are
9273: similar to stack overflows.
1.1 anton 9274:
9275: @item interpreting a word with undefined interpretation semantics:
9276: @cindex interpreting a word with undefined interpretation semantics
9277: @cindex Interpreting a compile-only word
9278: For some words, we have defined interpretation semantics. For the
9279: others: @code{-14 throw} (Interpreting a compile-only word).
9280:
9281: @item modifying the contents of the input buffer or a string literal:
9282: @cindex modifying the contents of the input buffer or a string literal
9283: These are located in writable memory and can be modified.
9284:
9285: @item overflow of the pictured numeric output string:
9286: @cindex overflow of the pictured numeric output string
9287: @cindex pictured numeric output string, overflow
1.24 anton 9288: @code{-17 throw} (Pictured numeric ouput string overflow).
1.1 anton 9289:
9290: @item parsed string overflow:
9291: @cindex parsed string overflow
9292: @code{PARSE} cannot overflow. @code{WORD} does not check for overflow.
9293:
9294: @item producing a result out of range:
9295: @cindex result out of range
9296: On two's complement machines, arithmetic is performed modulo
9297: 2**bits-per-cell for single arithmetic and 4**bits-per-cell for double
9298: arithmetic (with appropriate mapping for signed types). Division by zero
1.24 anton 9299: typically results in a @code{-10 throw} (divide by zero) or @code{-55
9300: throw} (floating point unidentified fault). @code{convert} and
9301: @code{>number} currently overflow silently.
1.1 anton 9302:
9303: @item reading from an empty data or return stack:
9304: @cindex stack empty
9305: @cindex stack underflow
1.24 anton 9306: @cindex return stack underflow
1.1 anton 9307: The data stack is checked by the outer (aka text) interpreter after
9308: every word executed. If it has underflowed, a @code{-4 throw} (Stack
9309: underflow) is performed. Apart from that, stacks may be checked or not,
1.24 anton 9310: depending on operating system, installation, and invocation. If they are
9311: caught by a check, they typically result in @code{-4 throw} (Stack
9312: underflow), @code{-6 throw} (Return stack underflow) or @code{-9 throw}
9313: (Invalid memory address), depending on the platform and which stack
9314: underflows and by how much. Note that even if the system uses checking
9315: (through the MMU), your program may have to underflow by a significant
9316: number of stack items to trigger the reaction (the reason for this is
9317: that the MMU, and therefore the checking, works with a page-size
9318: granularity). If there is no checking, the symptoms resulting from an
9319: underflow are similar to those from an overflow. Unbalanced return
9320: stack errors result in a variaty of symptoms, including @code{-9 throw}
9321: (Invalid memory address) and Illegal Instruction (typically @code{-260
9322: throw}).
1.1 anton 9323:
9324: @item unexpected end of the input buffer, resulting in an attempt to use a zero-length string as a name:
9325: @cindex unexpected end of the input buffer
9326: @cindex zero-length string as a name
9327: @cindex Attempt to use zero-length string as a name
9328: @code{Create} and its descendants perform a @code{-16 throw} (Attempt to
9329: use zero-length string as a name). Words like @code{'} probably will not
9330: find what they search. Note that it is possible to create zero-length
9331: names with @code{nextname} (should it not?).
9332:
9333: @item @code{>IN} greater than input buffer:
9334: @cindex @code{>IN} greater than input buffer
9335: The next invocation of a parsing word returns a string with length 0.
9336:
9337: @item @code{RECURSE} appears after @code{DOES>}:
9338: @cindex @code{RECURSE} appears after @code{DOES>}
9339: Compiles a recursive call to the defining word, not to the defined word.
9340:
9341: @item argument input source different than current input source for @code{RESTORE-INPUT}:
9342: @cindex argument input source different than current input source for @code{RESTORE-INPUT}
1.26 crook 9343: @cindex argument type mismatch, @code{RESTORE-INPUT}
1.1 anton 9344: @cindex @code{RESTORE-INPUT}, Argument type mismatch
9345: @code{-12 THROW}. Note that, once an input file is closed (e.g., because
9346: the end of the file was reached), its source-id may be
9347: reused. Therefore, restoring an input source specification referencing a
9348: closed file may lead to unpredictable results instead of a @code{-12
9349: THROW}.
9350:
9351: In the future, Gforth may be able to restore input source specifications
9352: from other than the current input source.
9353:
9354: @item data space containing definitions gets de-allocated:
9355: @cindex data space containing definitions gets de-allocated
9356: Deallocation with @code{allot} is not checked. This typically results in
9357: memory access faults or execution of illegal instructions.
9358:
9359: @item data space read/write with incorrect alignment:
9360: @cindex data space read/write with incorrect alignment
9361: @cindex alignment faults
1.26 crook 9362: @cindex address alignment exception
1.1 anton 9363: Processor-dependent. Typically results in a @code{-23 throw} (Address
1.12 anton 9364: alignment exception). Under Linux-Intel on a 486 or later processor with
1.1 anton 9365: alignment turned on, incorrect alignment results in a @code{-9 throw}
9366: (Invalid memory address). There are reportedly some processors with
1.12 anton 9367: alignment restrictions that do not report violations.
1.1 anton 9368:
9369: @item data space pointer not properly aligned, @code{,}, @code{C,}:
9370: @cindex data space pointer not properly aligned, @code{,}, @code{C,}
9371: Like other alignment errors.
9372:
9373: @item less than u+2 stack items (@code{PICK} and @code{ROLL}):
9374: Like other stack underflows.
9375:
9376: @item loop control parameters not available:
9377: @cindex loop control parameters not available
9378: Not checked. The counted loop words simply assume that the top of return
9379: stack items are loop control parameters and behave accordingly.
9380:
9381: @item most recent definition does not have a name (@code{IMMEDIATE}):
9382: @cindex most recent definition does not have a name (@code{IMMEDIATE})
9383: @cindex last word was headerless
9384: @code{abort" last word was headerless"}.
9385:
9386: @item name not defined by @code{VALUE} used by @code{TO}:
9387: @cindex name not defined by @code{VALUE} used by @code{TO}
9388: @cindex @code{TO} on non-@code{VALUE}s
9389: @cindex Invalid name argument, @code{TO}
9390: @code{-32 throw} (Invalid name argument) (unless name is a local or was
9391: defined by @code{CONSTANT}; in the latter case it just changes the constant).
9392:
9393: @item name not found (@code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]}):
9394: @cindex name not found (@code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]})
1.26 crook 9395: @cindex undefined word, @code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]}
1.1 anton 9396: @code{-13 throw} (Undefined word)
9397:
9398: @item parameters are not of the same type (@code{DO}, @code{?DO}, @code{WITHIN}):
9399: @cindex parameters are not of the same type (@code{DO}, @code{?DO}, @code{WITHIN})
9400: Gforth behaves as if they were of the same type. I.e., you can predict
9401: the behaviour by interpreting all parameters as, e.g., signed.
9402:
9403: @item @code{POSTPONE} or @code{[COMPILE]} applied to @code{TO}:
9404: @cindex @code{POSTPONE} or @code{[COMPILE]} applied to @code{TO}
9405: Assume @code{: X POSTPONE TO ; IMMEDIATE}. @code{X} performs the
9406: compilation semantics of @code{TO}.
9407:
9408: @item String longer than a counted string returned by @code{WORD}:
1.26 crook 9409: @cindex string longer than a counted string returned by @code{WORD}
1.1 anton 9410: @cindex @code{WORD}, string overflow
9411: Not checked. The string will be ok, but the count will, of course,
9412: contain only the least significant bits of the length.
9413:
9414: @item u greater than or equal to the number of bits in a cell (@code{LSHIFT}, @code{RSHIFT}):
9415: @cindex @code{LSHIFT}, large shift counts
9416: @cindex @code{RSHIFT}, large shift counts
9417: Processor-dependent. Typical behaviours are returning 0 and using only
9418: the low bits of the shift count.
9419:
9420: @item word not defined via @code{CREATE}:
9421: @cindex @code{>BODY} of non-@code{CREATE}d words
9422: @code{>BODY} produces the PFA of the word no matter how it was defined.
9423:
9424: @cindex @code{DOES>} of non-@code{CREATE}d words
9425: @code{DOES>} changes the execution semantics of the last defined word no
9426: matter how it was defined. E.g., @code{CONSTANT DOES>} is equivalent to
9427: @code{CREATE , DOES>}.
9428:
9429: @item words improperly used outside @code{<#} and @code{#>}:
9430: Not checked. As usual, you can expect memory faults.
9431:
9432: @end table
9433:
9434:
9435: @c ---------------------------------------------------------------------
9436: @node core-other, , core-ambcond, The Core Words
9437: @subsection Other system documentation
9438: @c ---------------------------------------------------------------------
9439: @cindex other system documentation, core words
9440: @cindex core words, other system documentation
9441:
9442: @table @i
9443: @item nonstandard words using @code{PAD}:
9444: @cindex @code{PAD} use by nonstandard words
9445: None.
9446:
9447: @item operator's terminal facilities available:
9448: @cindex operator's terminal facilities available
9449: After processing the command line, Gforth goes into interactive mode,
9450: and you can give commands to Gforth interactively. The actual facilities
9451: available depend on how you invoke Gforth.
9452:
9453: @item program data space available:
9454: @cindex program data space available
9455: @cindex data space available
9456: @code{UNUSED .} gives the remaining dictionary space. The total
9457: dictionary space can be specified with the @code{-m} switch
9458: (@pxref{Invoking Gforth}) when Gforth starts up.
9459:
9460: @item return stack space available:
9461: @cindex return stack space available
9462: You can compute the total return stack space in cells with
9463: @code{s" RETURN-STACK-CELLS" environment? drop .}. You can specify it at
9464: startup time with the @code{-r} switch (@pxref{Invoking Gforth}).
9465:
9466: @item stack space available:
9467: @cindex stack space available
9468: You can compute the total data stack space in cells with
9469: @code{s" STACK-CELLS" environment? drop .}. You can specify it at
9470: startup time with the @code{-d} switch (@pxref{Invoking Gforth}).
9471:
9472: @item system dictionary space required, in address units:
9473: @cindex system dictionary space required, in address units
9474: Type @code{here forthstart - .} after startup. At the time of this
9475: writing, this gives 80080 (bytes) on a 32-bit system.
9476: @end table
9477:
9478:
9479: @c =====================================================================
9480: @node The optional Block word set, The optional Double Number word set, The Core Words, ANS conformance
9481: @section The optional Block word set
9482: @c =====================================================================
9483: @cindex system documentation, block words
9484: @cindex block words, system documentation
9485:
9486: @menu
9487: * block-idef:: Implementation Defined Options
9488: * block-ambcond:: Ambiguous Conditions
9489: * block-other:: Other System Documentation
9490: @end menu
9491:
9492:
9493: @c ---------------------------------------------------------------------
9494: @node block-idef, block-ambcond, The optional Block word set, The optional Block word set
9495: @subsection Implementation Defined Options
9496: @c ---------------------------------------------------------------------
9497: @cindex implementation-defined options, block words
9498: @cindex block words, implementation-defined options
9499:
9500: @table @i
9501: @item the format for display by @code{LIST}:
9502: @cindex @code{LIST} display format
9503: First the screen number is displayed, then 16 lines of 64 characters,
9504: each line preceded by the line number.
9505:
9506: @item the length of a line affected by @code{\}:
9507: @cindex length of a line affected by @code{\}
9508: @cindex @code{\}, line length in blocks
9509: 64 characters.
9510: @end table
9511:
9512:
9513: @c ---------------------------------------------------------------------
9514: @node block-ambcond, block-other, block-idef, The optional Block word set
9515: @subsection Ambiguous conditions
9516: @c ---------------------------------------------------------------------
9517: @cindex block words, ambiguous conditions
9518: @cindex ambiguous conditions, block words
9519:
9520: @table @i
9521: @item correct block read was not possible:
9522: @cindex block read not possible
9523: Typically results in a @code{throw} of some OS-derived value (between
9524: -512 and -2048). If the blocks file was just not long enough, blanks are
9525: supplied for the missing portion.
9526:
9527: @item I/O exception in block transfer:
9528: @cindex I/O exception in block transfer
9529: @cindex block transfer, I/O exception
9530: Typically results in a @code{throw} of some OS-derived value (between
9531: -512 and -2048).
9532:
9533: @item invalid block number:
9534: @cindex invalid block number
9535: @cindex block number invalid
9536: @code{-35 throw} (Invalid block number)
9537:
9538: @item a program directly alters the contents of @code{BLK}:
9539: @cindex @code{BLK}, altering @code{BLK}
9540: The input stream is switched to that other block, at the same
9541: position. If the storing to @code{BLK} happens when interpreting
9542: non-block input, the system will get quite confused when the block ends.
9543:
9544: @item no current block buffer for @code{UPDATE}:
9545: @cindex @code{UPDATE}, no current block buffer
9546: @code{UPDATE} has no effect.
9547:
9548: @end table
9549:
9550: @c ---------------------------------------------------------------------
9551: @node block-other, , block-ambcond, The optional Block word set
9552: @subsection Other system documentation
9553: @c ---------------------------------------------------------------------
9554: @cindex other system documentation, block words
9555: @cindex block words, other system documentation
9556:
9557: @table @i
9558: @item any restrictions a multiprogramming system places on the use of buffer addresses:
9559: No restrictions (yet).
9560:
9561: @item the number of blocks available for source and data:
9562: depends on your disk space.
9563:
9564: @end table
9565:
9566:
9567: @c =====================================================================
9568: @node The optional Double Number word set, The optional Exception word set, The optional Block word set, ANS conformance
9569: @section The optional Double Number word set
9570: @c =====================================================================
9571: @cindex system documentation, double words
9572: @cindex double words, system documentation
9573:
9574: @menu
9575: * double-ambcond:: Ambiguous Conditions
9576: @end menu
9577:
9578:
9579: @c ---------------------------------------------------------------------
9580: @node double-ambcond, , The optional Double Number word set, The optional Double Number word set
9581: @subsection Ambiguous conditions
9582: @c ---------------------------------------------------------------------
9583: @cindex double words, ambiguous conditions
9584: @cindex ambiguous conditions, double words
9585:
9586: @table @i
1.29 crook 9587: @item @i{d} outside of range of @i{n} in @code{D>S}:
9588: @cindex @code{D>S}, @i{d} out of range of @i{n}
9589: The least significant cell of @i{d} is produced.
1.1 anton 9590:
9591: @end table
9592:
9593:
9594: @c =====================================================================
9595: @node The optional Exception word set, The optional Facility word set, The optional Double Number word set, ANS conformance
9596: @section The optional Exception word set
9597: @c =====================================================================
9598: @cindex system documentation, exception words
9599: @cindex exception words, system documentation
9600:
9601: @menu
9602: * exception-idef:: Implementation Defined Options
9603: @end menu
9604:
9605:
9606: @c ---------------------------------------------------------------------
9607: @node exception-idef, , The optional Exception word set, The optional Exception word set
9608: @subsection Implementation Defined Options
9609: @c ---------------------------------------------------------------------
9610: @cindex implementation-defined options, exception words
9611: @cindex exception words, implementation-defined options
9612:
9613: @table @i
9614: @item @code{THROW}-codes used in the system:
9615: @cindex @code{THROW}-codes used in the system
9616: The codes -256@minus{}-511 are used for reporting signals. The mapping
1.29 crook 9617: from OS signal numbers to throw codes is -256@minus{}@i{signal}. The
1.1 anton 9618: codes -512@minus{}-2047 are used for OS errors (for file and memory
9619: allocation operations). The mapping from OS error numbers to throw codes
9620: is -512@minus{}@code{errno}. One side effect of this mapping is that
9621: undefined OS errors produce a message with a strange number; e.g.,
9622: @code{-1000 THROW} results in @code{Unknown error 488} on my system.
9623: @end table
9624:
9625: @c =====================================================================
9626: @node The optional Facility word set, The optional File-Access word set, The optional Exception word set, ANS conformance
9627: @section The optional Facility word set
9628: @c =====================================================================
9629: @cindex system documentation, facility words
9630: @cindex facility words, system documentation
9631:
9632: @menu
9633: * facility-idef:: Implementation Defined Options
9634: * facility-ambcond:: Ambiguous Conditions
9635: @end menu
9636:
9637:
9638: @c ---------------------------------------------------------------------
9639: @node facility-idef, facility-ambcond, The optional Facility word set, The optional Facility word set
9640: @subsection Implementation Defined Options
9641: @c ---------------------------------------------------------------------
9642: @cindex implementation-defined options, facility words
9643: @cindex facility words, implementation-defined options
9644:
9645: @table @i
9646: @item encoding of keyboard events (@code{EKEY}):
9647: @cindex keyboard events, encoding in @code{EKEY}
9648: @cindex @code{EKEY}, encoding of keyboard events
9649: Not yet implemented.
9650:
9651: @item duration of a system clock tick:
9652: @cindex duration of a system clock tick
9653: @cindex clock tick duration
9654: System dependent. With respect to @code{MS}, the time is specified in
9655: microseconds. How well the OS and the hardware implement this, is
9656: another question.
9657:
9658: @item repeatability to be expected from the execution of @code{MS}:
9659: @cindex repeatability to be expected from the execution of @code{MS}
9660: @cindex @code{MS}, repeatability to be expected
9661: System dependent. On Unix, a lot depends on load. If the system is
9662: lightly loaded, and the delay is short enough that Gforth does not get
9663: swapped out, the performance should be acceptable. Under MS-DOS and
9664: other single-tasking systems, it should be good.
9665:
9666: @end table
9667:
9668:
9669: @c ---------------------------------------------------------------------
9670: @node facility-ambcond, , facility-idef, The optional Facility word set
9671: @subsection Ambiguous conditions
9672: @c ---------------------------------------------------------------------
9673: @cindex facility words, ambiguous conditions
9674: @cindex ambiguous conditions, facility words
9675:
9676: @table @i
9677: @item @code{AT-XY} can't be performed on user output device:
9678: @cindex @code{AT-XY} can't be performed on user output device
9679: Largely terminal dependent. No range checks are done on the arguments.
9680: No errors are reported. You may see some garbage appearing, you may see
9681: simply nothing happen.
9682:
9683: @end table
9684:
9685:
9686: @c =====================================================================
9687: @node The optional File-Access word set, The optional Floating-Point word set, The optional Facility word set, ANS conformance
9688: @section The optional File-Access word set
9689: @c =====================================================================
9690: @cindex system documentation, file words
9691: @cindex file words, system documentation
9692:
9693: @menu
9694: * file-idef:: Implementation Defined Options
9695: * file-ambcond:: Ambiguous Conditions
9696: @end menu
9697:
9698: @c ---------------------------------------------------------------------
9699: @node file-idef, file-ambcond, The optional File-Access word set, The optional File-Access word set
9700: @subsection Implementation Defined Options
9701: @c ---------------------------------------------------------------------
9702: @cindex implementation-defined options, file words
9703: @cindex file words, implementation-defined options
9704:
9705: @table @i
9706: @item file access methods used:
9707: @cindex file access methods used
9708: @code{R/O}, @code{R/W} and @code{BIN} work as you would
9709: expect. @code{W/O} translates into the C file opening mode @code{w} (or
9710: @code{wb}): The file is cleared, if it exists, and created, if it does
9711: not (with both @code{open-file} and @code{create-file}). Under Unix
9712: @code{create-file} creates a file with 666 permissions modified by your
9713: umask.
9714:
9715: @item file exceptions:
9716: @cindex file exceptions
9717: The file words do not raise exceptions (except, perhaps, memory access
9718: faults when you pass illegal addresses or file-ids).
9719:
9720: @item file line terminator:
9721: @cindex file line terminator
9722: System-dependent. Gforth uses C's newline character as line
9723: terminator. What the actual character code(s) of this are is
9724: system-dependent.
9725:
9726: @item file name format:
9727: @cindex file name format
9728: System dependent. Gforth just uses the file name format of your OS.
9729:
9730: @item information returned by @code{FILE-STATUS}:
9731: @cindex @code{FILE-STATUS}, returned information
9732: @code{FILE-STATUS} returns the most powerful file access mode allowed
9733: for the file: Either @code{R/O}, @code{W/O} or @code{R/W}. If the file
9734: cannot be accessed, @code{R/O BIN} is returned. @code{BIN} is applicable
9735: along with the returned mode.
9736:
9737: @item input file state after an exception when including source:
9738: @cindex exception when including source
9739: All files that are left via the exception are closed.
9740:
1.29 crook 9741: @item @i{ior} values and meaning:
9742: @cindex @i{ior} values and meaning
9743: The @i{ior}s returned by the file and memory allocation words are
1.1 anton 9744: intended as throw codes. They typically are in the range
9745: -512@minus{}-2047 of OS errors. The mapping from OS error numbers to
1.29 crook 9746: @i{ior}s is -512@minus{}@i{errno}.
1.1 anton 9747:
9748: @item maximum depth of file input nesting:
9749: @cindex maximum depth of file input nesting
9750: @cindex file input nesting, maximum depth
9751: limited by the amount of return stack, locals/TIB stack, and the number
9752: of open files available. This should not give you troubles.
9753:
9754: @item maximum size of input line:
9755: @cindex maximum size of input line
9756: @cindex input line size, maximum
9757: @code{/line}. Currently 255.
9758:
9759: @item methods of mapping block ranges to files:
9760: @cindex mapping block ranges to files
9761: @cindex files containing blocks
9762: @cindex blocks in files
9763: By default, blocks are accessed in the file @file{blocks.fb} in the
9764: current working directory. The file can be switched with @code{USE}.
9765:
9766: @item number of string buffers provided by @code{S"}:
9767: @cindex @code{S"}, number of string buffers
9768: 1
9769:
9770: @item size of string buffer used by @code{S"}:
9771: @cindex @code{S"}, size of string buffer
9772: @code{/line}. currently 255.
9773:
9774: @end table
9775:
9776: @c ---------------------------------------------------------------------
9777: @node file-ambcond, , file-idef, The optional File-Access word set
9778: @subsection Ambiguous conditions
9779: @c ---------------------------------------------------------------------
9780: @cindex file words, ambiguous conditions
9781: @cindex ambiguous conditions, file words
9782:
9783: @table @i
9784: @item attempting to position a file outside its boundaries:
9785: @cindex @code{REPOSITION-FILE}, outside the file's boundaries
9786: @code{REPOSITION-FILE} is performed as usual: Afterwards,
9787: @code{FILE-POSITION} returns the value given to @code{REPOSITION-FILE}.
9788:
9789: @item attempting to read from file positions not yet written:
9790: @cindex reading from file positions not yet written
9791: End-of-file, i.e., zero characters are read and no error is reported.
9792:
1.29 crook 9793: @item @i{file-id} is invalid (@code{INCLUDE-FILE}):
9794: @cindex @code{INCLUDE-FILE}, @i{file-id} is invalid
1.1 anton 9795: An appropriate exception may be thrown, but a memory fault or other
9796: problem is more probable.
9797:
1.29 crook 9798: @item I/O exception reading or closing @i{file-id} (@code{INCLUDE-FILE}, @code{INCLUDED}):
9799: @cindex @code{INCLUDE-FILE}, I/O exception reading or closing @i{file-id}
9800: @cindex @code{INCLUDED}, I/O exception reading or closing @i{file-id}
9801: The @i{ior} produced by the operation, that discovered the problem, is
1.1 anton 9802: thrown.
9803:
9804: @item named file cannot be opened (@code{INCLUDED}):
9805: @cindex @code{INCLUDED}, named file cannot be opened
1.29 crook 9806: The @i{ior} produced by @code{open-file} is thrown.
1.1 anton 9807:
9808: @item requesting an unmapped block number:
9809: @cindex unmapped block numbers
9810: There are no unmapped legal block numbers. On some operating systems,
9811: writing a block with a large number may overflow the file system and
9812: have an error message as consequence.
9813:
9814: @item using @code{source-id} when @code{blk} is non-zero:
9815: @cindex @code{SOURCE-ID}, behaviour when @code{BLK} is non-zero
9816: @code{source-id} performs its function. Typically it will give the id of
9817: the source which loaded the block. (Better ideas?)
9818:
9819: @end table
9820:
9821:
9822: @c =====================================================================
9823: @node The optional Floating-Point word set, The optional Locals word set, The optional File-Access word set, ANS conformance
9824: @section The optional Floating-Point word set
9825: @c =====================================================================
9826: @cindex system documentation, floating-point words
9827: @cindex floating-point words, system documentation
9828:
9829: @menu
9830: * floating-idef:: Implementation Defined Options
9831: * floating-ambcond:: Ambiguous Conditions
9832: @end menu
9833:
9834:
9835: @c ---------------------------------------------------------------------
9836: @node floating-idef, floating-ambcond, The optional Floating-Point word set, The optional Floating-Point word set
9837: @subsection Implementation Defined Options
9838: @c ---------------------------------------------------------------------
9839: @cindex implementation-defined options, floating-point words
9840: @cindex floating-point words, implementation-defined options
9841:
9842: @table @i
9843: @item format and range of floating point numbers:
9844: @cindex format and range of floating point numbers
9845: @cindex floating point numbers, format and range
9846: System-dependent; the @code{double} type of C.
9847:
1.29 crook 9848: @item results of @code{REPRESENT} when @i{float} is out of range:
9849: @cindex @code{REPRESENT}, results when @i{float} is out of range
1.1 anton 9850: System dependent; @code{REPRESENT} is implemented using the C library
9851: function @code{ecvt()} and inherits its behaviour in this respect.
9852:
9853: @item rounding or truncation of floating-point numbers:
9854: @cindex rounding of floating-point numbers
9855: @cindex truncation of floating-point numbers
9856: @cindex floating-point numbers, rounding or truncation
9857: System dependent; the rounding behaviour is inherited from the hosting C
9858: compiler. IEEE-FP-based (i.e., most) systems by default round to
9859: nearest, and break ties by rounding to even (i.e., such that the last
9860: bit of the mantissa is 0).
9861:
9862: @item size of floating-point stack:
9863: @cindex floating-point stack size
9864: @code{s" FLOATING-STACK" environment? drop .} gives the total size of
9865: the floating-point stack (in floats). You can specify this on startup
9866: with the command-line option @code{-f} (@pxref{Invoking Gforth}).
9867:
9868: @item width of floating-point stack:
9869: @cindex floating-point stack width
9870: @code{1 floats}.
9871:
9872: @end table
9873:
9874:
9875: @c ---------------------------------------------------------------------
9876: @node floating-ambcond, , floating-idef, The optional Floating-Point word set
9877: @subsection Ambiguous conditions
9878: @c ---------------------------------------------------------------------
9879: @cindex floating-point words, ambiguous conditions
9880: @cindex ambiguous conditions, floating-point words
9881:
9882: @table @i
9883: @item @code{df@@} or @code{df!} used with an address that is not double-float aligned:
9884: @cindex @code{df@@} or @code{df!} used with an address that is not double-float aligned
9885: System-dependent. Typically results in a @code{-23 THROW} like other
9886: alignment violations.
9887:
9888: @item @code{f@@} or @code{f!} used with an address that is not float aligned:
9889: @cindex @code{f@@} used with an address that is not float aligned
9890: @cindex @code{f!} used with an address that is not float aligned
9891: System-dependent. Typically results in a @code{-23 THROW} like other
9892: alignment violations.
9893:
9894: @item floating-point result out of range:
9895: @cindex floating-point result out of range
9896: System-dependent. Can result in a @code{-55 THROW} (Floating-point
9897: unidentified fault), or can produce a special value representing, e.g.,
9898: Infinity.
9899:
9900: @item @code{sf@@} or @code{sf!} used with an address that is not single-float aligned:
9901: @cindex @code{sf@@} or @code{sf!} used with an address that is not single-float aligned
9902: System-dependent. Typically results in an alignment fault like other
9903: alignment violations.
9904:
1.35 anton 9905: @item @code{base} is not decimal (@code{REPRESENT}, @code{F.}, @code{FE.}, @code{FS.}):
9906: @cindex @code{base} is not decimal (@code{REPRESENT}, @code{F.}, @code{FE.}, @code{FS.})
1.1 anton 9907: The floating-point number is converted into decimal nonetheless.
9908:
9909: @item Both arguments are equal to zero (@code{FATAN2}):
9910: @cindex @code{FATAN2}, both arguments are equal to zero
9911: System-dependent. @code{FATAN2} is implemented using the C library
9912: function @code{atan2()}.
9913:
1.29 crook 9914: @item Using @code{FTAN} on an argument @i{r1} where cos(@i{r1}) is zero:
9915: @cindex @code{FTAN} on an argument @i{r1} where cos(@i{r1}) is zero
9916: System-dependent. Anyway, typically the cos of @i{r1} will not be zero
1.1 anton 9917: because of small errors and the tan will be a very large (or very small)
9918: but finite number.
9919:
1.29 crook 9920: @item @i{d} cannot be presented precisely as a float in @code{D>F}:
9921: @cindex @code{D>F}, @i{d} cannot be presented precisely as a float
1.1 anton 9922: The result is rounded to the nearest float.
9923:
9924: @item dividing by zero:
9925: @cindex dividing by zero, floating-point
9926: @cindex floating-point dividing by zero
9927: @cindex floating-point unidentified fault, FP divide-by-zero
9928: @code{-55 throw} (Floating-point unidentified fault)
9929:
9930: @item exponent too big for conversion (@code{DF!}, @code{DF@@}, @code{SF!}, @code{SF@@}):
9931: @cindex exponent too big for conversion (@code{DF!}, @code{DF@@}, @code{SF!}, @code{SF@@})
9932: System dependent. On IEEE-FP based systems the number is converted into
9933: an infinity.
9934:
1.29 crook 9935: @item @i{float}<1 (@code{FACOSH}):
9936: @cindex @code{FACOSH}, @i{float}<1
1.1 anton 9937: @cindex floating-point unidentified fault, @code{FACOSH}
9938: @code{-55 throw} (Floating-point unidentified fault)
9939:
1.29 crook 9940: @item @i{float}=<-1 (@code{FLNP1}):
9941: @cindex @code{FLNP1}, @i{float}=<-1
1.1 anton 9942: @cindex floating-point unidentified fault, @code{FLNP1}
9943: @code{-55 throw} (Floating-point unidentified fault). On IEEE-FP systems
1.29 crook 9944: negative infinity is typically produced for @i{float}=-1.
1.1 anton 9945:
1.29 crook 9946: @item @i{float}=<0 (@code{FLN}, @code{FLOG}):
9947: @cindex @code{FLN}, @i{float}=<0
9948: @cindex @code{FLOG}, @i{float}=<0
1.1 anton 9949: @cindex floating-point unidentified fault, @code{FLN} or @code{FLOG}
9950: @code{-55 throw} (Floating-point unidentified fault). On IEEE-FP systems
1.29 crook 9951: negative infinity is typically produced for @i{float}=0.
1.1 anton 9952:
1.29 crook 9953: @item @i{float}<0 (@code{FASINH}, @code{FSQRT}):
9954: @cindex @code{FASINH}, @i{float}<0
9955: @cindex @code{FSQRT}, @i{float}<0
1.1 anton 9956: @cindex floating-point unidentified fault, @code{FASINH} or @code{FSQRT}
9957: @code{-55 throw} (Floating-point unidentified fault). @code{fasinh}
9958: produces values for these inputs on my Linux box (Bug in the C library?)
9959:
1.29 crook 9960: @item |@i{float}|>1 (@code{FACOS}, @code{FASIN}, @code{FATANH}):
9961: @cindex @code{FACOS}, |@i{float}|>1
9962: @cindex @code{FASIN}, |@i{float}|>1
9963: @cindex @code{FATANH}, |@i{float}|>1
1.1 anton 9964: @cindex floating-point unidentified fault, @code{FACOS}, @code{FASIN} or @code{FATANH}
9965: @code{-55 throw} (Floating-point unidentified fault).
9966:
1.29 crook 9967: @item integer part of float cannot be represented by @i{d} in @code{F>D}:
9968: @cindex @code{F>D}, integer part of float cannot be represented by @i{d}
1.1 anton 9969: @cindex floating-point unidentified fault, @code{F>D}
9970: @code{-55 throw} (Floating-point unidentified fault).
9971:
9972: @item string larger than pictured numeric output area (@code{f.}, @code{fe.}, @code{fs.}):
9973: @cindex string larger than pictured numeric output area (@code{f.}, @code{fe.}, @code{fs.})
9974: This does not happen.
9975: @end table
9976:
9977: @c =====================================================================
9978: @node The optional Locals word set, The optional Memory-Allocation word set, The optional Floating-Point word set, ANS conformance
9979: @section The optional Locals word set
9980: @c =====================================================================
9981: @cindex system documentation, locals words
9982: @cindex locals words, system documentation
9983:
9984: @menu
9985: * locals-idef:: Implementation Defined Options
9986: * locals-ambcond:: Ambiguous Conditions
9987: @end menu
9988:
9989:
9990: @c ---------------------------------------------------------------------
9991: @node locals-idef, locals-ambcond, The optional Locals word set, The optional Locals word set
9992: @subsection Implementation Defined Options
9993: @c ---------------------------------------------------------------------
9994: @cindex implementation-defined options, locals words
9995: @cindex locals words, implementation-defined options
9996:
9997: @table @i
9998: @item maximum number of locals in a definition:
9999: @cindex maximum number of locals in a definition
10000: @cindex locals, maximum number in a definition
10001: @code{s" #locals" environment? drop .}. Currently 15. This is a lower
10002: bound, e.g., on a 32-bit machine there can be 41 locals of up to 8
10003: characters. The number of locals in a definition is bounded by the size
10004: of locals-buffer, which contains the names of the locals.
10005:
10006: @end table
10007:
10008:
10009: @c ---------------------------------------------------------------------
10010: @node locals-ambcond, , locals-idef, The optional Locals word set
10011: @subsection Ambiguous conditions
10012: @c ---------------------------------------------------------------------
10013: @cindex locals words, ambiguous conditions
10014: @cindex ambiguous conditions, locals words
10015:
10016: @table @i
10017: @item executing a named local in interpretation state:
10018: @cindex local in interpretation state
10019: @cindex Interpreting a compile-only word, for a local
10020: Locals have no interpretation semantics. If you try to perform the
10021: interpretation semantics, you will get a @code{-14 throw} somewhere
10022: (Interpreting a compile-only word). If you perform the compilation
10023: semantics, the locals access will be compiled (irrespective of state).
10024:
1.29 crook 10025: @item @i{name} not defined by @code{VALUE} or @code{(LOCAL)} (@code{TO}):
1.1 anton 10026: @cindex name not defined by @code{VALUE} or @code{(LOCAL)} used by @code{TO}
10027: @cindex @code{TO} on non-@code{VALUE}s and non-locals
10028: @cindex Invalid name argument, @code{TO}
10029: @code{-32 throw} (Invalid name argument)
10030:
10031: @end table
10032:
10033:
10034: @c =====================================================================
10035: @node The optional Memory-Allocation word set, The optional Programming-Tools word set, The optional Locals word set, ANS conformance
10036: @section The optional Memory-Allocation word set
10037: @c =====================================================================
10038: @cindex system documentation, memory-allocation words
10039: @cindex memory-allocation words, system documentation
10040:
10041: @menu
10042: * memory-idef:: Implementation Defined Options
10043: @end menu
10044:
10045:
10046: @c ---------------------------------------------------------------------
10047: @node memory-idef, , The optional Memory-Allocation word set, The optional Memory-Allocation word set
10048: @subsection Implementation Defined Options
10049: @c ---------------------------------------------------------------------
10050: @cindex implementation-defined options, memory-allocation words
10051: @cindex memory-allocation words, implementation-defined options
10052:
10053: @table @i
1.29 crook 10054: @item values and meaning of @i{ior}:
10055: @cindex @i{ior} values and meaning
10056: The @i{ior}s returned by the file and memory allocation words are
1.1 anton 10057: intended as throw codes. They typically are in the range
10058: -512@minus{}-2047 of OS errors. The mapping from OS error numbers to
1.29 crook 10059: @i{ior}s is -512@minus{}@i{errno}.
1.1 anton 10060:
10061: @end table
10062:
10063: @c =====================================================================
10064: @node The optional Programming-Tools word set, The optional Search-Order word set, The optional Memory-Allocation word set, ANS conformance
10065: @section The optional Programming-Tools word set
10066: @c =====================================================================
10067: @cindex system documentation, programming-tools words
10068: @cindex programming-tools words, system documentation
10069:
10070: @menu
10071: * programming-idef:: Implementation Defined Options
10072: * programming-ambcond:: Ambiguous Conditions
10073: @end menu
10074:
10075:
10076: @c ---------------------------------------------------------------------
10077: @node programming-idef, programming-ambcond, The optional Programming-Tools word set, The optional Programming-Tools word set
10078: @subsection Implementation Defined Options
10079: @c ---------------------------------------------------------------------
10080: @cindex implementation-defined options, programming-tools words
10081: @cindex programming-tools words, implementation-defined options
10082:
10083: @table @i
10084: @item ending sequence for input following @code{;CODE} and @code{CODE}:
10085: @cindex @code{;CODE} ending sequence
10086: @cindex @code{CODE} ending sequence
10087: @code{END-CODE}
10088:
10089: @item manner of processing input following @code{;CODE} and @code{CODE}:
10090: @cindex @code{;CODE}, processing input
10091: @cindex @code{CODE}, processing input
10092: The @code{ASSEMBLER} vocabulary is pushed on the search order stack, and
10093: the input is processed by the text interpreter, (starting) in interpret
10094: state.
10095:
10096: @item search order capability for @code{EDITOR} and @code{ASSEMBLER}:
10097: @cindex @code{ASSEMBLER}, search order capability
10098: The ANS Forth search order word set.
10099:
10100: @item source and format of display by @code{SEE}:
10101: @cindex @code{SEE}, source and format of output
10102: The source for @code{see} is the intermediate code used by the inner
10103: interpreter. The current @code{see} tries to output Forth source code
10104: as well as possible.
10105:
10106: @end table
10107:
10108: @c ---------------------------------------------------------------------
10109: @node programming-ambcond, , programming-idef, The optional Programming-Tools word set
10110: @subsection Ambiguous conditions
10111: @c ---------------------------------------------------------------------
10112: @cindex programming-tools words, ambiguous conditions
10113: @cindex ambiguous conditions, programming-tools words
10114:
10115: @table @i
10116:
1.21 crook 10117: @item deleting the compilation word list (@code{FORGET}):
10118: @cindex @code{FORGET}, deleting the compilation word list
1.1 anton 10119: Not implemented (yet).
10120:
1.29 crook 10121: @item fewer than @i{u}+1 items on the control-flow stack (@code{CS-PICK}, @code{CS-ROLL}):
10122: @cindex @code{CS-PICK}, fewer than @i{u}+1 items on the control flow-stack
10123: @cindex @code{CS-ROLL}, fewer than @i{u}+1 items on the control flow-stack
1.1 anton 10124: @cindex control-flow stack underflow
10125: This typically results in an @code{abort"} with a descriptive error
10126: message (may change into a @code{-22 throw} (Control structure mismatch)
10127: in the future). You may also get a memory access error. If you are
10128: unlucky, this ambiguous condition is not caught.
10129:
1.29 crook 10130: @item @i{name} can't be found (@code{FORGET}):
10131: @cindex @code{FORGET}, @i{name} can't be found
1.1 anton 10132: Not implemented (yet).
10133:
1.29 crook 10134: @item @i{name} not defined via @code{CREATE}:
10135: @cindex @code{;CODE}, @i{name} not defined via @code{CREATE}
1.1 anton 10136: @code{;CODE} behaves like @code{DOES>} in this respect, i.e., it changes
10137: the execution semantics of the last defined word no matter how it was
10138: defined.
10139:
10140: @item @code{POSTPONE} applied to @code{[IF]}:
10141: @cindex @code{POSTPONE} applied to @code{[IF]}
10142: @cindex @code{[IF]} and @code{POSTPONE}
10143: After defining @code{: X POSTPONE [IF] ; IMMEDIATE}. @code{X} is
10144: equivalent to @code{[IF]}.
10145:
10146: @item reaching the end of the input source before matching @code{[ELSE]} or @code{[THEN]}:
10147: @cindex @code{[IF]}, end of the input source before matching @code{[ELSE]} or @code{[THEN]}
10148: Continue in the same state of conditional compilation in the next outer
10149: input source. Currently there is no warning to the user about this.
10150:
10151: @item removing a needed definition (@code{FORGET}):
10152: @cindex @code{FORGET}, removing a needed definition
10153: Not implemented (yet).
10154:
10155: @end table
10156:
10157:
10158: @c =====================================================================
10159: @node The optional Search-Order word set, , The optional Programming-Tools word set, ANS conformance
10160: @section The optional Search-Order word set
10161: @c =====================================================================
10162: @cindex system documentation, search-order words
10163: @cindex search-order words, system documentation
10164:
10165: @menu
10166: * search-idef:: Implementation Defined Options
10167: * search-ambcond:: Ambiguous Conditions
10168: @end menu
10169:
10170:
10171: @c ---------------------------------------------------------------------
10172: @node search-idef, search-ambcond, The optional Search-Order word set, The optional Search-Order word set
10173: @subsection Implementation Defined Options
10174: @c ---------------------------------------------------------------------
10175: @cindex implementation-defined options, search-order words
10176: @cindex search-order words, implementation-defined options
10177:
10178: @table @i
10179: @item maximum number of word lists in search order:
10180: @cindex maximum number of word lists in search order
10181: @cindex search order, maximum depth
10182: @code{s" wordlists" environment? drop .}. Currently 16.
10183:
10184: @item minimum search order:
10185: @cindex minimum search order
10186: @cindex search order, minimum
10187: @code{root root}.
10188:
10189: @end table
10190:
10191: @c ---------------------------------------------------------------------
10192: @node search-ambcond, , search-idef, The optional Search-Order word set
10193: @subsection Ambiguous conditions
10194: @c ---------------------------------------------------------------------
10195: @cindex search-order words, ambiguous conditions
10196: @cindex ambiguous conditions, search-order words
10197:
10198: @table @i
1.21 crook 10199: @item changing the compilation word list (during compilation):
10200: @cindex changing the compilation word list (during compilation)
10201: @cindex compilation word list, change before definition ends
10202: The word is entered into the word list that was the compilation word list
1.1 anton 10203: at the start of the definition. Any changes to the name field (e.g.,
10204: @code{immediate}) or the code field (e.g., when executing @code{DOES>})
10205: are applied to the latest defined word (as reported by @code{last} or
1.21 crook 10206: @code{lastxt}), if possible, irrespective of the compilation word list.
1.1 anton 10207:
10208: @item search order empty (@code{previous}):
10209: @cindex @code{previous}, search order empty
1.26 crook 10210: @cindex vocstack empty, @code{previous}
1.1 anton 10211: @code{abort" Vocstack empty"}.
10212:
10213: @item too many word lists in search order (@code{also}):
10214: @cindex @code{also}, too many word lists in search order
1.26 crook 10215: @cindex vocstack full, @code{also}
1.1 anton 10216: @code{abort" Vocstack full"}.
10217:
10218: @end table
10219:
10220: @c ***************************************************************
10221: @node Model, Integrating Gforth, ANS conformance, Top
10222: @chapter Model
10223:
10224: This chapter has yet to be written. It will contain information, on
10225: which internal structures you can rely.
10226:
10227: @c ***************************************************************
10228: @node Integrating Gforth, Emacs and Gforth, Model, Top
10229: @chapter Integrating Gforth into C programs
10230:
10231: This is not yet implemented.
10232:
10233: Several people like to use Forth as scripting language for applications
10234: that are otherwise written in C, C++, or some other language.
10235:
10236: The Forth system ATLAST provides facilities for embedding it into
10237: applications; unfortunately it has several disadvantages: most
10238: importantly, it is not based on ANS Forth, and it is apparently dead
10239: (i.e., not developed further and not supported). The facilities
1.21 crook 10240: provided by Gforth in this area are inspired by ATLAST's facilities, so
1.1 anton 10241: making the switch should not be hard.
10242:
10243: We also tried to design the interface such that it can easily be
10244: implemented by other Forth systems, so that we may one day arrive at a
10245: standardized interface. Such a standard interface would allow you to
10246: replace the Forth system without having to rewrite C code.
10247:
10248: You embed the Gforth interpreter by linking with the library
10249: @code{libgforth.a} (give the compiler the option @code{-lgforth}). All
10250: global symbols in this library that belong to the interface, have the
10251: prefix @code{forth_}. (Global symbols that are used internally have the
10252: prefix @code{gforth_}).
10253:
10254: You can include the declarations of Forth types and the functions and
10255: variables of the interface with @code{#include <forth.h>}.
10256:
10257: Types.
10258:
10259: Variables.
10260:
10261: Data and FP Stack pointer. Area sizes.
10262:
10263: functions.
10264:
10265: forth_init(imagefile)
10266: forth_evaluate(string) exceptions?
10267: forth_goto(address) (or forth_execute(xt)?)
10268: forth_continue() (a corountining mechanism)
10269:
10270: Adding primitives.
10271:
10272: No checking.
10273:
10274: Signals?
10275:
10276: Accessing the Stacks
10277:
1.26 crook 10278: @c ******************************************************************
1.1 anton 10279: @node Emacs and Gforth, Image Files, Integrating Gforth, Top
10280: @chapter Emacs and Gforth
10281: @cindex Emacs and Gforth
10282:
10283: @cindex @file{gforth.el}
10284: @cindex @file{forth.el}
10285: @cindex Rydqvist, Goran
10286: @cindex comment editing commands
10287: @cindex @code{\}, editing with Emacs
10288: @cindex debug tracer editing commands
10289: @cindex @code{~~}, removal with Emacs
10290: @cindex Forth mode in Emacs
10291: Gforth comes with @file{gforth.el}, an improved version of
10292: @file{forth.el} by Goran Rydqvist (included in the TILE package). The
1.26 crook 10293: improvements are:
10294:
10295: @itemize @bullet
10296: @item
10297: A better (but still not perfect) handling of indentation.
10298: @item
10299: Comment paragraph filling (@kbd{M-q})
10300: @item
10301: Commenting (@kbd{C-x \}) and uncommenting (@kbd{C-u C-x \}) of regions
10302: @item
10303: Removal of debugging tracers (@kbd{C-x ~}, @pxref{Debugging}).
10304: @end itemize
10305:
10306: I left the stuff I do not use alone, even though some of it only makes
10307: sense for TILE. To get a description of these features, enter Forth mode
10308: and type @kbd{C-h m}.
1.1 anton 10309:
10310: @cindex source location of error or debugging output in Emacs
10311: @cindex error output, finding the source location in Emacs
10312: @cindex debugging output, finding the source location in Emacs
10313: In addition, Gforth supports Emacs quite well: The source code locations
10314: given in error messages, debugging output (from @code{~~}) and failed
10315: assertion messages are in the right format for Emacs' compilation mode
10316: (@pxref{Compilation, , Running Compilations under Emacs, emacs, Emacs
10317: Manual}) so the source location corresponding to an error or other
10318: message is only a few keystrokes away (@kbd{C-x `} for the next error,
10319: @kbd{C-c C-c} for the error under the cursor).
10320:
10321: @cindex @file{TAGS} file
10322: @cindex @file{etags.fs}
10323: @cindex viewing the source of a word in Emacs
1.26 crook 10324: Also, if you @code{include} @file{etags.fs}, a new @file{TAGS} file will
10325: be produced (@pxref{Tags, , Tags Tables, emacs, Emacs Manual}) that
1.1 anton 10326: contains the definitions of all words defined afterwards. You can then
10327: find the source for a word using @kbd{M-.}. Note that emacs can use
10328: several tags files at the same time (e.g., one for the Gforth sources
10329: and one for your program, @pxref{Select Tags Table,,Selecting a Tags
10330: Table,emacs, Emacs Manual}). The TAGS file for the preloaded words is
10331: @file{$(datadir)/gforth/$(VERSION)/TAGS} (e.g.,
10332: @file{/usr/local/share/gforth/0.2.0/TAGS}).
10333:
10334: @cindex @file{.emacs}
10335: To get all these benefits, add the following lines to your @file{.emacs}
10336: file:
10337:
10338: @example
10339: (autoload 'forth-mode "gforth.el")
10340: (setq auto-mode-alist (cons '("\\.fs\\'" . forth-mode) auto-mode-alist))
10341: @end example
10342:
1.26 crook 10343: @c ******************************************************************
1.1 anton 10344: @node Image Files, Engine, Emacs and Gforth, Top
10345: @chapter Image Files
1.26 crook 10346: @cindex image file
10347: @cindex @file{.fi} files
1.1 anton 10348: @cindex precompiled Forth code
10349: @cindex dictionary in persistent form
10350: @cindex persistent form of dictionary
10351:
10352: An image file is a file containing an image of the Forth dictionary,
10353: i.e., compiled Forth code and data residing in the dictionary. By
10354: convention, we use the extension @code{.fi} for image files.
10355:
10356: @menu
1.18 anton 10357: * Image Licensing Issues:: Distribution terms for images.
10358: * Image File Background:: Why have image files?
1.29 crook 10359: * Non-Relocatable Image Files:: don't always work.
1.18 anton 10360: * Data-Relocatable Image Files:: are better.
1.29 crook 10361: * Fully Relocatable Image Files:: better yet.
1.18 anton 10362: * Stack and Dictionary Sizes:: Setting the default sizes for an image.
1.29 crook 10363: * Running Image Files:: @code{gforth -i @i{file}} or @i{file}.
1.18 anton 10364: * Modifying the Startup Sequence:: and turnkey applications.
1.1 anton 10365: @end menu
10366:
1.18 anton 10367: @node Image Licensing Issues, Image File Background, Image Files, Image Files
10368: @section Image Licensing Issues
10369: @cindex license for images
10370: @cindex image license
10371:
10372: An image created with @code{gforthmi} (@pxref{gforthmi}) or
10373: @code{savesystem} (@pxref{Non-Relocatable Image Files}) includes the
10374: original image; i.e., according to copyright law it is a derived work of
10375: the original image.
10376:
10377: Since Gforth is distributed under the GNU GPL, the newly created image
10378: falls under the GNU GPL, too. In particular, this means that if you
10379: distribute the image, you have to make all of the sources for the image
10380: available, including those you wrote. For details see @ref{License, ,
10381: GNU General Public License (Section 3)}.
10382:
10383: If you create an image with @code{cross} (@pxref{cross.fs}), the image
10384: contains only code compiled from the sources you gave it; if none of
10385: these sources is under the GPL, the terms discussed above do not apply
10386: to the image. However, if your image needs an engine (a gforth binary)
10387: that is under the GPL, you should make sure that you distribute both in
10388: a way that is at most a @emph{mere aggregation}, if you don't want the
10389: terms of the GPL to apply to the image.
10390:
10391: @node Image File Background, Non-Relocatable Image Files, Image Licensing Issues, Image Files
1.1 anton 10392: @section Image File Background
10393: @cindex image file background
10394:
10395: Our Forth system consists not only of primitives, but also of
10396: definitions written in Forth. Since the Forth compiler itself belongs to
10397: those definitions, it is not possible to start the system with the
10398: primitives and the Forth source alone. Therefore we provide the Forth
1.26 crook 10399: code as an image file in nearly executable form. When Gforth starts up,
10400: a C routine loads the image file into memory, optionally relocates the
10401: addresses, then sets up the memory (stacks etc.) according to
10402: information in the image file, and (finally) starts executing Forth
10403: code.
1.1 anton 10404:
10405: The image file variants represent different compromises between the
10406: goals of making it easy to generate image files and making them
10407: portable.
10408:
10409: @cindex relocation at run-time
1.26 crook 10410: Win32Forth 3.4 and Mitch Bradley's @code{cforth} use relocation at
1.1 anton 10411: run-time. This avoids many of the complications discussed below (image
10412: files are data relocatable without further ado), but costs performance
10413: (one addition per memory access).
10414:
10415: @cindex relocation at load-time
1.26 crook 10416: By contrast, the Gforth loader performs relocation at image load time. The
10417: loader also has to replace tokens that represent primitive calls with the
1.1 anton 10418: appropriate code-field addresses (or code addresses in the case of
10419: direct threading).
10420:
10421: There are three kinds of image files, with different degrees of
10422: relocatability: non-relocatable, data-relocatable, and fully relocatable
10423: image files.
10424:
10425: @cindex image file loader
10426: @cindex relocating loader
10427: @cindex loader for image files
10428: These image file variants have several restrictions in common; they are
10429: caused by the design of the image file loader:
10430:
10431: @itemize @bullet
10432: @item
10433: There is only one segment; in particular, this means, that an image file
10434: cannot represent @code{ALLOCATE}d memory chunks (and pointers to
1.26 crook 10435: them). The contents of the stacks are not represented, either.
1.1 anton 10436:
10437: @item
10438: The only kinds of relocation supported are: adding the same offset to
10439: all cells that represent data addresses; and replacing special tokens
10440: with code addresses or with pieces of machine code.
10441:
10442: If any complex computations involving addresses are performed, the
10443: results cannot be represented in the image file. Several applications that
10444: use such computations come to mind:
10445: @itemize @minus
10446: @item
10447: Hashing addresses (or data structures which contain addresses) for table
10448: lookup. If you use Gforth's @code{table}s or @code{wordlist}s for this
10449: purpose, you will have no problem, because the hash tables are
10450: recomputed automatically when the system is started. If you use your own
10451: hash tables, you will have to do something similar.
10452:
10453: @item
10454: There's a cute implementation of doubly-linked lists that uses
10455: @code{XOR}ed addresses. You could represent such lists as singly-linked
10456: in the image file, and restore the doubly-linked representation on
10457: startup.@footnote{In my opinion, though, you should think thrice before
10458: using a doubly-linked list (whatever implementation).}
10459:
10460: @item
10461: The code addresses of run-time routines like @code{docol:} cannot be
10462: represented in the image file (because their tokens would be replaced by
10463: machine code in direct threaded implementations). As a workaround,
10464: compute these addresses at run-time with @code{>code-address} from the
10465: executions tokens of appropriate words (see the definitions of
10466: @code{docol:} and friends in @file{kernel.fs}).
10467:
10468: @item
10469: On many architectures addresses are represented in machine code in some
10470: shifted or mangled form. You cannot put @code{CODE} words that contain
10471: absolute addresses in this form in a relocatable image file. Workarounds
10472: are representing the address in some relative form (e.g., relative to
10473: the CFA, which is present in some register), or loading the address from
10474: a place where it is stored in a non-mangled form.
10475: @end itemize
10476: @end itemize
10477:
10478: @node Non-Relocatable Image Files, Data-Relocatable Image Files, Image File Background, Image Files
10479: @section Non-Relocatable Image Files
10480: @cindex non-relocatable image files
1.26 crook 10481: @cindex image file, non-relocatable
1.1 anton 10482:
10483: These files are simple memory dumps of the dictionary. They are specific
10484: to the executable (i.e., @file{gforth} file) they were created
10485: with. What's worse, they are specific to the place on which the
10486: dictionary resided when the image was created. Now, there is no
10487: guarantee that the dictionary will reside at the same place the next
10488: time you start Gforth, so there's no guarantee that a non-relocatable
10489: image will work the next time (Gforth will complain instead of crashing,
10490: though).
10491:
10492: You can create a non-relocatable image file with
10493:
10494: doc-savesystem
10495:
10496: @node Data-Relocatable Image Files, Fully Relocatable Image Files, Non-Relocatable Image Files, Image Files
10497: @section Data-Relocatable Image Files
10498: @cindex data-relocatable image files
1.26 crook 10499: @cindex image file, data-relocatable
1.1 anton 10500:
10501: These files contain relocatable data addresses, but fixed code addresses
10502: (instead of tokens). They are specific to the executable (i.e.,
10503: @file{gforth} file) they were created with. For direct threading on some
10504: architectures (e.g., the i386), data-relocatable images do not work. You
10505: get a data-relocatable image, if you use @file{gforthmi} with a
10506: Gforth binary that is not doubly indirect threaded (@pxref{Fully
10507: Relocatable Image Files}).
10508:
10509: @node Fully Relocatable Image Files, Stack and Dictionary Sizes, Data-Relocatable Image Files, Image Files
10510: @section Fully Relocatable Image Files
10511: @cindex fully relocatable image files
1.26 crook 10512: @cindex image file, fully relocatable
1.1 anton 10513:
10514: @cindex @file{kern*.fi}, relocatability
10515: @cindex @file{gforth.fi}, relocatability
10516: These image files have relocatable data addresses, and tokens for code
10517: addresses. They can be used with different binaries (e.g., with and
10518: without debugging) on the same machine, and even across machines with
10519: the same data formats (byte order, cell size, floating point
10520: format). However, they are usually specific to the version of Gforth
10521: they were created with. The files @file{gforth.fi} and @file{kernl*.fi}
10522: are fully relocatable.
10523:
10524: There are two ways to create a fully relocatable image file:
10525:
10526: @menu
1.29 crook 10527: * gforthmi:: The normal way
1.1 anton 10528: * cross.fs:: The hard way
10529: @end menu
10530:
10531: @node gforthmi, cross.fs, Fully Relocatable Image Files, Fully Relocatable Image Files
10532: @subsection @file{gforthmi}
10533: @cindex @file{comp-i.fs}
10534: @cindex @file{gforthmi}
10535:
10536: You will usually use @file{gforthmi}. If you want to create an
1.29 crook 10537: image @i{file} that contains everything you would load by invoking
10538: Gforth with @code{gforth @i{options}}, you simply say:
1.1 anton 10539: @example
1.29 crook 10540: gforthmi @i{file} @i{options}
1.1 anton 10541: @end example
10542:
10543: E.g., if you want to create an image @file{asm.fi} that has the file
10544: @file{asm.fs} loaded in addition to the usual stuff, you could do it
10545: like this:
10546:
10547: @example
10548: gforthmi asm.fi asm.fs
10549: @end example
10550:
1.27 crook 10551: @file{gforthmi} is implemented as a sh script and works like this: It
10552: produces two non-relocatable images for different addresses and then
10553: compares them. Its output reflects this: first you see the output (if
10554: any) of the two Gforth invocations that produce the nonrelocatable image
10555: files, then you see the output of the comparing program: It displays the
10556: offset used for data addresses and the offset used for code addresses;
1.1 anton 10557: moreover, for each cell that cannot be represented correctly in the
10558: image files, it displays a line like the following one:
10559:
10560: @example
10561: 78DC BFFFFA50 BFFFFA40
10562: @end example
10563:
10564: This means that at offset $78dc from @code{forthstart}, one input image
10565: contains $bffffa50, and the other contains $bffffa40. Since these cells
10566: cannot be represented correctly in the output image, you should examine
10567: these places in the dictionary and verify that these cells are dead
10568: (i.e., not read before they are written).
10569:
1.27 crook 10570: If you type @file{gforthmi} with no arguments, it prints some usage
10571: instructions.
10572:
1.1 anton 10573: @cindex @code{savesystem} during @file{gforthmi}
10574: @cindex @code{bye} during @file{gforthmi}
10575: @cindex doubly indirect threaded code
10576: @cindex environment variable @code{GFORTHD}
10577: @cindex @code{GFORTHD} environment variable
10578: @cindex @code{gforth-ditc}
1.29 crook 10579: There are a few wrinkles: After processing the passed @i{options}, the
1.1 anton 10580: words @code{savesystem} and @code{bye} must be visible. A special doubly
10581: indirect threaded version of the @file{gforth} executable is used for
10582: creating the nonrelocatable images; you can pass the exact filename of
10583: this executable through the environment variable @code{GFORTHD}
10584: (default: @file{gforth-ditc}); if you pass a version that is not doubly
10585: indirect threaded, you will not get a fully relocatable image, but a
1.27 crook 10586: data-relocatable image (because there is no code address offset). The
10587: normal @file{gforth} executable is used for creating the relocatable
10588: image; you can pass the exact filename of this executable through the
10589: environment variable @code{GFORTH}.
1.1 anton 10590:
10591: @node cross.fs, , gforthmi, Fully Relocatable Image Files
10592: @subsection @file{cross.fs}
10593: @cindex @file{cross.fs}
10594: @cindex cross-compiler
10595: @cindex metacompiler
10596:
10597: You can also use @code{cross}, a batch compiler that accepts a Forth-like
10598: programming language. This @code{cross} language has to be documented
10599: yet.
10600:
10601: @cindex target compiler
10602: @code{cross} also allows you to create image files for machines with
10603: different data sizes and data formats than the one used for generating
10604: the image file. You can also use it to create an application image that
10605: does not contain a Forth compiler. These features are bought with
10606: restrictions and inconveniences in programming. E.g., addresses have to
10607: be stored in memory with special words (@code{A!}, @code{A,}, etc.) in
10608: order to make the code relocatable.
10609:
10610:
10611: @node Stack and Dictionary Sizes, Running Image Files, Fully Relocatable Image Files, Image Files
10612: @section Stack and Dictionary Sizes
10613: @cindex image file, stack and dictionary sizes
10614: @cindex dictionary size default
10615: @cindex stack size default
10616:
10617: If you invoke Gforth with a command line flag for the size
10618: (@pxref{Invoking Gforth}), the size you specify is stored in the
10619: dictionary. If you save the dictionary with @code{savesystem} or create
10620: an image with @file{gforthmi}, this size will become the default
10621: for the resulting image file. E.g., the following will create a
1.21 crook 10622: fully relocatable version of @file{gforth.fi} with a 1MB dictionary:
1.1 anton 10623:
10624: @example
10625: gforthmi gforth.fi -m 1M
10626: @end example
10627:
10628: In other words, if you want to set the default size for the dictionary
10629: and the stacks of an image, just invoke @file{gforthmi} with the
10630: appropriate options when creating the image.
10631:
10632: @cindex stack size, cache-friendly
10633: Note: For cache-friendly behaviour (i.e., good performance), you should
10634: make the sizes of the stacks modulo, say, 2K, somewhat different. E.g.,
10635: the default stack sizes are: data: 16k (mod 2k=0); fp: 15.5k (mod
10636: 2k=1.5k); return: 15k(mod 2k=1k); locals: 14.5k (mod 2k=0.5k).
10637:
10638: @node Running Image Files, Modifying the Startup Sequence, Stack and Dictionary Sizes, Image Files
10639: @section Running Image Files
10640: @cindex running image files
10641: @cindex invoking image files
10642: @cindex image file invocation
10643:
10644: @cindex -i, invoke image file
10645: @cindex --image file, invoke image file
1.29 crook 10646: You can invoke Gforth with an image file @i{image} instead of the
1.1 anton 10647: default @file{gforth.fi} with the @code{-i} flag (@pxref{Invoking Gforth}):
10648: @example
1.29 crook 10649: gforth -i @i{image}
1.1 anton 10650: @end example
10651:
10652: @cindex executable image file
1.26 crook 10653: @cindex image file, executable
1.1 anton 10654: If your operating system supports starting scripts with a line of the
10655: form @code{#! ...}, you just have to type the image file name to start
10656: Gforth with this image file (note that the file extension @code{.fi} is
1.29 crook 10657: just a convention). I.e., to run Gforth with the image file @i{image},
10658: you can just type @i{image} instead of @code{gforth -i @i{image}}.
1.27 crook 10659: This works because every @code{.fi} file starts with a line of this
10660: format:
10661:
10662: @example
10663: #! /usr/local/bin/gforth-0.4.0 -i
10664: @end example
10665:
10666: The file and pathname for the Gforth engine specified on this line is
10667: the specific Gforth executable that it was built against; i.e. the value
10668: of the environment variable @code{GFORTH} at the time that
10669: @file{gforthmi} was executed.
1.1 anton 10670:
1.27 crook 10671: You can make use of the same shell capability to make a Forth source
10672: file into an executable. For example, if you place this text in a file:
1.26 crook 10673:
10674: @example
10675: #! /usr/local/bin/gforth
10676:
10677: ." Hello, world" CR
10678: bye
10679: @end example
10680:
10681: @noindent
1.27 crook 10682: and then make the file executable (chmod +x in Unix), you can run it
1.26 crook 10683: directly from the command line. The sequence @code{#!} is used in two
10684: ways; firstly, it is recognised as a ``magic sequence'' by the operating
1.29 crook 10685: system@footnote{The Unix kernel actually recognises two types of files:
10686: executable files and files of data, where the data is processed by an
10687: interpreter that is specified on the ``interpreter line'' -- the first
10688: line of the file, starting with the sequence #!. There may be a small
10689: limit (e.g., 32) on the number of characters that may be specified on
10690: the interpreter line.} secondly it is treated as a comment character by
10691: Gforth. Because of the second usage, a space is required between
10692: @code{#!} and the path to the executable.
1.27 crook 10693:
10694: The disadvantage of this latter technique, compared with using
10695: @file{gforthmi}, is that it is slower; the Forth source code is compiled
10696: on-the-fly, each time the program is invoked.
10697:
1.26 crook 10698: @comment TODO describe the #! magic with reference to the Power Tools book.
10699:
1.1 anton 10700: doc-#!
10701:
10702: @node Modifying the Startup Sequence, , Running Image Files, Image Files
10703: @section Modifying the Startup Sequence
10704: @cindex startup sequence for image file
10705: @cindex image file initialization sequence
10706: @cindex initialization sequence of image file
10707:
10708: You can add your own initialization to the startup sequence through the
1.26 crook 10709: deferred word @code{'cold}. @code{'cold} is invoked just before the
10710: image-specific command line processing (by default, loading files and
10711: evaluating (@code{-e}) strings) starts.
1.1 anton 10712:
10713: A sequence for adding your initialization usually looks like this:
10714:
10715: @example
10716: :noname
10717: Defers 'cold \ do other initialization stuff (e.g., rehashing wordlists)
10718: ... \ your stuff
10719: ; IS 'cold
10720: @end example
10721:
10722: @cindex turnkey image files
1.26 crook 10723: @cindex image file, turnkey applications
1.1 anton 10724: You can make a turnkey image by letting @code{'cold} execute a word
10725: (your turnkey application) that never returns; instead, it exits Gforth
10726: via @code{bye} or @code{throw}.
10727:
10728: @cindex command-line arguments, access
10729: @cindex arguments on the command line, access
10730: You can access the (image-specific) command-line arguments through the
1.26 crook 10731: variables @code{argc} and @code{argv}. @code{arg} provides convenient
1.1 anton 10732: access to @code{argv}.
10733:
1.26 crook 10734: If @code{'cold} exits normally, Gforth processes the command-line
10735: arguments as files to be loaded and strings to be evaluated. Therefore,
10736: @code{'cold} should remove the arguments it has used in this case.
10737:
10738: doc-'cold
1.1 anton 10739: doc-argc
10740: doc-argv
10741: doc-arg
10742:
10743:
10744: @c ******************************************************************
1.13 pazsan 10745: @node Engine, Binding to System Library, Image Files, Top
1.1 anton 10746: @chapter Engine
10747: @cindex engine
10748: @cindex virtual machine
10749:
1.26 crook 10750: Reading this chapter is not necessary for programming with Gforth. It
1.1 anton 10751: may be helpful for finding your way in the Gforth sources.
10752:
10753: The ideas in this section have also been published in the papers
10754: @cite{ANS fig/GNU/??? Forth} (in German) by Bernd Paysan, presented at
10755: the Forth-Tagung '93 and @cite{A Portable Forth Engine} by M. Anton
10756: Ertl, presented at EuroForth '93; the latter is available at
10757: @*@url{http://www.complang.tuwien.ac.at/papers/ertl93.ps.Z}.
10758:
10759: @menu
10760: * Portability::
10761: * Threading::
10762: * Primitives::
10763: * Performance::
10764: @end menu
10765:
10766: @node Portability, Threading, Engine, Engine
10767: @section Portability
10768: @cindex engine portability
10769:
1.26 crook 10770: An important goal of the Gforth Project is availability across a wide
10771: range of personal machines. fig-Forth, and, to a lesser extent, F83,
10772: achieved this goal by manually coding the engine in assembly language
10773: for several then-popular processors. This approach is very
10774: labor-intensive and the results are short-lived due to progress in
10775: computer architecture.
1.1 anton 10776:
10777: @cindex C, using C for the engine
10778: Others have avoided this problem by coding in C, e.g., Mitch Bradley
10779: (cforth), Mikael Patel (TILE) and Dirk Zoller (pfe). This approach is
10780: particularly popular for UNIX-based Forths due to the large variety of
10781: architectures of UNIX machines. Unfortunately an implementation in C
10782: does not mix well with the goals of efficiency and with using
10783: traditional techniques: Indirect or direct threading cannot be expressed
10784: in C, and switch threading, the fastest technique available in C, is
10785: significantly slower. Another problem with C is that it is very
10786: cumbersome to express double integer arithmetic.
10787:
10788: @cindex GNU C for the engine
10789: @cindex long long
10790: Fortunately, there is a portable language that does not have these
10791: limitations: GNU C, the version of C processed by the GNU C compiler
10792: (@pxref{C Extensions, , Extensions to the C Language Family, gcc.info,
10793: GNU C Manual}). Its labels as values feature (@pxref{Labels as Values, ,
10794: Labels as Values, gcc.info, GNU C Manual}) makes direct and indirect
10795: threading possible, its @code{long long} type (@pxref{Long Long, ,
10796: Double-Word Integers, gcc.info, GNU C Manual}) corresponds to Forth's
10797: double numbers@footnote{Unfortunately, long longs are not implemented
10798: properly on all machines (e.g., on alpha-osf1, long longs are only 64
10799: bits, the same size as longs (and pointers), but they should be twice as
1.4 anton 10800: long according to @pxref{Long Long, , Double-Word Integers, gcc.info, GNU
1.1 anton 10801: C Manual}). So, we had to implement doubles in C after all. Still, on
10802: most machines we can use long longs and achieve better performance than
10803: with the emulation package.}. GNU C is available for free on all
10804: important (and many unimportant) UNIX machines, VMS, 80386s running
10805: MS-DOS, the Amiga, and the Atari ST, so a Forth written in GNU C can run
10806: on all these machines.
10807:
10808: Writing in a portable language has the reputation of producing code that
10809: is slower than assembly. For our Forth engine we repeatedly looked at
10810: the code produced by the compiler and eliminated most compiler-induced
10811: inefficiencies by appropriate changes in the source code.
10812:
10813: @cindex explicit register declarations
10814: @cindex --enable-force-reg, configuration flag
10815: @cindex -DFORCE_REG
10816: However, register allocation cannot be portably influenced by the
10817: programmer, leading to some inefficiencies on register-starved
10818: machines. We use explicit register declarations (@pxref{Explicit Reg
10819: Vars, , Variables in Specified Registers, gcc.info, GNU C Manual}) to
10820: improve the speed on some machines. They are turned on by using the
10821: configuration flag @code{--enable-force-reg} (@code{gcc} switch
10822: @code{-DFORCE_REG}). Unfortunately, this feature not only depends on the
10823: machine, but also on the compiler version: On some machines some
10824: compiler versions produce incorrect code when certain explicit register
10825: declarations are used. So by default @code{-DFORCE_REG} is not used.
10826:
10827: @node Threading, Primitives, Portability, Engine
10828: @section Threading
10829: @cindex inner interpreter implementation
10830: @cindex threaded code implementation
10831:
10832: @cindex labels as values
10833: GNU C's labels as values extension (available since @code{gcc-2.0},
10834: @pxref{Labels as Values, , Labels as Values, gcc.info, GNU C Manual})
1.29 crook 10835: makes it possible to take the address of @i{label} by writing
10836: @code{&&@i{label}}. This address can then be used in a statement like
10837: @code{goto *@i{address}}. I.e., @code{goto *&&x} is the same as
1.1 anton 10838: @code{goto x}.
10839:
1.26 crook 10840: @cindex @code{NEXT}, indirect threaded
1.1 anton 10841: @cindex indirect threaded inner interpreter
10842: @cindex inner interpreter, indirect threaded
1.26 crook 10843: With this feature an indirect threaded @code{NEXT} looks like:
1.1 anton 10844: @example
10845: cfa = *ip++;
10846: ca = *cfa;
10847: goto *ca;
10848: @end example
10849: @cindex instruction pointer
10850: For those unfamiliar with the names: @code{ip} is the Forth instruction
10851: pointer; the @code{cfa} (code-field address) corresponds to ANS Forths
10852: execution token and points to the code field of the next word to be
10853: executed; The @code{ca} (code address) fetched from there points to some
10854: executable code, e.g., a primitive or the colon definition handler
10855: @code{docol}.
10856:
1.26 crook 10857: @cindex @code{NEXT}, direct threaded
1.1 anton 10858: @cindex direct threaded inner interpreter
10859: @cindex inner interpreter, direct threaded
10860: Direct threading is even simpler:
10861: @example
10862: ca = *ip++;
10863: goto *ca;
10864: @end example
10865:
10866: Of course we have packaged the whole thing neatly in macros called
1.26 crook 10867: @code{NEXT} and @code{NEXT1} (the part of @code{NEXT} after fetching the cfa).
1.1 anton 10868:
10869: @menu
10870: * Scheduling::
10871: * Direct or Indirect Threaded?::
10872: * DOES>::
10873: @end menu
10874:
10875: @node Scheduling, Direct or Indirect Threaded?, Threading, Threading
10876: @subsection Scheduling
10877: @cindex inner interpreter optimization
10878:
10879: There is a little complication: Pipelined and superscalar processors,
10880: i.e., RISC and some modern CISC machines can process independent
10881: instructions while waiting for the results of an instruction. The
10882: compiler usually reorders (schedules) the instructions in a way that
10883: achieves good usage of these delay slots. However, on our first tries
10884: the compiler did not do well on scheduling primitives. E.g., for
10885: @code{+} implemented as
10886: @example
10887: n=sp[0]+sp[1];
10888: sp++;
10889: sp[0]=n;
10890: NEXT;
10891: @end example
1.26 crook 10892: the @code{NEXT} comes strictly after the other code, i.e., there is nearly no
1.1 anton 10893: scheduling. After a little thought the problem becomes clear: The
1.21 crook 10894: compiler cannot know that @code{sp} and @code{ip} point to different
10895: addresses (and the version of @code{gcc} we used would not know it even
10896: if it was possible), so it could not move the load of the cfa above the
10897: store to the TOS. Indeed the pointers could be the same, if code on or
10898: very near the top of stack were executed. In the interest of speed we
10899: chose to forbid this probably unused ``feature'' and helped the compiler
1.26 crook 10900: in scheduling: @code{NEXT} is divided into the loading part (@code{NEXT_P1})
1.21 crook 10901: and the goto part (@code{NEXT_P2}). @code{+} now looks like:
1.1 anton 10902: @example
10903: n=sp[0]+sp[1];
10904: sp++;
10905: NEXT_P1;
10906: sp[0]=n;
10907: NEXT_P2;
10908: @end example
10909: This can be scheduled optimally by the compiler.
10910:
10911: This division can be turned off with the switch @code{-DCISC_NEXT}. This
10912: switch is on by default on machines that do not profit from scheduling
10913: (e.g., the 80386), in order to preserve registers.
10914:
10915: @node Direct or Indirect Threaded?, DOES>, Scheduling, Threading
10916: @subsection Direct or Indirect Threaded?
10917: @cindex threading, direct or indirect?
10918:
10919: @cindex -DDIRECT_THREADED
10920: Both! After packaging the nasty details in macro definitions we
10921: realized that we could switch between direct and indirect threading by
10922: simply setting a compilation flag (@code{-DDIRECT_THREADED}) and
10923: defining a few machine-specific macros for the direct-threading case.
10924: On the Forth level we also offer access words that hide the
10925: differences between the threading methods (@pxref{Threading Words}).
10926:
10927: Indirect threading is implemented completely machine-independently.
10928: Direct threading needs routines for creating jumps to the executable
1.21 crook 10929: code (e.g. to @code{docol} or @code{dodoes}). These routines are inherently
10930: machine-dependent, but they do not amount to many source lines. Therefore,
10931: even porting direct threading to a new machine requires little effort.
1.1 anton 10932:
10933: @cindex --enable-indirect-threaded, configuration flag
10934: @cindex --enable-direct-threaded, configuration flag
10935: The default threading method is machine-dependent. You can enforce a
10936: specific threading method when building Gforth with the configuration
10937: flag @code{--enable-direct-threaded} or
10938: @code{--enable-indirect-threaded}. Note that direct threading is not
10939: supported on all machines.
10940:
10941: @node DOES>, , Direct or Indirect Threaded?, Threading
10942: @subsection DOES>
10943: @cindex @code{DOES>} implementation
10944:
1.26 crook 10945: @cindex @code{dodoes} routine
10946: @cindex @code{DOES>}-code
1.1 anton 10947: One of the most complex parts of a Forth engine is @code{dodoes}, i.e.,
10948: the chunk of code executed by every word defined by a
10949: @code{CREATE}...@code{DOES>} pair. The main problem here is: How to find
10950: the Forth code to be executed, i.e. the code after the
1.26 crook 10951: @code{DOES>} (the @code{DOES>}-code)? There are two solutions:
1.1 anton 10952:
1.21 crook 10953: In fig-Forth the code field points directly to the @code{dodoes} and the
1.26 crook 10954: @code{DOES>}code address is stored in the cell after the code address (i.e. at
1.29 crook 10955: @code{@i{CFA} cell+}). It may seem that this solution is illegal in
1.1 anton 10956: the Forth-79 and all later standards, because in fig-Forth this address
10957: lies in the body (which is illegal in these standards). However, by
10958: making the code field larger for all words this solution becomes legal
10959: again. We use this approach for the indirect threaded version and for
10960: direct threading on some machines. Leaving a cell unused in most words
10961: is a bit wasteful, but on the machines we are targeting this is hardly a
10962: problem. The other reason for having a code field size of two cells is
10963: to avoid having different image files for direct and indirect threaded
10964: systems (direct threaded systems require two-cell code fields on many
10965: machines).
10966:
1.26 crook 10967: @cindex @code{DOES>}-handler
1.1 anton 10968: The other approach is that the code field points or jumps to the cell
1.26 crook 10969: after @code{DOES>}. In this variant there is a jump to @code{dodoes} at
10970: this address (the @code{DOES>}-handler). @code{dodoes} can then get the
10971: @code{DOES>}-code address by computing the code address, i.e., the address of
1.1 anton 10972: the jump to dodoes, and add the length of that jump field. A variant of
10973: this is to have a call to @code{dodoes} after the @code{DOES>}; then the
10974: return address (which can be found in the return register on RISCs) is
1.26 crook 10975: the @code{DOES>}-code address. Since the two cells available in the code field
1.1 anton 10976: are used up by the jump to the code address in direct threading on many
10977: architectures, we use this approach for direct threading on these
10978: architectures. We did not want to add another cell to the code field.
10979:
10980: @node Primitives, Performance, Threading, Engine
10981: @section Primitives
10982: @cindex primitives, implementation
10983: @cindex virtual machine instructions, implementation
10984:
10985: @menu
10986: * Automatic Generation::
10987: * TOS Optimization::
10988: * Produced code::
10989: @end menu
10990:
10991: @node Automatic Generation, TOS Optimization, Primitives, Primitives
10992: @subsection Automatic Generation
10993: @cindex primitives, automatic generation
10994:
10995: @cindex @file{prims2x.fs}
10996: Since the primitives are implemented in a portable language, there is no
10997: longer any need to minimize the number of primitives. On the contrary,
10998: having many primitives has an advantage: speed. In order to reduce the
10999: number of errors in primitives and to make programming them easier, we
11000: provide a tool, the primitive generator (@file{prims2x.fs}), that
11001: automatically generates most (and sometimes all) of the C code for a
11002: primitive from the stack effect notation. The source for a primitive
11003: has the following form:
11004:
11005: @cindex primitive source format
11006: @format
1.29 crook 11007: @i{Forth-name} @i{stack-effect} @i{category} [@i{pronounc.}]
11008: [@code{""}@i{glossary entry}@code{""}]
11009: @i{C code}
1.1 anton 11010: [@code{:}
1.29 crook 11011: @i{Forth code}]
1.1 anton 11012: @end format
11013:
11014: The items in brackets are optional. The category and glossary fields
11015: are there for generating the documentation, the Forth code is there
11016: for manual implementations on machines without GNU C. E.g., the source
11017: for the primitive @code{+} is:
11018: @example
11019: + n1 n2 -- n core plus
11020: n = n1+n2;
11021: @end example
11022:
11023: This looks like a specification, but in fact @code{n = n1+n2} is C
11024: code. Our primitive generation tool extracts a lot of information from
11025: the stack effect notations@footnote{We use a one-stack notation, even
11026: though we have separate data and floating-point stacks; The separate
11027: notation can be generated easily from the unified notation.}: The number
11028: of items popped from and pushed on the stack, their type, and by what
11029: name they are referred to in the C code. It then generates a C code
11030: prelude and postlude for each primitive. The final C code for @code{+}
11031: looks like this:
11032:
11033: @example
11034: I_plus: /* + ( n1 n2 -- n ) */ /* label, stack effect */
11035: /* */ /* documentation */
11036: @{
11037: DEF_CA /* definition of variable ca (indirect threading) */
11038: Cell n1; /* definitions of variables */
11039: Cell n2;
11040: Cell n;
11041: n1 = (Cell) sp[1]; /* input */
11042: n2 = (Cell) TOS;
11043: sp += 1; /* stack adjustment */
11044: NAME("+") /* debugging output (with -DDEBUG) */
11045: @{
11046: n = n1+n2; /* C code taken from the source */
11047: @}
11048: NEXT_P1; /* NEXT part 1 */
11049: TOS = (Cell)n; /* output */
11050: NEXT_P2; /* NEXT part 2 */
11051: @}
11052: @end example
11053:
11054: This looks long and inefficient, but the GNU C compiler optimizes quite
11055: well and produces optimal code for @code{+} on, e.g., the R3000 and the
11056: HP RISC machines: Defining the @code{n}s does not produce any code, and
11057: using them as intermediate storage also adds no cost.
11058:
1.26 crook 11059: There are also other optimizations that are not illustrated by this
11060: example: assignments between simple variables are usually for free (copy
1.1 anton 11061: propagation). If one of the stack items is not used by the primitive
11062: (e.g. in @code{drop}), the compiler eliminates the load from the stack
11063: (dead code elimination). On the other hand, there are some things that
11064: the compiler does not do, therefore they are performed by
11065: @file{prims2x.fs}: The compiler does not optimize code away that stores
11066: a stack item to the place where it just came from (e.g., @code{over}).
11067:
11068: While programming a primitive is usually easy, there are a few cases
11069: where the programmer has to take the actions of the generator into
11070: account, most notably @code{?dup}, but also words that do not (always)
1.26 crook 11071: fall through to @code{NEXT}.
1.1 anton 11072:
11073: @node TOS Optimization, Produced code, Automatic Generation, Primitives
11074: @subsection TOS Optimization
11075: @cindex TOS optimization for primitives
11076: @cindex primitives, keeping the TOS in a register
11077:
11078: An important optimization for stack machine emulators, e.g., Forth
11079: engines, is keeping one or more of the top stack items in
1.29 crook 11080: registers. If a word has the stack effect @i{in1}...@i{inx} @code{--}
11081: @i{out1}...@i{outy}, keeping the top @i{n} items in registers
1.1 anton 11082: @itemize @bullet
11083: @item
1.29 crook 11084: is better than keeping @i{n-1} items, if @i{x>=n} and @i{y>=n},
1.1 anton 11085: due to fewer loads from and stores to the stack.
1.29 crook 11086: @item is slower than keeping @i{n-1} items, if @i{x<>y} and @i{x<n} and
11087: @i{y<n}, due to additional moves between registers.
1.1 anton 11088: @end itemize
11089:
11090: @cindex -DUSE_TOS
11091: @cindex -DUSE_NO_TOS
11092: In particular, keeping one item in a register is never a disadvantage,
11093: if there are enough registers. Keeping two items in registers is a
11094: disadvantage for frequent words like @code{?branch}, constants,
11095: variables, literals and @code{i}. Therefore our generator only produces
11096: code that keeps zero or one items in registers. The generated C code
11097: covers both cases; the selection between these alternatives is made at
11098: C-compile time using the switch @code{-DUSE_TOS}. @code{TOS} in the C
11099: code for @code{+} is just a simple variable name in the one-item case,
11100: otherwise it is a macro that expands into @code{sp[0]}. Note that the
11101: GNU C compiler tries to keep simple variables like @code{TOS} in
11102: registers, and it usually succeeds, if there are enough registers.
11103:
11104: @cindex -DUSE_FTOS
11105: @cindex -DUSE_NO_FTOS
11106: The primitive generator performs the TOS optimization for the
11107: floating-point stack, too (@code{-DUSE_FTOS}). For floating-point
11108: operations the benefit of this optimization is even larger:
11109: floating-point operations take quite long on most processors, but can be
11110: performed in parallel with other operations as long as their results are
11111: not used. If the FP-TOS is kept in a register, this works. If
11112: it is kept on the stack, i.e., in memory, the store into memory has to
11113: wait for the result of the floating-point operation, lengthening the
11114: execution time of the primitive considerably.
11115:
11116: The TOS optimization makes the automatic generation of primitives a
11117: bit more complicated. Just replacing all occurrences of @code{sp[0]} by
11118: @code{TOS} is not sufficient. There are some special cases to
11119: consider:
11120: @itemize @bullet
11121: @item In the case of @code{dup ( w -- w w )} the generator must not
11122: eliminate the store to the original location of the item on the stack,
11123: if the TOS optimization is turned on.
11124: @item Primitives with stack effects of the form @code{--}
1.29 crook 11125: @i{out1}...@i{outy} must store the TOS to the stack at the start.
11126: Likewise, primitives with the stack effect @i{in1}...@i{inx} @code{--}
1.1 anton 11127: must load the TOS from the stack at the end. But for the null stack
11128: effect @code{--} no stores or loads should be generated.
11129: @end itemize
11130:
11131: @node Produced code, , TOS Optimization, Primitives
11132: @subsection Produced code
11133: @cindex primitives, assembly code listing
11134:
11135: @cindex @file{engine.s}
11136: To see what assembly code is produced for the primitives on your machine
11137: with your compiler and your flag settings, type @code{make engine.s} and
11138: look at the resulting file @file{engine.s}.
11139:
11140: @node Performance, , Primitives, Engine
11141: @section Performance
11142: @cindex performance of some Forth interpreters
11143: @cindex engine performance
11144: @cindex benchmarking Forth systems
11145: @cindex Gforth performance
11146:
11147: On RISCs the Gforth engine is very close to optimal; i.e., it is usually
11148: impossible to write a significantly faster engine.
11149:
11150: On register-starved machines like the 386 architecture processors
11151: improvements are possible, because @code{gcc} does not utilize the
11152: registers as well as a human, even with explicit register declarations;
11153: e.g., Bernd Beuster wrote a Forth system fragment in assembly language
11154: and hand-tuned it for the 486; this system is 1.19 times faster on the
11155: Sieve benchmark on a 486DX2/66 than Gforth compiled with
11156: @code{gcc-2.6.3} with @code{-DFORCE_REG}.
11157:
11158: @cindex Win32Forth performance
11159: @cindex NT Forth performance
11160: @cindex eforth performance
11161: @cindex ThisForth performance
11162: @cindex PFE performance
11163: @cindex TILE performance
11164: However, this potential advantage of assembly language implementations
11165: is not necessarily realized in complete Forth systems: We compared
11166: Gforth (direct threaded, compiled with @code{gcc-2.6.3} and
11167: @code{-DFORCE_REG}) with Win32Forth 1.2093, LMI's NT Forth (Beta, May
11168: 1994) and Eforth (with and without peephole (aka pinhole) optimization
11169: of the threaded code); all these systems were written in assembly
11170: language. We also compared Gforth with three systems written in C:
11171: PFE-0.9.14 (compiled with @code{gcc-2.6.3} with the default
11172: configuration for Linux: @code{-O2 -fomit-frame-pointer -DUSE_REGS
1.21 crook 11173: -DUNROLL_NEXT}), ThisForth Beta (compiled with @code{gcc-2.6.3 -O3
11174: -fomit-frame-pointer}; ThisForth employs peephole optimization of the
1.1 anton 11175: threaded code) and TILE (compiled with @code{make opt}). We benchmarked
11176: Gforth, PFE, ThisForth and TILE on a 486DX2/66 under Linux. Kenneth
11177: O'Heskin kindly provided the results for Win32Forth and NT Forth on a
11178: 486DX2/66 with similar memory performance under Windows NT. Marcel
11179: Hendrix ported Eforth to Linux, then extended it to run the benchmarks,
11180: added the peephole optimizer, ran the benchmarks and reported the
11181: results.
11182:
11183: We used four small benchmarks: the ubiquitous Sieve; bubble-sorting and
11184: matrix multiplication come from the Stanford integer benchmarks and have
11185: been translated into Forth by Martin Fraeman; we used the versions
11186: included in the TILE Forth package, but with bigger data set sizes; and
11187: a recursive Fibonacci number computation for benchmarking calling
11188: performance. The following table shows the time taken for the benchmarks
11189: scaled by the time taken by Gforth (in other words, it shows the speedup
11190: factor that Gforth achieved over the other systems).
11191:
11192: @example
11193: relative Win32- NT eforth This-
11194: time Gforth Forth Forth eforth +opt PFE Forth TILE
11195: sieve 1.00 1.39 1.14 1.39 0.85 1.58 3.18 8.58
11196: bubble 1.00 1.31 1.41 1.48 0.88 1.50 3.88
11197: matmul 1.00 1.47 1.35 1.46 0.74 1.58 4.09
11198: fib 1.00 1.52 1.34 1.22 0.86 1.74 2.99 4.30
11199: @end example
11200:
1.26 crook 11201: You may be quite surprised by the good performance of Gforth when
11202: compared with systems written in assembly language. One important reason
11203: for the disappointing performance of these other systems is probably
11204: that they are not written optimally for the 486 (e.g., they use the
11205: @code{lods} instruction). In addition, Win32Forth uses a comfortable,
11206: but costly method for relocating the Forth image: like @code{cforth}, it
11207: computes the actual addresses at run time, resulting in two address
11208: computations per @code{NEXT} (@pxref{Image File Background}).
11209:
11210: Only Eforth with the peephole optimizer has a performance that is
11211: comparable to Gforth. The speedups achieved with peephole optimization
11212: of threaded code are quite remarkable. Adding a peephole optimizer to
11213: Gforth should cause similar speedups.
1.1 anton 11214:
11215: The speedup of Gforth over PFE, ThisForth and TILE can be easily
11216: explained with the self-imposed restriction of the latter systems to
11217: standard C, which makes efficient threading impossible (however, the
1.4 anton 11218: measured implementation of PFE uses a GNU C extension: @pxref{Global Reg
1.1 anton 11219: Vars, , Defining Global Register Variables, gcc.info, GNU C Manual}).
11220: Moreover, current C compilers have a hard time optimizing other aspects
11221: of the ThisForth and the TILE source.
11222:
1.26 crook 11223: The performance of Gforth on 386 architecture processors varies widely
11224: with the version of @code{gcc} used. E.g., @code{gcc-2.5.8} failed to
11225: allocate any of the virtual machine registers into real machine
11226: registers by itself and would not work correctly with explicit register
11227: declarations, giving a 1.3 times slower engine (on a 486DX2/66 running
11228: the Sieve) than the one measured above.
1.1 anton 11229:
1.26 crook 11230: Note that there have been several releases of Win32Forth since the
11231: release presented here, so the results presented above may have little
1.1 anton 11232: predictive value for the performance of Win32Forth today.
11233:
11234: @cindex @file{Benchres}
11235: In @cite{Translating Forth to Efficient C} by M. Anton Ertl and Martin
11236: Maierhofer (presented at EuroForth '95), an indirect threaded version of
11237: Gforth is compared with Win32Forth, NT Forth, PFE, and ThisForth; that
11238: version of Gforth is 2%@minus{}8% slower on a 486 than the direct
11239: threaded version used here. The paper available at
11240: @*@url{http://www.complang.tuwien.ac.at/papers/ertl&maierhofer95.ps.gz};
11241: it also contains numbers for some native code systems. You can find a
11242: newer version of these measurements at
11243: @url{http://www.complang.tuwien.ac.at/forth/performance.html}. You can
11244: find numbers for Gforth on various machines in @file{Benchres}.
11245:
1.26 crook 11246: @c ******************************************************************
1.13 pazsan 11247: @node Binding to System Library, Cross Compiler, Engine, Top
1.14 pazsan 11248: @chapter Binding to System Library
1.13 pazsan 11249:
11250: @node Cross Compiler, Bugs, Binding to System Library, Top
1.14 pazsan 11251: @chapter Cross Compiler
1.13 pazsan 11252:
11253: Cross Compiler
11254:
11255: @menu
11256: * Using the Cross Compiler::
11257: * How the Cross Compiler Works::
11258: @end menu
11259:
1.21 crook 11260: @node Using the Cross Compiler, How the Cross Compiler Works, Cross Compiler, Cross Compiler
1.14 pazsan 11261: @section Using the Cross Compiler
1.13 pazsan 11262:
1.21 crook 11263: @node How the Cross Compiler Works, , Using the Cross Compiler, Cross Compiler
1.14 pazsan 11264: @section How the Cross Compiler Works
1.13 pazsan 11265:
11266: @node Bugs, Origin, Cross Compiler, Top
1.21 crook 11267: @appendix Bugs
1.1 anton 11268: @cindex bug reporting
11269:
1.21 crook 11270: Known bugs are described in the file @file{BUGS} in the Gforth distribution.
1.1 anton 11271:
11272: If you find a bug, please send a bug report to
1.33 anton 11273: @email{bug-gforth@@gnu.org}. A bug report should include this
1.21 crook 11274: information:
11275:
11276: @itemize @bullet
11277: @item
11278: The Gforth version used (it is announced at the start of an
11279: interactive Gforth session).
11280: @item
11281: The machine and operating system (on Unix
11282: systems @code{uname -a} will report this information).
11283: @item
11284: The installation options (send the file @file{config.status}).
11285: @item
11286: A complete list of changes (if any) you (or your installer) have made to the
11287: Gforth sources.
11288: @item
11289: A program (or a sequence of keyboard commands) that reproduces the bug.
11290: @item
11291: A description of what you think constitutes the buggy behaviour.
11292: @end itemize
1.1 anton 11293:
11294: For a thorough guide on reporting bugs read @ref{Bug Reporting, , How
11295: to Report Bugs, gcc.info, GNU C Manual}.
11296:
11297:
1.21 crook 11298: @node Origin, Forth-related information, Bugs, Top
11299: @appendix Authors and Ancestors of Gforth
1.1 anton 11300:
11301: @section Authors and Contributors
11302: @cindex authors of Gforth
11303: @cindex contributors to Gforth
11304:
11305: The Gforth project was started in mid-1992 by Bernd Paysan and Anton
11306: Ertl. The third major author was Jens Wilke. Lennart Benschop (who was
11307: one of Gforth's first users, in mid-1993) and Stuart Ramsden inspired us
11308: with their continuous feedback. Lennart Benshop contributed
11309: @file{glosgen.fs}, while Stuart Ramsden has been working on automatic
11310: support for calling C libraries. Helpful comments also came from Paul
11311: Kleinrubatscher, Christian Pirker, Dirk Zoller, Marcel Hendrix, John
1.12 anton 11312: Wavrik, Barrie Stott, Marc de Groot, and Jorge Acerada. Since the
11313: release of Gforth-0.2.1 there were also helpful comments from many
11314: others; thank you all, sorry for not listing you here (but digging
1.23 crook 11315: through my mailbox to extract your names is on my to-do list). Since the
11316: release of Gforth-0.4.0 Neal Crook worked on the manual.
1.1 anton 11317:
11318: Gforth also owes a lot to the authors of the tools we used (GCC, CVS,
11319: and autoconf, among others), and to the creators of the Internet: Gforth
1.21 crook 11320: was developed across the Internet, and its authors did not meet
1.20 pazsan 11321: physically for the first 4 years of development.
1.1 anton 11322:
11323: @section Pedigree
1.26 crook 11324: @cindex pedigree of Gforth
1.1 anton 11325:
1.20 pazsan 11326: Gforth descends from bigFORTH (1993) and fig-Forth. Gforth and PFE (by
1.1 anton 11327: Dirk Zoller) will cross-fertilize each other. Of course, a significant
11328: part of the design of Gforth was prescribed by ANS Forth.
11329:
1.20 pazsan 11330: Bernd Paysan wrote bigFORTH, a descendent from TurboForth, an unreleased
1.1 anton 11331: 32 bit native code version of VolksForth for the Atari ST, written
11332: mostly by Dietrich Weineck.
11333:
11334: VolksForth descends from F83. It was written by Klaus Schleisiek, Bernd
11335: Pennemann, Georg Rehfeld and Dietrich Weineck for the C64 (called
11336: UltraForth there) in the mid-80s and ported to the Atari ST in 1986.
11337:
11338: Henry Laxen and Mike Perry wrote F83 as a model implementation of the
11339: Forth-83 standard. !! Pedigree? When?
11340:
11341: A team led by Bill Ragsdale implemented fig-Forth on many processors in
11342: 1979. Robert Selzer and Bill Ragsdale developed the original
11343: implementation of fig-Forth for the 6502 based on microForth.
11344:
11345: The principal architect of microForth was Dean Sanderson. microForth was
11346: FORTH, Inc.'s first off-the-shelf product. It was developed in 1976 for
11347: the 1802, and subsequently implemented on the 8080, the 6800 and the
11348: Z80.
11349:
11350: All earlier Forth systems were custom-made, usually by Charles Moore,
11351: who discovered (as he puts it) Forth during the late 60s. The first full
11352: Forth existed in 1971.
11353:
11354: A part of the information in this section comes from @cite{The Evolution
11355: of Forth} by Elizabeth D. Rather, Donald R. Colburn and Charles
11356: H. Moore, presented at the HOPL-II conference and preprinted in SIGPLAN
11357: Notices 28(3), 1993. You can find more historical and genealogical
11358: information about Forth there.
11359:
1.21 crook 11360: @node Forth-related information, Word Index, Origin, Top
11361: @appendix Other Forth-related information
11362: @cindex Forth-related information
11363:
11364: @menu
11365: * Internet resources::
11366: * Books::
11367: * The Forth Interest Group::
11368: * Conferences::
11369: @end menu
11370:
11371:
11372: @node Internet resources, Books, Forth-related information, Forth-related information
11373: @section Internet resources
1.26 crook 11374: @cindex internet resources
1.21 crook 11375:
11376: @cindex comp.lang.forth
11377: @cindex frequently asked questions
11378: There is an active newsgroup (comp.lang.forth) discussing Forth and
11379: Forth-related issues. A frequently-asked-questions (FAQ) list
11380: is posted to the newsgroup regulary, and archived at these sites:
11381:
11382: @itemize @bullet
11383: @item
11384: @url{ftp://rtfm.mit.edu/pub/usenet-by-group/comp.lang.forth/}
11385: @item
11386: @url{ftp://ftp.forth.org/pub/Forth/FAQ/}
11387: @end itemize
11388:
11389: The FAQ list should be considered mandatory reading before posting to
11390: the newsgroup.
11391:
11392: Here are some other web sites holding Forth-related material:
11393:
11394: @itemize @bullet
11395: @item
11396: @url{http://www.taygeta.com/forth.html} -- Skip Carter's Forth pages.
11397: @item
11398: @url{http://www.jwdt.com/~paysan/gforth.html} -- the Gforth home page.
11399: @item
11400: @url{http://www.minerva.com/uathena.htm} -- home of ANS Forth Standard.
11401: @item
11402: @url{http://dec.bournemouth.ac.uk/forth/index.html} -- the Forth
11403: Research page, including links to the Journal of Forth Application and
11404: Research (JFAR) and a searchable Forth bibliography.
11405: @end itemize
11406:
11407:
11408: @node Books, The Forth Interest Group, Internet resources, Forth-related information
11409: @section Books
1.26 crook 11410: @cindex books on Forth
1.21 crook 11411:
11412: As the Standard is relatively new, there are not many books out yet. It
11413: is not recommended to learn Forth by using Gforth and a book that is not
11414: written for ANS Forth, as you will not know your mistakes from the
11415: deviations of the book. However, books based on the Forth-83 standard
11416: should be ok, because ANS Forth is primarily an extension of Forth-83.
11417:
11418: @cindex standard document for ANS Forth
11419: @cindex ANS Forth document
11420: The definite reference if you want to write ANS Forth programs is, of
1.26 crook 11421: course, the ANS Forth document. It is available in printed form from the
1.21 crook 11422: National Standards Institute Sales Department (Tel.: USA (212) 642-4900;
11423: Fax.: USA (212) 302-1286) as document @cite{X3.215-1994} for about
11424: $200. You can also get it from Global Engineering Documents (Tel.: USA
11425: (800) 854-7179; Fax.: (303) 843-9880) for about $300.
11426:
11427: @cite{dpANS6}, the last draft of the standard, which was then submitted
11428: to ANSI for publication is available electronically and for free in some
11429: MS Word format, and it has been converted to HTML
11430: (@url{http://www.taygeta.com/forth/dpans.html}; this is my favourite
11431: format); this HTML version also includes the answers to Requests for
11432: Interpretation (RFIs). Some pointers to these versions can be found
11433: through @*@url{http://www.complang.tuwien.ac.at/projects/forth.html}.
11434:
1.26 crook 11435: @cindex introductory book on Forth
11436: @cindex book on Forth, introductory
1.21 crook 11437: @cindex Woehr, Jack: @cite{Forth: The New Model}
11438: @cindex @cite{Forth: The new model} (book)
11439: @cite{Forth: The New Model} by Jack Woehr (Prentice-Hall, 1993) is an
11440: introductory book based on a draft version of the standard. It does not
11441: cover the whole standard. It also contains interesting background
11442: information (Jack Woehr was in the ANS Forth Technical Committee). It is
11443: not appropriate for complete newbies, but programmers experienced in
11444: other languages should find it ok.
11445:
11446: @cindex Conklin, Edward K., and Elizabeth Rather: @cite{Forth Programmer's Handbook}
11447: @cindex Rather, Elizabeth and Edward K. Conklin: @cite{Forth Programmer's Handbook}
11448: @cindex @cite{Forth Programmer's Handbook} (book)
11449: @cite{Forth Programmer's Handbook} by Edward K. Conklin, Elizabeth
11450: D. Rather and the technical staff of Forth, Inc. (Forth, Inc., 1997;
11451: ISBN 0-9662156-0-5) contains little introductory material. The majority
11452: of the book is similar to @ref{Words}, but the book covers most of the
11453: standard words and some non-standard words (whereas this manual is
11454: quite incomplete). In addition, the book contains a chapter on
11455: programming style. The major drawback of this book is that it usually
11456: does not identify what is standard and what is specific to the Forth
11457: system described in the book (probably one of Forth, Inc.'s systems).
11458: Fortunately, many of the non-standard programming practices described in
11459: the book work in Gforth, too. Still, this drawback makes the book
11460: hardly more useful than a pre-ANS book.
11461:
11462: @node The Forth Interest Group, Conferences, Books, Forth-related information
11463: @section The Forth Interest Group
11464: @cindex Forth interest group (FIG)
11465:
11466: The Forth Interest Group (FIG) is a world-wide, non-profit,
1.26 crook 11467: member-supported organisation. It publishes a regular magazine,
11468: @var{FORTH Dimensions}, and offers other benefits of membership. You can
11469: contact the FIG through their office email address:
11470: @email{office@@forth.org} or by visiting their web site at
11471: @url{http://www.forth.org/}. This web site also includes links to FIG
11472: chapters in other countries and American cities
1.21 crook 11473: (@url{http://www.forth.org/chapters.html}).
11474:
11475: @node Conferences, , The Forth Interest Group, Forth-related information
11476: @section Conferences
11477: @cindex Conferences
11478:
11479: There are several regular conferences related to Forth. They are all
1.26 crook 11480: well-publicised in @var{FORTH Dimensions} and on the comp.lang.forth
11481: news group:
1.21 crook 11482:
11483: @itemize @bullet
11484: @item
11485: FORML -- the Forth modification laboratory convenes every year near
11486: Monterey, California.
11487: @item
11488: The Rochester Forth Conference -- an annual conference traditionally
11489: held in Rochester, New York.
11490: @item
11491: EuroForth -- this European conference takes place annually.
11492: @end itemize
11493:
11494:
11495: @node Word Index, Concept Index, Forth-related information, Top
1.1 anton 11496: @unnumbered Word Index
11497:
1.26 crook 11498: This index is a list of Forth words that have ``glossary'' entries
11499: within this manual. Each word is listed with its stack effect and
11500: wordset.
1.1 anton 11501:
11502: @printindex fn
11503:
11504: @node Concept Index, , Word Index, Top
11505: @unnumbered Concept and Word Index
11506:
1.26 crook 11507: Not all entries listed in this index are present verbatim in the
11508: text. This index also duplicates, in abbreviated form, all of the words
11509: listed in the Word Index (only the names are listed for the words here).
1.1 anton 11510:
11511: @printindex cp
11512:
11513: @contents
11514: @bye
11515:
FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>