1: \input texinfo @c -*-texinfo-*-
2: @comment The source is gforth.ds, from which gforth.texi is generated
3:
4: @comment TODO: nac29jan99 - a list of things to add in the next edit:
5: @comment 1. x-ref all ambiguous or implementation-defined features?
6: @comment 2. Describe the use of Auser Avariable AConstant A, etc.
7: @comment 3. words in miscellaneous section need a home.
8: @comment 4. search for TODO for other minor and major works required.
9: @comment 5. [rats] change all @var to @i in Forth source so that info
10: @comment file looks decent.
11: @c Not an improvement IMO - anton
12: @c and anyway, this should be taken up
13: @c with Karl Berry (the texinfo guy) - anton
14: @comment .. would be useful to have a word that identified all deferred words
15: @comment should semantics stuff in intro be moved to another section
16:
17:
18: @comment %**start of header (This is for running Texinfo on a region.)
19: @setfilename gforth.info
20: @settitle Gforth Manual
21: @dircategory GNU programming tools
22: @direntry
23: * Gforth: (gforth). A fast interpreter for the Forth language.
24: @end direntry
25: @comment @setchapternewpage odd
26: @comment TODO this gets left in by HTML converter
27: @macro progstyle {}
28: Programming style note:
29: @end macro
30:
31: @macro assignment {}
32: @table @i
33: @item Assignment:
34: @end macro
35: @macro endassignment {}
36: @end table
37: @end macro
38:
39: @comment %**end of header (This is for running Texinfo on a region.)
40:
41:
42: @comment ----------------------------------------------------------
43: @comment macros for beautifying glossary entries
44: @comment if these are used, need to strip them out for HTML converter
45: @comment else they get repeated verbatim in HTML output.
46: @comment .. not working yet.
47:
48: @macro GLOSS-START {}
49: @iftex
50: @ninerm
51: @end iftex
52: @end macro
53:
54: @macro GLOSS-END {}
55: @iftex
56: @rm
57: @end iftex
58: @end macro
59:
60: @comment ----------------------------------------------------------
61:
62:
63: @include version.texi
64:
65: @ifinfo
66: This file documents Gforth @value{VERSION}
67:
68: Copyright @copyright{} 1995-1999 Free Software Foundation, Inc.
69:
70: Permission is granted to make and distribute verbatim copies of
71: this manual provided the copyright notice and this permission notice
72: are preserved on all copies.
73:
74: @ignore
75: Permission is granted to process this file through TeX and print the
76: results, provided the printed document carries a copying permission
77: notice identical to this one except for the removal of this paragraph
78: (this paragraph not being relevant to the printed manual).
79:
80: @end ignore
81: Permission is granted to copy and distribute modified versions of this
82: manual under the conditions for verbatim copying, provided also that the
83: sections entitled "Distribution" and "General Public License" are
84: included exactly as in the original, and provided that the entire
85: resulting derived work is distributed under the terms of a permission
86: notice identical to this one.
87:
88: Permission is granted to copy and distribute translations of this manual
89: into another language, under the above conditions for modified versions,
90: except that the sections entitled "Distribution" and "General Public
91: License" may be included in a translation approved by the author instead
92: of in the original English.
93: @end ifinfo
94:
95: @finalout
96: @titlepage
97: @sp 10
98: @center @titlefont{Gforth Manual}
99: @sp 2
100: @center for version @value{VERSION}
101: @sp 2
102: @center Neal Crook
103: @center Anton Ertl
104: @center Bernd Paysan
105: @center Jens Wilke
106: @sp 3
107: @center This manual is permanently under construction and was last updated on 15-Mar-2000
108:
109: @comment The following two commands start the copyright page.
110: @page
111: @vskip 0pt plus 1filll
112: Copyright @copyright{} 1995--1999 Free Software Foundation, Inc.
113:
114: @comment !! Published by ... or You can get a copy of this manual ...
115:
116: Permission is granted to make and distribute verbatim copies of
117: this manual provided the copyright notice and this permission notice
118: are preserved on all copies.
119:
120: Permission is granted to copy and distribute modified versions of this
121: manual under the conditions for verbatim copying, provided also that the
122: sections entitled "Distribution" and "General Public License" are
123: included exactly as in the original, and provided that the entire
124: resulting derived work is distributed under the terms of a permission
125: notice identical to this one.
126:
127: Permission is granted to copy and distribute translations of this manual
128: into another language, under the above conditions for modified versions,
129: except that the sections entitled "Distribution" and "General Public
130: License" may be included in a translation approved by the author instead
131: of in the original English.
132: @end titlepage
133:
134: @node Top, License, (dir), (dir)
135: @ifinfo
136: Gforth is a free implementation of ANS Forth available on many
137: personal machines. This manual corresponds to version @value{VERSION}.
138: @end ifinfo
139:
140: @menu
141: * License:: The GPL
142: * Goals:: About the Gforth Project
143: * Gforth Environment:: Starting (and exiting) Gforth
144: * Tutorial:: Hands-on Forth Tutorial
145: * Introduction:: An introduction to ANS Forth
146: * Words:: Forth words available in Gforth
147: * Error messages:: How to interpret them
148: * Tools:: Programming tools
149: * ANS conformance:: Implementation-defined options etc.
150: * Model:: The abstract machine of Gforth
151: * Integrating Gforth:: Forth as scripting language for applications
152: * Emacs and Gforth:: The Gforth Mode
153: * Image Files:: @code{.fi} files contain compiled code
154: * Engine:: The inner interpreter and the primitives
155: * Binding to System Library::
156: * Cross Compiler:: The Cross Compiler
157: * Bugs:: How to report them
158: * Origin:: Authors and ancestors of Gforth
159: * Forth-related information:: Books and places to look on the WWW
160: * Word Index:: An item for each Forth word
161: * Name Index:: Forth words, only names listed
162: * Concept Index:: A menu covering many topics
163:
164: @detailmenu --- The Detailed Node Listing ---
165:
166: Goals of Gforth
167:
168: * Gforth Extensions Sinful?::
169:
170: Gforth Environment
171:
172: * Invoking Gforth:: Getting in
173: * Leaving Gforth:: Getting out
174: * Command-line editing::
175: * Upper and lower case::
176: * Environment variables:: that affect how Gforth starts up
177: * Gforth Files:: What gets installed and where
178: * Startup speed:: When 35ms is not fast enough ...
179:
180: Forth Tutorial
181:
182: * Starting Gforth Tutorial::
183: * Syntax Tutorial::
184: * Crash Course Tutorial::
185: * Stack Tutorial::
186: * Arithmetics Tutorial::
187: * Stack Manipulation Tutorial::
188: * Using files for Forth code Tutorial::
189: * Comments Tutorial::
190: * Colon Definitions Tutorial::
191: * Decompilation Tutorial::
192: * Stack-Effect Comments Tutorial::
193: * Types Tutorial::
194: * Factoring Tutorial::
195: * Designing the stack effect Tutorial::
196: * Local Variables Tutorial::
197: * Conditional execution Tutorial::
198: * Flags and Comparisons Tutorial::
199: * General Loops Tutorial::
200: * Counted loops Tutorial::
201: * Recursion Tutorial::
202: * Leaving definitions or loops Tutorial::
203: * Return Stack Tutorial::
204: * Memory Tutorial::
205: * Characters and Strings Tutorial::
206: * Alignment Tutorial::
207: * Interpretation and Compilation Semantics and Immediacy Tutorial::
208: * Execution Tokens Tutorial::
209: * Exceptions Tutorial::
210: * Defining Words Tutorial::
211: * Arrays and Records Tutorial::
212: * POSTPONE Tutorial::
213: * Literal Tutorial::
214: * Advanced macros Tutorial::
215: * Compilation Tokens Tutorial::
216: * Wordlists and Search Order Tutorial::
217:
218: An Introduction to ANS Forth
219:
220: * Introducing the Text Interpreter::
221: * Stacks and Postfix notation::
222: * Your first definition::
223: * How does that work?::
224: * Forth is written in Forth::
225: * Review - elements of a Forth system::
226: * Where to go next::
227: * Exercises::
228:
229: Forth Words
230:
231: * Notation::
232: * Comments::
233: * Boolean Flags::
234: * Arithmetic::
235: * Stack Manipulation::
236: * Memory::
237: * Control Structures::
238: * Defining Words::
239: * Interpretation and Compilation Semantics::
240: * Tokens for Words::
241: * The Text Interpreter::
242: * Word Lists::
243: * Environmental Queries::
244: * Files::
245: * Blocks::
246: * Other I/O::
247: * Programming Tools::
248: * Assembler and Code Words::
249: * Threading Words::
250: * Locals::
251: * Structures::
252: * Object-oriented Forth::
253: * Passing Commands to the OS::
254: * Keeping track of Time::
255: * Miscellaneous Words::
256:
257: Arithmetic
258:
259: * Single precision::
260: * Bitwise operations::
261: * Double precision:: Double-cell integer arithmetic
262: * Numeric comparison::
263: * Mixed precision:: Operations with single and double-cell integers
264: * Floating Point::
265:
266: Stack Manipulation
267:
268: * Data stack::
269: * Floating point stack::
270: * Return stack::
271: * Locals stack::
272: * Stack pointer manipulation::
273:
274: Memory
275:
276: * Memory model::
277: * Dictionary allocation::
278: * Heap Allocation::
279: * Memory Access::
280: * Address arithmetic::
281: * Memory Blocks::
282:
283: Control Structures
284:
285: * Selection:: IF ... ELSE ... ENDIF
286: * Simple Loops:: BEGIN ...
287: * Counted Loops:: DO
288: * Arbitrary control structures::
289: * Calls and returns::
290: * Exception Handling::
291:
292: Defining Words
293:
294: * CREATE::
295: * Variables:: Variables and user variables
296: * Constants::
297: * Values:: Initialised variables
298: * Colon Definitions::
299: * Anonymous Definitions:: Definitions without names
300: * User-defined Defining Words::
301: * Deferred words:: Allow forward references
302: * Aliases::
303: * Supplying names::
304:
305: Interpretation and Compilation Semantics
306:
307: * Combined words::
308:
309: The Text Interpreter
310:
311: * Input Sources::
312: * Number Conversion::
313: * Interpret/Compile states::
314: * Literals::
315: * Interpreter Directives::
316:
317: Word Lists
318:
319: * Why use word lists?::
320: * Word list examples::
321:
322: Files
323:
324: * Forth source files::
325: * General files::
326: * Search Paths::
327:
328: Search Paths
329:
330: * Forth Search Paths::
331: * General Search Paths::
332:
333: Other I/O
334:
335: * Simple numeric output:: Predefined formats
336: * Formatted numeric output:: Formatted (pictured) output
337: * String Formats:: How Forth stores strings in memory
338: * Displaying characters and strings:: Other stuff
339: * Input:: Input
340:
341: Programming Tools
342:
343: * Debugging:: Simple and quick.
344: * Assertions:: Making your programs self-checking.
345: * Singlestep Debugger:: Executing your program word by word.
346:
347: Locals
348:
349: * Gforth locals::
350: * ANS Forth locals::
351:
352: Gforth locals
353:
354: * Where are locals visible by name?::
355: * How long do locals live?::
356: * Programming Style::
357: * Implementation::
358:
359: Structures
360:
361: * Why explicit structure support?::
362: * Structure Usage::
363: * Structure Naming Convention::
364: * Structure Implementation::
365: * Structure Glossary::
366:
367: Object-oriented Forth
368:
369: * Why object-oriented programming?::
370: * Object-Oriented Terminology::
371: * Objects::
372: * OOF::
373: * Mini-OOF::
374: * Comparison with other object models::
375:
376: The @file{objects.fs} model
377:
378: * Properties of the Objects model::
379: * Basic Objects Usage::
380: * The Objects base class::
381: * Creating objects::
382: * Object-Oriented Programming Style::
383: * Class Binding::
384: * Method conveniences::
385: * Classes and Scoping::
386: * Dividing classes::
387: * Object Interfaces::
388: * Objects Implementation::
389: * Objects Glossary::
390:
391: The @file{oof.fs} model
392:
393: * Properties of the OOF model::
394: * Basic OOF Usage::
395: * The OOF base class::
396: * Class Declaration::
397: * Class Implementation::
398:
399: The @file{mini-oof.fs} model
400:
401: * Basic Mini-OOF Usage::
402: * Mini-OOF Example::
403: * Mini-OOF Implementation::
404: * Comparison with other object models::
405:
406: Tools
407:
408: * ANS Report:: Report the words used, sorted by wordset.
409:
410: ANS conformance
411:
412: * The Core Words::
413: * The optional Block word set::
414: * The optional Double Number word set::
415: * The optional Exception word set::
416: * The optional Facility word set::
417: * The optional File-Access word set::
418: * The optional Floating-Point word set::
419: * The optional Locals word set::
420: * The optional Memory-Allocation word set::
421: * The optional Programming-Tools word set::
422: * The optional Search-Order word set::
423:
424: The Core Words
425:
426: * core-idef:: Implementation Defined Options
427: * core-ambcond:: Ambiguous Conditions
428: * core-other:: Other System Documentation
429:
430: The optional Block word set
431:
432: * block-idef:: Implementation Defined Options
433: * block-ambcond:: Ambiguous Conditions
434: * block-other:: Other System Documentation
435:
436: The optional Double Number word set
437:
438: * double-ambcond:: Ambiguous Conditions
439:
440: The optional Exception word set
441:
442: * exception-idef:: Implementation Defined Options
443:
444: The optional Facility word set
445:
446: * facility-idef:: Implementation Defined Options
447: * facility-ambcond:: Ambiguous Conditions
448:
449: The optional File-Access word set
450:
451: * file-idef:: Implementation Defined Options
452: * file-ambcond:: Ambiguous Conditions
453:
454: The optional Floating-Point word set
455:
456: * floating-idef:: Implementation Defined Options
457: * floating-ambcond:: Ambiguous Conditions
458:
459: The optional Locals word set
460:
461: * locals-idef:: Implementation Defined Options
462: * locals-ambcond:: Ambiguous Conditions
463:
464: The optional Memory-Allocation word set
465:
466: * memory-idef:: Implementation Defined Options
467:
468: The optional Programming-Tools word set
469:
470: * programming-idef:: Implementation Defined Options
471: * programming-ambcond:: Ambiguous Conditions
472:
473: The optional Search-Order word set
474:
475: * search-idef:: Implementation Defined Options
476: * search-ambcond:: Ambiguous Conditions
477:
478: Image Files
479:
480: * Image Licensing Issues:: Distribution terms for images.
481: * Image File Background:: Why have image files?
482: * Non-Relocatable Image Files:: don't always work.
483: * Data-Relocatable Image Files:: are better.
484: * Fully Relocatable Image Files:: better yet.
485: * Stack and Dictionary Sizes:: Setting the default sizes for an image.
486: * Running Image Files:: @code{gforth -i @i{file}} or @i{file}.
487: * Modifying the Startup Sequence:: and turnkey applications.
488:
489: Fully Relocatable Image Files
490:
491: * gforthmi:: The normal way
492: * cross.fs:: The hard way
493:
494: Engine
495:
496: * Portability::
497: * Threading::
498: * Primitives::
499: * Performance::
500:
501: Threading
502:
503: * Scheduling::
504: * Direct or Indirect Threaded?::
505: * DOES>::
506:
507: Primitives
508:
509: * Automatic Generation::
510: * TOS Optimization::
511: * Produced code::
512:
513: Cross Compiler
514:
515: * Using the Cross Compiler::
516: * How the Cross Compiler Works::
517:
518: Other Forth-related information
519:
520: * Internet resources::
521: * Books::
522: * The Forth Interest Group::
523: * Conferences::
524:
525: @end detailmenu
526: @end menu
527:
528: @node License, Goals, Top, Top
529: @unnumbered GNU GENERAL PUBLIC LICENSE
530: @center Version 2, June 1991
531:
532: @display
533: Copyright @copyright{} 1989, 1991 Free Software Foundation, Inc.
534: 675 Mass Ave, Cambridge, MA 02139, USA
535:
536: Everyone is permitted to copy and distribute verbatim copies
537: of this license document, but changing it is not allowed.
538: @end display
539:
540: @unnumberedsec Preamble
541:
542: The licenses for most software are designed to take away your
543: freedom to share and change it. By contrast, the GNU General Public
544: License is intended to guarantee your freedom to share and change free
545: software---to make sure the software is free for all its users. This
546: General Public License applies to most of the Free Software
547: Foundation's software and to any other program whose authors commit to
548: using it. (Some other Free Software Foundation software is covered by
549: the GNU Library General Public License instead.) You can apply it to
550: your programs, too.
551:
552: When we speak of free software, we are referring to freedom, not
553: price. Our General Public Licenses are designed to make sure that you
554: have the freedom to distribute copies of free software (and charge for
555: this service if you wish), that you receive source code or can get it
556: if you want it, that you can change the software or use pieces of it
557: in new free programs; and that you know you can do these things.
558:
559: To protect your rights, we need to make restrictions that forbid
560: anyone to deny you these rights or to ask you to surrender the rights.
561: These restrictions translate to certain responsibilities for you if you
562: distribute copies of the software, or if you modify it.
563:
564: For example, if you distribute copies of such a program, whether
565: gratis or for a fee, you must give the recipients all the rights that
566: you have. You must make sure that they, too, receive or can get the
567: source code. And you must show them these terms so they know their
568: rights.
569:
570: We protect your rights with two steps: (1) copyright the software, and
571: (2) offer you this license which gives you legal permission to copy,
572: distribute and/or modify the software.
573:
574: Also, for each author's protection and ours, we want to make certain
575: that everyone understands that there is no warranty for this free
576: software. If the software is modified by someone else and passed on, we
577: want its recipients to know that what they have is not the original, so
578: that any problems introduced by others will not reflect on the original
579: authors' reputations.
580:
581: Finally, any free program is threatened constantly by software
582: patents. We wish to avoid the danger that redistributors of a free
583: program will individually obtain patent licenses, in effect making the
584: program proprietary. To prevent this, we have made it clear that any
585: patent must be licensed for everyone's free use or not licensed at all.
586:
587: The precise terms and conditions for copying, distribution and
588: modification follow.
589:
590: @iftex
591: @unnumberedsec TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION
592: @end iftex
593: @ifinfo
594: @center TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION
595: @end ifinfo
596:
597: @enumerate 0
598: @item
599: This License applies to any program or other work which contains
600: a notice placed by the copyright holder saying it may be distributed
601: under the terms of this General Public License. The ``Program'', below,
602: refers to any such program or work, and a ``work based on the Program''
603: means either the Program or any derivative work under copyright law:
604: that is to say, a work containing the Program or a portion of it,
605: either verbatim or with modifications and/or translated into another
606: language. (Hereinafter, translation is included without limitation in
607: the term ``modification''.) Each licensee is addressed as ``you''.
608:
609: Activities other than copying, distribution and modification are not
610: covered by this License; they are outside its scope. The act of
611: running the Program is not restricted, and the output from the Program
612: is covered only if its contents constitute a work based on the
613: Program (independent of having been made by running the Program).
614: Whether that is true depends on what the Program does.
615:
616: @item
617: You may copy and distribute verbatim copies of the Program's
618: source code as you receive it, in any medium, provided that you
619: conspicuously and appropriately publish on each copy an appropriate
620: copyright notice and disclaimer of warranty; keep intact all the
621: notices that refer to this License and to the absence of any warranty;
622: and give any other recipients of the Program a copy of this License
623: along with the Program.
624:
625: You may charge a fee for the physical act of transferring a copy, and
626: you may at your option offer warranty protection in exchange for a fee.
627:
628: @item
629: You may modify your copy or copies of the Program or any portion
630: of it, thus forming a work based on the Program, and copy and
631: distribute such modifications or work under the terms of Section 1
632: above, provided that you also meet all of these conditions:
633:
634: @enumerate a
635: @item
636: You must cause the modified files to carry prominent notices
637: stating that you changed the files and the date of any change.
638:
639: @item
640: You must cause any work that you distribute or publish, that in
641: whole or in part contains or is derived from the Program or any
642: part thereof, to be licensed as a whole at no charge to all third
643: parties under the terms of this License.
644:
645: @item
646: If the modified program normally reads commands interactively
647: when run, you must cause it, when started running for such
648: interactive use in the most ordinary way, to print or display an
649: announcement including an appropriate copyright notice and a
650: notice that there is no warranty (or else, saying that you provide
651: a warranty) and that users may redistribute the program under
652: these conditions, and telling the user how to view a copy of this
653: License. (Exception: if the Program itself is interactive but
654: does not normally print such an announcement, your work based on
655: the Program is not required to print an announcement.)
656: @end enumerate
657:
658: These requirements apply to the modified work as a whole. If
659: identifiable sections of that work are not derived from the Program,
660: and can be reasonably considered independent and separate works in
661: themselves, then this License, and its terms, do not apply to those
662: sections when you distribute them as separate works. But when you
663: distribute the same sections as part of a whole which is a work based
664: on the Program, the distribution of the whole must be on the terms of
665: this License, whose permissions for other licensees extend to the
666: entire whole, and thus to each and every part regardless of who wrote it.
667:
668: Thus, it is not the intent of this section to claim rights or contest
669: your rights to work written entirely by you; rather, the intent is to
670: exercise the right to control the distribution of derivative or
671: collective works based on the Program.
672:
673: In addition, mere aggregation of another work not based on the Program
674: with the Program (or with a work based on the Program) on a volume of
675: a storage or distribution medium does not bring the other work under
676: the scope of this License.
677:
678: @item
679: You may copy and distribute the Program (or a work based on it,
680: under Section 2) in object code or executable form under the terms of
681: Sections 1 and 2 above provided that you also do one of the following:
682:
683: @enumerate a
684: @item
685: Accompany it with the complete corresponding machine-readable
686: source code, which must be distributed under the terms of Sections
687: 1 and 2 above on a medium customarily used for software interchange; or,
688:
689: @item
690: Accompany it with a written offer, valid for at least three
691: years, to give any third party, for a charge no more than your
692: cost of physically performing source distribution, a complete
693: machine-readable copy of the corresponding source code, to be
694: distributed under the terms of Sections 1 and 2 above on a medium
695: customarily used for software interchange; or,
696:
697: @item
698: Accompany it with the information you received as to the offer
699: to distribute corresponding source code. (This alternative is
700: allowed only for noncommercial distribution and only if you
701: received the program in object code or executable form with such
702: an offer, in accord with Subsection b above.)
703: @end enumerate
704:
705: The source code for a work means the preferred form of the work for
706: making modifications to it. For an executable work, complete source
707: code means all the source code for all modules it contains, plus any
708: associated interface definition files, plus the scripts used to
709: control compilation and installation of the executable. However, as a
710: special exception, the source code distributed need not include
711: anything that is normally distributed (in either source or binary
712: form) with the major components (compiler, kernel, and so on) of the
713: operating system on which the executable runs, unless that component
714: itself accompanies the executable.
715:
716: If distribution of executable or object code is made by offering
717: access to copy from a designated place, then offering equivalent
718: access to copy the source code from the same place counts as
719: distribution of the source code, even though third parties are not
720: compelled to copy the source along with the object code.
721:
722: @item
723: You may not copy, modify, sublicense, or distribute the Program
724: except as expressly provided under this License. Any attempt
725: otherwise to copy, modify, sublicense or distribute the Program is
726: void, and will automatically terminate your rights under this License.
727: However, parties who have received copies, or rights, from you under
728: this License will not have their licenses terminated so long as such
729: parties remain in full compliance.
730:
731: @item
732: You are not required to accept this License, since you have not
733: signed it. However, nothing else grants you permission to modify or
734: distribute the Program or its derivative works. These actions are
735: prohibited by law if you do not accept this License. Therefore, by
736: modifying or distributing the Program (or any work based on the
737: Program), you indicate your acceptance of this License to do so, and
738: all its terms and conditions for copying, distributing or modifying
739: the Program or works based on it.
740:
741: @item
742: Each time you redistribute the Program (or any work based on the
743: Program), the recipient automatically receives a license from the
744: original licensor to copy, distribute or modify the Program subject to
745: these terms and conditions. You may not impose any further
746: restrictions on the recipients' exercise of the rights granted herein.
747: You are not responsible for enforcing compliance by third parties to
748: this License.
749:
750: @item
751: If, as a consequence of a court judgment or allegation of patent
752: infringement or for any other reason (not limited to patent issues),
753: conditions are imposed on you (whether by court order, agreement or
754: otherwise) that contradict the conditions of this License, they do not
755: excuse you from the conditions of this License. If you cannot
756: distribute so as to satisfy simultaneously your obligations under this
757: License and any other pertinent obligations, then as a consequence you
758: may not distribute the Program at all. For example, if a patent
759: license would not permit royalty-free redistribution of the Program by
760: all those who receive copies directly or indirectly through you, then
761: the only way you could satisfy both it and this License would be to
762: refrain entirely from distribution of the Program.
763:
764: If any portion of this section is held invalid or unenforceable under
765: any particular circumstance, the balance of the section is intended to
766: apply and the section as a whole is intended to apply in other
767: circumstances.
768:
769: It is not the purpose of this section to induce you to infringe any
770: patents or other property right claims or to contest validity of any
771: such claims; this section has the sole purpose of protecting the
772: integrity of the free software distribution system, which is
773: implemented by public license practices. Many people have made
774: generous contributions to the wide range of software distributed
775: through that system in reliance on consistent application of that
776: system; it is up to the author/donor to decide if he or she is willing
777: to distribute software through any other system and a licensee cannot
778: impose that choice.
779:
780: This section is intended to make thoroughly clear what is believed to
781: be a consequence of the rest of this License.
782:
783: @item
784: If the distribution and/or use of the Program is restricted in
785: certain countries either by patents or by copyrighted interfaces, the
786: original copyright holder who places the Program under this License
787: may add an explicit geographical distribution limitation excluding
788: those countries, so that distribution is permitted only in or among
789: countries not thus excluded. In such case, this License incorporates
790: the limitation as if written in the body of this License.
791:
792: @item
793: The Free Software Foundation may publish revised and/or new versions
794: of the General Public License from time to time. Such new versions will
795: be similar in spirit to the present version, but may differ in detail to
796: address new problems or concerns.
797:
798: Each version is given a distinguishing version number. If the Program
799: specifies a version number of this License which applies to it and ``any
800: later version'', you have the option of following the terms and conditions
801: either of that version or of any later version published by the Free
802: Software Foundation. If the Program does not specify a version number of
803: this License, you may choose any version ever published by the Free Software
804: Foundation.
805:
806: @item
807: If you wish to incorporate parts of the Program into other free
808: programs whose distribution conditions are different, write to the author
809: to ask for permission. For software which is copyrighted by the Free
810: Software Foundation, write to the Free Software Foundation; we sometimes
811: make exceptions for this. Our decision will be guided by the two goals
812: of preserving the free status of all derivatives of our free software and
813: of promoting the sharing and reuse of software generally.
814:
815: @iftex
816: @heading NO WARRANTY
817: @end iftex
818: @ifinfo
819: @center NO WARRANTY
820: @end ifinfo
821:
822: @item
823: BECAUSE THE PROGRAM IS LICENSED FREE OF CHARGE, THERE IS NO WARRANTY
824: FOR THE PROGRAM, TO THE EXTENT PERMITTED BY APPLICABLE LAW. EXCEPT WHEN
825: OTHERWISE STATED IN WRITING THE COPYRIGHT HOLDERS AND/OR OTHER PARTIES
826: PROVIDE THE PROGRAM ``AS IS'' WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESSED
827: OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
828: MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. THE ENTIRE RISK AS
829: TO THE QUALITY AND PERFORMANCE OF THE PROGRAM IS WITH YOU. SHOULD THE
830: PROGRAM PROVE DEFECTIVE, YOU ASSUME THE COST OF ALL NECESSARY SERVICING,
831: REPAIR OR CORRECTION.
832:
833: @item
834: IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING
835: WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MAY MODIFY AND/OR
836: REDISTRIBUTE THE PROGRAM AS PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES,
837: INCLUDING ANY GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING
838: OUT OF THE USE OR INABILITY TO USE THE PROGRAM (INCLUDING BUT NOT LIMITED
839: TO LOSS OF DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY
840: YOU OR THIRD PARTIES OR A FAILURE OF THE PROGRAM TO OPERATE WITH ANY OTHER
841: PROGRAMS), EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE
842: POSSIBILITY OF SUCH DAMAGES.
843: @end enumerate
844:
845: @iftex
846: @heading END OF TERMS AND CONDITIONS
847: @end iftex
848: @ifinfo
849: @center END OF TERMS AND CONDITIONS
850: @end ifinfo
851:
852: @page
853: @unnumberedsec How to Apply These Terms to Your New Programs
854:
855: If you develop a new program, and you want it to be of the greatest
856: possible use to the public, the best way to achieve this is to make it
857: free software which everyone can redistribute and change under these terms.
858:
859: To do so, attach the following notices to the program. It is safest
860: to attach them to the start of each source file to most effectively
861: convey the exclusion of warranty; and each file should have at least
862: the ``copyright'' line and a pointer to where the full notice is found.
863:
864: @smallexample
865: @var{one line to give the program's name and a brief idea of what it does.}
866: Copyright (C) 19@var{yy} @var{name of author}
867:
868: This program is free software; you can redistribute it and/or modify
869: it under the terms of the GNU General Public License as published by
870: the Free Software Foundation; either version 2 of the License, or
871: (at your option) any later version.
872:
873: This program is distributed in the hope that it will be useful,
874: but WITHOUT ANY WARRANTY; without even the implied warranty of
875: MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
876: GNU General Public License for more details.
877:
878: You should have received a copy of the GNU General Public License
879: along with this program; if not, write to the Free Software
880: Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
881: @end smallexample
882:
883: Also add information on how to contact you by electronic and paper mail.
884:
885: If the program is interactive, make it output a short notice like this
886: when it starts in an interactive mode:
887:
888: @smallexample
889: Gnomovision version 69, Copyright (C) 19@var{yy} @var{name of author}
890: Gnomovision comes with ABSOLUTELY NO WARRANTY; for details
891: type `show w'.
892: This is free software, and you are welcome to redistribute it
893: under certain conditions; type `show c' for details.
894: @end smallexample
895:
896: The hypothetical commands @samp{show w} and @samp{show c} should show
897: the appropriate parts of the General Public License. Of course, the
898: commands you use may be called something other than @samp{show w} and
899: @samp{show c}; they could even be mouse-clicks or menu items---whatever
900: suits your program.
901:
902: You should also get your employer (if you work as a programmer) or your
903: school, if any, to sign a ``copyright disclaimer'' for the program, if
904: necessary. Here is a sample; alter the names:
905:
906: @smallexample
907: Yoyodyne, Inc., hereby disclaims all copyright interest in the program
908: `Gnomovision' (which makes passes at compilers) written by James Hacker.
909:
910: @var{signature of Ty Coon}, 1 April 1989
911: Ty Coon, President of Vice
912: @end smallexample
913:
914: This General Public License does not permit incorporating your program into
915: proprietary programs. If your program is a subroutine library, you may
916: consider it more useful to permit linking proprietary applications with the
917: library. If this is what you want to do, use the GNU Library General
918: Public License instead of this License.
919:
920: @iftex
921: @unnumbered Preface
922: @cindex Preface
923: This manual documents Gforth. Some introductory material is provided for
924: readers who are unfamiliar with Forth or who are migrating to Gforth
925: from other Forth compilers. However, this manual is primarily a
926: reference manual.
927: @end iftex
928:
929: @comment TODO much more blurb here.
930:
931: @c ******************************************************************
932: @node Goals, Gforth Environment, License, Top
933: @comment node-name, next, previous, up
934: @chapter Goals of Gforth
935: @cindex goals of the Gforth project
936: The goal of the Gforth Project is to develop a standard model for
937: ANS Forth. This can be split into several subgoals:
938:
939: @itemize @bullet
940: @item
941: Gforth should conform to the ANS Forth Standard.
942: @item
943: It should be a model, i.e. it should define all the
944: implementation-dependent things.
945: @item
946: It should become standard, i.e. widely accepted and used. This goal
947: is the most difficult one.
948: @end itemize
949:
950: To achieve these goals Gforth should be
951: @itemize @bullet
952: @item
953: Similar to previous models (fig-Forth, F83)
954: @item
955: Powerful. It should provide for all the things that are considered
956: necessary today and even some that are not yet considered necessary.
957: @item
958: Efficient. It should not get the reputation of being exceptionally
959: slow.
960: @item
961: Free.
962: @item
963: Available on many machines/easy to port.
964: @end itemize
965:
966: Have we achieved these goals? Gforth conforms to the ANS Forth
967: standard. It may be considered a model, but we have not yet documented
968: which parts of the model are stable and which parts we are likely to
969: change. It certainly has not yet become a de facto standard, but it
970: appears to be quite popular. It has some similarities to and some
971: differences from previous models. It has some powerful features, but not
972: yet everything that we envisioned. We certainly have achieved our
973: execution speed goals (@pxref{Performance}). It is free and available
974: on many machines.
975:
976: @menu
977: * Gforth Extensions Sinful?::
978: @end menu
979:
980: @node Gforth Extensions Sinful?, , Goals, Goals
981: @comment node-name, next, previous, up
982: @section Is it a Sin to use Gforth Extensions?
983: @cindex Gforth extensions
984:
985: If you've been paying attention, you will have realised that there is an
986: ANS (American National Standard) for Forth. As you read through the rest
987: of this manual, you will see documentation for @i{Standard} words, and
988: documentation for some appealing Gforth @i{extensions}. You might ask
989: yourself the question: @i{``Given that there is a standard, would I be
990: committing a sin if I use (non-Standard) Gforth extensions?''}
991:
992: The answer to that question is somewhat pragmatic and somewhat
993: philosophical. Consider these points:
994:
995: @itemize @bullet
996: @item
997: A number of the Gforth extensions can be implemented in ANS Forth using
998: files provided in the @file{compat/} directory. These are mentioned in
999: the text in passing.
1000: @item
1001: Forth has a rich historical precedent for programmers taking advantage
1002: of implementation-dependent features of their tools (for example,
1003: relying on a knowledge of the dictionary structure). Sometimes these
1004: techniques are necessary to extract every last bit of performance from
1005: the hardware, sometimes they are just a programming shorthand.
1006: @item
1007: The best way to break the rules is to know what the rules are. To learn
1008: the rules, there is no substitute for studying the text of the Standard
1009: itself. In particular, Appendix A of the Standard (@var{Rationale})
1010: provides a valuable insight into the thought processes of the technical
1011: committee.
1012: @item
1013: The best reason to break a rule is because you have to; because it's
1014: more productive to do that, because it makes your code run fast enough
1015: or because you can see no Standard way to achieve what you want to
1016: achieve.
1017: @end itemize
1018:
1019: The tool @file{ans-report.fs} (@pxref{ANS Report}) makes it easy to
1020: analyse your program and determine what non-Standard definitions it
1021: relies upon.
1022:
1023:
1024: @c ******************************************************************
1025: @node Gforth Environment, Tutorial, Goals, Top
1026: @chapter Gforth Environment
1027: @cindex Gforth environment
1028:
1029: Note: ultimately, the Gforth man page will be auto-generated from the
1030: material in this chapter.
1031:
1032: @menu
1033: * Invoking Gforth:: Getting in
1034: * Leaving Gforth:: Getting out
1035: * Command-line editing::
1036: * Upper and lower case::
1037: * Environment variables:: that affect how Gforth starts up
1038: * Gforth Files:: What gets installed and where
1039: * Startup speed:: When 35ms is not fast enough ...
1040: @end menu
1041:
1042: @xref{Image Files} for related information about the creation of images.
1043:
1044: @comment ----------------------------------------------
1045: @node Invoking Gforth, Leaving Gforth, Gforth Environment, Gforth Environment
1046: @section Invoking Gforth
1047: @cindex invoking Gforth
1048: @cindex running Gforth
1049: @cindex command-line options
1050: @cindex options on the command line
1051: @cindex flags on the command line
1052:
1053: Gforth is made up of two parts; an executable ``engine'' (named
1054: @file{gforth} or @file{gforth-fast}) and an image file. To start it, you
1055: will usually just say @code{gforth} -- this automatically loads the
1056: default image file @file{gforth.fi}. In many other cases the default
1057: Gforth image will be invoked like this:
1058: @example
1059: gforth [file | -e forth-code] ...
1060: @end example
1061: @noindent
1062: This interprets the contents of the files and the Forth code in the order they
1063: are given.
1064:
1065: In addition to the @file{gforth} engine, there is also an engine called
1066: @file{gforth-fast}, which is faster, but gives less informative error
1067: messages (@pxref{Error messages}).
1068:
1069: In general, the command line looks like this:
1070:
1071: @example
1072: gforth[-fast] [engine options] [image options]
1073: @end example
1074:
1075: The engine options must come before the rest of the command
1076: line. They are:
1077:
1078: @table @code
1079: @cindex -i, command-line option
1080: @cindex --image-file, command-line option
1081: @item --image-file @i{file}
1082: @itemx -i @i{file}
1083: Loads the Forth image @i{file} instead of the default
1084: @file{gforth.fi} (@pxref{Image Files}).
1085:
1086: @cindex --appl-image, command-line option
1087: @item --appl-image @i{file}
1088: Loads the image @i{file} and leaves all further command-line arguments
1089: to the image (instead of processing them as options). This is useful
1090: for building executable application images on Unix, built with
1091: @code{gforthmi --application ...}.
1092:
1093: @cindex --path, command-line option
1094: @cindex -p, command-line option
1095: @item --path @i{path}
1096: @itemx -p @i{path}
1097: Uses @i{path} for searching the image file and Forth source code files
1098: instead of the default in the environment variable @code{GFORTHPATH} or
1099: the path specified at installation time (e.g.,
1100: @file{/usr/local/share/gforth/0.2.0:.}). A path is given as a list of
1101: directories, separated by @samp{:} (on Unix) or @samp{;} (on other OSs).
1102:
1103: @cindex --dictionary-size, command-line option
1104: @cindex -m, command-line option
1105: @cindex @i{size} parameters for command-line options
1106: @cindex size of the dictionary and the stacks
1107: @item --dictionary-size @i{size}
1108: @itemx -m @i{size}
1109: Allocate @i{size} space for the Forth dictionary space instead of
1110: using the default specified in the image (typically 256K). The
1111: @i{size} specification for this and subsequent options consists of
1112: an integer and a unit (e.g.,
1113: @code{4M}). The unit can be one of @code{b} (bytes), @code{e} (element
1114: size, in this case Cells), @code{k} (kilobytes), @code{M} (Megabytes),
1115: @code{G} (Gigabytes), and @code{T} (Terabytes). If no unit is specified,
1116: @code{e} is used.
1117:
1118: @cindex --data-stack-size, command-line option
1119: @cindex -d, command-line option
1120: @item --data-stack-size @i{size}
1121: @itemx -d @i{size}
1122: Allocate @i{size} space for the data stack instead of using the
1123: default specified in the image (typically 16K).
1124:
1125: @cindex --return-stack-size, command-line option
1126: @cindex -r, command-line option
1127: @item --return-stack-size @i{size}
1128: @itemx -r @i{size}
1129: Allocate @i{size} space for the return stack instead of using the
1130: default specified in the image (typically 15K).
1131:
1132: @cindex --fp-stack-size, command-line option
1133: @cindex -f, command-line option
1134: @item --fp-stack-size @i{size}
1135: @itemx -f @i{size}
1136: Allocate @i{size} space for the floating point stack instead of
1137: using the default specified in the image (typically 15.5K). In this case
1138: the unit specifier @code{e} refers to floating point numbers.
1139:
1140: @cindex --locals-stack-size, command-line option
1141: @cindex -l, command-line option
1142: @item --locals-stack-size @i{size}
1143: @itemx -l @i{size}
1144: Allocate @i{size} space for the locals stack instead of using the
1145: default specified in the image (typically 14.5K).
1146:
1147: @cindex -h, command-line option
1148: @cindex --help, command-line option
1149: @item --help
1150: @itemx -h
1151: Print a message about the command-line options
1152:
1153: @cindex -v, command-line option
1154: @cindex --version, command-line option
1155: @item --version
1156: @itemx -v
1157: Print version and exit
1158:
1159: @cindex --debug, command-line option
1160: @item --debug
1161: Print some information useful for debugging on startup.
1162:
1163: @cindex --offset-image, command-line option
1164: @item --offset-image
1165: Start the dictionary at a slightly different position than would be used
1166: otherwise (useful for creating data-relocatable images,
1167: @pxref{Data-Relocatable Image Files}).
1168:
1169: @cindex --no-offset-im, command-line option
1170: @item --no-offset-im
1171: Start the dictionary at the normal position.
1172:
1173: @cindex --clear-dictionary, command-line option
1174: @item --clear-dictionary
1175: Initialize all bytes in the dictionary to 0 before loading the image
1176: (@pxref{Data-Relocatable Image Files}).
1177:
1178: @cindex --die-on-signal, command-line-option
1179: @item --die-on-signal
1180: Normally Gforth handles most signals (e.g., the user interrupt SIGINT,
1181: or the segmentation violation SIGSEGV) by translating it into a Forth
1182: @code{THROW}. With this option, Gforth exits if it receives such a
1183: signal. This option is useful when the engine and/or the image might be
1184: severely broken (such that it causes another signal before recovering
1185: from the first); this option avoids endless loops in such cases.
1186: @end table
1187:
1188: @cindex loading files at startup
1189: @cindex executing code on startup
1190: @cindex batch processing with Gforth
1191: As explained above, the image-specific command-line arguments for the
1192: default image @file{gforth.fi} consist of a sequence of filenames and
1193: @code{-e @var{forth-code}} options that are interpreted in the sequence
1194: in which they are given. The @code{-e @var{forth-code}} or
1195: @code{--evaluate @var{forth-code}} option evaluates the Forth
1196: code. This option takes only one argument; if you want to evaluate more
1197: Forth words, you have to quote them or use @code{-e} several times. To exit
1198: after processing the command line (instead of entering interactive mode)
1199: append @code{-e bye} to the command line.
1200:
1201: @cindex versions, invoking other versions of Gforth
1202: If you have several versions of Gforth installed, @code{gforth} will
1203: invoke the version that was installed last. @code{gforth-@i{version}}
1204: invokes a specific version. If your environment contains the variable
1205: @code{GFORTHPATH}, you may want to override it by using the
1206: @code{--path} option.
1207:
1208: Not yet implemented:
1209: On startup the system first executes the system initialization file
1210: (unless the option @code{--no-init-file} is given; note that the system
1211: resulting from using this option may not be ANS Forth conformant). Then
1212: the user initialization file @file{.gforth.fs} is executed, unless the
1213: option @code{--no-rc} is given; this file is first searched in @file{.},
1214: then in @file{~}, then in the normal path (see above).
1215:
1216:
1217:
1218: @comment ----------------------------------------------
1219: @node Leaving Gforth, Command-line editing, Invoking Gforth, Gforth Environment
1220: @section Leaving Gforth
1221: @cindex Gforth - leaving
1222: @cindex leaving Gforth
1223:
1224: You can leave Gforth by typing @code{bye} or @kbd{Ctrl-d} (at the start
1225: of a line) or (if you invoked Gforth with the @code{--die-on-signal}
1226: option) @kbd{Ctrl-c}. When you leave Gforth, all of your definitions and
1227: data are discarded. @xref{Image Files} for ways of saving the state of
1228: the system before leaving Gforth.
1229:
1230: doc-bye
1231:
1232:
1233: @comment ----------------------------------------------
1234: @node Command-line editing, Upper and lower case, Leaving Gforth, Gforth Environment
1235: @section Command-line editing
1236: @cindex command-line editing
1237:
1238: Gforth maintains a history file that records every line that you type to
1239: the text interpreter. This file is preserved between sessions, and is
1240: used to provide a command-line recall facility; if you type @kbd{Ctrl-P}
1241: repeatedly you can recall successively older commands from this (or
1242: previous) session(s). The full list of command-line editing facilities is:
1243:
1244: @itemize @bullet
1245: @item
1246: @kbd{Ctrl-p} (``previous'') (or up-arrow) to recall successively older
1247: commands from the history buffer.
1248: @item
1249: @kbd{Ctrl-n} (``next'') (or down-arrow) to recall successively newer commands
1250: from the history buffer.
1251: @item
1252: @kbd{Ctrl-f} (or right-arrow) to move the cursor right, non-destructively.
1253: @item
1254: @kbd{Ctrl-b} (or left-arrow) to move the cursor left, non-destructively.
1255: @item
1256: @kbd{Ctrl-h} (backspace) to delete the character to the left of the cursor,
1257: closing up the line.
1258: @item
1259: @kbd{Ctrl-k} to delete (``kill'') from the cursor to the end of the line.
1260: @item
1261: @kbd{Ctrl-a} to move the cursor to the start of the line.
1262: @item
1263: @kbd{Ctrl-e} to move the cursor to the end of the line.
1264: @item
1265: @key{RET} (@kbd{Ctrl-m}) or @key{LFD} (@kbd{Ctrl-j}) to submit the current
1266: line.
1267: @item
1268: @key{TAB} to step through all possible full-word completions of the word
1269: currently being typed.
1270: @item
1271: @kbd{Ctrl-d} at the start of the line to terminate Gforth (gracefully,
1272: using @code{bye}).
1273: @end itemize
1274:
1275: When editing, displayable characters are inserted to the left of the
1276: cursor position; the line is always in ``insert'' (as opposed to
1277: ``overstrike'') mode.
1278:
1279: @cindex history file
1280: @cindex @file{.gforth-history}
1281: On Unix systems, the history file is @file{~/.gforth-history} by
1282: default@footnote{i.e. it is stored in the user's home directory.}. You
1283: can find out the name and location of your history file using:
1284:
1285: @example
1286: history-file type \ Unix-class systems
1287:
1288: history-file type \ Other systems
1289: history-dir type
1290: @end example
1291:
1292: If you enter long definitions by hand, you can use a text editor to
1293: paste them out of the history file into a Forth source file for reuse at
1294: a later time.
1295:
1296: Gforth never trims the size of the history file, so you should do this
1297: periodically, if necessary.
1298:
1299: @comment this is all defined in history.fs
1300: @comment NAC TODO the ctrl-D behaviour can either do a bye or a beep.. how is that option
1301: @comment chosen?
1302:
1303:
1304:
1305: @comment ----------------------------------------------
1306: @node Upper and lower case, Environment variables, Command-line editing, Gforth Environment
1307: @section Upper and lower case
1308: @cindex case-sensitivity
1309: @cindex upper and lower case
1310:
1311: Gforth is case-insensitive; you can enter definitions and invoke
1312: Standard words using upper, lower or mixed case (however,
1313: @pxref{core-idef, Implementation-defined options, Implementation-defined
1314: options}).
1315:
1316: ANS Forth only @i{requires} implementations to recognise Standard words
1317: when they are typed entirely in upper case. Therefore, a Standard
1318: program must use upper case for all Standard words. You can use whatever
1319: case you like for words that you define, but in a standard program you
1320: have to use the words in the same case that you defined them.
1321:
1322: Gforth supports case sensitivity through @code{table}s (case-sensitive
1323: wordlists, @pxref{Word Lists}).
1324:
1325: Two people have asked how to convert Gforth to case sensitivity; while
1326: we think this is a bad idea, you can change all wordlists into tables
1327: like this:
1328:
1329: @example
1330: ' table-find forth-wordlist wordlist-map @ !
1331: @end example
1332:
1333: Note that you now have to type the predefined words in the same case
1334: that we defined them, which are varying. You may want to convert them
1335: to your favourite case before doing this operation (I won't explain how,
1336: because if you are even contemplating to do this, you'd better have
1337: enough knowledge of Forth systems to know this already).
1338:
1339: @comment ----------------------------------------------
1340: @node Environment variables, Gforth Files, Upper and lower case, Gforth Environment
1341: @section Environment variables
1342: @cindex environment variables
1343:
1344: Gforth uses these environment variables:
1345:
1346: @itemize @bullet
1347: @item
1348: @cindex @code{GFORTHHIST} -- environment variable
1349: @code{GFORTHHIST} -- (Unix systems only) specifies the directory in which to
1350: open/create the history file, @file{.gforth-history}. Default:
1351: @code{$HOME}.
1352:
1353: @item
1354: @cindex @code{GFORTHPATH} -- environment variable
1355: @code{GFORTHPATH} -- specifies the path used when searching for the gforth image file and
1356: for Forth source-code files.
1357:
1358: @item
1359: @cindex @code{GFORTH} -- environment variable
1360: @code{GFORTH} -- used by @file{gforthmi} @xref{gforthmi}.
1361:
1362: @item
1363: @cindex @code{GFORTHD} -- environment variable
1364: @code{GFORTHD} -- used by @file{gforthmi} @xref{gforthmi}.
1365:
1366: @item
1367: @cindex @code{TMP}, @code{TEMP} - environment variable
1368: @code{TMP}, @code{TEMP} - (non-Unix systems only) used as a potential
1369: location for the history file.
1370: @end itemize
1371:
1372: @comment also POSIXELY_CORRECT LINES COLUMNS HOME but no interest in
1373: @comment mentioning these.
1374:
1375: All the Gforth environment variables default to sensible values if they
1376: are not set.
1377:
1378:
1379: @comment ----------------------------------------------
1380: @node Gforth Files, Startup speed, Environment variables, Gforth Environment
1381: @section Gforth files
1382: @cindex Gforth files
1383:
1384: When you install Gforth on a Unix system, it installs files in these
1385: locations by default:
1386:
1387: @itemize @bullet
1388: @item
1389: @file{/usr/local/bin/gforth}
1390: @item
1391: @file{/usr/local/bin/gforthmi}
1392: @item
1393: @file{/usr/local/man/man1/gforth.1} - man page.
1394: @item
1395: @file{/usr/local/info} - the Info version of this manual.
1396: @item
1397: @file{/usr/local/lib/gforth/<version>/...} - Gforth @file{.fi} files.
1398: @item
1399: @file{/usr/local/share/gforth/<version>/TAGS} - Emacs TAGS file.
1400: @item
1401: @file{/usr/local/share/gforth/<version>/...} - Gforth source files.
1402: @item
1403: @file{.../emacs/site-lisp/gforth.el} - Emacs gforth mode.
1404: @end itemize
1405:
1406: You can select different places for installation by using
1407: @code{configure} options (listed with @code{configure --help}).
1408:
1409: @comment ----------------------------------------------
1410: @node Startup speed, , Gforth Files, Gforth Environment
1411: @section Startup speed
1412: @cindex Startup speed
1413: @cindex speed, startup
1414:
1415: If Gforth is used for CGI scripts or in shell scripts, its startup
1416: speed may become a problem. On a 300MHz 21064a under Linux-2.2.13 with
1417: glibc-2.0.7, @code{gforth -e bye} takes about 24.6ms user and 11.3ms
1418: system time.
1419:
1420: If startup speed is a problem, you may consider the following ways to
1421: improve it; or you may consider ways to reduce the number of startups
1422: (e.g., Fast-CGI).
1423:
1424: The first step to improve startup speed is to statically link Gforth, by
1425: building it with @code{XLDFLAGS=-static}. This requires more memory for
1426: the code and will therefore slow down the first invocation, but
1427: subsequent invocations avoid the dynamic linking overhead. Another
1428: disadvantage is that Gforth won't profit from library upgrades. As a
1429: result, @code{gforth-static -e bye} takes about 17.1ms user and
1430: 8.2ms system time.
1431:
1432: The next step to improve startup speed is to use a non-relocatable image
1433: @ref{Non-Relocatable Image Files}. You can create this image with
1434: @code{gforth -e "savesystem gforthnr.fi bye"} and later use it with
1435: @code{gforth -i gforthnr.fi ...}. This avoids the relocation overhead
1436: and a part of the copy-on-write overhead. The disadvantage is that the
1437: nonrelocatable image does not work if the OS gives Gforth a different
1438: address for the dictionary, for whatever reason; so you better provide a
1439: fallback on a relocatable image. @code{gforth-static -i gforthnr.fi -e
1440: bye} takes about 15.3ms user and 7.5ms system time.
1441:
1442: The final step is to disable dictionary hashing in Gforth. Gforth
1443: builds the hash table on startup, which takes much of the startup
1444: overhead. You can do this by commenting out the @code{include hash.fs}
1445: in @file{startup.fs} and everything that requires @file{hash.fs} (at the
1446: moment @file{table.fs} and @file{ekey.fs}) and then doing @code{make}.
1447: The disadvantages are that functionality like @code{table} and
1448: @code{ekey} is missing and that text interpretation (e.g., compiling)
1449: now takes much longer. So, you should only use this method if there is
1450: no significant text interpretation to perform (the script should be
1451: compiled into the image, among other things). @code{gforth-static -i
1452: gforthnrnh.fi -e bye} takes about 2.1ms user and 6.1ms system time.
1453:
1454: @c ******************************************************************
1455: @node Tutorial, Introduction, Gforth Environment, Top
1456: @chapter Forth Tutorial
1457: @cindex Tutorial
1458: @cindex Forth Tutorial
1459:
1460: This tutorial can be used with any ANS-compliant Forth; any places that
1461: mention features specific to Gforth are marked as such and you can skip
1462: them, if you work with another Forth. This tutorial does not explain
1463: all features of Forth, just enough to get you started and give you some
1464: ideas about the facilities available in Forth. Read the rest of the
1465: manual and the standard when you are through this.
1466:
1467: The intended way to use this tutorial is that you work through it while
1468: sitting in front of the console, take a look at the examples and predict
1469: what they will do, then try them out; if the outcome is not as expected,
1470: find out why (e.g., by trying out variations of the example), so you
1471: understand what's going on. There are also some assignments that you
1472: should solve.
1473:
1474: This tutorial assumes that you have programmed before and know what,
1475: e.g., a loop is.
1476:
1477: @c !! explain compat library
1478:
1479: @menu
1480: * Starting Gforth Tutorial::
1481: * Syntax Tutorial::
1482: * Crash Course Tutorial::
1483: * Stack Tutorial::
1484: * Arithmetics Tutorial::
1485: * Stack Manipulation Tutorial::
1486: * Using files for Forth code Tutorial::
1487: * Comments Tutorial::
1488: * Colon Definitions Tutorial::
1489: * Decompilation Tutorial::
1490: * Stack-Effect Comments Tutorial::
1491: * Types Tutorial::
1492: * Factoring Tutorial::
1493: * Designing the stack effect Tutorial::
1494: * Local Variables Tutorial::
1495: * Conditional execution Tutorial::
1496: * Flags and Comparisons Tutorial::
1497: * General Loops Tutorial::
1498: * Counted loops Tutorial::
1499: * Recursion Tutorial::
1500: * Leaving definitions or loops Tutorial::
1501: * Return Stack Tutorial::
1502: * Memory Tutorial::
1503: * Characters and Strings Tutorial::
1504: * Alignment Tutorial::
1505: * Interpretation and Compilation Semantics and Immediacy Tutorial::
1506: * Execution Tokens Tutorial::
1507: * Exceptions Tutorial::
1508: * Defining Words Tutorial::
1509: * Arrays and Records Tutorial::
1510: * POSTPONE Tutorial::
1511: * Literal Tutorial::
1512: * Advanced macros Tutorial::
1513: * Compilation Tokens Tutorial::
1514: * Wordlists and Search Order Tutorial::
1515: @end menu
1516:
1517: @node Starting Gforth Tutorial, Syntax Tutorial, Tutorial, Tutorial
1518: @section Starting Gforth
1519:
1520: You can start Gforth by typing its name:
1521:
1522: @example
1523: gforth
1524: @end example
1525:
1526: That puts you into interactive mode; you can leave Gforth by typing
1527: @code{bye}. While in Gforth, you can edit the command line and access
1528: the command line history with cursor keys, similar to bash.
1529:
1530:
1531: @node Syntax Tutorial, Crash Course Tutorial, Starting Gforth Tutorial, Tutorial
1532: @section Syntax
1533:
1534: A @dfn{word} is a sequence of arbitrary characters (expcept white
1535: space). Words are separated by white space. E.g., each of the
1536: following lines contains exactly one word:
1537:
1538: @example
1539: word
1540: !@@#$%^&*()
1541: 1234567890
1542: 5!a
1543: @end example
1544:
1545: A frequent beginner's error is to leave away necessary white space,
1546: resulting in an error like @samp{Undefined word}; so if you see such an
1547: error, check if you have put spaces wherever necessary.
1548:
1549: @example
1550: ." hello, world" \ correct
1551: ."hello, world" \ gives an "Undefined word" error
1552: @end example
1553:
1554: Gforth and most other Forth systems ignores differences in case (it is
1555: case-insensitive), i.e., @samp{word} is the same as @samp{Word}. If
1556: your system is case-sensitive, you may have to type all the examples
1557: given here in upper case.
1558:
1559:
1560: @node Crash Course Tutorial, Stack Tutorial, Syntax Tutorial, Tutorial
1561: @section Crash Course
1562:
1563: Type
1564:
1565: @example
1566: 0 0 !
1567: here execute
1568: ' catch >body 20 erase abort
1569: ' (quit) >body 20 erase
1570: @end example
1571:
1572: The last two examples are guaranteed to destroy parts of Gforth (and
1573: most other systems), so you better leave Gforth afterwards (if it has
1574: not finished by itself). On some systems you may have to kill gforth
1575: from outside (e.g., in Unix with @code{kill}).
1576:
1577: Now that you know how to produce crashes (and that there's not much to
1578: them), let's learn how to produce meaningful programs.
1579:
1580:
1581: @node Stack Tutorial, Arithmetics Tutorial, Crash Course Tutorial, Tutorial
1582: @section Stack
1583:
1584: The most obvious feature of Forth is the stack. When you type in a
1585: number, it is pushed on the stack. You can display the content of the
1586: stack with @code{.s}.
1587:
1588: @example
1589: 1 2 .s
1590: 3 .s
1591: @end example
1592:
1593: @code{.s} displays the top-of-stack to the right, i.e., the numbers
1594: appear in @code{.s} output as they appeared in the input.
1595:
1596: You can print the top of stack element with @code{.}.
1597:
1598: @example
1599: 1 2 3 . . .
1600: @end example
1601:
1602: In general, words consume their stack arguments (@code{.s} is an
1603: exception).
1604:
1605: @assignment
1606: What does the stack contain after @code{5 6 7 .}?
1607: @endassignment
1608:
1609:
1610: @node Arithmetics Tutorial, Stack Manipulation Tutorial, Stack Tutorial, Tutorial
1611: @section Arithmetics
1612:
1613: The words @code{+}, @code{-}, @code{*}, @code{/}, and @code{mod} always
1614: operate on the top two stack items:
1615:
1616: @example
1617: 2 2 + .
1618: 2 1 - .
1619: 7 3 mod .
1620: @end example
1621:
1622: The operands of @code{-}, @code{/}, and @code{mod} are in the same order
1623: as in the corresponding infix expression (this is generally the case in
1624: Forth).
1625:
1626: Parentheses are superfluous (and not available), because the order of
1627: the words unambiguously determines the order of evaluation and the
1628: operands:
1629:
1630: @example
1631: 3 4 + 5 * .
1632: 3 4 5 * + .
1633: @end example
1634:
1635: @assignment
1636: What are the infix expressions corresponding to the Forth code above?
1637: Write @code{6-7*8+9} in Forth notation@footnote{This notation is also
1638: known as Postfix or RPN (Reverse Polish Notation).}.
1639: @endassignment
1640:
1641: To change the sign, use @code{negate}:
1642:
1643: @example
1644: 2 negate .
1645: @end example
1646:
1647: @assignment
1648: Convert -(-3)*4-5 to Forth.
1649: @endassignment
1650:
1651: @code{/mod} performs both @code{/} and @code{mod}.
1652:
1653: @example
1654: 7 3 /mod . .
1655: @end example
1656:
1657: @node Stack Manipulation Tutorial, Using files for Forth code Tutorial, Arithmetics Tutorial, Tutorial
1658: @section Stack Manipulation
1659:
1660: Stack manipulation words rearrange the data on the stack.
1661:
1662: @example
1663: 1 .s drop .s
1664: 1 .s dup .s drop drop .s
1665: 1 2 .s over .s drop drop drop
1666: 1 2 .s swap .s drop drop
1667: 1 2 3 .s rot .s drop drop drop
1668: @end example
1669:
1670: These are the most important stack manipulation words. There are also
1671: variants that manipulate twice as many stack items:
1672:
1673: @example
1674: 1 2 3 4 .s 2swap .s 2drop 2drop
1675: @end example
1676:
1677: Two more stack manipulation words are:
1678:
1679: @example
1680: 1 2 .s nip .s drop
1681: 1 2 .s tuck .s 2drop drop
1682: @end example
1683:
1684: @assignment
1685: Replace @code{nip} and @code{tuck} with combinations of other stack
1686: manipulation words.
1687:
1688: @example
1689: Given: How do you get:
1690: 1 2 3 3 2 1
1691: 1 2 3 1 2 3 2
1692: 1 2 3 1 2 3 3
1693: 1 2 3 1 3 3
1694: 1 2 3 2 1 3
1695: 1 2 3 4 4 3 2 1
1696: 1 2 3 1 2 3 1 2 3
1697: 1 2 3 4 1 2 3 4 1 2
1698: 1 2 3
1699: 1 2 3 1 2 3 4
1700: 1 2 3 1 3
1701: @end example
1702: @endassignment
1703:
1704: @example
1705: 5 dup * .
1706: @end example
1707:
1708: @assignment
1709: Write 17^3 and 17^4 in Forth, without writing @code{17} more than once.
1710: Write a piece of Forth code that expects two numbers on the stack
1711: (@var{a} and @var{b}, with @var{b} on top) and computes
1712: @code{(a-b)(a+1)}.
1713: @endassignment
1714:
1715: @node Using files for Forth code Tutorial, Comments Tutorial, Stack Manipulation Tutorial, Tutorial
1716: @section Using files for Forth code
1717:
1718: While working at the Forth command line is convenient for one-line
1719: examples and short one-off code, you probably want to store your source
1720: code in files for convenient editing and persistence. You can use your
1721: favourite editor (Gforth includes Emacs support, @pxref{Emacs and
1722: Gforth}) to create @var{file} and use
1723:
1724: @example
1725: s" @var{file}" included
1726: @end example
1727:
1728: to load it into your Forth system. The file name extension I use for
1729: Forth files is @samp{.fs}.
1730:
1731: You can easily start Gforth with some files loaded like this:
1732:
1733: @example
1734: gforth @var{file1} @var{file2}
1735: @end example
1736:
1737: If an error occurs during loading these files, Gforth terminates,
1738: whereas an error during @code{INCLUDED} within Gforth usually gives you
1739: a Gforth command line. Starting the Forth system every time gives you a
1740: clean start every time, without interference from the results of earlier
1741: tries.
1742:
1743: I often put all the tests in a file, then load the code and run the
1744: tests with
1745:
1746: @example
1747: gforth @var{code} @var{tests} -e bye
1748: @end example
1749:
1750: (often by performing this command with @kbd{C-x C-e} in Emacs). The
1751: @code{-e bye} ensures that Gforth terminates afterwards so that I can
1752: restart this command without ado.
1753:
1754: The advantage of this approach is that the tests can be repeated easily
1755: every time the program ist changed, making it easy to catch bugs
1756: introduced by the change.
1757:
1758:
1759: @node Comments Tutorial, Colon Definitions Tutorial, Using files for Forth code Tutorial, Tutorial
1760: @section Comments
1761:
1762: @example
1763: \ That's a comment; it ends at the end of the line
1764: ( Another comment; it ends here: ) .s
1765: @end example
1766:
1767: @code{\} and @code{(} are ordinary Forth words and therefore have to be
1768: separated with white space from the following text.
1769:
1770: @example
1771: \This gives an "Undefined word" error
1772: @end example
1773:
1774: The first @code{)} ends a comment started with @code{(}, so you cannot
1775: nest @code{(}-comments; and you cannot comment out text containing a
1776: @code{)} with @code{( ... )}@footnote{therefore it's a good idea to
1777: avoid @code{)} in word names.}.
1778:
1779: I use @code{\}-comments for descriptive text and for commenting out code
1780: of one or more line; I use @code{(}-comments for describing the stack
1781: effect, the stack contents, or for commenting out sub-line pieces of
1782: code.
1783:
1784: The Emacs mode @file{gforth.el} (@pxref{Emacs and Gforth}) supports
1785: these uses by commenting out a region with @kbd{C-x \}, uncommenting a
1786: region with @kbd{C-u C-x \}, and filling a @code{\}-commented region
1787: with @kbd{M-q}.
1788:
1789:
1790: @node Colon Definitions Tutorial, Decompilation Tutorial, Comments Tutorial, Tutorial
1791: @section Colon Definitions
1792:
1793: are similar to procedures and functions in other programming languages.
1794:
1795: @example
1796: : squared ( n -- n^2 )
1797: dup * ;
1798: 5 squared .
1799: 7 squared .
1800: @end example
1801:
1802: @code{:} starts the colon definition; its name is @code{squared}. The
1803: following comment describes its stack effect. The words @code{dup *}
1804: are not executed, but compiled into the definition. @code{;} ends the
1805: colon definition.
1806:
1807: The newly-defined word can be used like any other word, including using
1808: it in other definitions:
1809:
1810: @example
1811: : cubed ( n -- n^3 )
1812: dup squared * ;
1813: -5 cubed .
1814: : fourth-power ( n -- n^4 )
1815: squared squared ;
1816: 3 fourth-power .
1817: @end example
1818:
1819: @assignment
1820: Write colon definitions for @code{nip}, @code{tuck}, @code{negate}, and
1821: @code{/mod} in terms of other Forth words, and check if they work (hint:
1822: test your tests on the originals first). Don't let the
1823: @samp{redefined}-Messages spook you, they are just warnings.
1824: @endassignment
1825:
1826:
1827: @node Decompilation Tutorial, Stack-Effect Comments Tutorial, Colon Definitions Tutorial, Tutorial
1828: @section Decompilation
1829:
1830: You can decompile colon definitions with @code{see}:
1831:
1832: @example
1833: see squared
1834: see cubed
1835: @end example
1836:
1837: In Gforth @code{see} shows you a reconstruction of the source code from
1838: the executable code. Informations that were present in the source, but
1839: not in the executable code, are lost (e.g., comments).
1840:
1841: @node Stack-Effect Comments Tutorial, Types Tutorial, Decompilation Tutorial, Tutorial
1842: @section Stack-Effect Comments
1843:
1844: By convention the comment after the name of a definition describes the
1845: stack effect: The part in from of the @samp{--} describes the state of
1846: the stack before the execution of the definition, i.e., the parameters
1847: that are passed into the colon definition; the part behind the @samp{--}
1848: is the state of the stack after the execution of the definition, i.e.,
1849: the results of the definition. The stack comment only shows the top
1850: stack items that the definition accesses and/or changes.
1851:
1852: You should put a correct stack effect on every definition, even if it is
1853: just @code{( -- )}. You should also add some descriptive comment to
1854: more complicated words (I usually do this in the lines following
1855: @code{:}). If you don't do this, your code becomes unreadable (because
1856: you have to work through every definition before you can undertsand
1857: any).
1858:
1859: @assignment
1860: The stack effect of @code{swap} can be written like this: @code{x1 x2 --
1861: x2 x1}. Describe the stack effect of @code{-}, @code{drop}, @code{dup},
1862: @code{over}, @code{rot}, @code{nip}, and @code{tuck}. Hint: When you
1863: are done, you can compare your stack effects to this in this manual
1864: (@pxref{Word Index}).
1865: @endassignment
1866:
1867: Sometimes programmers put comments at various places in colon
1868: definitions that describe the contents of the stack at that place (stack
1869: comments); i.e., they are like the first part of a stack-effect
1870: comment. E.g.,
1871:
1872: @example
1873: : cubed ( n -- n^3 )
1874: dup squared ( n n^2 ) * ;
1875: @end example
1876:
1877: In this case the stack comment is pretty superfluous, because the word
1878: is simple enough. If you think it would be a good idea to add such a
1879: comment to increase readability, you should also consider factoring the
1880: word into several simpler words (@pxref{Factoring Tutorial,,
1881: Factoring}), which typically eliminates the need for the stack effect;
1882: however, if you decide not to refactor it, then having such a comment is
1883: better than not having it.
1884:
1885: The names of the stack items in stack-effect and stack comments in the
1886: standard, in this manual, and in many programs specify the type through
1887: a type prefix, similar to Fortran and Hungarian notation. The most
1888: frequent prefixes are:
1889:
1890: @table @code
1891: @item n
1892: signed integer
1893: @item u
1894: unsigned integer
1895: @item c
1896: character
1897: @item f
1898: Boolean flags, i.e. @code{false} or @code{true}.
1899: @item a-addr,a-
1900: Cell-aligned address
1901: @item c-addr,c-
1902: Char-aligned address (note that a Char may have two bytes in Windows NT)
1903: @item xt
1904: Execution token, same size as Cell
1905: @item w,x
1906: Cell, can contain an integer or an address. It usually takes 32, 64 or
1907: 16 bits (depending on your platform and Forth system). A cell is more
1908: commonly known as machine word, but the term @emph{word} already means
1909: something different in Forth.
1910: @item d
1911: signed double-cell integer
1912: @item ud
1913: unsigned double-cell integer
1914: @item r
1915: Float (on the FP stack)
1916: @end table
1917:
1918: You can find a more complete list in @ref{Notation}.
1919:
1920: @assignment
1921: Write stack-effect comments for all definitions you have written up to
1922: now.
1923: @endassignment
1924:
1925:
1926: @node Types Tutorial, Factoring Tutorial, Stack-Effect Comments Tutorial, Tutorial
1927: @section Types
1928:
1929: In Forth the names of the operations are not overloaded; so similar
1930: operations on different types need different names; e.g., @code{+} adds
1931: integers, and you have to use @code{f+} to add floating-point numbers.
1932: The following prefixes are often used for related operations on
1933: different types:
1934:
1935: @table @code
1936: @item (none)
1937: signed integer
1938: @item u
1939: unsigned integer
1940: @item c
1941: character
1942: @item d
1943: signed double-cell integer
1944: @item ud, du
1945: unsigned double-cell integer
1946: @item 2
1947: two cells (not-necessarily double-cell numbers)
1948: @item m, um
1949: mixed single-cell and double-cell operations
1950: @item f
1951: floating-point (note that in stack comments @samp{f} represents flags,
1952: and @samp{r} represents FP number).
1953: @end table
1954:
1955: If there are no differences between the signed and the unsigned variant
1956: (e.g., for @code{+}), there is only the prefix-less variant.
1957:
1958: Forth does not perform type checking, neither at compile time, nor at
1959: run time. If you use the wrong oeration, the data are interpreted
1960: incorrectly:
1961:
1962: @example
1963: -1 u.
1964: @end example
1965:
1966: If you have only experience with type-checked languages until now, and
1967: have heard how important type-checking is, don't panic! In my
1968: experience (and that of other Forthers), type errors in Forth code are
1969: usually easy to find (once you get used to it), the increased vigilance
1970: of the programmer tends to catch some harder errors in addition to most
1971: type errors, and you never have to work around the type system, so in
1972: most situations the lack of type-checking seems to be a win (projects to
1973: add type checking to Forth have not caught on).
1974:
1975:
1976: @node Factoring Tutorial, Designing the stack effect Tutorial, Types Tutorial, Tutorial
1977: @section Factoring
1978:
1979: If you try to write longer definitions, you will soon find it hard to
1980: keep track of the stack contents. Therefore, good Forth programmers
1981: tend to write only short definitions (e.g., three lines). The art of
1982: finding meaningful short definitions is known as factoring (as in
1983: factoring polynomials).
1984:
1985: Well-factored programs offer additional advantages: smaller, more
1986: general words, are easier to test and debug and can be reused more and
1987: better than larger, specialized words.
1988:
1989: So, if you run into difficulties with stack management, when writing
1990: code, try to define meaningful factors for the word, and define the word
1991: in terms of those. Even if a factor contains only two words, it is
1992: often helpful.
1993:
1994: Good factoring is not easy, and even experienced Forth programmers often
1995: don't find the right solution right away, but only when rewriting the
1996: program. So, if you don't come up with a good solution immediately,
1997: keep trying, don't despair.
1998:
1999: @c example !!
2000:
2001:
2002: @node Designing the stack effect Tutorial, Local Variables Tutorial, Factoring Tutorial, Tutorial
2003: @section Designing the stack effect
2004:
2005: In other languages you can use an arbitrary order of parameters for a
2006: function; and since ther is only one result, you don't have to deal with
2007: the order of results, either.
2008:
2009: In Forth (and other stack-based languages, e.g., Postscript) the
2010: parameter and result order of a definition is important and should be
2011: designed well. The general guideline is to design the stack effect such
2012: that the word is simple to use in most cases, even if that complicates
2013: the implementation of the word. Some concrete rules are:
2014:
2015: @itemize @bullet
2016:
2017: @item
2018: Words consume all of their parameters (e.g., @code{.}).
2019:
2020: @item
2021: If there is a convention on the order of parameters (e.g., from
2022: mathematics or another programming language), stick with it (e.g.,
2023: @code{-}).
2024:
2025: @item
2026: If one parameter usually requires only a short computation (e.g., it is
2027: a constant), pass it on the top of the stack. Conversely, parameters
2028: that usually require a long sequence of code to compute should be passed
2029: as the bottom (i.e., first) parameter. This makes the code easier to
2030: read, because reader does not need to keep track of the bottom item
2031: through a long sequence of code (or, alternatively, through stack
2032: manipulations). E.g., @code{!} (store, see @pxref{Memory}) expects the
2033: address on top of the stack because it is usually simpler to compute
2034: than the stored value (often the address is just a variable).
2035:
2036: @item
2037: Similarly, results that are usually consumed quickly should be returned
2038: on the top of stack, whereas a result that is often used in long
2039: computations should be passed as bottom result. E.g., the file words
2040: like @code{open-file} return the error code on the top of stack, because
2041: it is usually consumed quickly by @code{throw}; moreover, the error code
2042: has to be checked before doing anything with the other results.
2043:
2044: @end itemize
2045:
2046: These rules are just general guidelines, don't lose sight of the overall
2047: goal to make the words easy to use. E.g., if the convention rule
2048: conflicts with the computation-length rule, you might decide in favour
2049: of the convention if the word will be used rarely, and in favour of the
2050: computation-length rule if the word will be used frequently (because
2051: with frequent use the cost of breaking the computation-length rule would
2052: be quite high, and frequent use makes it easier to remember an
2053: unconventional order).
2054:
2055: @c example !! structure package
2056:
2057: @node Local Variables Tutorial, Conditional execution Tutorial, Designing the stack effect Tutorial, Tutorial
2058: @section Local Variables
2059:
2060: You can define local variables (@emph{locals}) in a colon definition:
2061:
2062: @example
2063: : swap @{ a b -- b a @}
2064: b a ;
2065: 1 2 swap .s 2drop
2066: @end example
2067:
2068: (If your Forth system does not support this syntax, include
2069: @file{compat/anslocals.fs} first).
2070:
2071: In this example @code{@{ a b -- b a @}} is the locals definition; it
2072: takes two cells from the stack, puts the top of stack in @code{b} and
2073: the next stack element in @code{a}. @code{--} starts a comment ending
2074: with @code{@}}. After the locals definition, using the name of the
2075: local will push its value on the stack. You can leave the comment
2076: part (@code{-- b a}) away:
2077:
2078: @example
2079: : swap ( x1 x2 -- x2 x1 )
2080: @{ a b @} b a ;
2081: @end example
2082:
2083: In Gforth you can have several locals definitions, anywhere in a colon
2084: definition; in contrast, in a standard program you can have only one
2085: locals definition per colon definition, and that locals definition must
2086: be outside any controll structure.
2087:
2088: With locals you can write slightly longer definitions without running
2089: into stack trouble. However, I recommend trying to write colon
2090: definitions without locals for exercise purposes to help you gain the
2091: essential factoring skills.
2092:
2093: @assignment
2094: Rewrite your definitions until now with locals
2095: @endassignment
2096:
2097:
2098: @node Conditional execution Tutorial, Flags and Comparisons Tutorial, Local Variables Tutorial, Tutorial
2099: @section Conditional execution
2100:
2101: In Forth you can use control structures only inside colon definitions.
2102: An @code{if}-structure looks like this:
2103:
2104: @example
2105: : abs ( n1 -- +n2 )
2106: dup 0 < if
2107: negate
2108: endif ;
2109: 5 abs .
2110: -5 abs .
2111: @end example
2112:
2113: @code{if} takes a flag from the stack. If the flag is non-zero (true),
2114: the following code is performed, otherwise execution continues after the
2115: @code{endif} (or @code{else}). @code{<} combares the top two stack
2116: elements and prioduces a flag:
2117:
2118: @example
2119: 1 2 < .
2120: 2 1 < .
2121: 1 1 < .
2122: @end example
2123:
2124: Actually the standard name for @code{endif} is @code{then}. This
2125: tutorial presents the examples using @code{endif}, because this is often
2126: less confusing for people familiar with other programming languages
2127: where @code{then} has a different meaning. If your system does not have
2128: @code{endif}, define it with
2129:
2130: @example
2131: : endif postpone then ; immediate
2132: @end example
2133:
2134: You can optionally use an @code{else}-part:
2135:
2136: @example
2137: : min ( n1 n2 -- n )
2138: 2dup < if
2139: drop
2140: else
2141: nip
2142: endif ;
2143: 2 3 min .
2144: 3 2 min .
2145: @end example
2146:
2147: @assignment
2148: Write @code{min} without @code{else}-part (hint: what's the definition
2149: of @code{nip}?).
2150: @endassignment
2151:
2152:
2153: @node Flags and Comparisons Tutorial, General Loops Tutorial, Conditional execution Tutorial, Tutorial
2154: @section Flags and Comparisons
2155:
2156: In a false-flag all bits are clear (0 when interpreted as integer). In
2157: a canonical true-flag all bits are set (-1 as a twos-complement signed
2158: integer); in many contexts (e.g., @code{if}) any non-zero value is
2159: treated as true flag.
2160:
2161: @example
2162: false .
2163: true .
2164: true hex u. decimal
2165: @end example
2166:
2167: Comparison words produce canonical flags:
2168:
2169: @example
2170: 1 1 = .
2171: 1 0= .
2172: 0 1 < .
2173: 0 0 < .
2174: -1 1 u< . \ type error, u< interprets -1 as large unsigned number
2175: -1 1 < .
2176: @end example
2177:
2178: Gforth supports all combinations of the prefixes @code{0 u d d0 du} (or
2179: none) and the comparisons @code{= <> < > <= >=}. Only a part of these
2180: combinations are standard (see the standard or !! the glossary for
2181: details).
2182:
2183: You can use @code{and or xor invert} can be used as operations on
2184: canonical flags. Actually they are bitwise operations:
2185:
2186: @example
2187: 1 2 and .
2188: 1 2 or .
2189: 1 3 xor .
2190: 1 invert .
2191: @end example
2192:
2193: You can convert a zero/non-zero flag into a canonical flag with
2194: @code{0<>} (and complement it on the way with @code{0=}).
2195:
2196: @example
2197: 1 0= .
2198: 1 0<> .
2199: @end example
2200:
2201: You can use the all-bits-set feature of canonicasl flags and the bitwise
2202: operation of the Boolean operations to avoid @code{if}s:
2203:
2204: @example
2205: : foo ( n1 -- n2 )
2206: 0= if
2207: 14
2208: else
2209: 0
2210: endif ;
2211: 0 foo .
2212: 1 foo .
2213:
2214: : foo ( n1 -- n2 )
2215: 0= 14 and ;
2216: 0 foo .
2217: 1 foo .
2218: @end example
2219:
2220: @assignment
2221: Write @code{min} without @code{if}.
2222: @endassignment
2223:
2224:
2225: @node General Loops Tutorial, Counted loops Tutorial, Flags and Comparisons Tutorial, Tutorial
2226: @section General Loops
2227:
2228: The endless loop is the most simple one:
2229:
2230: @example
2231: : endless ( -- )
2232: 0 begin
2233: dup . 1+
2234: again ;
2235: endless
2236: @end example
2237:
2238: Terminate this loop by pressing @kbd{Ctrl-C} (in Gforth). @code{begin}
2239: does nothing at run-time, @code{again} jumps back to @code{begin}.
2240:
2241: A loop with one exit at any place looks like this:
2242:
2243: @example
2244: : log2 ( +n1 -- n2 )
2245: \ logarithmus dualis of n1>0, rounded down to the next integer
2246: assert( dup 0> )
2247: 2/ 0 begin
2248: over 0> while
2249: 1+ swap 2/ swap
2250: repeat
2251: nip ;
2252: 7 log2 .
2253: 8 log2 .
2254: @end example
2255:
2256: At run-time @code{while} consumes a flag; if it is 0, execution
2257: continues behind the @code{repeat}; if the falg is non-zero, execution
2258: continues behind the @code{while}. @code{Repeat} jumps back to
2259: @code{begin}, just like @code{again}.
2260:
2261: In Forth there are many combinations/abbreviations, like @code{1+}.
2262: However, @code{2/} is not one of them; it shifts it's argument right by
2263: one bit (arithmetic shift right):
2264:
2265: @example
2266: -5 2 / .
2267: -5 2/ .
2268: @end example
2269:
2270: @code{assert(} is no standard word, but you can get it on systems other
2271: then Gforth by including @file{compat/assert.fs}. You can see what it
2272: does by trying
2273:
2274: @example
2275: 0 log2 .
2276: @end example
2277:
2278: Here's a loop with an exit at the end:
2279:
2280: @example
2281: : log2 ( +n1 -- n2 )
2282: \ logarithmus dualis of n1>0, rounded down to the next integer
2283: assert( dup 0 > )
2284: -1 begin
2285: 1+ swap 2/ swap
2286: over 0 <=
2287: until
2288: nip ;
2289: @end example
2290:
2291: @code{Until} consumes a flag; if it is non-zero, execution continues at
2292: the @code{begin}, otherwise after the @code{until}.
2293:
2294: @assignment
2295: Write a definition for computing the greatest common divisor.
2296: @endassignment
2297:
2298:
2299: @node Counted loops Tutorial, Recursion Tutorial, General Loops Tutorial, Tutorial
2300: @section Counted loops
2301:
2302: @example
2303: : ^ ( n1 u -- n )
2304: \ n = the uth power of u1
2305: 1 swap 0 u+do
2306: over *
2307: loop
2308: nip ;
2309: 3 2 ^ .
2310: 4 3 ^ .
2311: @end example
2312:
2313: @code{U+do} (from @file{compat/loops.fs}, if your Forth system doesn't
2314: have it) takes two numbers of the stack @code{( u3 u4 -- )}, and then
2315: performs the code between @code{u+do} and @code{loop} for @code{u3-u4}
2316: times (or not at all, if @code{u3-u4<0}).
2317:
2318: You can see the stack effect design rules at work in the stack effect of
2319: the loop start words: Since the start value of the loop is more
2320: frequently constant than the end value, the start value is passed on
2321: the top-of-stack.
2322:
2323: You can access the counter of a counted loop with @code{i}:
2324:
2325: @example
2326: : fac ( u -- u! )
2327: 1 swap 1+ 1 u+do
2328: i *
2329: loop ;
2330: 5 fac .
2331: 7 fac .
2332: @end example
2333:
2334: There is also @code{+do}, which expects signed numbers (important for
2335: deciding whether to enter the loop).
2336:
2337: @assignment
2338: Write a definition for computing the nth Fibonacci number.
2339: @endassignment
2340:
2341: !! +DO...+LOOP
2342: !! -DO...-LOOP
2343:
2344:
2345: @node Recursion Tutorial, Leaving definitions or loops Tutorial, Counted loops Tutorial, Tutorial
2346: @section Recursion
2347:
2348: Usually the name of a definition is not visible in the definition; but
2349: earlier definitions are usually visible:
2350:
2351: @example
2352: 1 0 / . \ "Floating-point unidentified fault" in Gforth on most platforms
2353: : / ( n1 n2 -- n )
2354: dup 0= if
2355: -10 throw \ report division by zero
2356: endif
2357: / \ old version
2358: ;
2359: 1 0 /
2360: @end example
2361:
2362: For recursive definitions you can use @code{recursive} (non-standard) or
2363: @code{recurse}:
2364:
2365: @example
2366: : fac1 ( n -- n! ) recursive
2367: dup 0> if
2368: dup 1- fac1 *
2369: else
2370: drop 1
2371: endif ;
2372: 7 fac1 .
2373:
2374: : fac2 ( n -- n! )
2375: dup 0> if
2376: dup 1- recurse *
2377: else
2378: drop 1
2379: endif ;
2380: 8 fac2 .
2381: @end example
2382:
2383: @assignment
2384: Write a recursive definition for computing the nth Fibonacci number.
2385: @endassignment
2386:
2387:
2388: @node Leaving definitions or loops Tutorial, Return Stack Tutorial, Recursion Tutorial, Tutorial
2389: @section Leaving definitions or loops
2390:
2391: @code{EXIT} exits the current definition right away. For every counted
2392: loop that is left in this way, an @code{UNLOOP} has to be performed
2393: before the @code{EXIT}:
2394:
2395: @c !! real examples
2396: @example
2397: : ...
2398: ... u+do
2399: ... if
2400: ... unloop exit
2401: endif
2402: ...
2403: loop
2404: ... ;
2405: @end example
2406:
2407: @code{LEAVE} leaves the innermost counted loop right away:
2408:
2409: @example
2410: : ...
2411: ... u+do
2412: ... if
2413: ... leave
2414: endif
2415: ...
2416: loop
2417: ... ;
2418: @end example
2419:
2420:
2421: @node Return Stack Tutorial, Memory Tutorial, Leaving definitions or loops Tutorial, Tutorial
2422: @section Return Stack
2423:
2424: In addition to the data stack Forth also has a second stack, the return
2425: stack; most Forth systems store the return addresses of procedure calls
2426: there (thus its name). Programmers can also use this stack:
2427:
2428: @example
2429: : foo ( n1 n2 -- )
2430: .s
2431: >r .s
2432: r@ .
2433: >r .s
2434: r@ .
2435: r> .
2436: r@ .
2437: r> . ;
2438: 1 2 foo
2439: @end example
2440:
2441: @code{>r} takes an element from the data stack and pushes it onto the
2442: return stack; conversely, @code{r>} moves an elementm from the return to
2443: the data stack; @code{r@@} pushes a copy of the top of the return stack
2444: on the return stack.
2445:
2446: Forth programmers usually use the return stack for storing data
2447: temporarily, if using the data stack alone would be too complex, and
2448: factoring and locals are not an option:
2449:
2450: @example
2451: : 2swap ( x1 x2 x3 x4 -- x3 x4 x1 x2 )
2452: rot >r rot r> ;
2453: @end example
2454:
2455: The return address of the definition and the loop control parameters of
2456: counted loops usually reside on the return stack, so you have to take
2457: all items, that you have pushed on the return stack in a colon
2458: definition or counted loop, from the return stack before the definition
2459: or loop ends. You cannot access items that you pushed on the return
2460: stack outside some definition or loop within the definition of loop.
2461:
2462: If you miscount the return stack items, this usually ends in a crash:
2463:
2464: @example
2465: : crash ( n -- )
2466: >r ;
2467: 5 crash
2468: @end example
2469:
2470: You cannot mix using locals and using the return stack (according to the
2471: standard; Gforth has no problem). However, they solve the same
2472: problems, so this shouldn't be an issue.
2473:
2474: @assignment
2475: Can you rewrite any of the definitions you wrote until now in a better
2476: way using the return stack?
2477: @endassignment
2478:
2479:
2480: @node Memory Tutorial, Characters and Strings Tutorial, Return Stack Tutorial, Tutorial
2481: @section Memory
2482:
2483: You can create a global variable @code{v} with
2484:
2485: @example
2486: variable v ( -- addr )
2487: @end example
2488:
2489: @code{v} pushes the address of a cell in memory on the stack. This cell
2490: was reserved by @code{variable}. You can use @code{!} (store) to store
2491: values into this cell and @code{@@} (fetch) to load the value from the
2492: stack into memory:
2493:
2494: @example
2495: v .
2496: 5 v ! .s
2497: v @ .
2498: @end example
2499:
2500: You can also reserve more memory:
2501:
2502: @example
2503: create v2 20 cells allot
2504: @end example
2505:
2506: creates a word @code{v2} and reserves 20 cells; the address pushed by
2507: @code{v2} points to the start of these 20 cells. You can use address
2508: arithmetic to access these cells:
2509:
2510: @example
2511: 3 v2 5 cells + !
2512: @end example
2513:
2514: You can reserve and initialize memory with @code{,}:
2515:
2516: @example
2517: create v3
2518: 5 , 4 , 3 , 2 , 1 ,
2519: v3 @ .
2520: v3 cell+ @ .
2521: v3 2 cells + @ .
2522: @end example
2523:
2524: @assignment
2525: Write a definition @code{vsum ( addr u -- n )} that computes the sum of
2526: @code{u} cells, with the first of these cells at @code{addr}, the next
2527: one at @code{addr cell+} etc.
2528: @endassignment
2529:
2530: You can also reserve memory without creating a new word:
2531:
2532: @example
2533: here 10 cells allot
2534: .s
2535: @end example
2536:
2537: @code{Here} pushes the start address of the memory area. You should
2538: store it somewhere, or you will have a hard time finding the memory area
2539: again.
2540:
2541: @code{Allot} manages dictionary memory. The dictionary memory contains
2542: the system's data structures for words etc. on Gforth and most other
2543: Forth systems. It is managed like a stack: You can free the memory that
2544: you have just @code{allot}ed with
2545:
2546: @example
2547: -10 cells allot
2548: @end example
2549:
2550: Note that you cannot do this if you have created a new word in the
2551: meantime (because then your @code{allot}ed memory is no longer on the
2552: top of the dictionary ``stack'').
2553:
2554: Alternatively, you can use @code{allocate} and @code{free} which allow
2555: freeing memory in any order:
2556:
2557: @example
2558: 10 cells allocate throw .s
2559: 20 cells allocate throw .s
2560: swap
2561: free throw
2562: free throw
2563: @end example
2564:
2565: The @code{throw}s deal with errors (e.g., out of memory).
2566:
2567: And there is also a garbage collector @url{!!}, which eliminates the
2568: need to @code{free} memory explicitly.
2569:
2570:
2571: @node Characters and Strings Tutorial, Alignment Tutorial, Memory Tutorial, Tutorial
2572: @section Characters and Strings
2573:
2574: On the stack characters take up a cell, like numbers. In memory they
2575: have their own size (one 8-bit byte on most systems), and therefore
2576: require their own words for memory access:
2577:
2578: @example
2579: create v4
2580: 104 c, 97 c, 108 c, 108 c, 111 c,
2581: v4 4 chars + c@ .
2582: @end example
2583:
2584: The preferred representation of strings on the stack is @code{addr
2585: u-count}, where @code{addr} is the address of the first character and
2586: @code{u-count} is the number of characters in the string.
2587:
2588: @example
2589: v4 5 type
2590: @end example
2591:
2592: You get a string constant with
2593:
2594: @example
2595: s" hello, world" .s
2596: type
2597: @end example
2598:
2599: Make sure you have a space between @code{s"} and the string; @code{s"}
2600: is a normal Forth word and must be delimited with white space (try what
2601: happens when you remove the space).
2602:
2603: However, this interpretive use of @code{s"} is quite restricted: the
2604: string exists only until the next call of @code{s"} (some Forth systems
2605: keep more than one of these strings, but usually they still have a
2606: limited lifetime.
2607:
2608: @example
2609: s" hello," s" world" .s
2610: type
2611: type
2612: @end example
2613:
2614: However, you can also use @code{s"} in a definition, and the resulting
2615: strings then live forever (well, as long as the definition):
2616:
2617: @example
2618: : foo s" hello," s" world" ;
2619: foo .s
2620: type
2621: type
2622: @end example
2623:
2624: @assignment
2625: @code{Emit ( c -- )} types @code{c} as character (not a number).
2626: Implement @code{type ( addr u -- )}.
2627: @endassignment
2628:
2629: @node Alignment Tutorial, Interpretation and Compilation Semantics and Immediacy Tutorial, Characters and Strings Tutorial, Tutorial
2630: @section Alignment
2631:
2632: On many processors cells have to be aligned in memory, if you want to
2633: access them with @code{@@} and @code{!} (and even if the processor does
2634: not require alignment, access to aligned cells are faster).
2635:
2636: @code{Create} aligns @code{here} (i.e., the place where the next
2637: allocation will occur, and that the @code{create}d word points to).
2638: Likewise, the memory produced by @code{allocate} starts at an aligned
2639: address. Adding a number of @code{cells} to an aligned address produces
2640: another aligned address.
2641:
2642: However, address arithmetic involving @code{char+} and @code{chars} can
2643: create an address that is not cell-aligned. @code{Aligned ( addr --
2644: a-addr )} produces the next aligned address:
2645:
2646: @example
2647: v3 char+ aligned .s @ .
2648: v3 char+ .s @ .
2649: @end example
2650:
2651: Similarly, @code{align} advances @code{here} to the next aligned
2652: address:
2653:
2654: @example
2655: create v5 97 c,
2656: here .
2657: align here .
2658: 1000 ,
2659: @end example
2660:
2661: Note that you should use aligned addresses even if your processor does
2662: not require them, if you want your program to be portable.
2663:
2664:
2665: @node Interpretation and Compilation Semantics and Immediacy Tutorial, Execution Tokens Tutorial, Alignment Tutorial, Tutorial
2666: @section Interpretation and Compilation Semantics and Immediacy
2667:
2668: When a word is compiled, it behaves differently from being interpreted.
2669: E.g., consider @code{+}:
2670:
2671: @example
2672: 1 2 + .
2673: : foo + ;
2674: @end example
2675:
2676: These two behaviours are known as compilation and interpretation
2677: semantics. For normal words (e.g., @code{+}), the compilation semantics
2678: is to append the interpretation semantics to the currently defined word
2679: (@code{foo} in the example above). I.e., when @code{foo} is executed
2680: later, the interpretation semantics of @code{+} (i.e., adding two
2681: numbers) will be performed.
2682:
2683: However, there are words with non-default compilation semantics, e.g.,
2684: the control-flow words like @code{if}. You can use @code{immediate} to
2685: change the compilation semantics of the last defined word to be equal to
2686: the interpretation semantics:
2687:
2688: @example
2689: : [FOO] ( -- )
2690: 5 . ; immediate
2691:
2692: [FOO]
2693: : bar ( -- )
2694: [FOO] ;
2695: bar
2696: see bar
2697: @end example
2698:
2699: Two conventions to mark words with non-default compilation semnatics are
2700: names with brackets (more frequently used) and to write them all in
2701: upper case (less frequently used).
2702:
2703: In Gforth (and many other systems) you can also remove the
2704: interpretation semantics with @code{compile-only} (the compilation
2705: semantics is derived from the original interpretation semantics):
2706:
2707: @example
2708: : flip ( -- )
2709: 6 . ; compile-only \ but not immediate
2710: flip
2711:
2712: : flop ( -- )
2713: flip ;
2714: flop
2715: @end example
2716:
2717: In this example the interpretation semantics of @code{flop} is equal to
2718: the original interpretation semantics of @code{flip}.
2719:
2720: The text interpreter has two states: in interpret state, it performs the
2721: interpretation semantics of words it encounters; in compile state, it
2722: performs the compilation semantics of these words.
2723:
2724: Among other things, @code{:} switches into compile state, and @code{;}
2725: switches back to interpret state. They contain the factors @code{]}
2726: (switch to compile state) and @code{[} (switch to interpret state), that
2727: do nothing but switch the state.
2728:
2729: @example
2730: : xxx ( -- )
2731: [ 5 . ]
2732: ;
2733:
2734: xxx
2735: see xxx
2736: @end example
2737:
2738: These brackets are also the source of the naming convention mentioned
2739: above.
2740:
2741:
2742: @node Execution Tokens Tutorial, Exceptions Tutorial, Interpretation and Compilation Semantics and Immediacy Tutorial, Tutorial
2743: @section Execution Tokens
2744:
2745: @code{' word} gives you the execution token (XT) of a word. The XT is a
2746: cell representing the interpretation semantics of a word. You can
2747: execute this semantics with @code{execute}:
2748:
2749: @example
2750: ' + .s
2751: 1 2 rot execute .
2752: @end example
2753:
2754: The XT is similar to a function pointer in C. However, parameter
2755: passing through the stack makes it a little more flexible:
2756:
2757: @example
2758: : map-array ( ... addr u xt -- ... )
2759: \ führt xt ( ... x -- ... ) für jedes Element des Arrays aus,
2760: \ das bei addr beginnt und u Elemente enthält
2761: @{ xt @}
2762: cells over + swap ?do
2763: i @ xt execute
2764: 1 cells +loop ;
2765:
2766: create a 3 , 4 , 2 , -1 , 4 ,
2767: a 5 ' . map-array .s
2768: 0 a 5 ' + map-array .
2769: s" max-n" environment? drop .s
2770: a 5 ' min map-array .
2771: @end example
2772:
2773: You can use map-array with the XTs of words that consume one element
2774: more than they produce. In theory you can also use it with other XTs,
2775: but the stack effect then depends on the size of the array, which is
2776: hard to understand.
2777:
2778: Since arrays are cell-sized, you can store them in memory and manipulate
2779: them on the stack like other cells. You can also compile the xt into a
2780: word with @code{compile,}:
2781:
2782: @example
2783: : foo1 ( n1 n2 -- n )
2784: [ ' + compile, ] ;
2785: see foo
2786: @end example
2787:
2788: This is non-standard, because @code{compile,} has no compilation
2789: semantics in the standard, but it works in good Forth systems. For the
2790: broken ones, use
2791:
2792: @example
2793: : [compile,] compile, ; immediate
2794:
2795: : foo1 ( n1 n2 -- n )
2796: [ ' + ] [compile,] ;
2797: see foo
2798: @end example
2799:
2800: @code{'} is a word with default compilation semantics; it parses the
2801: next word when its interpretation semantics are executed, not during
2802: compilation:
2803:
2804: @example
2805: : foo ( -- xt )
2806: ' ;
2807: see foo
2808: : bar ( ... "word" -- ... )
2809: ' execute ;
2810: see bar
2811: 1 2 bar +
2812: @end example
2813:
2814: You often want to parse a word during compilation and compile its XT so
2815: it will be pushed on the stack at run-time. @code{[']} does this:
2816:
2817: @example
2818: : xt-+ ( -- xt )
2819: ['] + ;
2820: see xt-+
2821: 1 2 xt-+ execute .
2822: @end example
2823:
2824: Many programmers tend to see @code{'} and the word it parses as one
2825: unit, and expect it to behave like @code{[']} when compiled, and are
2826: confused by the actual behaviour. If you are, just remember that the
2827: Forth system just takes @code{'} as one unit and has no idea that it is
2828: a parsing word (attempts to convenience programmers in this issue have
2829: usually resulted in even worse pitfalls, see
2830: @uref{http://www.complang.tuwien.ac.at/papers/ertl98.ps.gz}).
2831:
2832: Note that the state of the interpreter does not come into play when
2833: creating and executing xts. I.e., even when you execute @code{'} in
2834: compile state, it still gives you the interpretation semantics. And
2835: whatever that state is, @code{execute} performs the semantics
2836: represented by the xt (i.e., the interpretation semantics).
2837:
2838:
2839: @node Exceptions Tutorial, Defining Words Tutorial, Execution Tokens Tutorial, Tutorial
2840: @section Exceptions
2841:
2842: @code{throw ( n -- )} causes an exception unless n is zero.
2843:
2844: @example
2845: 100 throw .s
2846: 0 throw .s
2847: @end example
2848:
2849: @code{catch ( ... xt -- ... n )} behaves similar to @code{execute}, but
2850: it catches exceptions and pushes the number of the exception on the
2851: stack (or 0, if the xt executed without exception). If there was an
2852: exception, the stacks have the same depth as when entering @code{catch}:
2853:
2854: @example
2855: .s
2856: 3 0 ' / catch .s
2857: 3 2 ' / catch .s
2858: @end example
2859:
2860: @assignment
2861: Try the same with @code{execute} instead of @code{catch}.
2862: @endassignment
2863:
2864: @code{Throw} always jumps to the dynamically next enclosing
2865: @code{catch}, even if it has to leave several call levels to achieve
2866: this:
2867:
2868: @example
2869: : foo 100 throw ;
2870: : foo1 foo ." after foo" ;
2871: : bar ' foo1 catch ;
2872: bar
2873: @end example
2874:
2875: It is often important to restore a value upon leaving a definition, even
2876: if the definition is left through an exception. You can ensure this
2877: like this:
2878:
2879: @example
2880: : ...
2881: save-x
2882: ' word-changing-x catch ( ... n )
2883: restore-x
2884: ( ... n ) throw ;
2885: @end example
2886:
2887: Gforth provides an alternative syntax in addition to @code{cacth}:
2888: @code{try ... recover ... endtry}. If the code between @code{try} and
2889: @code{recover} has an exception, the stack depths are restored, the
2890: exception number is pushed on the stack, and the code between
2891: @code{recover} and @code{endtry} is performed. E.g., the definition for
2892: @code{catch} is
2893:
2894: @example
2895: : catch ( x1 .. xn xt -- y1 .. ym 0 / z1 .. zn error ) \ exception
2896: try
2897: execute 0
2898: recover
2899: nip
2900: endtry ;
2901: @end example
2902:
2903: The equivalent to the restoration code above is
2904:
2905: @example
2906: : ...
2907: save-x
2908: try
2909: word-changing-x
2910: end-try
2911: restore-x
2912: throw ;
2913: @end example
2914:
2915: As you can see, the @code{recover} part is optional.
2916:
2917:
2918: @node Defining Words Tutorial, Arrays and Records Tutorial, Exceptions Tutorial, Tutorial
2919: @section Defining Words
2920:
2921: @code{:}, @code{create}, and @code{variable} are definition words: They
2922: define other words. @code{Constant} is another definition word:
2923:
2924: @example
2925: 5 constant foo
2926: foo .
2927: @end example
2928:
2929: You can also use the prefixes @code{2} (double-cell) and @code{f}
2930: (floating point) with @code{variable} and @code{constant}.
2931:
2932: You can also define your own defining words. E.g.:
2933:
2934: @example
2935: : variable ( "name" -- )
2936: create 0 , ;
2937: @end example
2938:
2939: You can also define defining words that create words that do something
2940: other than just producing their address:
2941:
2942: @example
2943: : constant ( n "name" -- )
2944: create ,
2945: does> ( -- n )
2946: ( addr ) @ ;
2947:
2948: 5 constant foo
2949: foo .
2950: @end example
2951:
2952: The definition of @code{constant} above ends at the @code{does>}; i.e.,
2953: @code{does>} replaces @code{;}, but it also does something else: It
2954: changes the last defined word such that it pushes the address of the
2955: body of the word and then performs the code after the @code{does>}
2956: whenever it is called.
2957:
2958: In the example above, @code{constant} uses @code{,} to store 5 into the
2959: body of @code{foo}. When @code{foo} executes, it pushes the address of
2960: the body onto the stack, then (in the code after the @code{does>})
2961: fetches the 5 from there.
2962:
2963: The stack comment near the @code{does>} reflects the stack effect of the
2964: defined word, not the stack effect of the code after the @code{does>}
2965: (the difference is that the code expects the address of the body that
2966: the stack comment does not show).
2967:
2968: You can use these definition words to do factoring in cases that involve
2969: (other) definition words. E.g., a field offset is always added to an
2970: address. Instead of defining
2971:
2972: @example
2973: 2 cells constant offset-field1
2974: @end example
2975:
2976: and using this like
2977:
2978: @example
2979: ( addr ) offset-field1 +
2980: @end example
2981:
2982: you can define a definition word
2983:
2984: @example
2985: : simple-field ( n "name" -- )
2986: create ,
2987: does> ( n1 -- n1+n )
2988: ( addr ) @ + ;
2989: @end example
2990:
2991: Definition and use of field offsets now look like this:
2992:
2993: @example
2994: 2 cells simple-field field1
2995: ( addr ) field1
2996: @end example
2997:
2998: If you want to do something with the word without performing the code
2999: after the @code{does>}, you can access the body of a @code{create}d word
3000: with @code{>body ( xt -- addr )}:
3001:
3002: @example
3003: : value ( n "name" -- )
3004: create ,
3005: does> ( -- n1 )
3006: @ ;
3007: : to ( n "name" -- )
3008: ' >body ! ;
3009:
3010: 5 value foo
3011: foo .
3012: 7 to foo
3013: foo .
3014: @end example
3015:
3016: @assignment
3017: Define @code{defer ( "name" -- )}, which creates a word that stores an
3018: XT (at the start the XT of @code{abort}), and upon execution
3019: @code{execute}s the XT. Define @code{is ( xt "name" -- )} that stores
3020: @code{xt} into @code{name}, a word defined with @code{defer}. Indirect
3021: recursion is one application of @code{defer}.
3022: @endassignment
3023:
3024: @node Arrays and Records Tutorial, POSTPONE Tutorial, Defining Words Tutorial, Tutorial
3025: @section Arrays and Records
3026:
3027: Forth has no standard words for defining data structures such as arrays
3028: and records (structs in C terminology), but you can build them yourself
3029: based on address arithmetic. You can also define words for defining
3030: arrays and records (@pxref{Defining Words Tutorial,, Defining Words}).
3031:
3032: One of the first projects a Forth newcomer sets out upon when learning
3033: about defining words is an array defining word (possibly for
3034: n-dimensional arrays). Go ahead and do it, I did it, too; you will
3035: learn something from it. However, don't be disappointed when you later
3036: learn that you have little use for these words (inappropriate use would
3037: be even worse). I have not yet found a set of useful array words yet;
3038: the needs are just too diverse, and named, global arrays (the result of
3039: naive use of defining words) are often not flexible enough (e.g.,
3040: consider how to pass them as parameters).
3041:
3042: On the other hand, there is a useful set of record words, and it has
3043: been defined in @file{compat/struct.fs}; these words are predefined in
3044: Gforth. They are explained in depth elsewhere in this manual (see
3045: @pxref{Structures}). The @code{simple-field} example above is
3046: simplified variant of fields in this package.
3047:
3048:
3049: @node POSTPONE Tutorial, Literal Tutorial, Arrays and Records Tutorial, Tutorial
3050: @section @code{POSTPONE}
3051:
3052: You can compile the compilation semantics (instead of compiling the
3053: interpretation semantics) of a word with @code{POSTPONE}:
3054:
3055: @example
3056: : MY-+ ( Compilation: -- ; Run-time of compiled code: n1 n2 -- n )
3057: POSTPONE + ; immediate compile-only
3058: : foo ( n1 n2 -- n )
3059: MY-+ ;
3060: 1 2 foo .
3061: see foo
3062: @end example
3063:
3064: During the definition of @code{foo} the text interpreter performs the
3065: compilation semantics of @code{MY-+}, which performs the compilation
3066: semantics of @code{+}, i.e., it compiles @code{+} into @code{foo}.
3067:
3068: This example also displays separate stack comments for the compilation
3069: semantics and for the stack effect of the compiled code. For words with
3070: default compilation semantics these stack effects are usually not
3071: displayed; the stack effect of the compilation semantics is always
3072: @code{( -- )} for these words, the stack effect for the compiled code is
3073: the stack effect of the interpretation semantics.
3074:
3075: Note that the state of the interpreter does not come into play when
3076: performing the compilation semantics in this way. You can also perform
3077: it interpretively, e.g.:
3078:
3079: @example
3080: : foo2 ( n1 n2 -- n )
3081: [ MY-+ ] ;
3082: 1 2 foo .
3083: see foo
3084: @end example
3085:
3086: However, there are some broken Forth systems where this does not always
3087: work, and therefore this practice has been declared non-standard in
3088: 1999.
3089: @c !! repair.fs
3090:
3091: Here is another example for using @code{POSTPONE}:
3092:
3093: @example
3094: : MY-- ( Compilation: -- ; Run-time of compiled code: n1 n2 -- n )
3095: POSTPONE negate POSTPONE + ; immediate compile-only
3096: : bar ( n1 n2 -- n )
3097: MY-- ;
3098: 2 1 bar .
3099: see bar
3100: @end example
3101:
3102: You can define @code{ENDIF} in this way:
3103:
3104: @example
3105: : ENDIF ( Compilation: orig -- )
3106: POSTPONE then ; immediate
3107: @end example
3108:
3109: @assignment
3110: Write @code{MY-2DUP} that has compilation semantics equivalent to
3111: @code{2dup}, but compiles @code{over over}.
3112: @endassignment
3113:
3114: @node Literal Tutorial, Advanced macros Tutorial, POSTPONE Tutorial, Tutorial
3115: @section @code{Literal}
3116:
3117: You cannot @code{POSTPONE} numbers:
3118:
3119: @example
3120: : [FOO] POSTPONE 500 ; immediate
3121: @end example
3122:
3123: Instead, you can use @code{LITERAL (compilation: n --; run-time: -- n )}:
3124:
3125: @example
3126: : [FOO] ( compilation: --; run-time: -- n )
3127: 500 POSTPONE literal ; immediate
3128:
3129: : flip foo ;
3130: flip .
3131: see flip
3132: @end example
3133:
3134: @code{LITERAL} consumes a number at compile-time (when it's compilation
3135: semantics are executed) and pushes it at run-time (when the code it
3136: compiled is executed). A frequent use of @code{LITERAL} is to compile a
3137: number computed at compile time into the current word:
3138:
3139: @example
3140: : bar ( -- n )
3141: [ 2 2 + ] literal ;
3142: see bar
3143: @end example
3144:
3145: @assignment
3146: Write @code{]L} which allows writing the example above as @code{: bar (
3147: -- n ) [ 2 2 + ]L ;}
3148: @endassignment
3149:
3150:
3151: @node Advanced macros Tutorial, Compilation Tokens Tutorial, Literal Tutorial, Tutorial
3152: @section Advanced macros
3153:
3154: Reconsider @code{map-array} from @ref{Execution Tokens
3155: Tutorial,, Execution Tokens}. It frequently performs @code{execute}, a
3156: relatively expensive operation in some implementations. You can use
3157: @code{compile,} and @code{POSTPONE} to eliminate these @code{execute}s
3158: and produce a word that contains the word to be performed directly:
3159:
3160: @c use ]] ... [[
3161: @example
3162: : compile-map-array ( compilation: xt -- ; run-time: ... addr u -- ... )
3163: \ at run-time, execute xt ( ... x -- ... ) for each element of the
3164: \ array beginning at addr and containing u elements
3165: @{ xt @}
3166: POSTPONE cells POSTPONE over POSTPONE + POSTPONE swap POSTPONE ?do
3167: POSTPONE i POSTPONE @ xt compile,
3168: 1 cells POSTPONE literal POSTPONE +loop ;
3169:
3170: : sum-array ( addr u -- n )
3171: 0 rot rot [ ' + compile-map-array ] ;
3172: see sum-array
3173: a 5 sum-array .
3174: @end example
3175:
3176: You can use the full power of Forth for generating the code; here's an
3177: example where the code is generated in a loop:
3178:
3179: @example
3180: : compile-vmul-step ( compilation: n --; run-time: n1 addr1 -- n2 addr2 )
3181: \ n2=n1+(addr1)*n, addr2=addr1+cell
3182: POSTPONE tuck POSTPONE @
3183: POSTPONE literal POSTPONE * POSTPONE +
3184: POSTPONE swap POSTPONE cell+ ;
3185:
3186: : compile-vmul ( compilation: addr1 u -- ; run-time: addr2 -- n )
3187: \ n=v1*v2 (inneres Produkt), wobei die v_i als addr_i u repräsentiert sind
3188: 0 postpone literal postpone swap
3189: [ ' compile-vmul-step compile-map-array ]
3190: postpone drop ;
3191: see compile-vmul
3192:
3193: : a-vmul ( addr -- n )
3194: \ n=a*v, wobei v ein Vektor ist, der so lang ist wie a und bei addr anfängt
3195: [ a 5 compile-vmul ] ;
3196: see a-vmul
3197: a a-vmul .
3198: @end example
3199:
3200: This example uses @code{compile-map-array} to show off, but you could
3201: also use @code{map-array} instead (try it now).
3202:
3203: You can use this technique for efficient multiplication of large
3204: matrices. In matrix multiplication, you multiply every line of one
3205: matrix with every column of the other matrix. You can generate the code
3206: for one line once, and use it for every column. The only downside of
3207: this technique is that it is cumbersome to recover the memory consumed
3208: by the generated code when you are done (and in more complicated cases
3209: it is not possible portably).
3210:
3211: @node Compilation Tokens Tutorial, Wordlists and Search Order Tutorial, Advanced macros Tutorial, Tutorial
3212: @section Compilation Tokens
3213:
3214: This section is Gforth-specific. You can skip it.
3215:
3216: @code{' word compile,} compiles the interpretation semantics. For words
3217: with default compilation semantics this is the same as performing the
3218: compilation semantics. To represent the compilation semantics of other
3219: words (e.g., words like @code{if} that have no interpretation
3220: semantics), Gforth has the concept of a compilation token (CT,
3221: consisting of two cells), and words @code{comp'} and @code{[comp']}.
3222: You can perform the compilation semantics represented by a CT with
3223: @code{execute}:
3224:
3225: @example
3226: : foo2 ( n1 n2 -- n )
3227: [ comp' + execute ] ;
3228: see foo
3229: @end example
3230:
3231: You can compile the compilation semantics represented by a CT with
3232: @code{postpone,}:
3233:
3234: @example
3235: : foo3 ( -- )
3236: [ comp' + postpone, ] ;
3237: see foo3
3238: @end example
3239:
3240: @code{[ comp' wort postpone, ]} is equivalent to @code{POSTPONE word}.
3241: @code{comp'} is particularly useful for words that have no
3242: interpretation semantics:
3243:
3244: @example
3245: ' if
3246: comp' if .s
3247: @end example
3248:
3249:
3250: @node Wordlists and Search Order Tutorial, , Compilation Tokens Tutorial, Tutorial
3251: @section Wordlists and Search Order
3252:
3253: The dictionary is not just a memory area that allows you to allocate
3254: memory with @code{allot}, it also contains the Forth words, arranged in
3255: several wordlists. When searching for a word in a wordlist,
3256: conceptually you start searching at the youngest and proceed towards
3257: older words (in reality most systems nowadays use hash-tables); i.e., if
3258: you define a word with the same name as an older word, the new word
3259: shadows the older word.
3260:
3261: Which wordlists are searched in which order is determined by the search
3262: order. You can display the search order with @code{order}. It displays
3263: first the search order, starting with the wordlist searched first, then
3264: it displays the wordlist that will contain newly defined words.
3265:
3266: You can create a new, empty wordlist with @code{wordlist ( -- wid )}:
3267:
3268: @example
3269: wordlist constant mywords
3270: @end example
3271:
3272: @code{Set-current ( wid -- )} sets the wordlist that will contain newly
3273: defined words (the @emph{current} wordlist):
3274:
3275: @example
3276: mywords set-current
3277: order
3278: @end example
3279:
3280: Gforth does not display a name for the wordlist in @code{mywords}
3281: because this wordlist was created anonymously with @code{wordlist}.
3282:
3283: You can get the current wordlist with @code{get-current ( -- wid)}. If
3284: you want to put something into a specific wordlist without overall
3285: effect on the current wordlist, this typically looks like this:
3286:
3287: @example
3288: get-current mywords set-current ( wid )
3289: create someword
3290: ( wid ) set-current
3291: @end example
3292:
3293: You can write the search order with @code{set-order ( wid1 .. widn n --
3294: )} and read it with @code{get-order ( -- wid1 .. widn n )}. The first
3295: searched wordlist is topmost.
3296:
3297: @example
3298: get-order mywords swap 1+ set-order
3299: order
3300: @end example
3301:
3302: Yes, the order of wordlists in the output of @code{order} is reversed
3303: from stack comments and the output of @code{.s} and thus unintuitive.
3304:
3305: @assignment
3306: Define @code{>order ( wid -- )} with adds @code{wid} as first searched
3307: wordlist to the search order. Define @code{previous ( -- )}, which
3308: removes the first searched wordlist from the search order. Experiment
3309: with boundary conditions (you will see some crashes or situations that
3310: are hard or impossible to leave).
3311: @endassignment
3312:
3313: The search order is a powerful foundation for providing features similar
3314: to Modula-2 modules and C++ namespaces. However, trying to modularize
3315: programs in this way has disadvantages for debugging and reuse/factoring
3316: that overcome the advantages in my experience (I don't do huge projects,
3317: though). These disadvanategs are not so clear in other
3318: languages/programming environments, because these langauges are not so
3319: strong in debugging and reuse.
3320:
3321:
3322: @c ******************************************************************
3323: @node Introduction, Words, Tutorial, Top
3324: @comment node-name, next, previous, up
3325: @chapter An Introduction to ANS Forth
3326: @cindex Forth - an introduction
3327:
3328: The primary purpose of this manual is to document Gforth. However, since
3329: Forth is not a widely-known language and there is a lack of up-to-date
3330: teaching material, it seems worthwhile to provide some introductory
3331: material. @xref{Forth-related information} for other sources of Forth-related
3332: information.
3333:
3334: The examples in this section should work on any ANS Forth; the
3335: output shown was produced using Gforth. Each example attempts to
3336: reproduce the exact output that Gforth produces. If you try out the
3337: examples (and you should), what you should type is shown @kbd{like this}
3338: and Gforth's response is shown @code{like this}. The single exception is
3339: that, where the example shows @key{RET} it means that you should
3340: press the ``carriage return'' key. Unfortunately, some output formats for
3341: this manual cannot show the difference between @kbd{this} and
3342: @code{this} which will make trying out the examples harder (but not
3343: impossible).
3344:
3345: Forth is an unusual language. It provides an interactive development
3346: environment which includes both an interpreter and compiler. Forth
3347: programming style encourages you to break a problem down into many
3348: @cindex factoring
3349: small fragments (@dfn{factoring}), and then to develop and test each
3350: fragment interactively. Forth advocates assert that breaking the
3351: edit-compile-test cycle used by conventional programming languages can
3352: lead to great productivity improvements.
3353:
3354: @menu
3355: * Introducing the Text Interpreter::
3356: * Stacks and Postfix notation::
3357: * Your first definition::
3358: * How does that work?::
3359: * Forth is written in Forth::
3360: * Review - elements of a Forth system::
3361: * Where to go next::
3362: * Exercises::
3363: @end menu
3364:
3365: @comment ----------------------------------------------
3366: @node Introducing the Text Interpreter, Stacks and Postfix notation, Introduction, Introduction
3367: @section Introducing the Text Interpreter
3368: @cindex text interpreter
3369: @cindex outer interpreter
3370:
3371: @c IMO this is too detailed and the pace is too slow for
3372: @c an introduction. If you know German, take a look at
3373: @c http://www.complang.tuwien.ac.at/anton/lvas/skriptum-stack.html
3374: @c to see how I do it - anton
3375:
3376: @c nac-> Where I have accepted your comments 100% and modified the text
3377: @c accordingly, I have deleted your comments. Elsewhere I have added a
3378: @c response like this to attempt to rationalise what I have done. Of
3379: @c course, this is a very clumsy mechanism for something that would be
3380: @c done far more efficiently over a beer. Please delete any dialogue
3381: @c you consider closed.
3382:
3383: When you invoke the Forth image, you will see a startup banner printed
3384: and nothing else (if you have Gforth installed on your system, try
3385: invoking it now, by typing @kbd{gforth@key{RET}}). Forth is now running
3386: its command line interpreter, which is called the @dfn{Text Interpreter}
3387: (also known as the @dfn{Outer Interpreter}). (You will learn a lot
3388: about the text interpreter as you read through this chapter, but
3389: @pxref{The Text Interpreter} for more detail).
3390:
3391: Although it's not obvious, Forth is actually waiting for your
3392: input. Type a number and press the @key{RET} key:
3393:
3394: @example
3395: @kbd{45@key{RET}} ok
3396: @end example
3397:
3398: Rather than give you a prompt to invite you to input something, the text
3399: interpreter prints a status message @i{after} it has processed a line
3400: of input. The status message in this case (``@code{ ok}'' followed by
3401: carriage-return) indicates that the text interpreter was able to process
3402: all of your input successfully. Now type something illegal:
3403:
3404: @example
3405: @kbd{qwer341@key{RET}}
3406: :1: Undefined word
3407: qwer341
3408: ^^^^^^^
3409: $400D2BA8 Bounce
3410: $400DBDA8 no.extensions
3411: @end example
3412:
3413: The exact text, other than the ``Undefined word'' may differ slightly on
3414: your system, but the effect is the same; when the text interpreter
3415: detects an error, it discards any remaining text on a line, resets
3416: certain internal state and prints an error message. @xref{Error
3417: messages} for a detailed description of error messages.
3418:
3419: The text interpreter waits for you to press carriage-return, and then
3420: processes your input line. Starting at the beginning of the line, it
3421: breaks the line into groups of characters separated by spaces. For each
3422: group of characters in turn, it makes two attempts to do something:
3423:
3424: @itemize @bullet
3425: @item
3426: @cindex name dictionary
3427: It tries to treat it as a command. It does this by searching a @dfn{name
3428: dictionary}. If the group of characters matches an entry in the name
3429: dictionary, the name dictionary provides the text interpreter with
3430: information that allows the text interpreter perform some actions. In
3431: Forth jargon, we say that the group
3432: @cindex word
3433: @cindex definition
3434: @cindex execution token
3435: @cindex xt
3436: of characters names a @dfn{word}, that the dictionary search returns an
3437: @dfn{execution token (xt)} corresponding to the @dfn{definition} of the
3438: word, and that the text interpreter executes the xt. Often, the terms
3439: @dfn{word} and @dfn{definition} are used interchangeably.
3440: @item
3441: If the text interpreter fails to find a match in the name dictionary, it
3442: tries to treat the group of characters as a number in the current number
3443: base (when you start up Forth, the current number base is base 10). If
3444: the group of characters legitimately represents a number, the text
3445: interpreter pushes the number onto a stack (we'll learn more about that
3446: in the next section).
3447: @end itemize
3448:
3449: If the text interpreter is unable to do either of these things with any
3450: group of characters, it discards the group of characters and the rest of
3451: the line, then prints an error message. If the text interpreter reaches
3452: the end of the line without error, it prints the status message ``@code{ ok}''
3453: followed by carriage-return.
3454:
3455: This is the simplest command we can give to the text interpreter:
3456:
3457: @example
3458: @key{RET} ok
3459: @end example
3460:
3461: The text interpreter did everything we asked it to do (nothing) without
3462: an error, so it said that everything is ``@code{ ok}''. Try a slightly longer
3463: command:
3464:
3465: @example
3466: @kbd{12 dup fred dup@key{RET}}
3467: :1: Undefined word
3468: 12 dup fred dup
3469: ^^^^
3470: $400D2BA8 Bounce
3471: $400DBDA8 no.extensions
3472: @end example
3473:
3474: When you press the carriage-return key, the text interpreter starts to
3475: work its way along the line:
3476:
3477: @itemize @bullet
3478: @item
3479: When it gets to the space after the @code{2}, it takes the group of
3480: characters @code{12} and looks them up in the name
3481: dictionary@footnote{We can't tell if it found them or not, but assume
3482: for now that it did not}. There is no match for this group of characters
3483: in the name dictionary, so it tries to treat them as a number. It is
3484: able to do this successfully, so it puts the number, 12, ``on the stack''
3485: (whatever that means).
3486: @item
3487: The text interpreter resumes scanning the line and gets the next group
3488: of characters, @code{dup}. It looks it up in the name dictionary and
3489: (you'll have to take my word for this) finds it, and executes the word
3490: @code{dup} (whatever that means).
3491: @item
3492: Once again, the text interpreter resumes scanning the line and gets the
3493: group of characters @code{fred}. It looks them up in the name
3494: dictionary, but can't find them. It tries to treat them as a number, but
3495: they don't represent any legal number.
3496: @end itemize
3497:
3498: At this point, the text interpreter gives up and prints an error
3499: message. The error message shows exactly how far the text interpreter
3500: got in processing the line. In particular, it shows that the text
3501: interpreter made no attempt to do anything with the final character
3502: group, @code{dup}, even though we have good reason to believe that the
3503: text interpreter would have no problem looking that word up and
3504: executing it a second time.
3505:
3506:
3507: @comment ----------------------------------------------
3508: @node Stacks and Postfix notation, Your first definition, Introducing the Text Interpreter, Introduction
3509: @section Stacks, postfix notation and parameter passing
3510: @cindex text interpreter
3511: @cindex outer interpreter
3512:
3513: In procedural programming languages (like C and Pascal), the
3514: building-block of programs is the @dfn{function} or @dfn{procedure}. These
3515: functions or procedures are called with @dfn{explicit parameters}. For
3516: example, in C we might write:
3517:
3518: @example
3519: total = total + new_volume(length,height,depth);
3520: @end example
3521:
3522: @noindent
3523: where new_volume is a function-call to another piece of code, and total,
3524: length, height and depth are all variables. length, height and depth are
3525: parameters to the function-call.
3526:
3527: In Forth, the equivalent of the function or procedure is the
3528: @dfn{definition} and parameters are implicitly passed between
3529: definitions using a shared stack that is visible to the
3530: programmer. Although Forth does support variables, the existence of the
3531: stack means that they are used far less often than in most other
3532: programming languages. When the text interpreter encounters a number, it
3533: will place (@dfn{push}) it on the stack. There are several stacks (the
3534: actual number is implementation-dependent ...) and the particular stack
3535: used for any operation is implied unambiguously by the operation being
3536: performed. The stack used for all integer operations is called the @dfn{data
3537: stack} and, since this is the stack used most commonly, references to
3538: ``the data stack'' are often abbreviated to ``the stack''.
3539:
3540: The stacks have a last-in, first-out (LIFO) organisation. If you type:
3541:
3542: @example
3543: @kbd{1 2 3@key{RET}} ok
3544: @end example
3545:
3546: Then this instructs the text interpreter to placed three numbers on the
3547: (data) stack. An analogy for the behaviour of the stack is to take a
3548: pack of playing cards and deal out the ace (1), 2 and 3 into a pile on
3549: the table. The 3 was the last card onto the pile (``last-in'') and if
3550: you take a card off the pile then, unless you're prepared to fiddle a
3551: bit, the card that you take off will be the 3 (``first-out''). The
3552: number that will be first-out of the stack is called the @dfn{top of
3553: stack}, which
3554: @cindex TOS definition
3555: is often abbreviated to @dfn{TOS}.
3556:
3557: To understand how parameters are passed in Forth, consider the
3558: behaviour of the definition @code{+} (pronounced ``plus''). You will not
3559: be surprised to learn that this definition performs addition. More
3560: precisely, it adds two number together and produces a result. Where does
3561: it get the two numbers from? It takes the top two numbers off the
3562: stack. Where does it place the result? On the stack. You can act-out the
3563: behaviour of @code{+} with your playing cards like this:
3564:
3565: @itemize @bullet
3566: @item
3567: Pick up two cards from the stack on the table
3568: @item
3569: Stare at them intently and ask yourself ``what @i{is} the sum of these two
3570: numbers''
3571: @item
3572: Decide that the answer is 5
3573: @item
3574: Shuffle the two cards back into the pack and find a 5
3575: @item
3576: Put a 5 on the remaining ace that's on the table.
3577: @end itemize
3578:
3579: If you don't have a pack of cards handy but you do have Forth running,
3580: you can use the definition @code{.s} to show the current state of the stack,
3581: without affecting the stack. Type:
3582:
3583: @example
3584: @kbd{clearstack 1 2 3@key{RET}} ok
3585: @kbd{.s@key{RET}} <3> 1 2 3 ok
3586: @end example
3587:
3588: The text interpreter looks up the word @code{clearstack} and executes
3589: it; it tidies up the stack and removes any entries that may have been
3590: left on it by earlier examples. The text interpreter pushes each of the
3591: three numbers in turn onto the stack. Finally, the text interpreter
3592: looks up the word @code{.s} and executes it. The effect of executing
3593: @code{.s} is to print the ``<3>'' (the total number of items on the stack)
3594: followed by a list of all the items on the stack; the item on the far
3595: right-hand side is the TOS.
3596:
3597: You can now type:
3598:
3599: @example
3600: @kbd{+ .s@key{RET}} <2> 1 5 ok
3601: @end example
3602:
3603: @noindent
3604: which is correct; there are now 2 items on the stack and the result of
3605: the addition is 5.
3606:
3607: If you're playing with cards, try doing a second addition: pick up the
3608: two cards, work out that their sum is 6, shuffle them into the pack,
3609: look for a 6 and place that on the table. You now have just one item on
3610: the stack. What happens if you try to do a third addition? Pick up the
3611: first card, pick up the second card -- ah! There is no second card. This
3612: is called a @dfn{stack underflow} and consitutes an error. If you try to
3613: do the same thing with Forth it will report an error (probably a Stack
3614: Underflow or an Invalid Memory Address error).
3615:
3616: The opposite situation to a stack underflow is a @dfn{stack overflow},
3617: which simply accepts that there is a finite amount of storage space
3618: reserved for the stack. To stretch the playing card analogy, if you had
3619: enough packs of cards and you piled the cards up on the table, you would
3620: eventually be unable to add another card; you'd hit the ceiling. Gforth
3621: allows you to set the maximum size of the stacks. In general, the only
3622: time that you will get a stack overflow is because a definition has a
3623: bug in it and is generating data on the stack uncontrollably.
3624:
3625: There's one final use for the playing card analogy. If you model your
3626: stack using a pack of playing cards, the maximum number of items on
3627: your stack will be 52 (I assume you didn't use the Joker). The maximum
3628: @i{value} of any item on the stack is 13 (the King). In fact, the only
3629: possible numbers are positive integer numbers 1 through 13; you can't
3630: have (for example) 0 or 27 or 3.52 or -2. If you change the way you
3631: think about some of the cards, you can accommodate different
3632: numbers. For example, you could think of the Jack as representing 0,
3633: the Queen as representing -1 and the King as representing -2. Your
3634: @i{range} remains unchanged (you can still only represent a total of 13
3635: numbers) but the numbers that you can represent are -2 through 10.
3636:
3637: In that analogy, the limit was the amount of information that a single
3638: stack entry could hold, and Forth has a similar limit. In Forth, the
3639: size of a stack entry is called a @dfn{cell}. The actual size of a cell is
3640: implementation dependent and affects the maximum value that a stack
3641: entry can hold. A Standard Forth provides a cell size of at least
3642: 16-bits, and most desktop systems use a cell size of 32-bits.
3643:
3644: Forth does not do any type checking for you, so you are free to
3645: manipulate and combine stack items in any way you wish. A convenient way
3646: of treating stack items is as 2's complement signed integers, and that
3647: is what Standard words like @code{+} do. Therefore you can type:
3648:
3649: @example
3650: @kbd{-5 12 + .s@key{RET}} <1> 7 ok
3651: @end example
3652:
3653: If you use numbers and definitions like @code{+} in order to turn Forth
3654: into a great big pocket calculator, you will realise that it's rather
3655: different from a normal calculator. Rather than typing 2 + 3 = you had
3656: to type 2 3 + (ignore the fact that you had to use @code{.s} to see the
3657: result). The terminology used to describe this difference is to say that
3658: your calculator uses @dfn{Infix Notation} (parameters and operators are
3659: mixed) whilst Forth uses @dfn{Postfix Notation} (parameters and
3660: operators are separate), also called @dfn{Reverse Polish Notation}.
3661:
3662: Whilst postfix notation might look confusing to begin with, it has
3663: several important advantages:
3664:
3665: @itemize @bullet
3666: @item
3667: it is unambiguous
3668: @item
3669: it is more concise
3670: @item
3671: it fits naturally with a stack-based system
3672: @end itemize
3673:
3674: To examine these claims in more detail, consider these sums:
3675:
3676: @example
3677: 6 + 5 * 4 =
3678: 4 * 5 + 6 =
3679: @end example
3680:
3681: If you're just learning maths or your maths is very rusty, you will
3682: probably come up with the answer 44 for the first and 26 for the
3683: second. If you are a bit of a whizz at maths you will remember the
3684: @i{convention} that multiplication takes precendence over addition, and
3685: you'd come up with the answer 26 both times. To explain the answer 26
3686: to someone who got the answer 44, you'd probably rewrite the first sum
3687: like this:
3688:
3689: @example
3690: 6 + (5 * 4) =
3691: @end example
3692:
3693: If what you really wanted was to perform the addition before the
3694: multiplication, you would have to use parentheses to force it.
3695:
3696: If you did the first two sums on a pocket calculator you would probably
3697: get the right answers, unless you were very cautious and entered them using
3698: these keystroke sequences:
3699:
3700: 6 + 5 = * 4 =
3701: 4 * 5 = + 6 =
3702:
3703: Postfix notation is unambiguous because the order that the operators
3704: are applied is always explicit; that also means that parentheses are
3705: never required. The operators are @i{active} (the act of quoting the
3706: operator makes the operation occur) which removes the need for ``=''.
3707:
3708: The sum 6 + 5 * 4 can be written (in postfix notation) in two
3709: equivalent ways:
3710:
3711: @example
3712: 6 5 4 * + or:
3713: 5 4 * 6 +
3714: @end example
3715:
3716: An important thing that you should notice about this notation is that
3717: the @i{order} of the numbers does not change; if you want to subtract
3718: 2 from 10 you type @code{10 2 -}.
3719:
3720: The reason that Forth uses postfix notation is very simple to explain: it
3721: makes the implementation extremely simple, and it follows naturally from
3722: using the stack as a mechanism for passing parameters. Another way of
3723: thinking about this is to realise that all Forth definitions are
3724: @i{active}; they execute as they are encountered by the text
3725: interpreter. The result of this is that the syntax of Forth is trivially
3726: simple.
3727:
3728:
3729:
3730: @comment ----------------------------------------------
3731: @node Your first definition, How does that work?, Stacks and Postfix notation, Introduction
3732: @section Your first Forth definition
3733: @cindex first definition
3734:
3735: Until now, the examples we've seen have been trivial; we've just been
3736: using Forth as a bigger-than-pocket calculator. Also, each calculation
3737: we've shown has been a ``one-off'' -- to repeat it we'd need to type it in
3738: again@footnote{That's not quite true. If you press the up-arrow key on
3739: your keyboard you should be able to scroll back to any earlier command,
3740: edit it and re-enter it.} In this section we'll see how to add new
3741: words to Forth's vocabulary.
3742:
3743: The easiest way to create a new word is to use a @dfn{colon
3744: definition}. We'll define a few and try them out before worrying too
3745: much about how they work. Try typing in these examples; be careful to
3746: copy the spaces accurately:
3747:
3748: @example
3749: : add-two 2 + . ;
3750: : greet ." Hello and welcome" ;
3751: : demo 5 add-two ;
3752: @end example
3753:
3754: @noindent
3755: Now try them out:
3756:
3757: @example
3758: @kbd{greet@key{RET}} Hello and welcome ok
3759: @kbd{greet greet@key{RET}} Hello and welcomeHello and welcome ok
3760: @kbd{4 add-two@key{RET}} 6 ok
3761: @kbd{demo@key{RET}} 7 ok
3762: @kbd{9 greet demo add-two@key{RET}} Hello and welcome7 11 ok
3763: @end example
3764:
3765: The first new thing that we've introduced here is the pair of words
3766: @code{:} and @code{;}. These are used to start and terminate a new
3767: definition, respectively. The first word after the @code{:} is the name
3768: for the new definition.
3769:
3770: As you can see from the examples, a definition is built up of words that
3771: have already been defined; Forth makes no distinction between
3772: definitions that existed when you started the system up, and those that
3773: you define yourself.
3774:
3775: The examples also introduce the words @code{.} (dot), @code{."}
3776: (dot-quote) and @code{dup} (dewp). Dot takes the value from the top of
3777: the stack and displays it. It's like @code{.s} except that it only
3778: displays the top item of the stack and it is destructive; after it has
3779: executed, the number is no longer on the stack. There is always one
3780: space printed after the number, and no spaces before it. Dot-quote
3781: defines a string (a sequence of characters) that will be printed when
3782: the word is executed. The string can contain any printable characters
3783: except @code{"}. A @code{"} has a special function; it is not a Forth
3784: word but it acts as a delimiter (the way that delimiters work is
3785: described in the next section). Finally, @code{dup} duplicates the value
3786: at the top of the stack. Try typing @code{5 dup .s} to see what it does.
3787:
3788: We already know that the text interpreter searches through the
3789: dictionary to locate names. If you've followed the examples earlier, you
3790: will already have a definition called @code{add-two}. Lets try modifying
3791: it by typing in a new definition:
3792:
3793: @example
3794: @kbd{: add-two dup . ." + 2 =" 2 + . ;@key{RET}} redefined add-two ok
3795: @end example
3796:
3797: Forth recognised that we were defining a word that already exists, and
3798: printed a message to warn us of that fact. Let's try out the new
3799: definition:
3800:
3801: @example
3802: @kbd{9 add-two@key{RET}} 9 + 2 =11 ok
3803: @end example
3804:
3805: @noindent
3806: All that we've actually done here, though, is to create a new
3807: definition, with a particular name. The fact that there was already a
3808: definition with the same name did not make any difference to the way
3809: that the new definition was created (except that Forth printed a warning
3810: message). The old definition of add-two still exists (try @code{demo}
3811: again to see that this is true). Any new definition will use the new
3812: definition of @code{add-two}, but old definitions continue to use the
3813: version that already existed at the time that they were @code{compiled}.
3814:
3815: Before you go on to the next section, try defining and redefining some
3816: words of your own.
3817:
3818: @comment ----------------------------------------------
3819: @node How does that work?, Forth is written in Forth, Your first definition, Introduction
3820: @section How does that work?
3821: @cindex parsing words
3822:
3823: @c That's pretty deep (IMO way too deep) for an introduction. - anton
3824:
3825: @c Is it a good idea to talk about the interpretation semantics of a
3826: @c number? We don't have an xt to go along with it. - anton
3827:
3828: @c Now that I have eliminated execution semantics, I wonder if it would not
3829: @c be better to keep them (or add run-time semantics), to make it easier to
3830: @c explain what compilation semantics usually does. - anton
3831:
3832: @c nac-> I removed the term ``default compilation sematics'' from the
3833: @c introductory chapter. Removing ``execution semantics'' was making
3834: @c everything simpler to explain, then I think the use of this term made
3835: @c everything more complex again. I replaced it with ``default
3836: @c semantics'' (which is used elsewhere in the manual) by which I mean
3837: @c ``a definition that has neither the immediate nor the compile-only
3838: @c flag set''. I reworded big chunks of the ``how does that work''
3839: @c section (and, unusually for me, I think I even made it shorter!). See
3840: @c what you think -- I know I have not addressed your primary concern
3841: @c that it is too heavy-going for an introduction. From what I understood
3842: @c of your course notes it looks as though they might be a good framework.
3843: @c Things that I've tried to capture here are some things that came as a
3844: @c great revelation here when I first understood them. Also, I like the
3845: @c fact that a very simple code example shows up almost all of the issues
3846: @c that you need to understand to see how Forth works. That's unique and
3847: @c worthwhile to emphasise.
3848:
3849: Now we're going to take another look at the definition of @code{add-two}
3850: from the previous section. From our knowledge of the way that the text
3851: interpreter works, we would have expected this result when we tried to
3852: define @code{add-two}:
3853:
3854: @example
3855: @kbd{: add-two 2 + . ;@key{RET}}
3856: ^^^^^^^
3857: Error: Undefined word
3858: @end example
3859:
3860: The reason that this didn't happen is bound up in the way that @code{:}
3861: works. The word @code{:} does two special things. The first special
3862: thing that it does prevents the text interpreter from ever seeing the
3863: characters @code{add-two}. The text interpreter uses a variable called
3864: @cindex modifying >IN
3865: @code{>IN} (pronounced ``to-in'') to keep track of where it is in the
3866: input line. When it encounters the word @code{:} it behaves in exactly
3867: the same way as it does for any other word; it looks it up in the name
3868: dictionary, finds its xt and executes it. When @code{:} executes, it
3869: looks at the input buffer, finds the word @code{add-two} and advances the
3870: value of @code{>IN} to point past it. It then does some other stuff
3871: associated with creating the new definition (including creating an entry
3872: for @code{add-two} in the name dictionary). When the execution of @code{:}
3873: completes, control returns to the text interpreter, which is oblivious
3874: to the fact that it has been tricked into ignoring part of the input
3875: line.
3876:
3877: @cindex parsing words
3878: Words like @code{:} -- words that advance the value of @code{>IN} and so
3879: prevent the text interpreter from acting on the whole of the input line
3880: -- are called @dfn{parsing words}.
3881:
3882: @cindex @code{state} - effect on the text interpreter
3883: @cindex text interpreter - effect of state
3884: The second special thing that @code{:} does is change the value of a
3885: variable called @code{state}, which affects the way that the text
3886: interpreter behaves. When Gforth starts up, @code{state} has the value
3887: 0, and the text interpreter is said to be @dfn{interpreting}. During a
3888: colon definition (started with @code{:}), @code{state} is set to -1 and
3889: the text interpreter is said to be @dfn{compiling}.
3890:
3891: In this example, the text interpreter is compiling when it processes the
3892: string ``@code{2 + . ;}''. It still breaks the string down into
3893: character sequences in the same way. However, instead of pushing the
3894: number @code{2} onto the stack, it lays down (@dfn{compiles}) some magic
3895: into the definition of @code{add-two} that will make the number @code{2} get
3896: pushed onto the stack when @code{add-two} is @dfn{executed}. Similarly,
3897: the behaviours of @code{+} and @code{.} are also compiled into the
3898: definition.
3899:
3900: One category of words don't get compiled. These so-called @dfn{immediate
3901: words} get executed (performed @i{now}) regardless of whether the text
3902: interpreter is interpreting or compiling. The word @code{;} is an
3903: immediate word. Rather than being compiled into the definition, it
3904: executes. Its effect is to terminate the current definition, which
3905: includes changing the value of @code{state} back to 0.
3906:
3907: When you execute @code{add-two}, it has a @dfn{run-time effect} that is
3908: exactly the same as if you had typed @code{2 + . @key{RET}} outside of a
3909: definition.
3910:
3911: In Forth, every word or number can be described in terms of two
3912: properties:
3913:
3914: @itemize @bullet
3915: @item
3916: @cindex interpretation semantics
3917: Its @dfn{interpretation semantics} describe how it will behave when the
3918: text interpreter encounters it in @dfn{interpret} state. The
3919: interpretation semantics of a word are represented by an @dfn{execution
3920: token}.
3921: @item
3922: @cindex compilation semantics
3923: Its @dfn{compilation semantics} describe how it will behave when the
3924: text interpreter encounters it in @dfn{compile} state. The compilation
3925: semantics of a word are represented in an implementation-dependent way;
3926: Gforth uses a @dfn{compilation token}.
3927: @end itemize
3928:
3929: @noindent
3930: Numbers are always treated in a fixed way:
3931:
3932: @itemize @bullet
3933: @item
3934: When the number is @dfn{interpreted}, its behaviour is to push the
3935: number onto the stack.
3936: @item
3937: When the number is @dfn{compiled}, a piece of code is appended to the
3938: current definition that pushes the number when it runs. (In other words,
3939: the compilation semantics of a number are to postpone its interpretation
3940: semantics until the run-time of the definition that it is being compiled
3941: into.)
3942: @end itemize
3943:
3944: Words don't behave in such a regular way, but most have @i{default
3945: semantics} which means that they behave like this:
3946:
3947: @itemize @bullet
3948: @item
3949: The @dfn{interpretation semantics} of the word are to do something useful.
3950: @item
3951: The @dfn{compilation semantics} of the word are to append its
3952: @dfn{interpretation semantics} to the current definition (so that its
3953: run-time behaviour is to do something useful).
3954: @end itemize
3955:
3956: @cindex immediate words
3957: The actual behaviour of any particular word can be controlled by using
3958: the words @code{immediate} and @code{compile-only} when the word is
3959: defined. These words set flags in the name dictionary entry of the most
3960: recently defined word, and these flags are retrieved by the text
3961: interpreter when it finds the word in the name dictionary.
3962:
3963: A word that is marked as @dfn{immediate} has compilation semantics that
3964: are identical to its interpretation semantics. In other words, it
3965: behaves like this:
3966:
3967: @itemize @bullet
3968: @item
3969: The @dfn{interpretation semantics} of the word are to do something useful.
3970: @item
3971: The @dfn{compilation semantics} of the word are to do something useful
3972: (and actually the same thing); i.e., it is executed during compilation.
3973: @end itemize
3974:
3975: Marking a word as @dfn{compile-only} prohibits the text interpreter from
3976: performing the interpretation semantics of the word directly; an attempt
3977: to do so will generate an error. It is never necessary to use
3978: @code{compile-only} (and it is not even part of ANS Forth, though it is
3979: provided by many implementations) but it is good etiquette to apply it
3980: to a word that will not behave correctly (and might have unexpected
3981: side-effects) in interpret state. For example, it is only legal to use
3982: the conditional word @code{IF} within a definition. If you forget this
3983: and try to use it elsewhere, the fact that (in Gforth) it is marked as
3984: @code{compile-only} allows the text interpreter to generate a helpful
3985: error message rather than subjecting you to the consequences of your
3986: folly.
3987:
3988: This example shows the difference between an immediate and a
3989: non-immediate word:
3990:
3991: @example
3992: : show-state state @@ . ;
3993: : show-state-now show-state ; immediate
3994: : word1 show-state ;
3995: : word2 show-state-now ;
3996: @end example
3997:
3998: The word @code{immediate} after the definition of @code{show-state-now}
3999: makes that word an immediate word. These definitions introduce a new
4000: word: @code{@@} (pronounced ``fetch''). This word fetches the value of a
4001: variable, and leaves it on the stack. Therefore, the behaviour of
4002: @code{show-state} is to print a number that represents the current value
4003: of @code{state}.
4004:
4005: When you execute @code{word1}, it prints the number 0, indicating that
4006: the system is interpreting. When the text interpreter compiled the
4007: definition of @code{word1}, it encountered @code{show-state} whose
4008: compilation semantics are to append its interpretation semantics to the
4009: current definition. When you execute @code{word1}, it performs the
4010: interpretation semantics of @code{show-state}. At the time that @code{word1}
4011: (and therefore @code{show-state}) are executed, the system is
4012: interpreting.
4013:
4014: When you pressed @key{RET} after entering the definition of @code{word2},
4015: you should have seen the number -1 printed, followed by ``@code{
4016: ok}''. When the text interpreter compiled the definition of
4017: @code{word2}, it encountered @code{show-state-now}, an immediate word,
4018: whose compilation semantics are therefore to perform its interpretation
4019: semantics. It is executed straight away (even before the text
4020: interpreter has moved on to process another group of characters; the
4021: @code{;} in this example). The effect of executing it are to display the
4022: value of @code{state} @i{at the time that the definition of}
4023: @code{word2} @i{is being defined}. Printing -1 demonstrates that the
4024: system is compiling at this time. If you execute @code{word2} it does
4025: nothing at all.
4026:
4027: @cindex @code{."}, how it works
4028: Before leaving the subject of immediate words, consider the behaviour of
4029: @code{."} in the definition of @code{greet}, in the previous
4030: section. This word is both a parsing word and an immediate word. Notice
4031: that there is a space between @code{."} and the start of the text
4032: @code{Hello and welcome}, but that there is no space between the last
4033: letter of @code{welcome} and the @code{"} character. The reason for this
4034: is that @code{."} is a Forth word; it must have a space after it so that
4035: the text interpreter can identify it. The @code{"} is not a Forth word;
4036: it is a @dfn{delimiter}. The examples earlier show that, when the string
4037: is displayed, there is neither a space before the @code{H} nor after the
4038: @code{e}. Since @code{."} is an immediate word, it executes at the time
4039: that @code{greet} is defined. When it executes, its behaviour is to
4040: search forward in the input line looking for the delimiter. When it
4041: finds the delimiter, it updates @code{>IN} to point past the
4042: delimiter. It also compiles some magic code into the definition of
4043: @code{greet}; the xt of a run-time routine that prints a text string. It
4044: compiles the string @code{Hello and welcome} into memory so that it is
4045: available to be printed later. When the text interpreter gains control,
4046: the next word it finds in the input stream is @code{;} and so it
4047: terminates the definition of @code{greet}.
4048:
4049:
4050: @comment ----------------------------------------------
4051: @node Forth is written in Forth, Review - elements of a Forth system, How does that work?, Introduction
4052: @section Forth is written in Forth
4053: @cindex structure of Forth programs
4054:
4055: When you start up a Forth compiler, a large number of definitions
4056: already exist. In Forth, you develop a new application using bottom-up
4057: programming techniques to create new definitions that are defined in
4058: terms of existing definitions. As you create each definition you can
4059: test and debug it interactively.
4060:
4061: If you have tried out the examples in this section, you will probably
4062: have typed them in by hand; when you leave Gforth, your definitions will
4063: be lost. You can avoid this by using a text editor to enter Forth source
4064: code into a file, and then loading code from the file using
4065: @code{include} (@xref{Forth source files}). A Forth source file is
4066: processed by the text interpreter, just as though you had typed it in by
4067: hand@footnote{Actually, there are some subtle differences -- see
4068: @ref{The Text Interpreter}.}.
4069:
4070: Gforth also supports the traditional Forth alternative to using text
4071: files for program entry (@xref{Blocks}).
4072:
4073: In common with many, if not most, Forth compilers, most of Gforth is
4074: actually written in Forth. All of the @file{.fs} files in the
4075: installation directory@footnote{For example,
4076: @file{/usr/local/share/gforth...}} are Forth source files, which you can
4077: study to see examples of Forth programming.
4078:
4079: Gforth maintains a history file that records every line that you type to
4080: the text interpreter. This file is preserved between sessions, and is
4081: used to provide a command-line recall facility. If you enter long
4082: definitions by hand, you can use a text editor to paste them out of the
4083: history file into a Forth source file for reuse at a later time
4084: (@pxref{Command-line editing} for more information).
4085:
4086:
4087: @comment ----------------------------------------------
4088: @node Review - elements of a Forth system, Where to go next, Forth is written in Forth, Introduction
4089: @section Review - elements of a Forth system
4090: @cindex elements of a Forth system
4091:
4092: To summarise this chapter:
4093:
4094: @itemize @bullet
4095: @item
4096: Forth programs use @dfn{factoring} to break a problem down into small
4097: fragments called @dfn{words} or @dfn{definitions}.
4098: @item
4099: Forth program development is an interactive process.
4100: @item
4101: The main command loop that accepts input, and controls both
4102: interpretation and compilation, is called the @dfn{text interpreter}
4103: (also known as the @dfn{outer interpreter}).
4104: @item
4105: Forth has a very simple syntax, consisting of words and numbers
4106: separated by spaces or carriage-return characters. Any additional syntax
4107: is imposed by @dfn{parsing words}.
4108: @item
4109: Forth uses a stack to pass parameters between words. As a result, it
4110: uses postfix notation.
4111: @item
4112: To use a word that has previously been defined, the text interpreter
4113: searches for the word in the @dfn{name dictionary}.
4114: @item
4115: Words have @dfn{interpretation semantics} and @dfn{compilation semantics}.
4116: @item
4117: The text interpreter uses the value of @code{state} to select between
4118: the use of the @dfn{interpretation semantics} and the @dfn{compilation
4119: semantics} of a word that it encounters.
4120: @item
4121: The relationship between the @dfn{interpretation semantics} and
4122: @dfn{compilation semantics} for a word
4123: depend upon the way in which the word was defined (for example, whether
4124: it is an @dfn{immediate} word).
4125: @item
4126: Forth definitions can be implemented in Forth (called @dfn{high-level
4127: definitions}) or in some other way (usually a lower-level language and
4128: as a result often called @dfn{low-level definitions}, @dfn{code
4129: definitions} or @dfn{primitives}).
4130: @item
4131: Many Forth systems are implemented mainly in Forth.
4132: @end itemize
4133:
4134:
4135: @comment ----------------------------------------------
4136: @node Where to go next, Exercises, Review - elements of a Forth system, Introduction
4137: @section Where To Go Next
4138: @cindex where to go next
4139:
4140: Amazing as it may seem, if you have read (and understood) this far, you
4141: know almost all the fundamentals about the inner workings of a Forth
4142: system. You certainly know enough to be able to read and understand the
4143: rest of this manual and the ANS Forth document, to learn more about the
4144: facilities that Forth in general and Gforth in particular provide. Even
4145: scarier, you know almost enough to implement your own Forth system.
4146: However, that's not a good idea just yet... better to try writing some
4147: programs in Gforth.
4148:
4149: Forth has such a rich vocabulary that it can be hard to know where to
4150: start in learning it. This section suggests a few sets of words that are
4151: enough to write small but useful programs. Use the word index in this
4152: document to learn more about each word, then try it out and try to write
4153: small definitions using it. Start by experimenting with these words:
4154:
4155: @itemize @bullet
4156: @item
4157: Arithmetic: @code{+ - * / /MOD */ ABS INVERT}
4158: @item
4159: Comparison: @code{MIN MAX =}
4160: @item
4161: Logic: @code{AND OR XOR NOT}
4162: @item
4163: Stack manipulation: @code{DUP DROP SWAP OVER}
4164: @item
4165: Loops and decisions: @code{IF ELSE ENDIF ?DO I LOOP}
4166: @item
4167: Input/Output: @code{. ." EMIT CR KEY}
4168: @item
4169: Defining words: @code{: ; CREATE}
4170: @item
4171: Memory allocation words: @code{ALLOT ,}
4172: @item
4173: Tools: @code{SEE WORDS .S MARKER}
4174: @end itemize
4175:
4176: When you have mastered those, go on to:
4177:
4178: @itemize @bullet
4179: @item
4180: More defining words: @code{VARIABLE CONSTANT VALUE TO CREATE DOES>}
4181: @item
4182: Memory access: @code{@@ !}
4183: @end itemize
4184:
4185: When you have mastered these, there's nothing for it but to read through
4186: the whole of this manual and find out what you've missed.
4187:
4188: @comment ----------------------------------------------
4189: @node Exercises, , Where to go next, Introduction
4190: @section Exercises
4191: @cindex exercises
4192:
4193: TODO: provide a set of programming excercises linked into the stuff done
4194: already and into other sections of the manual. Provide solutions to all
4195: the exercises in a .fs file in the distribution.
4196:
4197: @c Get some inspiration from Starting Forth and Kelly&Spies.
4198:
4199: @c excercises:
4200: @c 1. take inches and convert to feet and inches.
4201: @c 2. take temperature and convert from fahrenheight to celcius;
4202: @c may need to care about symmetric vs floored??
4203: @c 3. take input line and do character substitution
4204: @c to encipher or decipher
4205: @c 4. as above but work on a file for in and out
4206: @c 5. take input line and convert to pig-latin
4207: @c
4208: @c thing of sets of things to exercise then come up with
4209: @c problems that need those things.
4210:
4211:
4212: @c ******************************************************************
4213: @node Words, Error messages, Introduction, Top
4214: @chapter Forth Words
4215: @cindex words
4216:
4217: @menu
4218: * Notation::
4219: * Comments::
4220: * Boolean Flags::
4221: * Arithmetic::
4222: * Stack Manipulation::
4223: * Memory::
4224: * Control Structures::
4225: * Defining Words::
4226: * Interpretation and Compilation Semantics::
4227: * Tokens for Words::
4228: * The Text Interpreter::
4229: * Word Lists::
4230: * Environmental Queries::
4231: * Files::
4232: * Blocks::
4233: * Other I/O::
4234: * Programming Tools::
4235: * Assembler and Code Words::
4236: * Threading Words::
4237: * Locals::
4238: * Structures::
4239: * Object-oriented Forth::
4240: * Passing Commands to the OS::
4241: * Keeping track of Time::
4242: * Miscellaneous Words::
4243: @end menu
4244:
4245: @node Notation, Comments, Words, Words
4246: @section Notation
4247: @cindex notation of glossary entries
4248: @cindex format of glossary entries
4249: @cindex glossary notation format
4250: @cindex word glossary entry format
4251:
4252: The Forth words are described in this section in the glossary notation
4253: that has become a de-facto standard for Forth texts, i.e.,
4254:
4255: @format
4256: @i{word} @i{Stack effect} @i{wordset} @i{pronunciation}
4257: @end format
4258: @i{Description}
4259:
4260: @table @var
4261: @item word
4262: The name of the word.
4263:
4264: @item Stack effect
4265: @cindex stack effect
4266: The stack effect is written in the notation @code{@i{before} --
4267: @i{after}}, where @i{before} and @i{after} describe the top of
4268: stack entries before and after the execution of the word. The rest of
4269: the stack is not touched by the word. The top of stack is rightmost,
4270: i.e., a stack sequence is written as it is typed in. Note that Gforth
4271: uses a separate floating point stack, but a unified stack
4272: notation. Also, return stack effects are not shown in @i{stack
4273: effect}, but in @i{Description}. The name of a stack item describes
4274: the type and/or the function of the item. See below for a discussion of
4275: the types.
4276:
4277: All words have two stack effects: A compile-time stack effect and a
4278: run-time stack effect. The compile-time stack-effect of most words is
4279: @i{ -- }. If the compile-time stack-effect of a word deviates from
4280: this standard behaviour, or the word does other unusual things at
4281: compile time, both stack effects are shown; otherwise only the run-time
4282: stack effect is shown.
4283:
4284: @cindex pronounciation of words
4285: @item pronunciation
4286: How the word is pronounced.
4287:
4288: @cindex wordset
4289: @item wordset
4290: The ANS Forth standard is divided into several word sets. A standard
4291: system need not support all of them. Therefore, in theory, the fewer
4292: word sets your program uses the more portable it will be. However, we
4293: suspect that most ANS Forth systems on personal machines will feature
4294: all word sets. Words that are not defined in ANS Forth have
4295: @code{gforth} or @code{gforth-internal} as word set. @code{gforth}
4296: describes words that will work in future releases of Gforth;
4297: @code{gforth-internal} words are more volatile. Environmental query
4298: strings are also displayed like words; you can recognize them by the
4299: @code{environment} in the word set field.
4300:
4301: @item Description
4302: A description of the behaviour of the word.
4303: @end table
4304:
4305: @cindex types of stack items
4306: @cindex stack item types
4307: The type of a stack item is specified by the character(s) the name
4308: starts with:
4309:
4310: @table @code
4311: @item f
4312: @cindex @code{f}, stack item type
4313: Boolean flags, i.e. @code{false} or @code{true}.
4314: @item c
4315: @cindex @code{c}, stack item type
4316: Char
4317: @item w
4318: @cindex @code{w}, stack item type
4319: Cell, can contain an integer or an address
4320: @item n
4321: @cindex @code{n}, stack item type
4322: signed integer
4323: @item u
4324: @cindex @code{u}, stack item type
4325: unsigned integer
4326: @item d
4327: @cindex @code{d}, stack item type
4328: double sized signed integer
4329: @item ud
4330: @cindex @code{ud}, stack item type
4331: double sized unsigned integer
4332: @item r
4333: @cindex @code{r}, stack item type
4334: Float (on the FP stack)
4335: @item a-
4336: @cindex @code{a_}, stack item type
4337: Cell-aligned address
4338: @item c-
4339: @cindex @code{c_}, stack item type
4340: Char-aligned address (note that a Char may have two bytes in Windows NT)
4341: @item f-
4342: @cindex @code{f_}, stack item type
4343: Float-aligned address
4344: @item df-
4345: @cindex @code{df_}, stack item type
4346: Address aligned for IEEE double precision float
4347: @item sf-
4348: @cindex @code{sf_}, stack item type
4349: Address aligned for IEEE single precision float
4350: @item xt
4351: @cindex @code{xt}, stack item type
4352: Execution token, same size as Cell
4353: @item wid
4354: @cindex @code{wid}, stack item type
4355: Word list ID, same size as Cell
4356: @item f83name
4357: @cindex @code{f83name}, stack item type
4358: Pointer to a name structure
4359: @item "
4360: @cindex @code{"}, stack item type
4361: string in the input stream (not on the stack). The terminating character
4362: is a blank by default. If it is not a blank, it is shown in @code{<>}
4363: quotes.
4364: @end table
4365:
4366: @node Comments, Boolean Flags, Notation, Words
4367: @section Comments
4368: @cindex comments
4369:
4370: Forth supports two styles of comment; the traditional @i{in-line} comment,
4371: @code{(} and its modern cousin, the @i{comment to end of line}; @code{\}.
4372:
4373:
4374: doc-(
4375: doc-\
4376: doc-\G
4377:
4378:
4379: @node Boolean Flags, Arithmetic, Comments, Words
4380: @section Boolean Flags
4381: @cindex Boolean flags
4382:
4383: A Boolean flag is cell-sized. A cell with all bits clear represents the
4384: flag @code{false} and a flag with all bits set represents the flag
4385: @code{true}. Words that check a flag (for example, @code{IF}) will treat
4386: a cell that has @i{any} bit set as @code{true}.
4387:
4388:
4389: doc-true
4390: doc-false
4391: doc-on
4392: doc-off
4393:
4394:
4395: @node Arithmetic, Stack Manipulation, Boolean Flags, Words
4396: @section Arithmetic
4397: @cindex arithmetic words
4398:
4399: @cindex division with potentially negative operands
4400: Forth arithmetic is not checked, i.e., you will not hear about integer
4401: overflow on addition or multiplication, you may hear about division by
4402: zero if you are lucky. The operator is written after the operands, but
4403: the operands are still in the original order. I.e., the infix @code{2-1}
4404: corresponds to @code{2 1 -}. Forth offers a variety of division
4405: operators. If you perform division with potentially negative operands,
4406: you do not want to use @code{/} or @code{/mod} with its undefined
4407: behaviour, but rather @code{fm/mod} or @code{sm/mod} (probably the
4408: former, @pxref{Mixed precision}).
4409: @comment TODO discuss the different division forms and the std approach
4410:
4411: @menu
4412: * Single precision::
4413: * Bitwise operations::
4414: * Double precision:: Double-cell integer arithmetic
4415: * Numeric comparison::
4416: * Mixed precision:: Operations with single and double-cell integers
4417: * Floating Point::
4418: @end menu
4419:
4420: @node Single precision, Bitwise operations, Arithmetic, Arithmetic
4421: @subsection Single precision
4422: @cindex single precision arithmetic words
4423:
4424: By default, numbers in Forth are single-precision integers that are 1
4425: cell in size. They can be signed or unsigned, depending upon how you
4426: treat them. @xref{Number Conversion} for the rules used by the text
4427: interpreter for recognising single-precision integers.
4428:
4429:
4430: doc-+
4431: doc-1+
4432: doc--
4433: doc-1-
4434: doc-*
4435: doc-/
4436: doc-mod
4437: doc-/mod
4438: doc-negate
4439: doc-abs
4440: doc-min
4441: doc-max
4442: doc-d>s
4443: doc-floored
4444:
4445:
4446: @node Bitwise operations, Double precision, Single precision, Arithmetic
4447: @subsection Bitwise operations
4448: @cindex bitwise operation words
4449:
4450:
4451: doc-and
4452: doc-or
4453: doc-xor
4454: doc-invert
4455: doc-lshift
4456: doc-rshift
4457: doc-2*
4458: doc-d2*
4459: doc-2/
4460: doc-d2/
4461:
4462:
4463: @node Double precision, Numeric comparison, Bitwise operations, Arithmetic
4464: @subsection Double precision
4465: @cindex double precision arithmetic words
4466:
4467: @xref{Number Conversion} for the rules used by the text interpreter for
4468: recognising double-precision integers.
4469:
4470: A double precision number is represented by a cell pair, with the most
4471: significant cell at the TOS. It is trivial to convert an unsigned
4472: single to an (unsigned) double; simply push a @code{0} onto the
4473: TOS. Since numbers are represented by Gforth using 2's complement
4474: arithmetic, converting a signed single to a (signed) double requires
4475: sign-extension across the most significant cell. This can be achieved
4476: using @code{s>d}. The moral of the story is that you cannot convert a
4477: number without knowing whether it represents an unsigned or a
4478: signed number.
4479:
4480:
4481: doc-s>d
4482: doc-d+
4483: doc-d-
4484: doc-dnegate
4485: doc-dabs
4486: doc-dmin
4487: doc-dmax
4488:
4489:
4490: @node Numeric comparison, Mixed precision, Double precision, Arithmetic
4491: @subsection Numeric comparison
4492: @cindex numeric comparison words
4493:
4494:
4495: doc-<
4496: doc-<=
4497: doc-<>
4498: doc-=
4499: doc->
4500: doc->=
4501:
4502: doc-0<
4503: doc-0<=
4504: doc-0<>
4505: doc-0=
4506: doc-0>
4507: doc-0>=
4508:
4509: doc-u<
4510: doc-u<=
4511: @c u<> and u= exist but are the same as <> and =
4512: @c doc-u<>
4513: @c doc-u=
4514: doc-u>
4515: doc-u>=
4516:
4517: doc-within
4518:
4519: doc-d<
4520: doc-d<=
4521: doc-d<>
4522: doc-d=
4523: doc-d>
4524: doc-d>=
4525:
4526: doc-d0<
4527: doc-d0<=
4528: doc-d0<>
4529: doc-d0=
4530: doc-d0>
4531: doc-d0>=
4532:
4533: doc-du<
4534: doc-du<=
4535: @c du<> and du= exist but are the same as d<> and d=
4536: @c doc-du<>
4537: @c doc-du=
4538: doc-du>
4539: doc-du>=
4540:
4541:
4542: @node Mixed precision, Floating Point, Numeric comparison, Arithmetic
4543: @subsection Mixed precision
4544: @cindex mixed precision arithmetic words
4545:
4546:
4547: doc-m+
4548: doc-*/
4549: doc-*/mod
4550: doc-m*
4551: doc-um*
4552: doc-m*/
4553: doc-um/mod
4554: doc-fm/mod
4555: doc-sm/rem
4556:
4557:
4558: @node Floating Point, , Mixed precision, Arithmetic
4559: @subsection Floating Point
4560: @cindex floating point arithmetic words
4561:
4562: @xref{Number Conversion} for the rules used by the text interpreter for
4563: recognising floating-point numbers.
4564:
4565: Gforth has a separate floating point
4566: stack, but the documentation uses the unified notation.
4567:
4568: @cindex floating-point arithmetic, pitfalls
4569: Floating point numbers have a number of unpleasant surprises for the
4570: unwary (e.g., floating point addition is not associative) and even a few
4571: for the wary. You should not use them unless you know what you are doing
4572: or you don't care that the results you get are totally bogus. If you
4573: want to learn about the problems of floating point numbers (and how to
4574: avoid them), you might start with @cite{David Goldberg, What Every
4575: Computer Scientist Should Know About Floating-Point Arithmetic, ACM
4576: Computing Surveys 23(1):5@minus{}48, March 1991}
4577: (@uref{http://www.validgh.com/goldberg/paper.ps}).
4578:
4579:
4580: doc-d>f
4581: doc-f>d
4582: doc-f+
4583: doc-f-
4584: doc-f*
4585: doc-f/
4586: doc-fnegate
4587: doc-fabs
4588: doc-fmax
4589: doc-fmin
4590: doc-floor
4591: doc-fround
4592: doc-f**
4593: doc-fsqrt
4594: doc-fexp
4595: doc-fexpm1
4596: doc-fln
4597: doc-flnp1
4598: doc-flog
4599: doc-falog
4600: doc-f2*
4601: doc-f2/
4602: doc-1/f
4603: doc-precision
4604: doc-set-precision
4605:
4606: @cindex angles in trigonometric operations
4607: @cindex trigonometric operations
4608: Angles in floating point operations are given in radians (a full circle
4609: has 2 pi radians).
4610:
4611: doc-fsin
4612: doc-fcos
4613: doc-fsincos
4614: doc-ftan
4615: doc-fasin
4616: doc-facos
4617: doc-fatan
4618: doc-fatan2
4619: doc-fsinh
4620: doc-fcosh
4621: doc-ftanh
4622: doc-fasinh
4623: doc-facosh
4624: doc-fatanh
4625: doc-pi
4626:
4627: @cindex equality of floats
4628: @cindex floating-point comparisons
4629: One particular problem with floating-point arithmetic is that comparison
4630: for equality often fails when you would expect it to succeed. For this
4631: reason approximate equality is often preferred (but you still have to
4632: know what you are doing). The comparison words are:
4633:
4634: doc-f~rel
4635: doc-f~abs
4636: doc-f=
4637: doc-f~
4638: doc-f<>
4639:
4640: doc-f<
4641: doc-f<=
4642: doc-f>
4643: doc-f>=
4644:
4645: doc-f0<
4646: doc-f0<=
4647: doc-f0<>
4648: doc-f0=
4649: doc-f0>
4650: doc-f0>=
4651:
4652:
4653: @node Stack Manipulation, Memory, Arithmetic, Words
4654: @section Stack Manipulation
4655: @cindex stack manipulation words
4656:
4657: @cindex floating-point stack in the standard
4658: Gforth maintains a number of separate stacks:
4659:
4660: @cindex data stack
4661: @cindex parameter stack
4662: @itemize @bullet
4663: @item
4664: A data stack (also known as the @dfn{parameter stack}) -- for
4665: characters, cells, addresses, and double cells.
4666:
4667: @cindex floating-point stack
4668: @item
4669: A floating point stack -- for holding floating point (FP) numbers.
4670:
4671: @cindex return stack
4672: @item
4673: A return stack -- for holding the return addresses of colon
4674: definitions and other (non-FP) data.
4675:
4676: @cindex locals stack
4677: @item
4678: A locals stack -- for holding local variables.
4679: @end itemize
4680:
4681: @menu
4682: * Data stack::
4683: * Floating point stack::
4684: * Return stack::
4685: * Locals stack::
4686: * Stack pointer manipulation::
4687: @end menu
4688:
4689: @node Data stack, Floating point stack, Stack Manipulation, Stack Manipulation
4690: @subsection Data stack
4691: @cindex data stack manipulation words
4692: @cindex stack manipulations words, data stack
4693:
4694:
4695: doc-drop
4696: doc-nip
4697: doc-dup
4698: doc-over
4699: doc-tuck
4700: doc-swap
4701: doc-pick
4702: doc-rot
4703: doc--rot
4704: doc-?dup
4705: doc-roll
4706: doc-2drop
4707: doc-2nip
4708: doc-2dup
4709: doc-2over
4710: doc-2tuck
4711: doc-2swap
4712: doc-2rot
4713:
4714:
4715: @node Floating point stack, Return stack, Data stack, Stack Manipulation
4716: @subsection Floating point stack
4717: @cindex floating-point stack manipulation words
4718: @cindex stack manipulation words, floating-point stack
4719:
4720: Whilst every sane Forth has a separate floating-point stack, it is not
4721: strictly required; an ANS Forth system could theoretically keep
4722: floating-point numbers on the data stack. As an additional difficulty,
4723: you don't know how many cells a floating-point number takes. It is
4724: reportedly possible to write words in a way that they work also for a
4725: unified stack model, but we do not recommend trying it. Instead, just
4726: say that your program has an environmental dependency on a separate
4727: floating-point stack.
4728:
4729: doc-floating-stack
4730:
4731: doc-fdrop
4732: doc-fnip
4733: doc-fdup
4734: doc-fover
4735: doc-ftuck
4736: doc-fswap
4737: doc-fpick
4738: doc-frot
4739:
4740:
4741: @node Return stack, Locals stack, Floating point stack, Stack Manipulation
4742: @subsection Return stack
4743: @cindex return stack manipulation words
4744: @cindex stack manipulation words, return stack
4745:
4746: @cindex return stack and locals
4747: @cindex locals and return stack
4748: A Forth system is allowed to keep local variables on the
4749: return stack. This is reasonable, as local variables usually eliminate
4750: the need to use the return stack explicitly. So, if you want to produce
4751: a standard compliant program and you are using local variables in a
4752: word, forget about return stack manipulations in that word (refer to the
4753: standard document for the exact rules).
4754:
4755: doc->r
4756: doc-r>
4757: doc-r@
4758: doc-rdrop
4759: doc-2>r
4760: doc-2r>
4761: doc-2r@
4762: doc-2rdrop
4763:
4764:
4765: @node Locals stack, Stack pointer manipulation, Return stack, Stack Manipulation
4766: @subsection Locals stack
4767:
4768: Gforth uses an extra locals stack. It is described, along with the
4769: reasons for its existence, in @ref{Implementation,Implementation of locals}.
4770:
4771: @node Stack pointer manipulation, , Locals stack, Stack Manipulation
4772: @subsection Stack pointer manipulation
4773: @cindex stack pointer manipulation words
4774:
4775: @c removed s0 r0 l0 -- they are obsolete aliases for sp0 rp0 lp0
4776: doc-sp0
4777: doc-sp@
4778: doc-sp!
4779: doc-fp0
4780: doc-fp@
4781: doc-fp!
4782: doc-rp0
4783: doc-rp@
4784: doc-rp!
4785: doc-lp0
4786: doc-lp@
4787: doc-lp!
4788:
4789:
4790: @node Memory, Control Structures, Stack Manipulation, Words
4791: @section Memory
4792: @cindex memory words
4793:
4794: @menu
4795: * Memory model::
4796: * Dictionary allocation::
4797: * Heap Allocation::
4798: * Memory Access::
4799: * Address arithmetic::
4800: * Memory Blocks::
4801: @end menu
4802:
4803: @node Memory model, Dictionary allocation, Memory, Memory
4804: @subsection ANS Forth and Gforth memory models
4805:
4806: @c The ANS Forth description is a mess (e.g., is the heap part of
4807: @c the dictionary?), so let's not stick to closely with it.
4808:
4809: ANS Forth considers a Forth system as consisting of several memories, of
4810: which only @dfn{data space} is managed and accessible with the memory
4811: words. Memory not necessarily in data space includes the stacks, the
4812: code (called code space) and the headers (called name space). In Gforth
4813: everything is in data space, but the code for the primitives is usually
4814: read-only.
4815:
4816: Data space is divided into a number of areas: The (data space portion of
4817: the) dictionary@footnote{Sometimes, the term @dfn{dictionary} is used to
4818: refer to the search data structure embodied in word lists and headers,
4819: because it is used for looking up names, just as you would in a
4820: conventional dictionary.}, the heap, and a number of system-allocated
4821: buffers.
4822:
4823: In ANS Forth data space is also divided into contiguous regions. You
4824: can only use address arithmetic within a contiguous region, not between
4825: them. Usually each allocation gives you one contiguous region, but the
4826: dictionary allocation words have additional rules (@pxref{Dictionary
4827: allocation}).
4828:
4829: Gforth provides one big address space, and address arithmetic can be
4830: performed between any addresses. However, in the dictionary headers or
4831: code are interleaved with data, so almost the only contiguous data space
4832: regions there are those described by ANS Forth as contiguous; but you
4833: can be sure that the dictionary is allocated towards increasing
4834: addresses even between contiguous regions. The memory order of
4835: allocations in the heap is platform-dependent (and possibly different
4836: from one run to the next).
4837:
4838: @subsubsection ANS Forth dictionary details
4839:
4840: This section is just informative, you can skip it if you are in a hurry.
4841:
4842: When you create a colon definition, the text interpreter compiles the
4843: code for the definition into the code space and compiles the name
4844: of the definition into the header space, together with other
4845: information about the definition (such as its execution token).
4846:
4847: When you create a variable, the execution of @code{Variable} will
4848: compile some code, assign one cell in data space, and compile the name
4849: of the variable into the header space.
4850:
4851: @cindex memory regions - relationship between them
4852: ANS Forth does not specify the relationship between the three memory
4853: regions, and specifies that a Standard program must not access code or
4854: data space directly -- it may only access data space directly. In
4855: addition, the Standard defines what relationships you may and may not
4856: rely on when allocating regions in data space. These constraints are
4857: simply a reflection of the many diverse techniques that are used to
4858: implement Forth systems; understanding and following the requirements of
4859: the Standard allows you to write portable programs -- programs that run
4860: in the same way on any of these diverse systems. Another way of looking
4861: at this is to say that ANS Forth was designed to permit compliant Forth
4862: systems to be implemented in many diverse ways.
4863:
4864: @cindex memory regions - how they are assigned
4865: Here are some examples of ways in which name, code and data spaces
4866: might be assigned in different Forth implementations:
4867:
4868: @itemize @bullet
4869: @item
4870: For a Forth system that runs from RAM under a general-purpose operating
4871: system, it can be convenient to interleave name, code and data spaces in
4872: a single contiguous memory region. This organisation can be
4873: memory-efficient (for example, because the relationship between the name
4874: dictionary entry and the associated code space entry can be
4875: implicit, rather than requiring an explicit memory pointer to reference
4876: from the header space and the code space). This is the
4877: organisation used by Gforth, as this example@footnote{The addresses
4878: in the example have been truncated to fit it onto the page, and the
4879: addresses and data shown will not match the output from your system} shows:
4880: @example
4881: hex
4882: variable fred 123456 fred !
4883: variable jim abcd jim !
4884: : foo + / - ;
4885: ' fred 10 - 50 dump
4886: ..80: 5C 46 0E 40 84 66 72 65 - 64 20 20 20 20 20 20 20 \F.@.fred
4887: ..90: D0 9B 04 08 00 00 00 00 - 56 34 12 00 80 46 0E 40 ........V4...F.@
4888: ..A0: 83 6A 69 6D 20 20 20 20 - D0 9B 04 08 00 00 00 00 .jim ........
4889: ..B0: CD AB 00 00 9C 46 0E 40 - 83 66 6F 6F 20 20 20 20 .....F.@.foo
4890: ..C0: 80 9B 04 08 00 00 00 00 - E4 2E 05 08 0C 2F 05 08 ............./..
4891: @end example
4892:
4893: @item
4894: For a high-performance system running on a modern RISC processor with a
4895: modified Harvard architecture (one that has a unified main memory but
4896: separate instruction and data caches), it is desirable to separate
4897: processor instructions from processor data. This encourages a high cache
4898: density and therefore a high cache hit rate. The Forth code space
4899: is not necessarily made up entirely of processor instructions; its
4900: nature is dependent upon the Forth implementation.
4901:
4902: @item
4903: A Forth compiler that runs on a segmented 8086 processor could be
4904: designed to interleave the name, code and data spaces within a single
4905: 64Kbyte segment. A more common implementation choice is to use a
4906: separate 64Kbyte segment for each region, which provides more memory
4907: overall but provides an address map in which only the data space is
4908: accessible.
4909:
4910: @item
4911: Microprocessors exist that run Forth (or many of the primitives required
4912: to implement the Forth virtual machine efficiently) directly. On these
4913: processors, the relationship between name, code and data spaces may be
4914: imposed as a side-effect of the architecture of the processor.
4915:
4916: @item
4917: A Forth compiler that executes from ROM on an embedded system needs its
4918: data space separated from the name and code spaces so that the data
4919: space can be mapped to a RAM area.
4920:
4921: @item
4922: A Forth compiler that runs on an embedded system may have a requirement
4923: for a small memory footprint. On such a system it can be useful to
4924: separate the header space from the data and code spaces; once the
4925: application has been compiled, the header space is no longer
4926: required@footnote{more strictly speaking, most applications can be
4927: designed so that this is the case}. The header space can be deleted
4928: entirely, or could be stored in memory on a remote @i{host} system for
4929: debug and development purposes. In the latter case, the compiler running
4930: on the @i{target} system could implement a protocol across a
4931: communication link that would allow it to interrogate the header space.
4932: @end itemize
4933:
4934: @node Dictionary allocation, Heap Allocation, Memory model, Memory
4935: @subsection Dictionary allocation
4936: @cindex reserving data space
4937: @cindex data space - reserving some
4938:
4939: Dictionary allocation is a stack-oriented allocation scheme, i.e., if
4940: you want to deallocate X, you also deallocate everything
4941: allocated after X.
4942:
4943: The allocations using the words below are contiguous and grow the region
4944: towards increasing addresses. Other words that allocate dictionary
4945: memory of any kind (i.e., defining words including @code{:noname}) end
4946: the contiguous region and start a new one.
4947:
4948: In ANS Forth only @code{create}d words are guaranteed to produce an
4949: address that is the start of the following contiguous region. In
4950: particular, the cell allocated by @code{variable} is not guaranteed to
4951: be contiguous with following @code{allot}ed memory.
4952:
4953: You can deallocate memory by using @code{allot} with a negative argument
4954: (with some restrictions, see @code{allot}). For larger deallocations use
4955: @code{marker}.
4956:
4957:
4958: doc-here
4959: doc-unused
4960: doc-allot
4961: doc-c,
4962: doc-f,
4963: doc-,
4964: doc-2,
4965: @cindex user space
4966: doc-udp
4967: doc-uallot
4968:
4969: Memory accesses have to be aligned (@pxref{Address arithmetic}). So of
4970: course you should allocate memory in an aligned way, too. I.e., before
4971: allocating allocating a cell, @code{here} must be cell-aligned, etc.
4972: The words below align @code{here} if it is not already. Basically it is
4973: only already aligned for a type, if the last allocation was a multiple
4974: of the size of this type and if @code{here} was aligned for this type
4975: before.
4976:
4977: After freshly @code{create}ing a word, @code{here} is @code{align}ed in
4978: ANS Forth (@code{maxalign}ed in Gforth).
4979:
4980: doc-align
4981: doc-falign
4982: doc-sfalign
4983: doc-dfalign
4984: doc-maxalign
4985: doc-cfalign
4986:
4987:
4988: @node Heap Allocation, Memory Access, Dictionary allocation, Memory
4989: @subsection Heap allocation
4990: @cindex heap allocation
4991: @cindex dynamic allocation of memory
4992: @cindex memory-allocation word set
4993:
4994: Heap allocation supports deallocation of allocated memory in any
4995: order. Dictionary allocation is not affected by it (i.e., it does not
4996: end a contiguous region). In Gforth, these words are implemented using
4997: the standard C library calls malloc(), free() and resize().
4998:
4999: doc-allocate
5000: doc-free
5001: doc-resize
5002:
5003:
5004: @node Memory Access, Address arithmetic, Heap Allocation, Memory
5005: @subsection Memory Access
5006: @cindex memory access words
5007:
5008:
5009: doc-@
5010: doc-!
5011: doc-+!
5012: doc-c@
5013: doc-c!
5014: doc-2@
5015: doc-2!
5016: doc-f@
5017: doc-f!
5018: doc-sf@
5019: doc-sf!
5020: doc-df@
5021: doc-df!
5022:
5023: @node Address arithmetic, Memory Blocks, Memory Access, Memory
5024: @subsection Address arithmetic
5025: @cindex address arithmetic words
5026:
5027: Address arithmetic is the foundation on which data structures like
5028: arrays, records (@pxref{Structures}) and objects (@pxref{Object-oriented
5029: Forth}) are built.
5030:
5031: ANS Forth does not specify the sizes of the data types. Instead, it
5032: offers a number of words for computing sizes and doing address
5033: arithmetic. Address arithmetic is performed in terms of address units
5034: (aus); on most systems the address unit is one byte. Note that a
5035: character may have more than one au, so @code{chars} is no noop (on
5036: systems where it is a noop, it compiles to nothing).
5037:
5038: @cindex alignment of addresses for types
5039: ANS Forth also defines words for aligning addresses for specific
5040: types. Many computers require that accesses to specific data types
5041: must only occur at specific addresses; e.g., that cells may only be
5042: accessed at addresses divisible by 4. Even if a machine allows unaligned
5043: accesses, it can usually perform aligned accesses faster.
5044:
5045: For the performance-conscious: alignment operations are usually only
5046: necessary during the definition of a data structure, not during the
5047: (more frequent) accesses to it.
5048:
5049: ANS Forth defines no words for character-aligning addresses. This is not
5050: an oversight, but reflects the fact that addresses that are not
5051: char-aligned have no use in the standard and therefore will not be
5052: created.
5053:
5054: @cindex @code{CREATE} and alignment
5055: ANS Forth guarantees that addresses returned by @code{CREATE}d words
5056: are cell-aligned; in addition, Gforth guarantees that these addresses
5057: are aligned for all purposes.
5058:
5059: Note that the ANS Forth word @code{char} has nothing to do with address
5060: arithmetic.
5061:
5062:
5063: doc-chars
5064: doc-char+
5065: doc-cells
5066: doc-cell+
5067: doc-cell
5068: doc-aligned
5069: doc-floats
5070: doc-float+
5071: doc-float
5072: doc-faligned
5073: doc-sfloats
5074: doc-sfloat+
5075: doc-sfaligned
5076: doc-dfloats
5077: doc-dfloat+
5078: doc-dfaligned
5079: doc-maxaligned
5080: doc-cfaligned
5081: doc-address-unit-bits
5082:
5083:
5084: @node Memory Blocks, , Address arithmetic, Memory
5085: @subsection Memory Blocks
5086: @cindex memory block words
5087: @cindex character strings - moving and copying
5088:
5089: Memory blocks often represent character strings; @xref{String Formats}
5090: for ways of storing character strings in memory. @xref{Displaying
5091: characters and strings} for other string-processing words.
5092:
5093: Some of these words work on address units. Others work on character
5094: units (increments of @code{CHAR}), and expect a @code{CHAR}-aligned
5095: address. Choose the correct operation depending upon your data type.
5096:
5097: When copying characters between overlapping memory regions, choose
5098: carefully between @code{cmove} and @code{cmove>}.
5099:
5100: You can only use any of these words @i{portably} to access data space.
5101:
5102: @comment TODO - think the naming of the arguments is wrong for move
5103: @comment well, really it seems to be the Standard that's wrong; it
5104: @comment describes MOVE as a word that requires a CELL-aligned source
5105: @comment and destination address but a xtranfer count that need not
5106: @comment be a multiple of CELL.
5107:
5108: doc-move
5109: doc-erase
5110: doc-cmove
5111: doc-cmove>
5112: doc-fill
5113: doc-blank
5114: doc-compare
5115: doc-search
5116: doc--trailing
5117: doc-/string
5118:
5119:
5120: @comment TODO examples
5121:
5122:
5123: @node Control Structures, Defining Words, Memory, Words
5124: @section Control Structures
5125: @cindex control structures
5126:
5127: Control structures in Forth cannot be used interpretively, only in a
5128: colon definition@footnote{To be precise, they have no interpretation
5129: semantics (@pxref{Interpretation and Compilation Semantics}).}. We do
5130: not like this limitation, but have not seen a satisfying way around it
5131: yet, although many schemes have been proposed.
5132:
5133: @menu
5134: * Selection:: IF ... ELSE ... ENDIF
5135: * Simple Loops:: BEGIN ...
5136: * Counted Loops:: DO
5137: * Arbitrary control structures::
5138: * Calls and returns::
5139: * Exception Handling::
5140: @end menu
5141:
5142: @node Selection, Simple Loops, Control Structures, Control Structures
5143: @subsection Selection
5144: @cindex selection control structures
5145: @cindex control structures for selection
5146:
5147: @c what's the purpose of all these @i? Maybe we should define a macro
5148: @c so we can produce logical markup. - anton
5149:
5150: @c nac-> When I started working on the manual, a mixture of @i and @var
5151: @c were used inconsistently in code examples and \Glossary entries. These
5152: @c two behave differently in info format so I decided to standardize on @i.
5153: @c Logical markup would be better but texi isn't really upto it, and
5154: @c texi2html just ignores macros.
5155: @c nac02dec1999-> update: the latest texinfo release can spit out html
5156: @c and it handles macros, so we could do some logical markup. Unfortunately
5157: @c texinfo will not split html output, which would be a big pain if you
5158: @c wanted to put the document on the web, which would be nice.
5159:
5160: @cindex @code{IF} control structure
5161: @example
5162: @i{flag}
5163: IF
5164: @i{code}
5165: ENDIF
5166: @end example
5167: @noindent
5168:
5169: If @i{flag} is non-zero (as far as @code{IF} etc. are concerned, a cell
5170: with any bit set represents truth) @i{code} is executed.
5171:
5172: @example
5173: @i{flag}
5174: IF
5175: @i{code1}
5176: ELSE
5177: @i{code2}
5178: ENDIF
5179: @end example
5180:
5181: If @var{flag} is true, @i{code1} is executed, otherwise @i{code2} is
5182: executed.
5183:
5184: You can use @code{THEN} instead of @code{ENDIF}. Indeed, @code{THEN} is
5185: standard, and @code{ENDIF} is not, although it is quite popular. We
5186: recommend using @code{ENDIF}, because it is less confusing for people
5187: who also know other languages (and is not prone to reinforcing negative
5188: prejudices against Forth in these people). Adding @code{ENDIF} to a
5189: system that only supplies @code{THEN} is simple:
5190: @example
5191: : ENDIF POSTPONE THEN ; immediate
5192: @end example
5193:
5194: [According to @cite{Webster's New Encyclopedic Dictionary}, @dfn{then
5195: (adv.)} has the following meanings:
5196: @quotation
5197: ... 2b: following next after in order ... 3d: as a necessary consequence
5198: (if you were there, then you saw them).
5199: @end quotation
5200: Forth's @code{THEN} has the meaning 2b, whereas @code{THEN} in Pascal
5201: and many other programming languages has the meaning 3d.]
5202:
5203: Gforth also provides the words @code{?DUP-IF} and @code{?DUP-0=-IF}, so
5204: you can avoid using @code{?dup}. Using these alternatives is also more
5205: efficient than using @code{?dup}. Definitions in ANS Forth
5206: for @code{ENDIF}, @code{?DUP-IF} and @code{?DUP-0=-IF} are provided in
5207: @file{compat/control.fs}.
5208:
5209: @cindex @code{CASE} control structure
5210: @example
5211: @i{n}
5212: CASE
5213: @i{n1} OF @i{code1} ENDOF
5214: @i{n2} OF @i{code2} ENDOF
5215: @dots{}
5216: ENDCASE
5217: @end example
5218:
5219: Executes the first @i{codei}, where the @i{ni} is equal to
5220: @i{n}. A default case can be added by simply writing the code after
5221: the last @code{ENDOF}. It may use @i{n}, which is on top of the stack,
5222: but must not consume it.
5223:
5224: @node Simple Loops, Counted Loops, Selection, Control Structures
5225: @subsection Simple Loops
5226: @cindex simple loops
5227: @cindex loops without count
5228:
5229: @cindex @code{WHILE} loop
5230: @example
5231: BEGIN
5232: @i{code1}
5233: @i{flag}
5234: WHILE
5235: @i{code2}
5236: REPEAT
5237: @end example
5238:
5239: @i{code1} is executed and @i{flag} is computed. If it is true,
5240: @i{code2} is executed and the loop is restarted; If @i{flag} is
5241: false, execution continues after the @code{REPEAT}.
5242:
5243: @cindex @code{UNTIL} loop
5244: @example
5245: BEGIN
5246: @i{code}
5247: @i{flag}
5248: UNTIL
5249: @end example
5250:
5251: @i{code} is executed. The loop is restarted if @code{flag} is false.
5252:
5253: @cindex endless loop
5254: @cindex loops, endless
5255: @example
5256: BEGIN
5257: @i{code}
5258: AGAIN
5259: @end example
5260:
5261: This is an endless loop.
5262:
5263: @node Counted Loops, Arbitrary control structures, Simple Loops, Control Structures
5264: @subsection Counted Loops
5265: @cindex counted loops
5266: @cindex loops, counted
5267: @cindex @code{DO} loops
5268:
5269: The basic counted loop is:
5270: @example
5271: @i{limit} @i{start}
5272: ?DO
5273: @i{body}
5274: LOOP
5275: @end example
5276:
5277: This performs one iteration for every integer, starting from @i{start}
5278: and up to, but excluding @i{limit}. The counter, or @i{index}, can be
5279: accessed with @code{i}. For example, the loop:
5280: @example
5281: 10 0 ?DO
5282: i .
5283: LOOP
5284: @end example
5285: @noindent
5286: prints @code{0 1 2 3 4 5 6 7 8 9}
5287:
5288: The index of the innermost loop can be accessed with @code{i}, the index
5289: of the next loop with @code{j}, and the index of the third loop with
5290: @code{k}.
5291:
5292:
5293: doc-i
5294: doc-j
5295: doc-k
5296:
5297:
5298: The loop control data are kept on the return stack, so there are some
5299: restrictions on mixing return stack accesses and counted loop words. In
5300: particuler, if you put values on the return stack outside the loop, you
5301: cannot read them inside the loop@footnote{well, not in a way that is
5302: portable.}. If you put values on the return stack within a loop, you
5303: have to remove them before the end of the loop and before accessing the
5304: index of the loop.
5305:
5306: There are several variations on the counted loop:
5307:
5308: @itemize @bullet
5309: @item
5310: @code{LEAVE} leaves the innermost counted loop immediately; execution
5311: continues after the associated @code{LOOP} or @code{NEXT}. For example:
5312:
5313: @example
5314: 10 0 ?DO i DUP . 3 = IF LEAVE THEN LOOP
5315: @end example
5316: prints @code{0 1 2 3}
5317:
5318:
5319: @item
5320: @code{UNLOOP} prepares for an abnormal loop exit, e.g., via
5321: @code{EXIT}. @code{UNLOOP} removes the loop control parameters from the
5322: return stack so @code{EXIT} can get to its return address. For example:
5323:
5324: @example
5325: : demo 10 0 ?DO i DUP . 3 = IF UNLOOP EXIT THEN LOOP ." Done" ;
5326: @end example
5327: prints @code{0 1 2 3}
5328:
5329:
5330: @item
5331: If @i{start} is greater than @i{limit}, a @code{?DO} loop is entered
5332: (and @code{LOOP} iterates until they become equal by wrap-around
5333: arithmetic). This behaviour is usually not what you want. Therefore,
5334: Gforth offers @code{+DO} and @code{U+DO} (as replacements for
5335: @code{?DO}), which do not enter the loop if @i{start} is greater than
5336: @i{limit}; @code{+DO} is for signed loop parameters, @code{U+DO} for
5337: unsigned loop parameters.
5338:
5339: @item
5340: @code{?DO} can be replaced by @code{DO}. @code{DO} always enters
5341: the loop, independent of the loop parameters. Do not use @code{DO}, even
5342: if you know that the loop is entered in any case. Such knowledge tends
5343: to become invalid during maintenance of a program, and then the
5344: @code{DO} will make trouble.
5345:
5346: @item
5347: @code{LOOP} can be replaced with @code{@i{n} +LOOP}; this updates the
5348: index by @i{n} instead of by 1. The loop is terminated when the border
5349: between @i{limit-1} and @i{limit} is crossed. E.g.:
5350:
5351: @example
5352: 4 0 +DO i . 2 +LOOP
5353: @end example
5354: @noindent
5355: prints @code{0 2}
5356:
5357: @example
5358: 4 1 +DO i . 2 +LOOP
5359: @end example
5360: @noindent
5361: prints @code{1 3}
5362:
5363:
5364: @cindex negative increment for counted loops
5365: @cindex counted loops with negative increment
5366: The behaviour of @code{@i{n} +LOOP} is peculiar when @i{n} is negative:
5367:
5368: @example
5369: -1 0 ?DO i . -1 +LOOP
5370: @end example
5371: @noindent
5372: prints @code{0 -1}
5373:
5374: @example
5375: 0 0 ?DO i . -1 +LOOP
5376: @end example
5377: prints nothing.
5378:
5379: Therefore we recommend avoiding @code{@i{n} +LOOP} with negative
5380: @i{n}. One alternative is @code{@i{u} -LOOP}, which reduces the
5381: index by @i{u} each iteration. The loop is terminated when the border
5382: between @i{limit+1} and @i{limit} is crossed. Gforth also provides
5383: @code{-DO} and @code{U-DO} for down-counting loops. E.g.:
5384:
5385: @example
5386: -2 0 -DO i . 1 -LOOP
5387: @end example
5388: @noindent
5389: prints @code{0 -1}
5390:
5391: @example
5392: -1 0 -DO i . 1 -LOOP
5393: @end example
5394: @noindent
5395: prints @code{0}
5396:
5397: @example
5398: 0 0 -DO i . 1 -LOOP
5399: @end example
5400: @noindent
5401: prints nothing.
5402:
5403: @end itemize
5404:
5405: Unfortunately, @code{+DO}, @code{U+DO}, @code{-DO}, @code{U-DO} and
5406: @code{-LOOP} are not defined in ANS Forth. However, an implementation
5407: for these words that uses only standard words is provided in
5408: @file{compat/loops.fs}.
5409:
5410:
5411: @cindex @code{FOR} loops
5412: Another counted loop is:
5413: @example
5414: @i{n}
5415: FOR
5416: @i{body}
5417: NEXT
5418: @end example
5419: This is the preferred loop of native code compiler writers who are too
5420: lazy to optimize @code{?DO} loops properly. This loop structure is not
5421: defined in ANS Forth. In Gforth, this loop iterates @i{n+1} times;
5422: @code{i} produces values starting with @i{n} and ending with 0. Other
5423: Forth systems may behave differently, even if they support @code{FOR}
5424: loops. To avoid problems, don't use @code{FOR} loops.
5425:
5426: @node Arbitrary control structures, Calls and returns, Counted Loops, Control Structures
5427: @subsection Arbitrary control structures
5428: @cindex control structures, user-defined
5429:
5430: @cindex control-flow stack
5431: ANS Forth permits and supports using control structures in a non-nested
5432: way. Information about incomplete control structures is stored on the
5433: control-flow stack. This stack may be implemented on the Forth data
5434: stack, and this is what we have done in Gforth.
5435:
5436: @cindex @code{orig}, control-flow stack item
5437: @cindex @code{dest}, control-flow stack item
5438: An @i{orig} entry represents an unresolved forward branch, a @i{dest}
5439: entry represents a backward branch target. A few words are the basis for
5440: building any control structure possible (except control structures that
5441: need storage, like calls, coroutines, and backtracking).
5442:
5443:
5444: doc-if
5445: doc-ahead
5446: doc-then
5447: doc-begin
5448: doc-until
5449: doc-again
5450: doc-cs-pick
5451: doc-cs-roll
5452:
5453:
5454: The Standard words @code{CS-PICK} and @code{CS-ROLL} allow you to
5455: manipulate the control-flow stack in a portable way. Without them, you
5456: would need to know how many stack items are occupied by a control-flow
5457: entry (many systems use one cell. In Gforth they currently take three,
5458: but this may change in the future).
5459:
5460: Some standard control structure words are built from these words:
5461:
5462:
5463: doc-else
5464: doc-while
5465: doc-repeat
5466:
5467:
5468: @noindent
5469: Gforth adds some more control-structure words:
5470:
5471:
5472: doc-endif
5473: doc-?dup-if
5474: doc-?dup-0=-if
5475:
5476:
5477: @noindent
5478: Counted loop words constitute a separate group of words:
5479:
5480:
5481: doc-?do
5482: doc-+do
5483: doc-u+do
5484: doc--do
5485: doc-u-do
5486: doc-do
5487: doc-for
5488: doc-loop
5489: doc-+loop
5490: doc--loop
5491: doc-next
5492: doc-leave
5493: doc-?leave
5494: doc-unloop
5495: doc-done
5496:
5497:
5498: The standard does not allow using @code{CS-PICK} and @code{CS-ROLL} on
5499: @i{do-sys}. Gforth allows it, but it's your job to ensure that for
5500: every @code{?DO} etc. there is exactly one @code{UNLOOP} on any path
5501: through the definition (@code{LOOP} etc. compile an @code{UNLOOP} on the
5502: fall-through path). Also, you have to ensure that all @code{LEAVE}s are
5503: resolved (by using one of the loop-ending words or @code{DONE}).
5504:
5505: @noindent
5506: Another group of control structure words are:
5507:
5508:
5509: doc-case
5510: doc-endcase
5511: doc-of
5512: doc-endof
5513:
5514:
5515: @i{case-sys} and @i{of-sys} cannot be processed using @code{CS-PICK} and
5516: @code{CS-ROLL}.
5517:
5518: @subsubsection Programming Style
5519: @cindex control structures programming style
5520: @cindex programming style, arbitrary control structures
5521:
5522: In order to ensure readability we recommend that you do not create
5523: arbitrary control structures directly, but define new control structure
5524: words for the control structure you want and use these words in your
5525: program. For example, instead of writing:
5526:
5527: @example
5528: BEGIN
5529: ...
5530: IF [ 1 CS-ROLL ]
5531: ...
5532: AGAIN THEN
5533: @end example
5534:
5535: @noindent
5536: we recommend defining control structure words, e.g.,
5537:
5538: @example
5539: : WHILE ( DEST -- ORIG DEST )
5540: POSTPONE IF
5541: 1 CS-ROLL ; immediate
5542:
5543: : REPEAT ( orig dest -- )
5544: POSTPONE AGAIN
5545: POSTPONE THEN ; immediate
5546: @end example
5547:
5548: @noindent
5549: and then using these to create the control structure:
5550:
5551: @example
5552: BEGIN
5553: ...
5554: WHILE
5555: ...
5556: REPEAT
5557: @end example
5558:
5559: That's much easier to read, isn't it? Of course, @code{REPEAT} and
5560: @code{WHILE} are predefined, so in this example it would not be
5561: necessary to define them.
5562:
5563: @node Calls and returns, Exception Handling, Arbitrary control structures, Control Structures
5564: @subsection Calls and returns
5565: @cindex calling a definition
5566: @cindex returning from a definition
5567:
5568: @cindex recursive definitions
5569: A definition can be called simply be writing the name of the definition
5570: to be called. Normally a definition is invisible during its own
5571: definition. If you want to write a directly recursive definition, you
5572: can use @code{recursive} to make the current definition visible, or
5573: @code{recurse} to call the current definition directly.
5574:
5575:
5576: doc-recursive
5577: doc-recurse
5578:
5579:
5580: @comment TODO add example of the two recursion methods
5581: @quotation
5582: @progstyle
5583: I prefer using @code{recursive} to @code{recurse}, because calling the
5584: definition by name is more descriptive (if the name is well-chosen) than
5585: the somewhat cryptic @code{recurse}. E.g., in a quicksort
5586: implementation, it is much better to read (and think) ``now sort the
5587: partitions'' than to read ``now do a recursive call''.
5588: @end quotation
5589:
5590: For mutual recursion, use @code{Defer}red words, like this:
5591:
5592: @example
5593: Defer foo
5594:
5595: : bar ( ... -- ... )
5596: ... foo ... ;
5597:
5598: :noname ( ... -- ... )
5599: ... bar ... ;
5600: IS foo
5601: @end example
5602:
5603: Deferred words are discussed in more detail in @ref{Deferred words}.
5604:
5605: The current definition returns control to the calling definition when
5606: the end of the definition is reached or @code{EXIT} is encountered.
5607:
5608: doc-exit
5609: doc-;s
5610:
5611:
5612: @node Exception Handling, , Calls and returns, Control Structures
5613: @subsection Exception Handling
5614: @cindex exceptions
5615:
5616: If your program detects a fatal error condition, the simplest action
5617: that it can take is to @code{quit}. This resets the return stack and
5618: restarts the text interpreter, but does not print any error message.
5619:
5620: The next stage in severity is to execute @code{abort}, which has the
5621: same effect as @code{quit}, with the addition that it resets the data
5622: stack.
5623:
5624: A slightly more sophisticated approach is use use @code{abort"}, which
5625: compiles a string to be used as an error message and does a conditional
5626: @code{abort} at run-time. For example:
5627:
5628: @example
5629: @kbd{: checker abort" That flag was true" ." A false flag" ;@key{RET}} ok
5630: @kbd{0 checker@key{RET}} A false flag ok
5631: @kbd{1 checker@key{RET}}
5632: :1: That flag was true
5633: 1 checker
5634: ^^^^^^^
5635: $400D1648 throw
5636: $400E4660
5637: @end example
5638:
5639: These simple techniques allow a program to react to a fatal error
5640: condition, but they are not exactly user-friendly. The ANS Forth
5641: Exception word set provides the pair of words @code{throw} and
5642: @code{catch}, which can be used to provide sophisticated error-handling.
5643:
5644: @code{catch} has a similar behaviour to @code{execute}, in that it takes
5645: an @i{xt} as a parameter and starts execution of the xt. However,
5646: before passing control to the xt, @code{catch} pushes an
5647: @dfn{exception frame} onto the @dfn{exception stack}. This exception
5648: frame is used to restore the system to a known state if a detected error
5649: occurs during the execution of the xt. A typical way to use @code{catch}
5650: would be:
5651:
5652: @example
5653: ... ['] foo catch IF ...
5654: @end example
5655:
5656: @c TOS is undefined. - anton
5657:
5658: @c nac-> TODO -- I need to look at this example again.
5659:
5660: Whilst @code{foo} executes, it can call other words to any level of
5661: nesting, as usual. If @code{foo} (and all the words that it calls)
5662: execute successfully, control will ultimately pass to the word following
5663: the @code{catch}, and there will be a 0 at TOS. However, if any word
5664: detects an error, it can terminate the execution of @code{foo} by
5665: pushing a non-zero error code onto the stack and then performing a
5666: @code{throw}. The execution of @code{throw} will pass control to the
5667: word following the @code{catch}, but this time the TOS will hold the
5668: error code. Therefore, the @code{IF} in the example can be used to
5669: determine whether @code{foo} executed successfully.
5670:
5671: This simple example shows how you can use @code{throw} and @code{catch}
5672: to ``take over'' exception handling from the system:
5673: @example
5674: : my-div ['] / catch if ." DIVIDE ERROR" else ." OK.. " . then ;
5675: @end example
5676:
5677: The next example is more sophisticated and shows a multi-level
5678: @code{throw} and @code{catch}. To understand this example, start at the
5679: definition of @code{top-level} and work backwards:
5680:
5681: @example
5682: : lowest-level ( -- c )
5683: key dup 27 = if
5684: 1 throw \ ESCAPE key pressed
5685: else
5686: ." lowest-level successful" CR
5687: then
5688: ;
5689:
5690: : lower-level ( -- c )
5691: lowest-level
5692: \ at this level consider a CTRL-U to be a fatal error
5693: dup 21 = if \ CTRL-U
5694: 2 throw
5695: else
5696: ." lower-level successful" CR
5697: then
5698: ;
5699:
5700: : low-level ( -- c )
5701: ['] lower-level catch
5702: ?dup if
5703: \ error occurred - do we recognise it?
5704: dup 1 = if
5705: \ ESCAPE key pressed.. pretend it was an E
5706: [char] E
5707: else throw \ propogate the error upwards
5708: then
5709: then
5710: ." low-level successfull" CR
5711: ;
5712:
5713: : top-level ( -- )
5714: CR ['] low-level catch \ CATCH is used like EXECUTE
5715: ?dup if \ error occurred..
5716: ." Error " . ." occurred - contact your supplier"
5717: else
5718: ." The '" emit ." ' key was pressed" CR
5719: then
5720: ;
5721: @end example
5722:
5723: The ANS Forth document assigns @code{throw} codes thus:
5724:
5725: @itemize @bullet
5726: @item
5727: codes in the range -1 -- -255 are reserved to be assigned by the
5728: Standard. Assignments for codes in the range -1 -- -58 are currently
5729: documented in the Standard. In particular, @code{-1 throw} is equivalent
5730: to @code{abort} and @code{-2 throw} is equivalent to @code{abort"}.
5731: @item
5732: codes in the range -256 -- -4095 are reserved to be assigned by the system.
5733: @item
5734: all other codes may be assigned by programs.
5735: @end itemize
5736:
5737: Gforth provides the word @code{exception} as a mechanism for assigning
5738: system throw codes to applications. This allows multiple applications to
5739: co-exist in memory without any clash of @code{throw} codes. A definition
5740: of @code{exception} in ANS Forth is provided in
5741: @file{compat/exception.fs}.
5742:
5743:
5744: doc-quit
5745: doc-abort
5746: doc-abort"
5747:
5748: doc-catch
5749: doc-throw
5750: doc---exception-exception
5751:
5752:
5753:
5754: @c -------------------------------------------------------------
5755: @node Defining Words, Interpretation and Compilation Semantics, Control Structures, Words
5756: @section Defining Words
5757: @cindex defining words
5758:
5759: Defining words are used to extend Forth by creating new entries in the dictionary.
5760:
5761: @menu
5762: * CREATE::
5763: * Variables:: Variables and user variables
5764: * Constants::
5765: * Values:: Initialised variables
5766: * Colon Definitions::
5767: * Anonymous Definitions:: Definitions without names
5768: * User-defined Defining Words::
5769: * Deferred words:: Allow forward references
5770: * Aliases::
5771: * Supplying names::
5772: @end menu
5773:
5774: @node CREATE, Variables, Defining Words, Defining Words
5775: @subsection @code{CREATE}
5776: @cindex simple defining words
5777: @cindex defining words, simple
5778:
5779: Defining words are used to create new entries in the dictionary. The
5780: simplest defining word is @code{CREATE}. @code{CREATE} is used like
5781: this:
5782:
5783: @example
5784: CREATE new-word1
5785: @end example
5786:
5787: @code{CREATE} is a parsing word that generates a dictionary entry for
5788: @code{new-word1}. When @code{new-word1} is executed, all that it does is
5789: leave an address on the stack. The address represents the value of
5790: the data space pointer (@code{HERE}) at the time that @code{new-word1}
5791: was defined. Therefore, @code{CREATE} is a way of associating a name
5792: with the address of a region of memory.
5793:
5794: doc-create
5795:
5796: By extending this example to reserve some memory in data space, we end
5797: up with a @i{variable}. Here are two different ways to do it:
5798:
5799: @example
5800: CREATE new-word2 1 cells allot \ reserve 1 cell - initial value undefined
5801: CREATE new-word3 4 , \ reserve 1 cell and initialise it (to 4)
5802: @end example
5803:
5804: The variable can be examined and modified using @code{@@} (``fetch'') and
5805: @code{!} (``store'') like this:
5806:
5807: @example
5808: new-word2 @@ . \ get address, fetch from it and display
5809: 1234 new-word2 ! \ new value, get address, store to it
5810: @end example
5811:
5812: @cindex arrays
5813: A similar mechanism can be used to create arrays. For example, an
5814: 80-character text input buffer:
5815:
5816: @example
5817: CREATE text-buf 80 chars allot
5818:
5819: text-buf 0 chars c@@ \ the 1st character (offset 0)
5820: text-buf 3 chars c@@ \ the 4th character (offset 3)
5821: @end example
5822:
5823: You can build arbitrarily complex data structures by allocating
5824: appropriate areas of memory. @xref{Structures} for further discussions
5825: of this, and to learn about some Gforth tools that make it easier.
5826:
5827:
5828: @node Variables, Constants, CREATE, Defining Words
5829: @subsection Variables
5830: @cindex variables
5831:
5832: The previous section showed how a sequence of commands could be used to
5833: generate a variable. As a final refinement, the whole code sequence can
5834: be wrapped up in a defining word (pre-empting the subject of the next
5835: section), making it easier to create new variables:
5836:
5837: @example
5838: : myvariableX ( "name" -- a-addr ) CREATE 1 cells allot ;
5839: : myvariable0 ( "name" -- a-addr ) CREATE 0 , ;
5840:
5841: myvariableX foo \ variable foo starts off with an unknown value
5842: myvariable0 joe \ whilst joe is initialised to 0
5843:
5844: 45 3 * foo ! \ set foo to 135
5845: 1234 joe ! \ set joe to 1234
5846: 3 joe +! \ increment joe by 3.. to 1237
5847: @end example
5848:
5849: Not surprisingly, there is no need to define @code{myvariable}, since
5850: Forth already has a definition @code{Variable}. ANS Forth does not
5851: require a @code{Variable} to be initialised when it is created (i.e., it
5852: behaves like @code{myvariableX}). In contrast, Gforth's @code{Variable}
5853: initialises the variable to 0 (i.e., it behaves exactly like
5854: @code{myvariable0}). Forth also provides @code{2Variable} and
5855: @code{fvariable} for double and floating-point variables, respectively
5856: -- both are initialised to 0 in Gforth. If you use a @code{Variable} to
5857: store a boolean, you can use @code{on} and @code{off} to toggle its
5858: state.
5859:
5860: doc-variable
5861: doc-2variable
5862: doc-fvariable
5863:
5864: @cindex user variables
5865: @cindex user space
5866: The defining word @code{User} behaves in the same way as @code{Variable}.
5867: The difference is that it reserves space in @i{user (data) space} rather
5868: than normal data space. In a Forth system that has a multi-tasker, each
5869: task has its own set of user variables.
5870:
5871: doc-user
5872:
5873: @comment TODO is that stuff about user variables strictly correct? Is it
5874: @comment just terminal tasks that have user variables?
5875: @comment should document tasker.fs (with some examples) elsewhere
5876: @comment in this manual, then expand on user space and user variables.
5877:
5878:
5879: @node Constants, Values, Variables, Defining Words
5880: @subsection Constants
5881: @cindex constants
5882:
5883: @code{Constant} allows you to declare a fixed value and refer to it by
5884: name. For example:
5885:
5886: @example
5887: 12 Constant INCHES-PER-FOOT
5888: 3E+08 fconstant SPEED-O-LIGHT
5889: @end example
5890:
5891: A @code{Variable} can be both read and written, so its run-time
5892: behaviour is to supply an address through which its current value can be
5893: manipulated. In contrast, the value of a @code{Constant} cannot be
5894: changed once it has been declared@footnote{Well, often it can be -- but
5895: not in a Standard, portable way. It's safer to use a @code{Value} (read
5896: on).} so it's not necessary to supply the address -- it is more
5897: efficient to return the value of the constant directly. That's exactly
5898: what happens; the run-time effect of a constant is to put its value on
5899: the top of the stack (@ref{User-defined Defining Words} describes one
5900: way of implementing @code{Constant}).
5901:
5902: Gforth also provides @code{2Constant} and @code{fconstant} for defining
5903: double and floating-point constants, respectively.
5904:
5905: doc-constant
5906: doc-2constant
5907: doc-fconstant
5908:
5909: @c that's too deep, and it's not necessarily true for all ANS Forths. - anton
5910: @c nac-> How could that not be true in an ANS Forth? You can't define a
5911: @c constant, use it and then delete the definition of the constant..
5912: @c I agree that it's rather deep, but IMO it is an important difference
5913: @c relative to other programming languages.. often it's annoying: it
5914: @c certainly changes my programming style relative to C.
5915:
5916: Constants in Forth behave differently from their equivalents in other
5917: programming languages. In other languages, a constant (such as an EQU in
5918: assembler or a #define in C) only exists at compile-time; in the
5919: executable program the constant has been translated into an absolute
5920: number and, unless you are using a symbolic debugger, it's impossible to
5921: know what abstract thing that number represents. In Forth a constant has
5922: an entry in the header space and remains there after the code that uses
5923: it has been defined. In fact, it must remain in the dictionary since it
5924: has run-time duties to perform. For example:
5925:
5926: @example
5927: 12 Constant INCHES-PER-FOOT
5928: : FEET-TO-INCHES ( n1 -- n2 ) INCHES-PER-FOOT * ;
5929: @end example
5930:
5931: @cindex in-lining of constants
5932: When @code{FEET-TO-INCHES} is executed, it will in turn execute the xt
5933: associated with the constant @code{INCHES-PER-FOOT}. If you use
5934: @code{see} to decompile the definition of @code{FEET-TO-INCHES}, you can
5935: see that it makes a call to @code{INCHES-PER-FOOT}. Some Forth compilers
5936: attempt to optimise constants by in-lining them where they are used. You
5937: can force Gforth to in-line a constant like this:
5938:
5939: @example
5940: : FEET-TO-INCHES ( n1 -- n2 ) [ INCHES-PER-FOOT ] LITERAL * ;
5941: @end example
5942:
5943: If you use @code{see} to decompile @i{this} version of
5944: @code{FEET-TO-INCHES}, you can see that @code{INCHES-PER-FOOT} is no
5945: longer present. @xref{Interpret/Compile states} and @ref{Literals} on
5946: how this works.
5947:
5948: In-lining constants in this way might improve execution time
5949: fractionally, and can ensure that a constant is now only referenced at
5950: compile-time. However, the definition of the constant still remains in
5951: the dictionary. Some Forth compilers provide a mechanism for controlling
5952: a second dictionary for holding transient words such that this second
5953: dictionary can be deleted later in order to recover memory
5954: space. However, there is no standard way of doing this.
5955:
5956:
5957: @node Values, Colon Definitions, Constants, Defining Words
5958: @subsection Values
5959: @cindex values
5960:
5961: A @code{Value} is like a @code{Variable} but with two important
5962: differences:
5963:
5964: @itemize @bullet
5965: @item
5966: A @code{Value} is initialised when it is declared; like a
5967: @code{Constant} but unlike a @code{Variable}.
5968: @item
5969: A @code{Value} returns its value rather than its address when it is
5970: executed; i.e., it has the same run-time behaviour as @code{Constant}.
5971: @end itemize
5972:
5973: A @code{Value} needs an additional word, @code{TO} to allow its value to
5974: be changed. Here are some examples:
5975:
5976: @example
5977: 12 Value APPLES \ Define APPLES with an initial value of 12
5978: 34 TO APPLES \ Change the value of APPLES. TO is a parsing word
5979: APPLES \ puts 34 on the top of the stack.
5980: @end example
5981:
5982: doc-value
5983: doc-to
5984:
5985:
5986: @node Colon Definitions, Anonymous Definitions, Values, Defining Words
5987: @subsection Colon Definitions
5988: @cindex colon definitions
5989:
5990: @example
5991: : name ( ... -- ... )
5992: word1 word2 word3 ;
5993: @end example
5994:
5995: @noindent
5996: Creates a word called @code{name} that, upon execution, executes
5997: @code{word1 word2 word3}. @code{name} is a @dfn{(colon) definition}.
5998:
5999: The explanation above is somewhat superficial. @xref{Your first
6000: definition} for simple examples of colon definitions, then
6001: @xref{Interpretation and Compilation Semantics} for an in-depth
6002: discussion of some of the issues involved.
6003:
6004: doc-:
6005: doc-;
6006:
6007:
6008: @node Anonymous Definitions, User-defined Defining Words, Colon Definitions, Defining Words
6009: @subsection Anonymous Definitions
6010: @cindex colon definitions
6011: @cindex defining words without name
6012:
6013: Sometimes you want to define an @dfn{anonymous word}; a word without a
6014: name. You can do this with:
6015:
6016: doc-:noname
6017:
6018: This leaves the execution token for the word on the stack after the
6019: closing @code{;}. Here's an example in which a deferred word is
6020: initialised with an @code{xt} from an anonymous colon definition:
6021:
6022: @example
6023: Defer deferred
6024: :noname ( ... -- ... )
6025: ... ;
6026: IS deferred
6027: @end example
6028:
6029: @noindent
6030: Gforth provides an alternative way of doing this, using two separate
6031: words:
6032:
6033: doc-noname
6034: @cindex execution token of last defined word
6035: doc-lastxt
6036:
6037: @noindent
6038: The previous example can be rewritten using @code{noname} and
6039: @code{lastxt}:
6040:
6041: @example
6042: Defer deferred
6043: noname : ( ... -- ... )
6044: ... ;
6045: lastxt IS deferred
6046: @end example
6047:
6048: @noindent
6049: @code{noname} works with any defining word, not just @code{:}.
6050:
6051: @code{lastxt} also works when the last word was not defined as
6052: @code{noname}. It also has the useful property that is is valid as soon
6053: as the header for a definition has been built. Thus:
6054:
6055: @example
6056: lastxt . : foo [ lastxt . ] ; ' foo .
6057: @end example
6058:
6059: @noindent
6060: prints 3 numbers; the last two are the same.
6061:
6062:
6063: @node User-defined Defining Words, Deferred words, Anonymous Definitions, Defining Words
6064: @subsection User-defined Defining Words
6065: @cindex user-defined defining words
6066: @cindex defining words, user-defined
6067:
6068: You can create a new defining word by wrapping defining-time code around
6069: an existing defining word and putting the sequence in a colon
6070: definition. For example, suppose that you have a word @code{stats} that
6071: gathers statistics about colon definitions given the @i{xt} of the
6072: definition, and you want every colon definition in your application to
6073: make a call to @code{stats}. You can define and use a new version of
6074: @code{:} like this:
6075:
6076: @example
6077: : stats ( xt -- ) DUP ." (Gathering statistics for " . ." )"
6078: ... ; \ other code
6079:
6080: : my: : lastxt postpone literal ['] stats compile, ;
6081:
6082: my: foo + - ;
6083: @end example
6084:
6085: When @code{foo} is defined using @code{my:} these steps occur:
6086:
6087: @itemize @bullet
6088: @item
6089: @code{my:} is executed.
6090: @item
6091: The @code{:} within the definition (the one between @code{my:} and
6092: @code{lastxt}) is executed, and does just what it always does; it parses
6093: the input stream for a name, builds a dictionary header for the name
6094: @code{foo} and switches @code{state} from interpret to compile.
6095: @item
6096: The word @code{lastxt} is executed. It puts the @i{xt} for the word that is
6097: being defined -- @code{foo} -- onto the stack.
6098: @item
6099: The code that was produced by @code{postpone literal} is executed; this
6100: causes the value on the stack to be compiled as a literal in the code
6101: area of @code{foo}.
6102: @item
6103: The code @code{['] stats} compiles a literal into the definition of
6104: @code{my:}. When @code{compile,} is executed, that literal -- the
6105: execution token for @code{stats} -- is layed down in the code area of
6106: @code{foo} , following the literal@footnote{Strictly speaking, the
6107: mechanism that @code{compile,} uses to convert an @i{xt} into something
6108: in the code area is implementation-dependent. A threaded implementation
6109: might spit out the execution token directly whilst another
6110: implementation might spit out a native code sequence.}.
6111: @item
6112: At this point, the execution of @code{my:} is complete, and control
6113: returns to the text interpreter. The text interpreter is in compile
6114: state, so subsequent text @code{+ -} is compiled into the definition of
6115: @code{foo} and the @code{;} terminates the definition as always.
6116: @end itemize
6117:
6118: You can use @code{see} to decompile a word that was defined using
6119: @code{my:} and see how it is different from a normal @code{:}
6120: definition. For example:
6121:
6122: @example
6123: : bar + - ; \ like foo but using : rather than my:
6124: see bar
6125: : bar
6126: + - ;
6127: see foo
6128: : foo
6129: 107645672 stats + - ;
6130:
6131: \ use ' stats . to show that 107645672 is the xt for stats
6132: @end example
6133:
6134: You can use techniques like this to make new defining words in terms of
6135: @i{any} existing defining word.
6136:
6137:
6138: @cindex defining defining words
6139: @cindex @code{CREATE} ... @code{DOES>}
6140: If you want the words defined with your defining words to behave
6141: differently from words defined with standard defining words, you can
6142: write your defining word like this:
6143:
6144: @example
6145: : def-word ( "name" -- )
6146: CREATE @i{code1}
6147: DOES> ( ... -- ... )
6148: @i{code2} ;
6149:
6150: def-word name
6151: @end example
6152:
6153: @cindex child words
6154: This fragment defines a @dfn{defining word} @code{def-word} and then
6155: executes it. When @code{def-word} executes, it @code{CREATE}s a new
6156: word, @code{name}, and executes the code @i{code1}. The code @i{code2}
6157: is not executed at this time. The word @code{name} is sometimes called a
6158: @dfn{child} of @code{def-word}.
6159:
6160: When you execute @code{name}, the address of the body of @code{name} is
6161: put on the data stack and @i{code2} is executed (the address of the body
6162: of @code{name} is the address @code{HERE} returns immediately after the
6163: @code{CREATE}).
6164:
6165: @cindex atavism in child words
6166: You can use @code{def-word} to define a set of child words that behave
6167: differently, though atavistically; they all have a common run-time
6168: behaviour determined by @i{code2}. Typically, the @i{code1} sequence
6169: builds a data area in the body of the child word. The structure of the
6170: data is common to all children of @code{def-word}, but the data values
6171: are specific -- and private -- to each child word. When a child word is
6172: executed, the address of its private data area is passed as a parameter
6173: on TOS to be used and manipulated@footnote{It is legitimate both to read
6174: and write to this data area.} by @i{code2}.
6175:
6176: The two fragments of code that make up the defining words act (are
6177: executed) at two completely separate times:
6178:
6179: @itemize @bullet
6180: @item
6181: At @i{define time}, the defining word executes @i{code1} to generate a
6182: child word
6183: @item
6184: At @i{child execution time}, when a child word is invoked, @i{code2}
6185: is executed, using parameters (data) that are private and specific to
6186: the child word.
6187: @end itemize
6188:
6189: Another way of understanding the behaviour of @code{def-word} and
6190: @code{name} is to say that, if you make the following definitions:
6191: @example
6192: : def-word1 ( "name" -- )
6193: CREATE @i{code1} ;
6194:
6195: : action1 ( ... -- ... )
6196: @i{code2} ;
6197:
6198: def-word1 name1
6199: @end example
6200:
6201: @noindent
6202: Then using @code{name1 action1} is equivalent to using @code{name}.
6203:
6204: The classic example is that you can define @code{CONSTANT} in this way:
6205:
6206: @example
6207: : CONSTANT ( w "name" -- )
6208: CREATE ,
6209: DOES> ( -- w )
6210: @@ ;
6211: @end example
6212:
6213: @comment There is a beautiful description of how this works and what
6214: @comment it does in the Forthwrite 100th edition.. as well as an elegant
6215: @comment commentary on the Counting Fruits problem.
6216:
6217: When you create a constant with @code{5 CONSTANT five}, a set of
6218: define-time actions take place; first a new word @code{five} is created,
6219: then the value 5 is laid down in the body of @code{five} with
6220: @code{,}. When @code{five} is executed, the address of the body is put on
6221: the stack, and @code{@@} retrieves the value 5. The word @code{five} has
6222: no code of its own; it simply contains a data field and a pointer to the
6223: code that follows @code{DOES>} in its defining word. That makes words
6224: created in this way very compact.
6225:
6226: The final example in this section is intended to remind you that space
6227: reserved in @code{CREATE}d words is @i{data} space and therefore can be
6228: both read and written by a Standard program@footnote{Exercise: use this
6229: example as a starting point for your own implementation of @code{Value}
6230: and @code{TO} -- if you get stuck, investigate the behaviour of @code{'} and
6231: @code{[']}.}:
6232:
6233: @example
6234: : foo ( "name" -- )
6235: CREATE -1 ,
6236: DOES> ( -- )
6237: @@ . ;
6238:
6239: foo first-word
6240: foo second-word
6241:
6242: 123 ' first-word >BODY !
6243: @end example
6244:
6245: If @code{first-word} had been a @code{CREATE}d word, we could simply
6246: have executed it to get the address of its data field. However, since it
6247: was defined to have @code{DOES>} actions, its execution semantics are to
6248: perform those @code{DOES>} actions. To get the address of its data field
6249: it's necessary to use @code{'} to get its xt, then @code{>BODY} to
6250: translate the xt into the address of the data field. When you execute
6251: @code{first-word}, it will display @code{123}. When you execute
6252: @code{second-word} it will display @code{-1}.
6253:
6254: @cindex stack effect of @code{DOES>}-parts
6255: @cindex @code{DOES>}-parts, stack effect
6256: In the examples above the stack comment after the @code{DOES>} specifies
6257: the stack effect of the defined words, not the stack effect of the
6258: following code (the following code expects the address of the body on
6259: the top of stack, which is not reflected in the stack comment). This is
6260: the convention that I use and recommend (it clashes a bit with using
6261: locals declarations for stack effect specification, though).
6262:
6263: @subsubsection Applications of @code{CREATE..DOES>}
6264: @cindex @code{CREATE} ... @code{DOES>}, applications
6265:
6266: You may wonder how to use this feature. Here are some usage patterns:
6267:
6268: @cindex factoring similar colon definitions
6269: When you see a sequence of code occurring several times, and you can
6270: identify a meaning, you will factor it out as a colon definition. When
6271: you see similar colon definitions, you can factor them using
6272: @code{CREATE..DOES>}. E.g., an assembler usually defines several words
6273: that look very similar:
6274: @example
6275: : ori, ( reg-target reg-source n -- )
6276: 0 asm-reg-reg-imm ;
6277: : andi, ( reg-target reg-source n -- )
6278: 1 asm-reg-reg-imm ;
6279: @end example
6280:
6281: @noindent
6282: This could be factored with:
6283: @example
6284: : reg-reg-imm ( op-code -- )
6285: CREATE ,
6286: DOES> ( reg-target reg-source n -- )
6287: @@ asm-reg-reg-imm ;
6288:
6289: 0 reg-reg-imm ori,
6290: 1 reg-reg-imm andi,
6291: @end example
6292:
6293: @cindex currying
6294: Another view of @code{CREATE..DOES>} is to consider it as a crude way to
6295: supply a part of the parameters for a word (known as @dfn{currying} in
6296: the functional language community). E.g., @code{+} needs two
6297: parameters. Creating versions of @code{+} with one parameter fixed can
6298: be done like this:
6299: @example
6300: : curry+ ( n1 -- )
6301: CREATE ,
6302: DOES> ( n2 -- n1+n2 )
6303: @@ + ;
6304:
6305: 3 curry+ 3+
6306: -2 curry+ 2-
6307: @end example
6308:
6309: @subsubsection The gory details of @code{CREATE..DOES>}
6310: @cindex @code{CREATE} ... @code{DOES>}, details
6311:
6312: doc-does>
6313:
6314: @cindex @code{DOES>} in a separate definition
6315: This means that you need not use @code{CREATE} and @code{DOES>} in the
6316: same definition; you can put the @code{DOES>}-part in a separate
6317: definition. This allows us to, e.g., select among different @code{DOES>}-parts:
6318: @example
6319: : does1
6320: DOES> ( ... -- ... )
6321: ... ;
6322:
6323: : does2
6324: DOES> ( ... -- ... )
6325: ... ;
6326:
6327: : def-word ( ... -- ... )
6328: create ...
6329: IF
6330: does1
6331: ELSE
6332: does2
6333: ENDIF ;
6334: @end example
6335:
6336: In this example, the selection of whether to use @code{does1} or
6337: @code{does2} is made at compile-time; at the time that the child word is
6338: @code{CREATE}d.
6339:
6340: @cindex @code{DOES>} in interpretation state
6341: In a standard program you can apply a @code{DOES>}-part only if the last
6342: word was defined with @code{CREATE}. In Gforth, the @code{DOES>}-part
6343: will override the behaviour of the last word defined in any case. In a
6344: standard program, you can use @code{DOES>} only in a colon
6345: definition. In Gforth, you can also use it in interpretation state, in a
6346: kind of one-shot mode; for example:
6347: @example
6348: CREATE name ( ... -- ... )
6349: @i{initialization}
6350: DOES>
6351: @i{code} ;
6352: @end example
6353:
6354: @noindent
6355: is equivalent to the standard:
6356: @example
6357: :noname
6358: DOES>
6359: @i{code} ;
6360: CREATE name EXECUTE ( ... -- ... )
6361: @i{initialization}
6362: @end example
6363:
6364:
6365: doc->body
6366:
6367:
6368: @node Deferred words, Aliases, User-defined Defining Words, Defining Words
6369: @subsection Deferred words
6370: @cindex deferred words
6371:
6372: The defining word @code{Defer} allows you to define a word by name
6373: without defining its behaviour; the definition of its behaviour is
6374: deferred. Here are two situation where this can be useful:
6375:
6376: @itemize @bullet
6377: @item
6378: Where you want to allow the behaviour of a word to be altered later, and
6379: for all precompiled references to the word to change when its behaviour
6380: is changed.
6381: @item
6382: For mutual recursion; @xref{Calls and returns}.
6383: @end itemize
6384:
6385: In the following example, @code{foo} always invokes the version of
6386: @code{greet} that prints ``@code{Good morning}'' whilst @code{bar}
6387: always invokes the version that prints ``@code{Hello}''. There is no way
6388: of getting @code{foo} to use the later version without re-ordering the
6389: source code and recompiling it.
6390:
6391: @example
6392: : greet ." Good morning" ;
6393: : foo ... greet ... ;
6394: : greet ." Hello" ;
6395: : bar ... greet ... ;
6396: @end example
6397:
6398: This problem can be solved by defining @code{greet} as a @code{Defer}red
6399: word. The behaviour of a @code{Defer}red word can be defined and
6400: redefined at any time by using @code{IS} to associate the xt of a
6401: previously-defined word with it. The previous example becomes:
6402:
6403: @example
6404: Defer greet
6405: : foo ... greet ... ;
6406: : bar ... greet ... ;
6407: : greet1 ." Good morning" ;
6408: : greet2 ." Hello" ;
6409: ' greet2 <IS> greet \ make greet behave like greet2
6410: @end example
6411:
6412: A deferred word can be used to improve the statistics-gathering example
6413: from @ref{User-defined Defining Words}; rather than edit the
6414: application's source code to change every @code{:} to a @code{my:}, do
6415: this:
6416:
6417: @example
6418: : real: : ; \ retain access to the original
6419: defer : \ redefine as a deferred word
6420: ' my: IS : \ use special version of :
6421: \
6422: \ load application here
6423: \
6424: ' real: IS : \ go back to the original
6425: @end example
6426:
6427:
6428: One thing to note is that @code{<IS>} consumes its name when it is
6429: executed. If you want to specify the name at compile time, use
6430: @code{[IS]}:
6431:
6432: @example
6433: : set-greet ( xt -- )
6434: [IS] greet ;
6435:
6436: ' greet1 set-greet
6437: @end example
6438:
6439: A deferred word can only inherit default semantics from the xt (because
6440: that is all that an xt can represent -- @pxref{Tokens for Words} for
6441: more discussion of this). However, the semantics of the deferred word
6442: itself can be modified at the time that it is defined. For example:
6443:
6444: @example
6445: : bar .... ; compile-only
6446: Defer fred immediate
6447: Defer jim
6448:
6449: ' bar <IS> jim \ jim has default semantics
6450: ' bar <IS> fred \ fred is immediate
6451: @end example
6452:
6453: doc-defer
6454: doc-<is>
6455: doc-[is]
6456: doc-is
6457: @comment TODO document these: what's defers [is]
6458: doc-what's
6459: doc-defers
6460:
6461: @c Use @code{words-deferred} to see a list of deferred words.
6462:
6463: Definitions in ANS Forth for @code{defer}, @code{<is>} and @code{[is]}
6464: are provided in @file{compat/defer.fs}.
6465:
6466:
6467: @node Aliases, Supplying names, Deferred words, Defining Words
6468: @subsection Aliases
6469: @cindex aliases
6470:
6471: The defining word @code{Alias} allows you to define a word by name that
6472: has the same behaviour as some other word. Here are two situation where
6473: this can be useful:
6474:
6475: @itemize @bullet
6476: @item
6477: When you want access to a word's definition from a different word list
6478: (for an example of this, see the definition of the @code{Root} word list
6479: in the Gforth source).
6480: @item
6481: When you want to create a synonym; a definition that can be known by
6482: either of two names (for example, @code{THEN} and @code{ENDIF} are
6483: aliases).
6484: @end itemize
6485:
6486: The word whose behaviour the alias is to inherit is represented by an
6487: xt. Therefore, the alias only inherits default semantics from its
6488: ancestor. The semantics of the alias itself can be modified at the time
6489: that it is defined. For example:
6490:
6491: @example
6492: : foo ... ; immediate
6493:
6494: ' foo Alias bar \ bar is not an immediate word
6495: ' foo Alias fooby immediate \ fooby is an immediate word
6496: @end example
6497:
6498: Words that are aliases have the same xt, different headers in the
6499: dictionary, and consequently different name tokens (@pxref{Tokens for
6500: Words}) and possibly different immediate flags. An alias can only have
6501: default or immediate compilation semantics; you can define aliases for
6502: combined words with @code{interpret/compile:} -- see @ref{Combined words}.
6503:
6504: doc-alias
6505:
6506:
6507: @node Supplying names, , Aliases, Defining Words
6508: @subsection Supplying the name of a defined word
6509: @cindex names for defined words
6510: @cindex defining words, name given in a string
6511:
6512: By default, a defining word takes the name for the defined word from the
6513: input stream. Sometimes you want to supply the name from a string. You
6514: can do this with:
6515:
6516: doc-nextname
6517:
6518: For example:
6519:
6520: @example
6521: s" foo" nextname create
6522: @end example
6523:
6524: @noindent
6525: is equivalent to:
6526:
6527: @example
6528: create foo
6529: @end example
6530:
6531: @noindent
6532: @code{nextname} works with any defining word, not just @code{:}.
6533:
6534:
6535: @node Interpretation and Compilation Semantics, Tokens for Words, Defining Words, Words
6536: @section Interpretation and Compilation Semantics
6537: @cindex semantics, interpretation and compilation
6538:
6539: @cindex interpretation semantics
6540: The @dfn{interpretation semantics} of a word are what the text
6541: interpreter does when it encounters the word in interpret state. It also
6542: appears in some other contexts, e.g., the execution token returned by
6543: @code{' @i{word}} identifies the interpretation semantics of
6544: @i{word} (in other words, @code{' @i{word} execute} is equivalent to
6545: interpret-state text interpretation of @code{@i{word}}).
6546:
6547: @cindex compilation semantics
6548: The @dfn{compilation semantics} of a word are what the text interpreter
6549: does when it encounters the word in compile state. It also appears in
6550: other contexts, e.g, @code{POSTPONE @i{word}} compiles@footnote{In
6551: standard terminology, ``appends to the current definition''.} the
6552: compilation semantics of @i{word}.
6553:
6554: @cindex execution semantics
6555: The standard also talks about @dfn{execution semantics}. They are used
6556: only for defining the interpretation and compilation semantics of many
6557: words. By default, the interpretation semantics of a word are to
6558: @code{execute} its execution semantics, and the compilation semantics of
6559: a word are to @code{compile,} its execution semantics.@footnote{In
6560: standard terminology: The default interpretation semantics are its
6561: execution semantics; the default compilation semantics are to append its
6562: execution semantics to the execution semantics of the current
6563: definition.}
6564:
6565: @comment TODO expand, make it co-operate with new sections on text interpreter.
6566:
6567: @cindex immediate words
6568: @cindex compile-only words
6569: You can change the semantics of the most-recently defined word:
6570:
6571:
6572: doc-immediate
6573: doc-compile-only
6574: doc-restrict
6575:
6576:
6577: Note that ticking (@code{'}) a compile-only word gives an error
6578: (``Interpreting a compile-only word'').
6579:
6580: @menu
6581: * Combined words::
6582: @end menu
6583:
6584: @node Combined words, , Interpretation and Compilation Semantics, Interpretation and Compilation Semantics
6585: @subsection Combined Words
6586: @cindex combined words
6587:
6588: Gforth allows you to define @dfn{combined words} -- words that have an
6589: arbitrary combination of interpretation and compilation semantics.
6590:
6591:
6592: doc-interpret/compile:
6593:
6594:
6595: This feature was introduced for implementing @code{TO} and @code{S"}. I
6596: recommend that you do not define such words, as cute as they may be:
6597: they make it hard to get at both parts of the word in some contexts.
6598: E.g., assume you want to get an execution token for the compilation
6599: part. Instead, define two words, one that embodies the interpretation
6600: part, and one that embodies the compilation part. Once you have done
6601: that, you can define a combined word with @code{interpret/compile:} for
6602: the convenience of your users.
6603:
6604: You might try to use this feature to provide an optimizing
6605: implementation of the default compilation semantics of a word. For
6606: example, by defining:
6607: @example
6608: :noname
6609: foo bar ;
6610: :noname
6611: POSTPONE foo POSTPONE bar ;
6612: interpret/compile: opti-foobar
6613: @end example
6614:
6615: @noindent
6616: as an optimizing version of:
6617:
6618: @example
6619: : foobar
6620: foo bar ;
6621: @end example
6622:
6623: Unfortunately, this does not work correctly with @code{[compile]},
6624: because @code{[compile]} assumes that the compilation semantics of all
6625: @code{interpret/compile:} words are non-default. I.e., @code{[compile]
6626: opti-foobar} would compile compilation semantics, whereas
6627: @code{[compile] foobar} would compile interpretation semantics.
6628:
6629: @cindex state-smart words (are a bad idea)
6630: Some people try to use @dfn{state-smart} words to emulate the feature provided
6631: by @code{interpret/compile:} (words are state-smart if they check
6632: @code{STATE} during execution). E.g., they would try to code
6633: @code{foobar} like this:
6634:
6635: @example
6636: : foobar
6637: STATE @@
6638: IF ( compilation state )
6639: POSTPONE foo POSTPONE bar
6640: ELSE
6641: foo bar
6642: ENDIF ; immediate
6643: @end example
6644:
6645: Although this works if @code{foobar} is only processed by the text
6646: interpreter, it does not work in other contexts (like @code{'} or
6647: @code{POSTPONE}). E.g., @code{' foobar} will produce an execution token
6648: for a state-smart word, not for the interpretation semantics of the
6649: original @code{foobar}; when you execute this execution token (directly
6650: with @code{EXECUTE} or indirectly through @code{COMPILE,}) in compile
6651: state, the result will not be what you expected (i.e., it will not
6652: perform @code{foo bar}). State-smart words are a bad idea. Simply don't
6653: write them@footnote{For a more detailed discussion of this topic, see
6654: @cite{@code{State}-smartness -- Why it is Evil and How to Exorcise it} by Anton
6655: Ertl; presented at EuroForth '98 and available from
6656: @*@uref{http://www.complang.tuwien.ac.at/papers/ertl98.ps.gz}}!
6657:
6658: @cindex defining words with arbitrary semantics combinations
6659: It is also possible to write defining words that define words with
6660: arbitrary combinations of interpretation and compilation semantics. In
6661: general, they look like this:
6662:
6663: @example
6664: : def-word
6665: create-interpret/compile
6666: @i{code1}
6667: interpretation>
6668: @i{code2}
6669: <interpretation
6670: compilation>
6671: @i{code3}
6672: <compilation ;
6673: @end example
6674:
6675: For a @i{word} defined with @code{def-word}, the interpretation
6676: semantics are to push the address of the body of @i{word} and perform
6677: @i{code2}, and the compilation semantics are to push the address of
6678: the body of @i{word} and perform @i{code3}. E.g., @code{constant}
6679: can also be defined like this (except that the defined constants don't
6680: behave correctly when @code{[compile]}d):
6681:
6682: @example
6683: : constant ( n "name" -- )
6684: create-interpret/compile
6685: ,
6686: interpretation> ( -- n )
6687: @@
6688: <interpretation
6689: compilation> ( compilation. -- ; run-time. -- n )
6690: @@ postpone literal
6691: <compilation ;
6692: @end example
6693:
6694:
6695: doc-create-interpret/compile
6696: doc-interpretation>
6697: doc-<interpretation
6698: doc-compilation>
6699: doc-<compilation
6700:
6701:
6702: Words defined with @code{interpret/compile:} and
6703: @code{create-interpret/compile} have an extended header structure that
6704: differs from other words; however, unless you try to access them with
6705: plain address arithmetic, you should not notice this. Words for
6706: accessing the header structure usually know how to deal with this; e.g.,
6707: @code{'} @i{word} @code{>body} also gives you the body of a word created
6708: with @code{create-interpret/compile}.
6709:
6710:
6711: doc-postpone
6712:
6713: @comment TODO -- expand glossary text for POSTPONE
6714:
6715:
6716: @c -------------------------------------------------------------
6717: @node Tokens for Words, The Text Interpreter, Interpretation and Compilation Semantics, Words
6718: @section Tokens for Words
6719: @cindex tokens for words
6720:
6721: This section describes the creation and use of tokens that represent
6722: words.
6723:
6724: Named words have information stored in their header space entries to
6725: indicate any non-default semantics (@pxref{Interpretation and
6726: Compilation Semantics}). The semantics can be modified, using
6727: @code{immediate} and/or @code{compile-only}, at the time that the words
6728: are defined. Unnamed words have (by definition) no header space
6729: entry, and therefore must have default semantics.
6730:
6731: Named words have interpretation and compilation semantics. Unnamed words
6732: just have execution semantics.
6733:
6734: @cindex xt
6735: @cindex execution token
6736: The execution semantics of an unnamed word are represented by an
6737: @dfn{execution token} (@i{xt}). As explained in @ref{Supplying names},
6738: the execution token of the last word defined can be produced with
6739: @code{lastxt}.
6740:
6741: The interpretation semantics of a named word are also represented by an
6742: execution token. You can produce the execution token using @code{'} or
6743: @code{[']}. A simple example shows the difference between the two:
6744:
6745: @example
6746: : greet ( -- ) ." Hello" ;
6747: : foo ( -- xt ) ['] greet execute ; \ ['] parses greet at compile-time
6748: : bar ( -- ) ' execute ; \ ' parses at run-time
6749:
6750: \ the next four lines all do the same thing
6751: foo
6752: bar greet
6753: greet
6754: ' greet EXECUTE
6755: @end example
6756:
6757: An execution token occupies one cell.
6758: @cindex code field address
6759: @cindex CFA
6760: In Gforth, the abstract data type @i{execution token} is implemented
6761: as a code field address (CFA).
6762: @comment TODO note that the standard does not say what it represents..
6763: @comment and you cannot necessarily compile it in all Forths (eg native
6764: @comment compilers?).
6765:
6766: For literals, use @code{'} in interpreted code and @code{[']} in
6767: compiled code. Gforth's @code{'} and @code{[']} behave somewhat
6768: unusually by complaining about compile-only words. To get the execution
6769: token for a compile-only word @i{name}, use @code{COMP' @i{name} DROP}
6770: or @code{[COMP'] @i{name} DROP}.
6771:
6772: @cindex compilation token
6773: The compilation semantics of a named word are represented by a
6774: @dfn{compilation token} consisting of two cells: @i{w xt}. The top cell
6775: @i{xt} is an execution token. The compilation semantics represented by
6776: the compilation token can be performed with @code{execute}, which
6777: consumes the whole compilation token, with an additional stack effect
6778: determined by the represented compilation semantics.
6779:
6780: At present, the @i{w} part of a compilation token is an execution token,
6781: and the @i{xt} part represents either @code{execute} or
6782: @code{compile,}@footnote{Depending upon the compilation semantics of the
6783: word. If the word has default compilation semantics, the @i{xt} will
6784: represent @code{compile,}. Otherwise (e.g., for immediate words), the
6785: @i{xt} will represent @code{execute}.}. However, don't rely on that
6786: knowledge, unless necessary; future versions of Gforth may introduce
6787: unusual compilation tokens (e.g., a compilation token that represents
6788: the compilation semantics of a literal).
6789:
6790: You can compile the compilation semantics with @code{postpone,}. I.e.,
6791: @code{COMP' @i{word} postpone,} is equivalent to @code{postpone
6792: @i{word}}.
6793:
6794: @cindex name token
6795: @cindex name field address
6796: @cindex NFA
6797: Named words are also represented by the @dfn{name token}, (@i{nt}). In
6798: Gforth, the abstract data type @emph{name token} is implemented as a
6799: name field address (NFA).
6800:
6801:
6802: doc-execute
6803: doc-perform
6804: doc-compile,
6805: doc-[']
6806: doc-'
6807: doc-[comp']
6808: doc-comp'
6809: doc-postpone,
6810:
6811: doc-find-name
6812: doc-name>int
6813: doc-name?int
6814: doc-name>comp
6815: doc-name>string
6816:
6817:
6818: @c ----------------------------------------------------------
6819: @node The Text Interpreter, Word Lists, Tokens for Words, Words
6820: @section The Text Interpreter
6821: @cindex interpreter - outer
6822: @cindex text interpreter
6823: @cindex outer interpreter
6824:
6825: @c Should we really describe all these ugly details? IMO the text
6826: @c interpreter should be much cleaner, but that may not be possible within
6827: @c ANS Forth. - anton
6828: @c nac-> I wanted to explain how it works to show how you can exploit
6829: @c it in your own programs. When I was writing a cross-compiler, figuring out
6830: @c some of these gory details was very helpful to me. None of the textbooks
6831: @c I've seen cover it, and the most modern Forth textbook -- Forth Inc's,
6832: @c seems to positively avoid going into too much detail for some of
6833: @c the internals.
6834:
6835: The text interpreter@footnote{This is an expanded version of the
6836: material in @ref{Introducing the Text Interpreter}.} is an endless loop
6837: that processes input from the current input device. It is also called
6838: the outer interpreter, in contrast to the inner interpreter
6839: (@pxref{Engine}) which executes the compiled Forth code on interpretive
6840: implementations.
6841:
6842: @cindex interpret state
6843: @cindex compile state
6844: The text interpreter operates in one of two states: @dfn{interpret
6845: state} and @dfn{compile state}. The current state is defined by the
6846: aptly-named variable, @code{state}.
6847:
6848: This section starts by describing how the text interpreter behaves when
6849: it is in interpret state, processing input from the user input device --
6850: the keyboard. This is the mode that a Forth system is in after it starts
6851: up.
6852:
6853: @cindex input buffer
6854: @cindex terminal input buffer
6855: The text interpreter works from an area of memory called the @dfn{input
6856: buffer}@footnote{When the text interpreter is processing input from the
6857: keyboard, this area of memory is called the @dfn{terminal input buffer}
6858: (TIB) and is addressed by the (obsolescent) words @code{TIB} and
6859: @code{#TIB}.}, which stores your keyboard input when you press the
6860: @key{RET} key. Starting at the beginning of the input buffer, it skips
6861: leading spaces (called @dfn{delimiters}) then parses a string (a
6862: sequence of non-space characters) until it reaches either a space
6863: character or the end of the buffer. Having parsed a string, it makes two
6864: attempts to process it:
6865:
6866: @cindex dictionary
6867: @itemize @bullet
6868: @item
6869: It looks for the string in a @dfn{dictionary} of definitions. If the
6870: string is found, the string names a @dfn{definition} (also known as a
6871: @dfn{word}) and the dictionary search returns information that allows
6872: the text interpreter to perform the word's @dfn{interpretation
6873: semantics}. In most cases, this simply means that the word will be
6874: executed.
6875: @item
6876: If the string is not found in the dictionary, the text interpreter
6877: attempts to treat it as a number, using the rules described in
6878: @ref{Number Conversion}. If the string represents a legal number in the
6879: current radix, the number is pushed onto a parameter stack (the data
6880: stack for integers, the floating-point stack for floating-point
6881: numbers).
6882: @end itemize
6883:
6884: If both attempts fail, or if the word is found in the dictionary but has
6885: no interpretation semantics@footnote{This happens if the word was
6886: defined as @code{COMPILE-ONLY}.} the text interpreter discards the
6887: remainder of the input buffer, issues an error message and waits for
6888: more input. If one of the attempts succeeds, the text interpreter
6889: repeats the parsing process until the whole of the input buffer has been
6890: processed, at which point it prints the status message ``@code{ ok}''
6891: and waits for more input.
6892:
6893: @cindex parse area
6894: The text interpreter keeps track of its position in the input buffer by
6895: updating a variable called @code{>IN} (pronounced ``to-in''). The value
6896: of @code{>IN} starts out as 0, indicating an offset of 0 from the start
6897: of the input buffer. The region from offset @code{>IN @@} to the end of
6898: the input buffer is called the @dfn{parse area}@footnote{In other words,
6899: the text interpreter processes the contents of the input buffer by
6900: parsing strings from the parse area until the parse area is empty.}.
6901: This example shows how @code{>IN} changes as the text interpreter parses
6902: the input buffer:
6903:
6904: @example
6905: : remaining >IN @@ SOURCE 2 PICK - -ROT + SWAP
6906: CR ." ->" TYPE ." <-" ; IMMEDIATE
6907:
6908: 1 2 3 remaining + remaining .
6909:
6910: : foo 1 2 3 remaining SWAP remaining ;
6911: @end example
6912:
6913: @noindent
6914: The result is:
6915:
6916: @example
6917: ->+ remaining .<-
6918: ->.<-5 ok
6919:
6920: ->SWAP remaining ;-<
6921: ->;<- ok
6922: @end example
6923:
6924: @cindex parsing words
6925: The value of @code{>IN} can also be modified by a word in the input
6926: buffer that is executed by the text interpreter. This means that a word
6927: can ``trick'' the text interpreter into either skipping a section of the
6928: input buffer@footnote{This is how parsing words work.} or into parsing a
6929: section twice. For example:
6930:
6931: @example
6932: : lat ." <<lat>>" ;
6933: : flat ." <<flat>>" >IN DUP @@ 3 - SWAP ! ;
6934: @end example
6935:
6936: @noindent
6937: When @code{flat} is executed, this output is produced@footnote{Exercise
6938: for the reader: what would happen if the @code{3} were replaced with
6939: @code{4}?}:
6940:
6941: @example
6942: <<flat>><<lat>>
6943: @end example
6944:
6945: @noindent
6946: Two important notes about the behaviour of the text interpreter:
6947:
6948: @itemize @bullet
6949: @item
6950: It processes each input string to completion before parsing additional
6951: characters from the input buffer.
6952: @item
6953: It treats the input buffer as a read-only region (and so must your code).
6954: @end itemize
6955:
6956: @noindent
6957: When the text interpreter is in compile state, its behaviour changes in
6958: these ways:
6959:
6960: @itemize @bullet
6961: @item
6962: If a parsed string is found in the dictionary, the text interpreter will
6963: perform the word's @dfn{compilation semantics}. In most cases, this
6964: simply means that the execution semantics of the word will be appended
6965: to the current definition.
6966: @item
6967: When a number is encountered, it is compiled into the current definition
6968: (as a literal) rather than being pushed onto a parameter stack.
6969: @item
6970: If an error occurs, @code{state} is modified to put the text interpreter
6971: back into interpret state.
6972: @item
6973: Each time a line is entered from the keyboard, Gforth prints
6974: ``@code{ compiled}'' rather than `` @code{ok}''.
6975: @end itemize
6976:
6977: @cindex text interpreter - input sources
6978: When the text interpreter is using an input device other than the
6979: keyboard, its behaviour changes in these ways:
6980:
6981: @itemize @bullet
6982: @item
6983: When the parse area is empty, the text interpreter attempts to refill
6984: the input buffer from the input source. When the input source is
6985: exhausted, the input source is set back to the user input device.
6986: @item
6987: It doesn't print out ``@code{ ok}'' or ``@code{ compiled}'' messages each
6988: time the parse area is emptied.
6989: @item
6990: If an error occurs, the input source is set back to the user input
6991: device.
6992: @end itemize
6993:
6994: @ref{Input Sources} describes this in more detail.
6995:
6996:
6997: doc->in
6998: doc-source
6999:
7000: doc-tib
7001: doc-#tib
7002:
7003:
7004: @menu
7005: * Input Sources::
7006: * Number Conversion::
7007: * Interpret/Compile states::
7008: * Literals::
7009: * Interpreter Directives::
7010: @end menu
7011:
7012: @node Input Sources, Number Conversion, The Text Interpreter, The Text Interpreter
7013: @subsection Input Sources
7014: @cindex input sources
7015: @cindex text interpreter - input sources
7016:
7017: By default, the text interpreter processes input from the user input
7018: device (the keyboard) when Forth starts up. The text interpreter can
7019: process input from any of these sources:
7020:
7021: @itemize @bullet
7022: @item
7023: The user input device -- the keyboard.
7024: @item
7025: A file, using the words described in @ref{Forth source files}.
7026: @item
7027: A block, using the words described in @ref{Blocks}.
7028: @item
7029: A text string, using @code{evaluate}.
7030: @end itemize
7031:
7032: A program can identify the current input device from the values of
7033: @code{source-id} and @code{blk}.
7034:
7035:
7036: doc-source-id
7037: doc-blk
7038:
7039: doc-save-input
7040: doc-restore-input
7041:
7042: doc-evaluate
7043:
7044:
7045:
7046: @node Number Conversion, Interpret/Compile states, Input Sources, The Text Interpreter
7047: @subsection Number Conversion
7048: @cindex number conversion
7049: @cindex double-cell numbers, input format
7050: @cindex input format for double-cell numbers
7051: @cindex single-cell numbers, input format
7052: @cindex input format for single-cell numbers
7053: @cindex floating-point numbers, input format
7054: @cindex input format for floating-point numbers
7055:
7056: This section describes the rules that the text interpreter uses when it
7057: tries to convert a string into a number.
7058:
7059: Let <digit> represent any character that is a legal digit in the current
7060: number base@footnote{For example, 0-9 when the number base is decimal or
7061: 0-9, A-F when the number base is hexadecimal.}.
7062:
7063: Let <decimal digit> represent any character in the range 0-9.
7064:
7065: Let @{@i{a b}@} represent the @i{optional} presence of any of the characters
7066: in the braces (@i{a} or @i{b} or neither).
7067:
7068: Let * represent any number of instances of the previous character
7069: (including none).
7070:
7071: Let any other character represent itself.
7072:
7073: @noindent
7074: Now, the conversion rules are:
7075:
7076: @itemize @bullet
7077: @item
7078: A string of the form <digit><digit>* is treated as a single-precision
7079: (cell-sized) positive integer. Examples are 0 123 6784532 32343212343456 42
7080: @item
7081: A string of the form -<digit><digit>* is treated as a single-precision
7082: (cell-sized) negative integer, and is represented using 2's-complement
7083: arithmetic. Examples are -45 -5681 -0
7084: @item
7085: A string of the form <digit><digit>*.<digit>* is treated as a double-precision
7086: (double-cell-sized) positive integer. Examples are 3465. 3.465 34.65
7087: (all three of these represent the same number).
7088: @item
7089: A string of the form -<digit><digit>*.<digit>* is treated as a
7090: double-precision (double-cell-sized) negative integer, and is
7091: represented using 2's-complement arithmetic. Examples are -3465. -3.465
7092: -34.65 (all three of these represent the same number).
7093: @item
7094: A string of the form @{+ -@}<decimal digit>@{.@}<decimal digit>*@{e
7095: E@}@{+ -@}<decimal digit><decimal digit>* is treated as a floating-point
7096: number. Examples are 1e 1e0 1.e 1.e0 +1e+0 (which all represent the same
7097: number) +12.E-4
7098: @end itemize
7099:
7100: By default, the number base used for integer number conversion is given
7101: by the contents of the variable @code{base}. Note that a lot of
7102: confusion can result from unexpected values of @code{base}. If you
7103: change @code{base} anywhere, make sure to save the old value and restore
7104: it afterwards. In general I recommend keeping @code{base} decimal, and
7105: using the prefixes described below for the popular non-decimal bases.
7106:
7107: doc-dpl
7108: doc-base
7109: doc-hex
7110: doc-decimal
7111:
7112:
7113: @cindex '-prefix for character strings
7114: @cindex &-prefix for decimal numbers
7115: @cindex %-prefix for binary numbers
7116: @cindex $-prefix for hexadecimal numbers
7117: Gforth allows you to override the value of @code{base} by using a
7118: prefix@footnote{Some Forth implementations provide a similar scheme by
7119: implementing @code{$} etc. as parsing words that process the subsequent
7120: number in the input stream and push it onto the stack. For example, see
7121: @cite{Number Conversion and Literals}, by Wil Baden; Forth Dimensions
7122: 20(3) pages 26--27. In such implementations, unlike in Gforth, a space
7123: is required between the prefix and the number.} before the first digit
7124: of an (integer) number. Four prefixes are supported:
7125:
7126: @itemize @bullet
7127: @item
7128: @code{&} -- decimal
7129: @item
7130: @code{%} -- binary
7131: @item
7132: @code{$} -- hexadecimal
7133: @item
7134: @code{'} -- base @code{max-char+1}
7135: @end itemize
7136:
7137: Here are some examples, with the equivalent decimal number shown after
7138: in braces:
7139:
7140: -$41 (-65), %1001101 (205), %1001.0001 (145 - a double-precision number),
7141: 'AB (16706; ascii A is 65, ascii B is 66, number is 65*256 + 66),
7142: 'ab (24930; ascii a is 97, ascii B is 98, number is 97*256 + 98),
7143: &905 (905), $abc (2478), $ABC (2478).
7144:
7145: @cindex number conversion - traps for the unwary
7146: @noindent
7147: Number conversion has a number of traps for the unwary:
7148:
7149: @itemize @bullet
7150: @item
7151: You cannot determine the current number base using the code sequence
7152: @code{base @@ .} -- the number base is always 10 in the current number
7153: base. Instead, use something like @code{base @@ dec.}
7154: @item
7155: If the number base is set to a value greater than 14 (for example,
7156: hexadecimal), the number 123E4 is ambiguous; the conversion rules allow
7157: it to be intepreted as either a single-precision integer or a
7158: floating-point number (Gforth treats it as an integer). The ambiguity
7159: can be resolved by explicitly stating the sign of the mantissa and/or
7160: exponent: 123E+4 or +123E4 -- if the number base is decimal, no
7161: ambiguity arises; either representation will be treated as a
7162: floating-point number.
7163: @item
7164: There is a word @code{bin} but it does @i{not} set the number base!
7165: It is used to specify file types.
7166: @item
7167: ANS Forth requires the @code{.} of a double-precision number to
7168: be the final character in the string. Allowing the @code{.} to be
7169: anywhere after the first digit is a Gforth extension.
7170: @item
7171: The number conversion process does not check for overflow.
7172: @item
7173: In Gforth, number conversion to floating-point numbers always use base
7174: 10, irrespective of the value of @code{base}. In ANS Forth,
7175: conversion to floating-point numbers whilst the value of
7176: @code{base} is not 10 is an ambiguous condition.
7177: @end itemize
7178:
7179: @ref{Input} describes words that you can use to read numbers into your
7180: programs.
7181:
7182: @node Interpret/Compile states, Literals, Number Conversion, The Text Interpreter
7183: @subsection Interpret/Compile states
7184: @cindex Interpret/Compile states
7185:
7186: A standard program is not permitted to change @code{state}
7187: explicitly. However, it can change @code{state} implicitly, using the
7188: words @code{[} and @code{]}. When @code{[} is executed it switches
7189: @code{state} to interpret state, and therefore the text interpreter
7190: starts interpreting. When @code{]} is executed it switches @code{state}
7191: to compile state and therefore the text interpreter starts
7192: compiling. The most common usage for these words is for switching into
7193: interpret state and back from within a colon definition; this technique
7194: can be used to compile a literal (@pxref{Literals} for an example) or
7195: for conditional compilation (@pxref{Interpreter Directives} for an
7196: example).
7197:
7198:
7199: @c This is a bad example: It's non-standard, and it's not necessary.
7200: @c However, I can't think of a good example for switching into compile
7201: @c state when there is no current word (@code{state}-smart words are not a
7202: @c good reason). So maybe we should use an example for switching into
7203: @c interpret @code{state} in a colon def. - anton
7204: @c nac-> I agree. I started out by putting in the example, then realised
7205: @c that it was non-ANS, so wrote more words around it. I hope this
7206: @c re-written version is acceptable to you. I do want to keep the example
7207: @c as it is helpful for showing what is and what is not portable, particularly
7208: @c where it outlaws a style in common use.
7209:
7210:
7211: @code{[} and @code{]} also give you the ability to switch into compile
7212: state and back, but we cannot think of any useful Standard application
7213: for this ability. Pre-ANS Forth textbooks have examples like this:
7214:
7215: @example
7216: : AA ." this is A" ;
7217: : BB ." this is B" ;
7218: : CC ." this is C" ;
7219:
7220: create table ] aa bb cc [
7221:
7222: : go ( n -- ) \ n is offset into table.. 0 for 1st entry
7223: cells table + @ execute ;
7224: @end example
7225:
7226: This example builds a jump table; @code{0 go} will display ``@code{this
7227: is A}''. Using @code{[} and @code{]} in this example is equivalent to
7228: defining @code{table} like this:
7229:
7230: @example
7231: create table ' aa COMPILE, ' bb COMPILE, ' cc COMPILE,
7232: @end example
7233:
7234: The problem with this code is that the definition of @code{table} is not
7235: portable -- it @i{compile}s execution tokens into code space. Whilst it
7236: @i{may} work on systems where code space and data space co-incide, the
7237: Standard only allows data space to be assigned for a @code{CREATE}d
7238: word. In addition, the Standard only allows @code{@@} to access data
7239: space, whilst this example is using it to access code space. The only
7240: portable, Standard way to build this table is to build it in data space,
7241: like this:
7242:
7243: @example
7244: create table ' aa , ' bb , ' cc ,
7245: @end example
7246:
7247: doc-state
7248: doc-[
7249: doc-]
7250:
7251:
7252: @node Literals, Interpreter Directives, Interpret/Compile states, The Text Interpreter
7253: @subsection Literals
7254: @cindex Literals
7255:
7256: Often, you want to use a number within a colon definition. When you do
7257: this, the text interpreter automatically compiles the number as a
7258: @i{literal}. A literal is a number whose run-time effect is to be pushed
7259: onto the stack. If you had to do some maths to generate the number, you
7260: might write it like this:
7261:
7262: @example
7263: : HOUR-TO-SEC ( n1 -- n2 )
7264: 60 * \ to minutes
7265: 60 * ; \ to seconds
7266: @end example
7267:
7268: It is very clear what this definition is doing, but it's inefficient
7269: since it is performing 2 multiples at run-time. An alternative would be
7270: to write:
7271:
7272: @example
7273: : HOUR-TO-SEC ( n1 -- n2 )
7274: 3600 * ; \ to seconds
7275: @end example
7276:
7277: Which does the same thing, and has the advantage of using a single
7278: multiply. Ideally, we'd like the efficiency of the second with the
7279: readability of the first.
7280:
7281: @code{Literal} allows us to achieve that. It takes a number from the
7282: stack and lays it down in the current definition just as though the
7283: number had been typed directly into the definition. Our first attempt
7284: might look like this:
7285:
7286: @example
7287: 60 \ mins per hour
7288: 60 * \ seconds per minute
7289: : HOUR-TO-SEC ( n1 -- n2 )
7290: Literal * ; \ to seconds
7291: @end example
7292:
7293: But this produces the error message @code{unstructured}. What happened?
7294: The stack notation for @code{:} is (@i{ -- colon-sys}) and the size of
7295: @i{colon-sys} is implementation-defined. In other words, once we start a
7296: colon definition we can't portably access anything that was on the stack
7297: before the definition began@footnote{@cite{Two Problems in ANS Forth},
7298: by Thomas Worthington; Forth Dimensions 20(2) pages 32--34 describes
7299: some situations where you might want to access stack items above
7300: colon-sys, and provides a solution to the problem.}. The correct way of
7301: solving this problem in this instance is to use @code{[ ]} like this:
7302:
7303: @example
7304: : HOUR-TO-SEC ( n1 -- n2 )
7305: [ 60 \ minutes per hour
7306: 60 * ] \ seconds per minute
7307: LITERAL * ; \ to seconds
7308: @end example
7309:
7310:
7311: doc-literal
7312: doc-]L
7313: doc-2literal
7314: doc-fliteral
7315:
7316:
7317: @node Interpreter Directives, , Literals, The Text Interpreter
7318: @subsection Interpreter Directives
7319: @cindex interpreter directives
7320:
7321: These words are usually used in interpret state; typically to control
7322: which parts of a source file are processed by the text
7323: interpreter. There are only a few ANS Forth Standard words, but Gforth
7324: supplements these with a rich set of immediate control structure words
7325: to compensate for the fact that the non-immediate versions can only be
7326: used in compile state (@pxref{Control Structures}). Typical usages:
7327:
7328: @example
7329: FALSE Constant ASSEMBLER
7330: .
7331: .
7332: ASSEMBLER [IF]
7333: : ASSEMBLER-FEATURE
7334: ...
7335: ;
7336: [ENDIF]
7337: .
7338: .
7339: : SEE
7340: ... \ general-purpose SEE code
7341: [ ASSEMBLER [IF] ]
7342: ... \ assembler-specific SEE code
7343: [ [ENDIF] ]
7344: ;
7345: @end example
7346:
7347:
7348: doc-[IF]
7349: doc-[ELSE]
7350: doc-[THEN]
7351: doc-[ENDIF]
7352:
7353: doc-[IFDEF]
7354: doc-[IFUNDEF]
7355:
7356: doc-[?DO]
7357: doc-[DO]
7358: doc-[FOR]
7359: doc-[LOOP]
7360: doc-[+LOOP]
7361: doc-[NEXT]
7362:
7363: doc-[BEGIN]
7364: doc-[UNTIL]
7365: doc-[AGAIN]
7366: doc-[WHILE]
7367: doc-[REPEAT]
7368:
7369:
7370: @c -------------------------------------------------------------
7371: @node Word Lists, Environmental Queries, The Text Interpreter, Words
7372: @section Word Lists
7373: @cindex word lists
7374: @cindex header space
7375:
7376: A wordlist is a list of named words; you can add new words and look up
7377: words by name (and you can remove words in a restricted way with
7378: markers). Every named (and @code{reveal}ed) word is in one wordlist.
7379:
7380: @cindex search order stack
7381: The text interpreter searches the wordlists present in the search order
7382: (a stack of wordlists), from the top to the bottom. Within each
7383: wordlist, the search starts conceptually at the newest word; i.e., if
7384: two words in a wordlist have the same name, the newer word is found.
7385:
7386: @cindex compilation word list
7387: New words are added to the @dfn{compilation wordlist} (aka current
7388: wordlist).
7389:
7390: @cindex wid
7391: A word list is identified by a cell-sized word list identifier (@i{wid})
7392: in much the same way as a file is identified by a file handle. The
7393: numerical value of the wid has no (portable) meaning, and might change
7394: from session to session.
7395:
7396: The ANS Forth ``Search order'' word set is intended to provide a set of
7397: low-level tools that allow various different schemes to be
7398: implemented. Gforth provides @code{vocabulary}, a traditional Forth
7399: word. @file{compat/vocabulary.fs} provides an implementation in ANS
7400: Forth.
7401:
7402: @comment TODO: locals section refers to here, saying that every word list (aka
7403: @comment vocabulary) has its own methods for searching etc. Need to document that.
7404:
7405: @comment TODO: document markers, reveal, tables, mappedwordlist
7406:
7407: @comment the gforthman- prefix is used to pick out the true definition of a
7408: @comment word from the source files, rather than some alias.
7409:
7410: doc-forth-wordlist
7411: doc-definitions
7412: doc-get-current
7413: doc-set-current
7414: doc-get-order
7415: doc---gforthman-set-order
7416: doc-wordlist
7417: doc-table
7418: doc-push-order
7419: doc-previous
7420: doc-also
7421: doc---gforthman-forth
7422: doc-only
7423: doc---gforthman-order
7424:
7425: doc-find
7426: doc-search-wordlist
7427:
7428: doc-words
7429: doc-vlist
7430: @c doc-words-deferred
7431:
7432: doc-mappedwordlist
7433: doc-root
7434: doc-vocabulary
7435: doc-seal
7436: doc-vocs
7437: doc-current
7438: doc-context
7439:
7440:
7441: @menu
7442: * Why use word lists?::
7443: * Word list examples::
7444: @end menu
7445:
7446: @node Why use word lists?, Word list examples, Word Lists, Word Lists
7447: @subsection Why use word lists?
7448: @cindex word lists - why use them?
7449:
7450: Here are some reasons for using multiple word lists:
7451:
7452: @itemize @bullet
7453: @item
7454: To improve compilation speed by reducing the number of header space
7455: entries that must be searched. This is achieved by creating a new
7456: word list that contains all of the definitions that are used in the
7457: definition of a Forth system but which would not usually be used by
7458: programs running on that system. That word list would be on the search
7459: list when the Forth system was compiled but would be removed from the
7460: search list for normal operation. This can be a useful technique for
7461: low-performance systems (for example, 8-bit processors in embedded
7462: systems) but is unlikely to be necessary in high-performance desktop
7463: systems.
7464: @item
7465: To prevent a set of words from being used outside the context in which
7466: they are valid. Two classic examples of this are an integrated editor
7467: (all of the edit commands are defined in a separate word list; the
7468: search order is set to the editor word list when the editor is invoked;
7469: the old search order is restored when the editor is terminated) and an
7470: integrated assembler (the op-codes for the machine are defined in a
7471: separate word list which is used when a @code{CODE} word is defined).
7472: @item
7473: To prevent a name-space clash between multiple definitions with the same
7474: name. For example, when building a cross-compiler you might have a word
7475: @code{IF} that generates conditional code for your target system. By
7476: placing this definition in a different word list you can control whether
7477: the host system's @code{IF} or the target system's @code{IF} get used in
7478: any particular context by controlling the order of the word lists on the
7479: search order stack.
7480: @end itemize
7481:
7482: @node Word list examples, , Why use word lists?, Word Lists
7483: @subsection Word list examples
7484: @cindex word lists - examples
7485:
7486: Here is an example of creating and using a new wordlist using ANS
7487: Forth Standard words:
7488:
7489: @example
7490: wordlist constant my-new-words-wordlist
7491: : my-new-words get-order nip my-new-words-wordlist swap set-order ;
7492:
7493: \ add it to the search order
7494: also my-new-words
7495:
7496: \ alternatively, add it to the search order and make it
7497: \ the compilation word list
7498: also my-new-words definitions
7499: \ type "order" to see the problem
7500: @end example
7501:
7502: The problem with this example is that @code{order} has no way to
7503: associate the name @code{my-new-words} with the wid of the word list (in
7504: Gforth, @code{order} and @code{vocs} will display @code{???} for a wid
7505: that has no associated name). There is no Standard way of associating a
7506: name with a wid.
7507:
7508: In Gforth, this example can be re-coded using @code{vocabulary}, which
7509: associates a name with a wid:
7510:
7511: @example
7512: vocabulary my-new-words
7513:
7514: \ add it to the search order
7515: also my-new-words
7516:
7517: \ alternatively, add it to the search order and make it
7518: \ the compilation word list
7519: my-new-words definitions
7520: \ type "order" to see that the problem is solved
7521: @end example
7522:
7523: @c -------------------------------------------------------------
7524: @node Environmental Queries, Files, Word Lists, Words
7525: @section Environmental Queries
7526: @cindex environmental queries
7527:
7528: ANS Forth introduced the idea of ``environmental queries'' as a way
7529: for a program running on a system to determine certain characteristics of the system.
7530: The Standard specifies a number of strings that might be recognised by a system.
7531:
7532: The Standard requires that the header space used for environmental queries
7533: be distinct from the header space used for definitions.
7534:
7535: Typically, environmental queries are supported by creating a set of
7536: definitions in a word list that is @i{only} used during environmental
7537: queries; that is what Gforth does. There is no Standard way of adding
7538: definitions to the set of recognised environmental queries, but any
7539: implementation that supports the loading of optional word sets must have
7540: some mechanism for doing this (after loading the word set, the
7541: associated environmental query string must return @code{true}). In
7542: Gforth, the word list used to honour environmental queries can be
7543: manipulated just like any other word list.
7544:
7545:
7546: doc-environment?
7547: doc-environment-wordlist
7548:
7549: doc-gforth
7550: doc-os-class
7551:
7552:
7553: Note that, whilst the documentation for (e.g.) @code{gforth} shows it
7554: returning two items on the stack, querying it using @code{environment?}
7555: will return an additional item; the @code{true} flag that shows that the
7556: string was recognised.
7557:
7558: @comment TODO Document the standard strings or note where they are documented herein
7559:
7560: Here are some examples of using environmental queries:
7561:
7562: @example
7563: s" address-unit-bits" environment? 0=
7564: [IF]
7565: cr .( environmental attribute address-units-bits unknown... ) cr
7566: [THEN]
7567:
7568: s" block" environment? [IF] DROP include block.fs [THEN]
7569:
7570: s" gforth" environment? [IF] 2DROP include compat/vocabulary.fs [THEN]
7571:
7572: s" gforth" environment? [IF] .( Gforth version ) TYPE
7573: [ELSE] .( Not Gforth..) [THEN]
7574: @end example
7575:
7576:
7577: Here is an example of adding a definition to the environment word list:
7578:
7579: @example
7580: get-current environment-wordlist set-current
7581: true constant block
7582: true constant block-ext
7583: set-current
7584: @end example
7585:
7586: You can see what definitions are in the environment word list like this:
7587:
7588: @example
7589: get-order 1+ environment-wordlist swap set-order words previous
7590: @end example
7591:
7592:
7593: @c -------------------------------------------------------------
7594: @node Files, Blocks, Environmental Queries, Words
7595: @section Files
7596: @cindex files
7597: @cindex I/O - file-handling
7598:
7599: Gforth provides facilities for accessing files that are stored in the
7600: host operating system's file-system. Files that are processed by Gforth
7601: can be divided into two categories:
7602:
7603: @itemize @bullet
7604: @item
7605: Files that are processed by the Text Interpreter (@dfn{Forth source files}).
7606: @item
7607: Files that are processed by some other program (@dfn{general files}).
7608: @end itemize
7609:
7610: doc-loadfilename
7611: doc-sourcefilename
7612: doc-sourceline#
7613:
7614: @menu
7615: * Forth source files::
7616: * General files::
7617: * Search Paths::
7618: @end menu
7619:
7620:
7621: @c -------------------------------------------------------------
7622: @node Forth source files, General files, Files, Files
7623: @subsection Forth source files
7624: @cindex including files
7625: @cindex Forth source files
7626:
7627: The simplest way to interpret the contents of a file is to use one of
7628: these two formats:
7629:
7630: @example
7631: include mysource.fs
7632: s" mysource.fs" included
7633: @end example
7634:
7635: Sometimes you want to include a file only if it is not included already
7636: (by, say, another source file). In that case, you can use one of these
7637: three formats:
7638:
7639: @example
7640: require mysource.fs
7641: needs mysource.fs
7642: s" mysource.fs" required
7643: @end example
7644:
7645: @cindex stack effect of included files
7646: @cindex including files, stack effect
7647: It is good practice to write your source files such that interpreting them
7648: does not change the stack. Source files designed in this way can be used with
7649: @code{required} and friends without complications. For example:
7650:
7651: @example
7652: 1 require foo.fs drop
7653: @end example
7654:
7655:
7656: doc-include-file
7657: doc-included
7658: doc-included?
7659: doc-include
7660: doc-required
7661: doc-require
7662: doc-needs
7663: doc-init-included-files
7664:
7665:
7666: A definition in ANS Forth for @code{required} is provided in
7667: @file{compat/required.fs}.
7668:
7669: @c -------------------------------------------------------------
7670: @node General files, Search Paths, Forth source files, Files
7671: @subsection General files
7672: @cindex general files
7673: @cindex file-handling
7674:
7675: Files are opened/created by name and type. The following types are
7676: recognised:
7677:
7678:
7679: doc-r/o
7680: doc-r/w
7681: doc-w/o
7682: doc-bin
7683:
7684:
7685: When a file is opened/created, it returns a file identifier,
7686: @i{wfileid} that is used for all other file commands. All file
7687: commands also return a status value, @i{wior}, that is 0 for a
7688: successful operation and an implementation-defined non-zero value in the
7689: case of an error.
7690:
7691:
7692: doc-open-file
7693: doc-create-file
7694:
7695: doc-close-file
7696: doc-delete-file
7697: doc-rename-file
7698: doc-read-file
7699: doc-read-line
7700: doc-write-file
7701: doc-write-line
7702: doc-emit-file
7703: doc-flush-file
7704:
7705: doc-file-status
7706: doc-file-position
7707: doc-reposition-file
7708: doc-file-size
7709: doc-resize-file
7710:
7711:
7712: @c ---------------------------------------------------------
7713: @node Search Paths, , General files, Files
7714: @subsection Search Paths
7715: @cindex path for @code{included}
7716: @cindex file search path
7717: @cindex @code{include} search path
7718: @cindex search path for files
7719:
7720: If you specify an absolute filename (i.e., a filename starting with
7721: @file{/} or @file{~}, or with @file{:} in the second position (as in
7722: @samp{C:...})) for @code{included} and friends, that file is included
7723: just as you would expect.
7724:
7725: For relative filenames, Gforth uses a search path similar to Forth's
7726: search order (@pxref{Word Lists}). It tries to find the given filename
7727: in the directories present in the path, and includes the first one it
7728: finds. There are separate search paths for Forth source files and
7729: general files.
7730:
7731: If the search path contains the directory @file{.} (as it should), this
7732: refers to the directory that the present file was @code{included}
7733: from. This allows files to include other files relative to their own
7734: position (irrespective of the current working directory or the absolute
7735: position). This feature is essential for libraries consisting of
7736: several files, where a file may include other files from the library.
7737: It corresponds to @code{#include "..."} in C. If the current input
7738: source is not a file, @file{.} refers to the directory of the innermost
7739: file being included, or, if there is no file being included, to the
7740: current working directory.
7741:
7742: Use @file{~+} to refer to the current working directory (as in the
7743: @code{bash}).
7744:
7745: If the filename starts with @file{./}, the search path is not searched
7746: (just as with absolute filenames), and the @file{.} has the same meaning
7747: as described above.
7748:
7749: @menu
7750: * Forth Search Paths::
7751: * General Search Paths::
7752: @end menu
7753:
7754: @c ---------------------------------------------------------
7755: @node Forth Search Paths, General Search Paths, Search Paths, Search Paths
7756: @subsubsection Forth Search Paths
7757: @cindex search path control - Forth
7758:
7759: The search path is initialized when you start Gforth (@pxref{Invoking
7760: Gforth}). You can display it and change it using these words:
7761:
7762:
7763: doc-.fpath
7764: doc-fpath+
7765: doc-fpath=
7766: doc-open-fpath-file
7767:
7768:
7769: @noindent
7770: Here is an example of using @code{fpath} and @code{require}:
7771:
7772: @example
7773: fpath= /usr/lib/forth/|./
7774: require timer.fs
7775: @end example
7776:
7777: @c ---------------------------------------------------------
7778: @node General Search Paths, , Forth Search Paths, Search Paths
7779: @subsubsection General Search Paths
7780: @cindex search path control - for user applications
7781:
7782: Your application may need to search files in several directories, like
7783: @code{included} does. To facilitate this, Gforth allows you to define
7784: and use your own search paths, by providing generic equivalents of the
7785: Forth search path words:
7786:
7787:
7788: doc-.path
7789: doc-path+
7790: doc-path=
7791: doc-open-path-file
7792:
7793:
7794: Here's an example of creating a search path:
7795:
7796: @example
7797: \ Make a buffer for the path:
7798: create mypath 100 chars , \ maximum length (is checked)
7799: 0 , \ real len
7800: 100 chars allot \ space for path
7801: @end example
7802:
7803: @c -------------------------------------------------------------
7804: @node Blocks, Other I/O, Files, Words
7805: @section Blocks
7806: @cindex I/O - blocks
7807: @cindex blocks
7808:
7809: When you run Gforth on a modern desk-top computer, it runs under the
7810: control of an operating system which provides certain services. One of
7811: these services is @var{file services}, which allows Forth source code
7812: and data to be stored in files and read into Gforth (@pxref{Files}).
7813:
7814: Traditionally, Forth has been an important programming language on
7815: systems where it has interfaced directly to the underlying hardware with
7816: no intervening operating system. Forth provides a mechanism, called
7817: @dfn{blocks}, for accessing mass storage on such systems.
7818:
7819: A block is a 1024-byte data area, which can be used to hold data or
7820: Forth source code. No structure is imposed on the contents of the
7821: block. A block is identified by its number; blocks are numbered
7822: contiguously from 1 to an implementation-defined maximum.
7823:
7824: A typical system that used blocks but no operating system might use a
7825: single floppy-disk drive for mass storage, with the disks formatted to
7826: provide 256-byte sectors. Blocks would be implemented by assigning the
7827: first four sectors of the disk to block 1, the second four sectors to
7828: block 2 and so on, up to the limit of the capacity of the disk. The disk
7829: would not contain any file system information, just the set of blocks.
7830:
7831: @cindex blocks file
7832: On systems that do provide file services, blocks are typically
7833: implemented by storing a sequence of blocks within a single @dfn{blocks
7834: file}. The size of the blocks file will be an exact multiple of 1024
7835: bytes, corresponding to the number of blocks it contains. This is the
7836: mechanism that Gforth uses.
7837:
7838: @cindex @file{blocks.fb}
7839: Only 1 blocks file can be open at a time. If you use block words without
7840: having specified a blocks file, Gforth defaults to the blocks file
7841: @file{blocks.fb}. Gforth uses the Forth search path when attempting to
7842: locate a blocks file (@pxref{Forth Search Paths}).
7843:
7844: @cindex block buffers
7845: When you read and write blocks under program control, Gforth uses a
7846: number of @dfn{block buffers} as intermediate storage. These buffers are
7847: not used when you use @code{load} to interpret the contents of a block.
7848:
7849: The behaviour of the block buffers is directly analagous to that of a
7850: cache. Each block buffer has three states:
7851:
7852: @itemize @bullet
7853: @item
7854: Unassigned
7855: @item
7856: Assigned-clean
7857: @item
7858: Assigned-dirty
7859: @end itemize
7860:
7861: Initially, all block buffers are @i{unassigned}. In order to access a
7862: block, the block (specified by its block number) must be assigned to a
7863: block buffer.
7864:
7865: The assignment of a block to a block buffer is performed by @code{block}
7866: or @code{buffer}. Use @code{block} when you wish to modify the existing
7867: contents of a block. Use @code{buffer} when you don't care about the
7868: existing contents of the block@footnote{The ANS Forth definition of
7869: @code{buffer} is intended not to cause disk I/O; if the data associated
7870: with the particular block is already stored in a block buffer due to an
7871: earlier @code{block} command, @code{buffer} will return that block
7872: buffer and the existing contents of the block will be
7873: available. Otherwise, @code{buffer} will simply assign a new, empty
7874: block buffer for the block.}.
7875:
7876: Once a block has been assigned to a block buffer using @code{block} or
7877: @code{buffer}, that block buffer becomes the @i{current block buffer}
7878: and its state changes to @i{assigned-clean}. Data may only be
7879: manipulated (read or written) within the current block buffer.
7880:
7881: When the contents of the current block buffer has been modified it is
7882: necessary, @emph{before calling @code{block} or @code{buffer} again}, to
7883: either abandon the changes (by doing nothing) or commit the changes,
7884: using @code{update}. Using @code{update} does not change the blocks
7885: file; it simply changes a block buffer's state to @i{assigned-dirty}.
7886:
7887: The word @code{flush} causes all @i{assigned-dirty} blocks to be
7888: written back to the blocks file on disk. Leaving Gforth using @code{bye}
7889: also causes a @code{flush} to be performed.
7890:
7891: In Gforth, @code{block} and @code{buffer} use a @i{direct-mapped}
7892: algorithm to assign a block buffer to a block. That means that any
7893: particular block can only be assigned to one specific block buffer,
7894: called (for the particular operation) the @i{victim buffer}. If the
7895: victim buffer is @i{unassigned} or @i{assigned-clean} it is allocated to
7896: the new block immediately. If it is @i{assigned-dirty} its current
7897: contents are written back to the blocks file on disk before it is
7898: allocated to the new block.
7899:
7900: Although no structure is imposed on the contents of a block, it is
7901: traditional to display the contents as 16 lines each of 64 characters. A
7902: block provides a single, continuous stream of input (for example, it
7903: acts as a single parse area) -- there are no end-of-line characters
7904: within a block, and no end-of-file character at the end of a
7905: block. There are two consequences of this:
7906:
7907: @itemize @bullet
7908: @item
7909: The last character of one line wraps straight into the first character
7910: of the following line
7911: @item
7912: The word @code{\} -- comment to end of line -- requires special
7913: treatment; in the context of a block it causes all characters until the
7914: end of the current 64-character ``line'' to be ignored.
7915: @end itemize
7916:
7917: In Gforth, when you use @code{block} with a non-existent block number,
7918: the current blocks file will be extended to the appropriate size and the
7919: block buffer will be initialised with spaces.
7920:
7921: Gforth includes a simple block editor (type @code{use blocked.fb 0 list}
7922: for details) but doesn't encourage the use of blocks; the mechanism is
7923: only provided for backward compatibility -- ANS Forth requires blocks to
7924: be available when files are.
7925:
7926: Common techniques that are used when working with blocks include:
7927:
7928: @itemize @bullet
7929: @item
7930: A screen editor that allows you to edit blocks without leaving the Forth
7931: environment.
7932: @item
7933: Shadow screens; where every code block has an associated block
7934: containing comments (for example: code in odd block numbers, comments in
7935: even block numbers). Typically, the block editor provides a convenient
7936: mechanism to toggle between code and comments.
7937: @item
7938: Load blocks; a single block (typically block 1) contains a number of
7939: @code{thru} commands which @code{load} the whole of the application.
7940: @end itemize
7941:
7942: See Frank Sergeant's Pygmy Forth to see just how well blocks can be
7943: integrated into a Forth programming environment.
7944:
7945: @comment TODO what about errors on open-blocks?
7946:
7947: doc-open-blocks
7948: doc-use
7949: doc-get-block-fid
7950: doc-block-position
7951:
7952: doc-scr
7953: doc-list
7954:
7955: doc---gforthman-block
7956: doc-buffer
7957:
7958: doc-update
7959: doc-updated?
7960: doc-save-buffers
7961: doc-empty-buffers
7962: doc-empty-buffer
7963: doc-flush
7964:
7965: doc-load
7966: doc-thru
7967: doc-+load
7968: doc-+thru
7969: doc---gforthman--->
7970: doc-block-included
7971:
7972:
7973: @c -------------------------------------------------------------
7974: @node Other I/O, Programming Tools, Blocks, Words
7975: @section Other I/O
7976: @cindex I/O - keyboard and display
7977:
7978: @menu
7979: * Simple numeric output:: Predefined formats
7980: * Formatted numeric output:: Formatted (pictured) output
7981: * String Formats:: How Forth stores strings in memory
7982: * Displaying characters and strings:: Other stuff
7983: * Input:: Input
7984: @end menu
7985:
7986: @node Simple numeric output, Formatted numeric output, Other I/O, Other I/O
7987: @subsection Simple numeric output
7988: @cindex numeric output - simple/free-format
7989:
7990: The simplest output functions are those that display numbers from the
7991: data or floating-point stacks. Floating-point output is always displayed
7992: using base 10. Numbers displayed from the data stack use the value stored
7993: in @code{base}.
7994:
7995:
7996: doc-.
7997: doc-dec.
7998: doc-hex.
7999: doc-u.
8000: doc-.r
8001: doc-u.r
8002: doc-d.
8003: doc-ud.
8004: doc-d.r
8005: doc-ud.r
8006: doc-f.
8007: doc-fe.
8008: doc-fs.
8009:
8010:
8011: Examples of printing the number 1234.5678E23 in the different floating-point output
8012: formats are shown below:
8013:
8014: @example
8015: f. 123456779999999000000000000.
8016: fe. 123.456779999999E24
8017: fs. 1.23456779999999E26
8018: @end example
8019:
8020:
8021: @node Formatted numeric output, String Formats, Simple numeric output, Other I/O
8022: @subsection Formatted numeric output
8023: @cindex formatted numeric output
8024: @cindex pictured numeric output
8025: @cindex numeric output - formatted
8026:
8027: Forth traditionally uses a technique called @dfn{pictured numeric
8028: output} for formatted printing of integers. In this technique, digits
8029: are extracted from the number (using the current output radix defined by
8030: @code{base}), converted to ASCII codes and appended to a string that is
8031: built in a scratch-pad area of memory (@pxref{core-idef,
8032: Implementation-defined options, Implementation-defined
8033: options}). Arbitrary characters can be appended to the string during the
8034: extraction process. The completed string is specified by an address
8035: and length and can be manipulated (@code{TYPE}ed, copied, modified)
8036: under program control.
8037:
8038: All of the words described in the previous section for simple numeric
8039: output are implemented in Gforth using pictured numeric output.
8040:
8041: Three important things to remember about pictured numeric output:
8042:
8043: @itemize @bullet
8044: @item
8045: It always operates on double-precision numbers; to display a
8046: single-precision number, convert it first (@pxref{Double precision} for
8047: ways of doing this).
8048: @item
8049: It always treats the double-precision number as though it were
8050: unsigned. The examples below show ways of printing signed numbers.
8051: @item
8052: The string is built up from right to left; least significant digit first.
8053: @end itemize
8054:
8055:
8056: doc-<#
8057: doc-<<#
8058: doc-#
8059: doc-#s
8060: doc-hold
8061: doc-sign
8062: doc-#>
8063: doc-#>>
8064:
8065: doc-represent
8066:
8067:
8068: @noindent
8069: Here are some examples of using pictured numeric output:
8070:
8071: @example
8072: : my-u. ( u -- )
8073: \ Simplest use of pns.. behaves like Standard u.
8074: 0 \ convert to unsigned double
8075: <# \ start conversion
8076: #s \ convert all digits
8077: #> \ complete conversion
8078: TYPE SPACE ; \ display, with trailing space
8079:
8080: : cents-only ( u -- )
8081: 0 \ convert to unsigned double
8082: <# \ start conversion
8083: # # \ convert two least-significant digits
8084: #> \ complete conversion, discard other digits
8085: TYPE SPACE ; \ display, with trailing space
8086:
8087: : dollars-and-cents ( u -- )
8088: 0 \ convert to unsigned double
8089: <# \ start conversion
8090: # # \ convert two least-significant digits
8091: [char] . hold \ insert decimal point
8092: #s \ convert remaining digits
8093: [char] $ hold \ append currency symbol
8094: #> \ complete conversion
8095: TYPE SPACE ; \ display, with trailing space
8096:
8097: : my-. ( n -- )
8098: \ handling negatives.. behaves like Standard .
8099: s>d \ convert to signed double
8100: swap over dabs \ leave sign byte followed by unsigned double
8101: <# \ start conversion
8102: #s \ convert all digits
8103: rot sign \ get at sign byte, append "-" if needed
8104: #> \ complete conversion
8105: TYPE SPACE ; \ display, with trailing space
8106:
8107: : account. ( n -- )
8108: \ accountants don't like minus signs, they use braces
8109: \ for negative numbers
8110: s>d \ convert to signed double
8111: swap over dabs \ leave sign byte followed by unsigned double
8112: <# \ start conversion
8113: 2 pick \ get copy of sign byte
8114: 0< IF [char] ) hold THEN \ right-most character of output
8115: #s \ convert all digits
8116: rot \ get at sign byte
8117: 0< IF [char] ( hold THEN
8118: #> \ complete conversion
8119: TYPE SPACE ; \ display, with trailing space
8120: @end example
8121:
8122: Here are some examples of using these words:
8123:
8124: @example
8125: 1 my-u. 1
8126: hex -1 my-u. decimal FFFFFFFF
8127: 1 cents-only 01
8128: 1234 cents-only 34
8129: 2 dollars-and-cents $0.02
8130: 1234 dollars-and-cents $12.34
8131: 123 my-. 123
8132: -123 my. -123
8133: 123 account. 123
8134: -456 account. (456)
8135: @end example
8136:
8137:
8138: @node String Formats, Displaying characters and strings, Formatted numeric output, Other I/O
8139: @subsection String Formats
8140: @cindex strings - see character strings
8141: @cindex character strings - formats
8142: @cindex I/O - see character strings
8143:
8144: Forth commonly uses two different methods for representing character
8145: strings:
8146:
8147: @itemize @bullet
8148: @item
8149: @cindex address of counted string
8150: @cindex counted string
8151: As a @dfn{counted string}, represented by a @i{c-addr}. The char
8152: addressed by @i{c-addr} contains a character-count, @i{n}, of the
8153: string and the string occupies the subsequent @i{n} char addresses in
8154: memory.
8155: @item
8156: As cell pair on the stack; @i{c-addr u}, where @i{u} is the length
8157: of the string in characters, and @i{c-addr} is the address of the
8158: first byte of the string.
8159: @end itemize
8160:
8161: ANS Forth encourages the use of the second format when representing
8162: strings on the stack, whilst conceeding that the counted string format
8163: remains useful as a way of storing strings in memory.
8164:
8165:
8166: doc-count
8167:
8168:
8169: @xref{Memory Blocks} for words that move, copy and search
8170: for strings. @xref{Displaying characters and strings,} for words that
8171: display characters and strings.
8172:
8173:
8174: @node Displaying characters and strings, Input, String Formats, Other I/O
8175: @subsection Displaying characters and strings
8176: @cindex characters - compiling and displaying
8177: @cindex character strings - compiling and displaying
8178:
8179: This section starts with a glossary of Forth words and ends with a set
8180: of examples.
8181:
8182:
8183: doc-bl
8184: doc-space
8185: doc-spaces
8186: doc-emit
8187: doc-toupper
8188: doc-."
8189: doc-.(
8190: doc-type
8191: doc-typewhite
8192: doc-cr
8193: @cindex cursor control
8194: doc-at-xy
8195: doc-page
8196: doc-s"
8197: doc-c"
8198: doc-char
8199: doc-[char]
8200: doc-sliteral
8201:
8202:
8203: @noindent
8204: As an example, consider the following text, stored in a file @file{test.fs}:
8205:
8206: @example
8207: .( text-1)
8208: : my-word
8209: ." text-2" cr
8210: .( text-3)
8211: ;
8212:
8213: ." text-4"
8214:
8215: : my-char
8216: [char] ALPHABET emit
8217: char emit
8218: ;
8219: @end example
8220:
8221: When you load this code into Gforth, the following output is generated:
8222:
8223: @example
8224: @kbd{include test.fs @key{RET}} text-1text-3text-4 ok
8225: @end example
8226:
8227: @itemize @bullet
8228: @item
8229: Messages @code{text-1} and @code{text-3} are displayed because @code{.(}
8230: is an immediate word; it behaves in the same way whether it is used inside
8231: or outside a colon definition.
8232: @item
8233: Message @code{text-4} is displayed because of Gforth's added interpretation
8234: semantics for @code{."}.
8235: @item
8236: Message @code{text-2} is @i{not} displayed, because the text interpreter
8237: performs the compilation semantics for @code{."} within the definition of
8238: @code{my-word}.
8239: @end itemize
8240:
8241: Here are some examples of executing @code{my-word} and @code{my-char}:
8242:
8243: @example
8244: @kbd{my-word @key{RET}} text-2
8245: ok
8246: @kbd{my-char fred @key{RET}} Af ok
8247: @kbd{my-char jim @key{RET}} Aj ok
8248: @end example
8249:
8250: @itemize @bullet
8251: @item
8252: Message @code{text-2} is displayed because of the run-time behaviour of
8253: @code{."}.
8254: @item
8255: @code{[char]} compiles the ``A'' from ``ALPHABET'' and puts its display code
8256: on the stack at run-time. @code{emit} always displays the character
8257: when @code{my-char} is executed.
8258: @item
8259: @code{char} parses a string at run-time and the second @code{emit} displays
8260: the first character of the string.
8261: @item
8262: If you type @code{see my-char} you can see that @code{[char]} discarded
8263: the text ``LPHABET'' and only compiled the display code for ``A'' into the
8264: definition of @code{my-char}.
8265: @end itemize
8266:
8267:
8268:
8269: @node Input, , Displaying characters and strings, Other I/O
8270: @subsection Input
8271: @cindex input
8272: @cindex I/O - see input
8273: @cindex parsing a string
8274:
8275: @xref{String Formats} for ways of storing character strings in memory.
8276:
8277: @comment TODO examples for >number >float accept key key? pad parse word refill
8278: @comment then index them
8279:
8280:
8281: doc-key
8282: doc-key?
8283: doc-ekey
8284: doc-ekey?
8285: doc-ekey>char
8286: doc->number
8287: doc->float
8288: doc-accept
8289: doc-pad
8290: doc-parse
8291: doc-word
8292: doc-sword
8293: doc-(name)
8294: doc-refill
8295: @comment obsolescent words..
8296: doc-convert
8297: doc-query
8298: doc-expect
8299: doc-span
8300:
8301:
8302:
8303: @c -------------------------------------------------------------
8304: @node Programming Tools, Assembler and Code Words, Other I/O, Words
8305: @section Programming Tools
8306: @cindex programming tools
8307:
8308: @menu
8309: * Debugging:: Simple and quick.
8310: * Assertions:: Making your programs self-checking.
8311: * Singlestep Debugger:: Executing your program word by word.
8312: @end menu
8313:
8314: @node Debugging, Assertions, Programming Tools, Programming Tools
8315: @subsection Debugging
8316: @cindex debugging
8317:
8318: Languages with a slow edit/compile/link/test development loop tend to
8319: require sophisticated tracing/stepping debuggers to facilate
8320: productive debugging.
8321:
8322: A much better (faster) way in fast-compiling languages is to add
8323: printing code at well-selected places, let the program run, look at
8324: the output, see where things went wrong, add more printing code, etc.,
8325: until the bug is found.
8326:
8327: The simple debugging aids provided in @file{debugs.fs}
8328: are meant to support this style of debugging. In addition, there are
8329: words for non-destructively inspecting the stack and memory:
8330:
8331:
8332: doc-.s
8333: doc-f.s
8334:
8335:
8336: There is a word @code{.r} but it does @i{not} display the return
8337: stack! It is used for formatted numeric output.
8338:
8339:
8340: doc-depth
8341: doc-fdepth
8342: doc-clearstack
8343: doc-?
8344: doc-dump
8345:
8346:
8347: The word @code{~~} prints debugging information (by default the source
8348: location and the stack contents). It is easy to insert. If you use Emacs
8349: it is also easy to remove (@kbd{C-x ~} in the Emacs Forth mode to
8350: query-replace them with nothing). The deferred words
8351: @code{printdebugdata} and @code{printdebugline} control the output of
8352: @code{~~}. The default source location output format works well with
8353: Emacs' compilation mode, so you can step through the program at the
8354: source level using @kbd{C-x `} (the advantage over a stepping debugger
8355: is that you can step in any direction and you know where the crash has
8356: happened or where the strange data has occurred).
8357:
8358: The default actions of @code{~~} clobber the contents of the pictured
8359: numeric output string, so you should not use @code{~~}, e.g., between
8360: @code{<#} and @code{#>}.
8361:
8362:
8363: doc-~~
8364: doc-printdebugdata
8365: doc-printdebugline
8366:
8367: doc-see
8368: doc-marker
8369:
8370:
8371: Here's an example of using @code{marker} at the start of a source file
8372: that you are debugging; it ensures that you only ever have one copy of
8373: the file's definitions compiled at any time:
8374:
8375: @example
8376: [IFDEF] my-code
8377: my-code
8378: [ENDIF]
8379:
8380: marker my-code
8381: init-included-files
8382:
8383: \ .. definitions start here
8384: \ .
8385: \ .
8386: \ end
8387: @end example
8388:
8389:
8390:
8391: @node Assertions, Singlestep Debugger, Debugging, Programming Tools
8392: @subsection Assertions
8393: @cindex assertions
8394:
8395: It is a good idea to make your programs self-checking, especially if you
8396: make an assumption that may become invalid during maintenance (for
8397: example, that a certain field of a data structure is never zero). Gforth
8398: supports @dfn{assertions} for this purpose. They are used like this:
8399:
8400: @example
8401: assert( @i{flag} )
8402: @end example
8403:
8404: The code between @code{assert(} and @code{)} should compute a flag, that
8405: should be true if everything is alright and false otherwise. It should
8406: not change anything else on the stack. The overall stack effect of the
8407: assertion is @code{( -- )}. E.g.
8408:
8409: @example
8410: assert( 1 1 + 2 = ) \ what we learn in school
8411: assert( dup 0<> ) \ assert that the top of stack is not zero
8412: assert( false ) \ this code should not be reached
8413: @end example
8414:
8415: The need for assertions is different at different times. During
8416: debugging, we want more checking, in production we sometimes care more
8417: for speed. Therefore, assertions can be turned off, i.e., the assertion
8418: becomes a comment. Depending on the importance of an assertion and the
8419: time it takes to check it, you may want to turn off some assertions and
8420: keep others turned on. Gforth provides several levels of assertions for
8421: this purpose:
8422:
8423:
8424: doc-assert0(
8425: doc-assert1(
8426: doc-assert2(
8427: doc-assert3(
8428: doc-assert(
8429: doc-)
8430:
8431:
8432: The variable @code{assert-level} specifies the highest assertions that
8433: are turned on. I.e., at the default @code{assert-level} of one,
8434: @code{assert0(} and @code{assert1(} assertions perform checking, while
8435: @code{assert2(} and @code{assert3(} assertions are treated as comments.
8436:
8437: The value of @code{assert-level} is evaluated at compile-time, not at
8438: run-time. Therefore you cannot turn assertions on or off at run-time;
8439: you have to set the @code{assert-level} appropriately before compiling a
8440: piece of code. You can compile different pieces of code at different
8441: @code{assert-level}s (e.g., a trusted library at level 1 and
8442: newly-written code at level 3).
8443:
8444:
8445: doc-assert-level
8446:
8447:
8448: If an assertion fails, a message compatible with Emacs' compilation mode
8449: is produced and the execution is aborted (currently with @code{ABORT"}.
8450: If there is interest, we will introduce a special throw code. But if you
8451: intend to @code{catch} a specific condition, using @code{throw} is
8452: probably more appropriate than an assertion).
8453:
8454: Definitions in ANS Forth for these assertion words are provided
8455: in @file{compat/assert.fs}.
8456:
8457:
8458: @node Singlestep Debugger, , Assertions, Programming Tools
8459: @subsection Singlestep Debugger
8460: @cindex singlestep Debugger
8461: @cindex debugging Singlestep
8462:
8463: When you create a new word there's often the need to check whether it
8464: behaves correctly or not. You can do this by typing @code{dbg
8465: badword}. A debug session might look like this:
8466:
8467: @example
8468: : badword 0 DO i . LOOP ; ok
8469: 2 dbg badword
8470: : badword
8471: Scanning code...
8472:
8473: Nesting debugger ready!
8474:
8475: 400D4738 8049BC4 0 -> [ 2 ] 00002 00000
8476: 400D4740 8049F68 DO -> [ 0 ]
8477: 400D4744 804A0C8 i -> [ 1 ] 00000
8478: 400D4748 400C5E60 . -> 0 [ 0 ]
8479: 400D474C 8049D0C LOOP -> [ 0 ]
8480: 400D4744 804A0C8 i -> [ 1 ] 00001
8481: 400D4748 400C5E60 . -> 1 [ 0 ]
8482: 400D474C 8049D0C LOOP -> [ 0 ]
8483: 400D4758 804B384 ; -> ok
8484: @end example
8485:
8486: Each line displayed is one step. You always have to hit return to
8487: execute the next word that is displayed. If you don't want to execute
8488: the next word in a whole, you have to type @kbd{n} for @code{nest}. Here is
8489: an overview what keys are available:
8490:
8491: @table @i
8492:
8493: @item @key{RET}
8494: Next; Execute the next word.
8495:
8496: @item n
8497: Nest; Single step through next word.
8498:
8499: @item u
8500: Unnest; Stop debugging and execute rest of word. If we got to this word
8501: with nest, continue debugging with the calling word.
8502:
8503: @item d
8504: Done; Stop debugging and execute rest.
8505:
8506: @item s
8507: Stop; Abort immediately.
8508:
8509: @end table
8510:
8511: Debugging large application with this mechanism is very difficult, because
8512: you have to nest very deeply into the program before the interesting part
8513: begins. This takes a lot of time.
8514:
8515: To do it more directly put a @code{BREAK:} command into your source code.
8516: When program execution reaches @code{BREAK:} the single step debugger is
8517: invoked and you have all the features described above.
8518:
8519: If you have more than one part to debug it is useful to know where the
8520: program has stopped at the moment. You can do this by the
8521: @code{BREAK" string"} command. This behaves like @code{BREAK:} except that
8522: string is typed out when the ``breakpoint'' is reached.
8523:
8524:
8525: doc-dbg
8526: doc-break:
8527: doc-break"
8528:
8529:
8530:
8531: @c -------------------------------------------------------------
8532: @node Assembler and Code Words, Threading Words, Programming Tools, Words
8533: @section Assembler and Code Words
8534: @cindex assembler
8535: @cindex code words
8536:
8537: Gforth provides some words for defining primitives (words written in
8538: machine code), and for defining the machine-code equivalent of
8539: @code{DOES>}-based defining words. However, the machine-independent
8540: nature of Gforth poses a few problems: First of all, Gforth runs on
8541: several architectures, so it can provide no standard assembler. What's
8542: worse is that the register allocation not only depends on the processor,
8543: but also on the @code{gcc} version and options used.
8544:
8545: The words that Gforth offers encapsulate some system dependences (e.g.,
8546: the header structure), so a system-independent assembler may be used in
8547: Gforth. If you do not have an assembler, you can compile machine code
8548: directly with @code{,} and @code{c,}@footnote{This isn't portable,
8549: because these words emit stuff in @i{data} space; it works because
8550: Gforth has unified code/data spaces. Assembler isn't likely to be
8551: portable anyway.}.
8552:
8553:
8554: doc-assembler
8555: doc-init-asm
8556: doc-code
8557: doc-end-code
8558: doc-;code
8559: doc-flush-icache
8560:
8561:
8562: If @code{flush-icache} does not work correctly, @code{code} words
8563: etc. will not work (reliably), either.
8564:
8565: The typical usage of these @code{code} words can be shown most easily by
8566: analogy to the equivalent high-level defining words:
8567:
8568: @example
8569: : foo code foo
8570: <high-level Forth words> <assembler>
8571: ; end-code
8572:
8573: : bar : bar
8574: <high-level Forth words> <high-level Forth words>
8575: CREATE CREATE
8576: <high-level Forth words> <high-level Forth words>
8577: DOES> ;code
8578: <high-level Forth words> <assembler>
8579: ; end-code
8580: @end example
8581:
8582: @code{flush-icache} is always present. The other words are rarely used
8583: and reside in @code{code.fs}, which is usually not loaded. You can load
8584: it with @code{require code.fs}.
8585:
8586: @cindex registers of the inner interpreter
8587: In the assembly code you will want to refer to the inner interpreter's
8588: registers (e.g., the data stack pointer) and you may want to use other
8589: registers for temporary storage. Unfortunately, the register allocation
8590: is installation-dependent.
8591:
8592: The easiest solution is to use explicit register declarations
8593: (@pxref{Explicit Reg Vars, , Variables in Specified Registers, gcc.info,
8594: GNU C Manual}) for all of the inner interpreter's registers: You have to
8595: compile Gforth with @code{-DFORCE_REG} (configure option
8596: @code{--enable-force-reg}) and the appropriate declarations must be
8597: present in the @code{machine.h} file (see @code{mips.h} for an example;
8598: you can find a full list of all declarable register symbols with
8599: @code{grep register engine.c}). If you give explicit registers to all
8600: variables that are declared at the beginning of @code{engine()}, you
8601: should be able to use the other caller-saved registers for temporary
8602: storage. Alternatively, you can use the @code{gcc} option
8603: @code{-ffixed-REG} (@pxref{Code Gen Options, , Options for Code
8604: Generation Conventions, gcc.info, GNU C Manual}) to reserve a register
8605: (however, this restriction on register allocation may slow Gforth
8606: significantly).
8607:
8608: If this solution is not viable (e.g., because @code{gcc} does not allow
8609: you to explicitly declare all the registers you need), you have to find
8610: out by looking at the code where the inner interpreter's registers
8611: reside and which registers can be used for temporary storage. You can
8612: get an assembly listing of the engine's code with @code{make engine.s}.
8613:
8614: In any case, it is good practice to abstract your assembly code from the
8615: actual register allocation. E.g., if the data stack pointer resides in
8616: register @code{$17}, create an alias for this register called @code{sp},
8617: and use that in your assembly code.
8618:
8619: @cindex code words, portable
8620: Another option for implementing normal and defining words efficiently
8621: is to add the desired functionality to the source of Gforth. For normal
8622: words you just have to edit @file{primitives} (@pxref{Automatic
8623: Generation}). Defining words (equivalent to @code{;CODE} words, for fast
8624: defined words) may require changes in @file{engine.c}, @file{kernel.fs},
8625: @file{prims2x.fs}, and possibly @file{cross.fs}.
8626:
8627:
8628: @c -------------------------------------------------------------
8629: @node Threading Words, Locals, Assembler and Code Words, Words
8630: @section Threading Words
8631: @cindex threading words
8632:
8633: @cindex code address
8634: These words provide access to code addresses and other threading stuff
8635: in Gforth (and, possibly, other interpretive Forths). It more or less
8636: abstracts away the differences between direct and indirect threading
8637: (and, for direct threading, the machine dependences). However, at
8638: present this wordset is still incomplete. It is also pretty low-level;
8639: some day it will hopefully be made unnecessary by an internals wordset
8640: that abstracts implementation details away completely.
8641:
8642:
8643: doc-threading-method
8644: doc->code-address
8645: doc->does-code
8646: doc-code-address!
8647: doc-does-code!
8648: doc-does-handler!
8649: doc-/does-handler
8650:
8651:
8652: The code addresses produced by various defining words are produced by
8653: the following words:
8654:
8655:
8656: doc-docol:
8657: doc-docon:
8658: doc-dovar:
8659: doc-douser:
8660: doc-dodefer:
8661: doc-dofield:
8662:
8663:
8664: You can recognize words defined by a @code{CREATE}...@code{DOES>} word
8665: with @code{>does-code}. If the word was defined in that way, the value
8666: returned is non-zero and identifies the @code{DOES>} used by the
8667: defining word.
8668: @comment TODO should that be ``identifies the xt of the DOES> ??''
8669:
8670: @c -------------------------------------------------------------
8671: @node Locals, Structures, Threading Words, Words
8672: @section Locals
8673: @cindex locals
8674:
8675: Local variables can make Forth programming more enjoyable and Forth
8676: programs easier to read. Unfortunately, the locals of ANS Forth are
8677: laden with restrictions. Therefore, we provide not only the ANS Forth
8678: locals wordset, but also our own, more powerful locals wordset (we
8679: implemented the ANS Forth locals wordset through our locals wordset).
8680:
8681: The ideas in this section have also been published in the paper
8682: @cite{Automatic Scoping of Local Variables} by M. Anton Ertl, presented
8683: at EuroForth '94; it is available at
8684: @*@uref{http://www.complang.tuwien.ac.at/papers/ertl94l.ps.gz}.
8685:
8686: @menu
8687: * Gforth locals::
8688: * ANS Forth locals::
8689: @end menu
8690:
8691: @node Gforth locals, ANS Forth locals, Locals, Locals
8692: @subsection Gforth locals
8693: @cindex Gforth locals
8694: @cindex locals, Gforth style
8695:
8696: Locals can be defined with
8697:
8698: @example
8699: @{ local1 local2 ... -- comment @}
8700: @end example
8701: or
8702: @example
8703: @{ local1 local2 ... @}
8704: @end example
8705:
8706: E.g.,
8707: @example
8708: : max @{ n1 n2 -- n3 @}
8709: n1 n2 > if
8710: n1
8711: else
8712: n2
8713: endif ;
8714: @end example
8715:
8716: The similarity of locals definitions with stack comments is intended. A
8717: locals definition often replaces the stack comment of a word. The order
8718: of the locals corresponds to the order in a stack comment and everything
8719: after the @code{--} is really a comment.
8720:
8721: This similarity has one disadvantage: It is too easy to confuse locals
8722: declarations with stack comments, causing bugs and making them hard to
8723: find. However, this problem can be avoided by appropriate coding
8724: conventions: Do not use both notations in the same program. If you do,
8725: they should be distinguished using additional means, e.g. by position.
8726:
8727: @cindex types of locals
8728: @cindex locals types
8729: The name of the local may be preceded by a type specifier, e.g.,
8730: @code{F:} for a floating point value:
8731:
8732: @example
8733: : CX* @{ F: Ar F: Ai F: Br F: Bi -- Cr Ci @}
8734: \ complex multiplication
8735: Ar Br f* Ai Bi f* f-
8736: Ar Bi f* Ai Br f* f+ ;
8737: @end example
8738:
8739: @cindex flavours of locals
8740: @cindex locals flavours
8741: @cindex value-flavoured locals
8742: @cindex variable-flavoured locals
8743: Gforth currently supports cells (@code{W:}, @code{W^}), doubles
8744: (@code{D:}, @code{D^}), floats (@code{F:}, @code{F^}) and characters
8745: (@code{C:}, @code{C^}) in two flavours: a value-flavoured local (defined
8746: with @code{W:}, @code{D:} etc.) produces its value and can be changed
8747: with @code{TO}. A variable-flavoured local (defined with @code{W^} etc.)
8748: produces its address (which becomes invalid when the variable's scope is
8749: left). E.g., the standard word @code{emit} can be defined in terms of
8750: @code{type} like this:
8751:
8752: @example
8753: : emit @{ C^ char* -- @}
8754: char* 1 type ;
8755: @end example
8756:
8757: @cindex default type of locals
8758: @cindex locals, default type
8759: A local without type specifier is a @code{W:} local. Both flavours of
8760: locals are initialized with values from the data or FP stack.
8761:
8762: Currently there is no way to define locals with user-defined data
8763: structures, but we are working on it.
8764:
8765: Gforth allows defining locals everywhere in a colon definition. This
8766: poses the following questions:
8767:
8768: @menu
8769: * Where are locals visible by name?::
8770: * How long do locals live?::
8771: * Programming Style::
8772: * Implementation::
8773: @end menu
8774:
8775: @node Where are locals visible by name?, How long do locals live?, Gforth locals, Gforth locals
8776: @subsubsection Where are locals visible by name?
8777: @cindex locals visibility
8778: @cindex visibility of locals
8779: @cindex scope of locals
8780:
8781: Basically, the answer is that locals are visible where you would expect
8782: it in block-structured languages, and sometimes a little longer. If you
8783: want to restrict the scope of a local, enclose its definition in
8784: @code{SCOPE}...@code{ENDSCOPE}.
8785:
8786:
8787: doc-scope
8788: doc-endscope
8789:
8790:
8791: These words behave like control structure words, so you can use them
8792: with @code{CS-PICK} and @code{CS-ROLL} to restrict the scope in
8793: arbitrary ways.
8794:
8795: If you want a more exact answer to the visibility question, here's the
8796: basic principle: A local is visible in all places that can only be
8797: reached through the definition of the local@footnote{In compiler
8798: construction terminology, all places dominated by the definition of the
8799: local.}. In other words, it is not visible in places that can be reached
8800: without going through the definition of the local. E.g., locals defined
8801: in @code{IF}...@code{ENDIF} are visible until the @code{ENDIF}, locals
8802: defined in @code{BEGIN}...@code{UNTIL} are visible after the
8803: @code{UNTIL} (until, e.g., a subsequent @code{ENDSCOPE}).
8804:
8805: The reasoning behind this solution is: We want to have the locals
8806: visible as long as it is meaningful. The user can always make the
8807: visibility shorter by using explicit scoping. In a place that can
8808: only be reached through the definition of a local, the meaning of a
8809: local name is clear. In other places it is not: How is the local
8810: initialized at the control flow path that does not contain the
8811: definition? Which local is meant, if the same name is defined twice in
8812: two independent control flow paths?
8813:
8814: This should be enough detail for nearly all users, so you can skip the
8815: rest of this section. If you really must know all the gory details and
8816: options, read on.
8817:
8818: In order to implement this rule, the compiler has to know which places
8819: are unreachable. It knows this automatically after @code{AHEAD},
8820: @code{AGAIN}, @code{EXIT} and @code{LEAVE}; in other cases (e.g., after
8821: most @code{THROW}s), you can use the word @code{UNREACHABLE} to tell the
8822: compiler that the control flow never reaches that place. If
8823: @code{UNREACHABLE} is not used where it could, the only consequence is
8824: that the visibility of some locals is more limited than the rule above
8825: says. If @code{UNREACHABLE} is used where it should not (i.e., if you
8826: lie to the compiler), buggy code will be produced.
8827:
8828:
8829: doc-unreachable
8830:
8831:
8832: Another problem with this rule is that at @code{BEGIN}, the compiler
8833: does not know which locals will be visible on the incoming
8834: back-edge. All problems discussed in the following are due to this
8835: ignorance of the compiler (we discuss the problems using @code{BEGIN}
8836: loops as examples; the discussion also applies to @code{?DO} and other
8837: loops). Perhaps the most insidious example is:
8838: @example
8839: AHEAD
8840: BEGIN
8841: x
8842: [ 1 CS-ROLL ] THEN
8843: @{ x @}
8844: ...
8845: UNTIL
8846: @end example
8847:
8848: This should be legal according to the visibility rule. The use of
8849: @code{x} can only be reached through the definition; but that appears
8850: textually below the use.
8851:
8852: From this example it is clear that the visibility rules cannot be fully
8853: implemented without major headaches. Our implementation treats common
8854: cases as advertised and the exceptions are treated in a safe way: The
8855: compiler makes a reasonable guess about the locals visible after a
8856: @code{BEGIN}; if it is too pessimistic, the
8857: user will get a spurious error about the local not being defined; if the
8858: compiler is too optimistic, it will notice this later and issue a
8859: warning. In the case above the compiler would complain about @code{x}
8860: being undefined at its use. You can see from the obscure examples in
8861: this section that it takes quite unusual control structures to get the
8862: compiler into trouble, and even then it will often do fine.
8863:
8864: If the @code{BEGIN} is reachable from above, the most optimistic guess
8865: is that all locals visible before the @code{BEGIN} will also be
8866: visible after the @code{BEGIN}. This guess is valid for all loops that
8867: are entered only through the @code{BEGIN}, in particular, for normal
8868: @code{BEGIN}...@code{WHILE}...@code{REPEAT} and
8869: @code{BEGIN}...@code{UNTIL} loops and it is implemented in our
8870: compiler. When the branch to the @code{BEGIN} is finally generated by
8871: @code{AGAIN} or @code{UNTIL}, the compiler checks the guess and
8872: warns the user if it was too optimistic:
8873: @example
8874: IF
8875: @{ x @}
8876: BEGIN
8877: \ x ?
8878: [ 1 cs-roll ] THEN
8879: ...
8880: UNTIL
8881: @end example
8882:
8883: Here, @code{x} lives only until the @code{BEGIN}, but the compiler
8884: optimistically assumes that it lives until the @code{THEN}. It notices
8885: this difference when it compiles the @code{UNTIL} and issues a
8886: warning. The user can avoid the warning, and make sure that @code{x}
8887: is not used in the wrong area by using explicit scoping:
8888: @example
8889: IF
8890: SCOPE
8891: @{ x @}
8892: ENDSCOPE
8893: BEGIN
8894: [ 1 cs-roll ] THEN
8895: ...
8896: UNTIL
8897: @end example
8898:
8899: Since the guess is optimistic, there will be no spurious error messages
8900: about undefined locals.
8901:
8902: If the @code{BEGIN} is not reachable from above (e.g., after
8903: @code{AHEAD} or @code{EXIT}), the compiler cannot even make an
8904: optimistic guess, as the locals visible after the @code{BEGIN} may be
8905: defined later. Therefore, the compiler assumes that no locals are
8906: visible after the @code{BEGIN}. However, the user can use
8907: @code{ASSUME-LIVE} to make the compiler assume that the same locals are
8908: visible at the BEGIN as at the point where the top control-flow stack
8909: item was created.
8910:
8911:
8912: doc-assume-live
8913:
8914:
8915: @noindent
8916: E.g.,
8917: @example
8918: @{ x @}
8919: AHEAD
8920: ASSUME-LIVE
8921: BEGIN
8922: x
8923: [ 1 CS-ROLL ] THEN
8924: ...
8925: UNTIL
8926: @end example
8927:
8928: Other cases where the locals are defined before the @code{BEGIN} can be
8929: handled by inserting an appropriate @code{CS-ROLL} before the
8930: @code{ASSUME-LIVE} (and changing the control-flow stack manipulation
8931: behind the @code{ASSUME-LIVE}).
8932:
8933: Cases where locals are defined after the @code{BEGIN} (but should be
8934: visible immediately after the @code{BEGIN}) can only be handled by
8935: rearranging the loop. E.g., the ``most insidious'' example above can be
8936: arranged into:
8937: @example
8938: BEGIN
8939: @{ x @}
8940: ... 0=
8941: WHILE
8942: x
8943: REPEAT
8944: @end example
8945:
8946: @node How long do locals live?, Programming Style, Where are locals visible by name?, Gforth locals
8947: @subsubsection How long do locals live?
8948: @cindex locals lifetime
8949: @cindex lifetime of locals
8950:
8951: The right answer for the lifetime question would be: A local lives at
8952: least as long as it can be accessed. For a value-flavoured local this
8953: means: until the end of its visibility. However, a variable-flavoured
8954: local could be accessed through its address far beyond its visibility
8955: scope. Ultimately, this would mean that such locals would have to be
8956: garbage collected. Since this entails un-Forth-like implementation
8957: complexities, I adopted the same cowardly solution as some other
8958: languages (e.g., C): The local lives only as long as it is visible;
8959: afterwards its address is invalid (and programs that access it
8960: afterwards are erroneous).
8961:
8962: @node Programming Style, Implementation, How long do locals live?, Gforth locals
8963: @subsubsection Programming Style
8964: @cindex locals programming style
8965: @cindex programming style, locals
8966:
8967: The freedom to define locals anywhere has the potential to change
8968: programming styles dramatically. In particular, the need to use the
8969: return stack for intermediate storage vanishes. Moreover, all stack
8970: manipulations (except @code{PICK}s and @code{ROLL}s with run-time
8971: determined arguments) can be eliminated: If the stack items are in the
8972: wrong order, just write a locals definition for all of them; then
8973: write the items in the order you want.
8974:
8975: This seems a little far-fetched and eliminating stack manipulations is
8976: unlikely to become a conscious programming objective. Still, the number
8977: of stack manipulations will be reduced dramatically if local variables
8978: are used liberally (e.g., compare @code{max} in @ref{Gforth locals} with
8979: a traditional implementation of @code{max}).
8980:
8981: This shows one potential benefit of locals: making Forth programs more
8982: readable. Of course, this benefit will only be realized if the
8983: programmers continue to honour the principle of factoring instead of
8984: using the added latitude to make the words longer.
8985:
8986: @cindex single-assignment style for locals
8987: Using @code{TO} can and should be avoided. Without @code{TO},
8988: every value-flavoured local has only a single assignment and many
8989: advantages of functional languages apply to Forth. I.e., programs are
8990: easier to analyse, to optimize and to read: It is clear from the
8991: definition what the local stands for, it does not turn into something
8992: different later.
8993:
8994: E.g., a definition using @code{TO} might look like this:
8995: @example
8996: : strcmp @{ addr1 u1 addr2 u2 -- n @}
8997: u1 u2 min 0
8998: ?do
8999: addr1 c@@ addr2 c@@ -
9000: ?dup-if
9001: unloop exit
9002: then
9003: addr1 char+ TO addr1
9004: addr2 char+ TO addr2
9005: loop
9006: u1 u2 - ;
9007: @end example
9008: Here, @code{TO} is used to update @code{addr1} and @code{addr2} at
9009: every loop iteration. @code{strcmp} is a typical example of the
9010: readability problems of using @code{TO}. When you start reading
9011: @code{strcmp}, you think that @code{addr1} refers to the start of the
9012: string. Only near the end of the loop you realize that it is something
9013: else.
9014:
9015: This can be avoided by defining two locals at the start of the loop that
9016: are initialized with the right value for the current iteration.
9017: @example
9018: : strcmp @{ addr1 u1 addr2 u2 -- n @}
9019: addr1 addr2
9020: u1 u2 min 0
9021: ?do @{ s1 s2 @}
9022: s1 c@@ s2 c@@ -
9023: ?dup-if
9024: unloop exit
9025: then
9026: s1 char+ s2 char+
9027: loop
9028: 2drop
9029: u1 u2 - ;
9030: @end example
9031: Here it is clear from the start that @code{s1} has a different value
9032: in every loop iteration.
9033:
9034: @node Implementation, , Programming Style, Gforth locals
9035: @subsubsection Implementation
9036: @cindex locals implementation
9037: @cindex implementation of locals
9038:
9039: @cindex locals stack
9040: Gforth uses an extra locals stack. The most compelling reason for
9041: this is that the return stack is not float-aligned; using an extra stack
9042: also eliminates the problems and restrictions of using the return stack
9043: as locals stack. Like the other stacks, the locals stack grows toward
9044: lower addresses. A few primitives allow an efficient implementation:
9045:
9046:
9047: doc-@local#
9048: doc-f@local#
9049: doc-laddr#
9050: doc-lp+!#
9051: doc-lp!
9052: doc->l
9053: doc-f>l
9054:
9055:
9056: In addition to these primitives, some specializations of these
9057: primitives for commonly occurring inline arguments are provided for
9058: efficiency reasons, e.g., @code{@@local0} as specialization of
9059: @code{@@local#} for the inline argument 0. The following compiling words
9060: compile the right specialized version, or the general version, as
9061: appropriate:
9062:
9063:
9064: doc-compile-@local
9065: doc-compile-f@local
9066: doc-compile-lp+!
9067:
9068:
9069: Combinations of conditional branches and @code{lp+!#} like
9070: @code{?branch-lp+!#} (the locals pointer is only changed if the branch
9071: is taken) are provided for efficiency and correctness in loops.
9072:
9073: A special area in the dictionary space is reserved for keeping the
9074: local variable names. @code{@{} switches the dictionary pointer to this
9075: area and @code{@}} switches it back and generates the locals
9076: initializing code. @code{W:} etc.@ are normal defining words. This
9077: special area is cleared at the start of every colon definition.
9078:
9079: @cindex word list for defining locals
9080: A special feature of Gforth's dictionary is used to implement the
9081: definition of locals without type specifiers: every word list (aka
9082: vocabulary) has its own methods for searching
9083: etc. (@pxref{Word Lists}). For the present purpose we defined a word list
9084: with a special search method: When it is searched for a word, it
9085: actually creates that word using @code{W:}. @code{@{} changes the search
9086: order to first search the word list containing @code{@}}, @code{W:} etc.,
9087: and then the word list for defining locals without type specifiers.
9088:
9089: The lifetime rules support a stack discipline within a colon
9090: definition: The lifetime of a local is either nested with other locals
9091: lifetimes or it does not overlap them.
9092:
9093: At @code{BEGIN}, @code{IF}, and @code{AHEAD} no code for locals stack
9094: pointer manipulation is generated. Between control structure words
9095: locals definitions can push locals onto the locals stack. @code{AGAIN}
9096: is the simplest of the other three control flow words. It has to
9097: restore the locals stack depth of the corresponding @code{BEGIN}
9098: before branching. The code looks like this:
9099: @format
9100: @code{lp+!#} current-locals-size @minus{} dest-locals-size
9101: @code{branch} <begin>
9102: @end format
9103:
9104: @code{UNTIL} is a little more complicated: If it branches back, it
9105: must adjust the stack just like @code{AGAIN}. But if it falls through,
9106: the locals stack must not be changed. The compiler generates the
9107: following code:
9108: @format
9109: @code{?branch-lp+!#} <begin> current-locals-size @minus{} dest-locals-size
9110: @end format
9111: The locals stack pointer is only adjusted if the branch is taken.
9112:
9113: @code{THEN} can produce somewhat inefficient code:
9114: @format
9115: @code{lp+!#} current-locals-size @minus{} orig-locals-size
9116: <orig target>:
9117: @code{lp+!#} orig-locals-size @minus{} new-locals-size
9118: @end format
9119: The second @code{lp+!#} adjusts the locals stack pointer from the
9120: level at the @i{orig} point to the level after the @code{THEN}. The
9121: first @code{lp+!#} adjusts the locals stack pointer from the current
9122: level to the level at the orig point, so the complete effect is an
9123: adjustment from the current level to the right level after the
9124: @code{THEN}.
9125:
9126: @cindex locals information on the control-flow stack
9127: @cindex control-flow stack items, locals information
9128: In a conventional Forth implementation a dest control-flow stack entry
9129: is just the target address and an orig entry is just the address to be
9130: patched. Our locals implementation adds a word list to every orig or dest
9131: item. It is the list of locals visible (or assumed visible) at the point
9132: described by the entry. Our implementation also adds a tag to identify
9133: the kind of entry, in particular to differentiate between live and dead
9134: (reachable and unreachable) orig entries.
9135:
9136: A few unusual operations have to be performed on locals word lists:
9137:
9138:
9139: doc-common-list
9140: doc-sub-list?
9141: doc-list-size
9142:
9143:
9144: Several features of our locals word list implementation make these
9145: operations easy to implement: The locals word lists are organised as
9146: linked lists; the tails of these lists are shared, if the lists
9147: contain some of the same locals; and the address of a name is greater
9148: than the address of the names behind it in the list.
9149:
9150: Another important implementation detail is the variable
9151: @code{dead-code}. It is used by @code{BEGIN} and @code{THEN} to
9152: determine if they can be reached directly or only through the branch
9153: that they resolve. @code{dead-code} is set by @code{UNREACHABLE},
9154: @code{AHEAD}, @code{EXIT} etc., and cleared at the start of a colon
9155: definition, by @code{BEGIN} and usually by @code{THEN}.
9156:
9157: Counted loops are similar to other loops in most respects, but
9158: @code{LEAVE} requires special attention: It performs basically the same
9159: service as @code{AHEAD}, but it does not create a control-flow stack
9160: entry. Therefore the information has to be stored elsewhere;
9161: traditionally, the information was stored in the target fields of the
9162: branches created by the @code{LEAVE}s, by organizing these fields into a
9163: linked list. Unfortunately, this clever trick does not provide enough
9164: space for storing our extended control flow information. Therefore, we
9165: introduce another stack, the leave stack. It contains the control-flow
9166: stack entries for all unresolved @code{LEAVE}s.
9167:
9168: Local names are kept until the end of the colon definition, even if
9169: they are no longer visible in any control-flow path. In a few cases
9170: this may lead to increased space needs for the locals name area, but
9171: usually less than reclaiming this space would cost in code size.
9172:
9173:
9174: @node ANS Forth locals, , Gforth locals, Locals
9175: @subsection ANS Forth locals
9176: @cindex locals, ANS Forth style
9177:
9178: The ANS Forth locals wordset does not define a syntax for locals, but
9179: words that make it possible to define various syntaxes. One of the
9180: possible syntaxes is a subset of the syntax we used in the Gforth locals
9181: wordset, i.e.:
9182:
9183: @example
9184: @{ local1 local2 ... -- comment @}
9185: @end example
9186: @noindent
9187: or
9188: @example
9189: @{ local1 local2 ... @}
9190: @end example
9191:
9192: The order of the locals corresponds to the order in a stack comment. The
9193: restrictions are:
9194:
9195: @itemize @bullet
9196: @item
9197: Locals can only be cell-sized values (no type specifiers are allowed).
9198: @item
9199: Locals can be defined only outside control structures.
9200: @item
9201: Locals can interfere with explicit usage of the return stack. For the
9202: exact (and long) rules, see the standard. If you don't use return stack
9203: accessing words in a definition using locals, you will be all right. The
9204: purpose of this rule is to make locals implementation on the return
9205: stack easier.
9206: @item
9207: The whole definition must be in one line.
9208: @end itemize
9209:
9210: Locals defined in this way behave like @code{VALUE}s
9211: (@xref{Values}). I.e., they are initialized from the stack. Using their
9212: name produces their value. Their value can be changed using @code{TO}.
9213:
9214: Since this syntax is supported by Gforth directly, you need not do
9215: anything to use it. If you want to port a program using this syntax to
9216: another ANS Forth system, use @file{compat/anslocal.fs} to implement the
9217: syntax on the other system.
9218:
9219: Note that a syntax shown in the standard, section A.13 looks
9220: similar, but is quite different in having the order of locals
9221: reversed. Beware!
9222:
9223: The ANS Forth locals wordset itself consists of a word:
9224:
9225:
9226: doc-(local)
9227:
9228:
9229: The ANS Forth locals extension wordset defines a syntax using @code{locals|}, but it is so
9230: awful that we strongly recommend not to use it. We have implemented this
9231: syntax to make porting to Gforth easy, but do not document it here. The
9232: problem with this syntax is that the locals are defined in an order
9233: reversed with respect to the standard stack comment notation, making
9234: programs harder to read, and easier to misread and miswrite. The only
9235: merit of this syntax is that it is easy to implement using the ANS Forth
9236: locals wordset.
9237:
9238:
9239: @c ----------------------------------------------------------
9240: @node Structures, Object-oriented Forth, Locals, Words
9241: @section Structures
9242: @cindex structures
9243: @cindex records
9244:
9245: This section presents the structure package that comes with Gforth. A
9246: version of the package implemented in ANS Forth is available in
9247: @file{compat/struct.fs}. This package was inspired by a posting on
9248: comp.lang.forth in 1989 (unfortunately I don't remember, by whom;
9249: possibly John Hayes). A version of this section has been published in
9250: ???. Marcel Hendrix provided helpful comments.
9251:
9252: @menu
9253: * Why explicit structure support?::
9254: * Structure Usage::
9255: * Structure Naming Convention::
9256: * Structure Implementation::
9257: * Structure Glossary::
9258: @end menu
9259:
9260: @node Why explicit structure support?, Structure Usage, Structures, Structures
9261: @subsection Why explicit structure support?
9262:
9263: @cindex address arithmetic for structures
9264: @cindex structures using address arithmetic
9265: If we want to use a structure containing several fields, we could simply
9266: reserve memory for it, and access the fields using address arithmetic
9267: (@pxref{Address arithmetic}). As an example, consider a structure with
9268: the following fields
9269:
9270: @table @code
9271: @item a
9272: is a float
9273: @item b
9274: is a cell
9275: @item c
9276: is a float
9277: @end table
9278:
9279: Given the (float-aligned) base address of the structure we get the
9280: address of the field
9281:
9282: @table @code
9283: @item a
9284: without doing anything further.
9285: @item b
9286: with @code{float+}
9287: @item c
9288: with @code{float+ cell+ faligned}
9289: @end table
9290:
9291: It is easy to see that this can become quite tiring.
9292:
9293: Moreover, it is not very readable, because seeing a
9294: @code{cell+} tells us neither which kind of structure is
9295: accessed nor what field is accessed; we have to somehow infer the kind
9296: of structure, and then look up in the documentation, which field of
9297: that structure corresponds to that offset.
9298:
9299: Finally, this kind of address arithmetic also causes maintenance
9300: troubles: If you add or delete a field somewhere in the middle of the
9301: structure, you have to find and change all computations for the fields
9302: afterwards.
9303:
9304: So, instead of using @code{cell+} and friends directly, how
9305: about storing the offsets in constants:
9306:
9307: @example
9308: 0 constant a-offset
9309: 0 float+ constant b-offset
9310: 0 float+ cell+ faligned c-offset
9311: @end example
9312:
9313: Now we can get the address of field @code{x} with @code{x-offset
9314: +}. This is much better in all respects. Of course, you still
9315: have to change all later offset definitions if you add a field. You can
9316: fix this by declaring the offsets in the following way:
9317:
9318: @example
9319: 0 constant a-offset
9320: a-offset float+ constant b-offset
9321: b-offset cell+ faligned constant c-offset
9322: @end example
9323:
9324: Since we always use the offsets with @code{+}, we could use a defining
9325: word @code{cfield} that includes the @code{+} in the action of the
9326: defined word:
9327:
9328: @example
9329: : cfield ( n "name" -- )
9330: create ,
9331: does> ( name execution: addr1 -- addr2 )
9332: @@ + ;
9333:
9334: 0 cfield a
9335: 0 a float+ cfield b
9336: 0 b cell+ faligned cfield c
9337: @end example
9338:
9339: Instead of @code{x-offset +}, we now simply write @code{x}.
9340:
9341: The structure field words now can be used quite nicely. However,
9342: their definition is still a bit cumbersome: We have to repeat the
9343: name, the information about size and alignment is distributed before
9344: and after the field definitions etc. The structure package presented
9345: here addresses these problems.
9346:
9347: @node Structure Usage, Structure Naming Convention, Why explicit structure support?, Structures
9348: @subsection Structure Usage
9349: @cindex structure usage
9350:
9351: @cindex @code{field} usage
9352: @cindex @code{struct} usage
9353: @cindex @code{end-struct} usage
9354: You can define a structure for a (data-less) linked list with:
9355: @example
9356: struct
9357: cell% field list-next
9358: end-struct list%
9359: @end example
9360:
9361: With the address of the list node on the stack, you can compute the
9362: address of the field that contains the address of the next node with
9363: @code{list-next}. E.g., you can determine the length of a list
9364: with:
9365:
9366: @example
9367: : list-length ( list -- n )
9368: \ "list" is a pointer to the first element of a linked list
9369: \ "n" is the length of the list
9370: 0 BEGIN ( list1 n1 )
9371: over
9372: WHILE ( list1 n1 )
9373: 1+ swap list-next @@ swap
9374: REPEAT
9375: nip ;
9376: @end example
9377:
9378: You can reserve memory for a list node in the dictionary with
9379: @code{list% %allot}, which leaves the address of the list node on the
9380: stack. For the equivalent allocation on the heap you can use @code{list%
9381: %alloc} (or, for an @code{allocate}-like stack effect (i.e., with ior),
9382: use @code{list% %allocate}). You can get the the size of a list
9383: node with @code{list% %size} and its alignment with @code{list%
9384: %alignment}.
9385:
9386: Note that in ANS Forth the body of a @code{create}d word is
9387: @code{aligned} but not necessarily @code{faligned};
9388: therefore, if you do a:
9389: @example
9390: create @emph{name} foo% %allot
9391: @end example
9392:
9393: @noindent
9394: then the memory alloted for @code{foo%} is
9395: guaranteed to start at the body of @code{@emph{name}} only if
9396: @code{foo%} contains only character, cell and double fields.
9397:
9398: @cindex structures containing structures
9399: You can include a structure @code{foo%} as a field of
9400: another structure, like this:
9401: @example
9402: struct
9403: ...
9404: foo% field ...
9405: ...
9406: end-struct ...
9407: @end example
9408:
9409: @cindex structure extension
9410: @cindex extended records
9411: Instead of starting with an empty structure, you can extend an
9412: existing structure. E.g., a plain linked list without data, as defined
9413: above, is hardly useful; You can extend it to a linked list of integers,
9414: like this:@footnote{This feature is also known as @emph{extended
9415: records}. It is the main innovation in the Oberon language; in other
9416: words, adding this feature to Modula-2 led Wirth to create a new
9417: language, write a new compiler etc. Adding this feature to Forth just
9418: required a few lines of code.}
9419:
9420: @example
9421: list%
9422: cell% field intlist-int
9423: end-struct intlist%
9424: @end example
9425:
9426: @code{intlist%} is a structure with two fields:
9427: @code{list-next} and @code{intlist-int}.
9428:
9429: @cindex structures containing arrays
9430: You can specify an array type containing @emph{n} elements of
9431: type @code{foo%} like this:
9432:
9433: @example
9434: foo% @emph{n} *
9435: @end example
9436:
9437: You can use this array type in any place where you can use a normal
9438: type, e.g., when defining a @code{field}, or with
9439: @code{%allot}.
9440:
9441: @cindex first field optimization
9442: The first field is at the base address of a structure and the word
9443: for this field (e.g., @code{list-next}) actually does not change
9444: the address on the stack. You may be tempted to leave it away in the
9445: interest of run-time and space efficiency. This is not necessary,
9446: because the structure package optimizes this case and compiling such
9447: words does not generate any code. So, in the interest of readability
9448: and maintainability you should include the word for the field when
9449: accessing the field.
9450:
9451: @node Structure Naming Convention, Structure Implementation, Structure Usage, Structures
9452: @subsection Structure Naming Convention
9453: @cindex structure naming convention
9454:
9455: The field names that come to (my) mind are often quite generic, and,
9456: if used, would cause frequent name clashes. E.g., many structures
9457: probably contain a @code{counter} field. The structure names
9458: that come to (my) mind are often also the logical choice for the names
9459: of words that create such a structure.
9460:
9461: Therefore, I have adopted the following naming conventions:
9462:
9463: @itemize @bullet
9464: @cindex field naming convention
9465: @item
9466: The names of fields are of the form
9467: @code{@emph{struct}-@emph{field}}, where
9468: @code{@emph{struct}} is the basic name of the structure, and
9469: @code{@emph{field}} is the basic name of the field. You can
9470: think of field words as converting the (address of the)
9471: structure into the (address of the) field.
9472:
9473: @cindex structure naming convention
9474: @item
9475: The names of structures are of the form
9476: @code{@emph{struct}%}, where
9477: @code{@emph{struct}} is the basic name of the structure.
9478: @end itemize
9479:
9480: This naming convention does not work that well for fields of extended
9481: structures; e.g., the integer list structure has a field
9482: @code{intlist-int}, but has @code{list-next}, not
9483: @code{intlist-next}.
9484:
9485: @node Structure Implementation, Structure Glossary, Structure Naming Convention, Structures
9486: @subsection Structure Implementation
9487: @cindex structure implementation
9488: @cindex implementation of structures
9489:
9490: The central idea in the implementation is to pass the data about the
9491: structure being built on the stack, not in some global
9492: variable. Everything else falls into place naturally once this design
9493: decision is made.
9494:
9495: The type description on the stack is of the form @emph{align
9496: size}. Keeping the size on the top-of-stack makes dealing with arrays
9497: very simple.
9498:
9499: @code{field} is a defining word that uses @code{Create}
9500: and @code{DOES>}. The body of the field contains the offset
9501: of the field, and the normal @code{DOES>} action is simply:
9502:
9503: @example
9504: @@ +
9505: @end example
9506:
9507: @noindent
9508: i.e., add the offset to the address, giving the stack effect
9509: @i{addr1 -- addr2} for a field.
9510:
9511: @cindex first field optimization, implementation
9512: This simple structure is slightly complicated by the optimization
9513: for fields with offset 0, which requires a different
9514: @code{DOES>}-part (because we cannot rely on there being
9515: something on the stack if such a field is invoked during
9516: compilation). Therefore, we put the different @code{DOES>}-parts
9517: in separate words, and decide which one to invoke based on the
9518: offset. For a zero offset, the field is basically a noop; it is
9519: immediate, and therefore no code is generated when it is compiled.
9520:
9521: @node Structure Glossary, , Structure Implementation, Structures
9522: @subsection Structure Glossary
9523: @cindex structure glossary
9524:
9525:
9526: doc-%align
9527: doc-%alignment
9528: doc-%alloc
9529: doc-%allocate
9530: doc-%allot
9531: doc-cell%
9532: doc-char%
9533: doc-dfloat%
9534: doc-double%
9535: doc-end-struct
9536: doc-field
9537: doc-float%
9538: doc-naligned
9539: doc-sfloat%
9540: doc-%size
9541: doc-struct
9542:
9543:
9544: @c -------------------------------------------------------------
9545: @node Object-oriented Forth, Passing Commands to the OS, Structures, Words
9546: @section Object-oriented Forth
9547:
9548: Gforth comes with three packages for object-oriented programming:
9549: @file{objects.fs}, @file{oof.fs}, and @file{mini-oof.fs}; none of them
9550: is preloaded, so you have to @code{include} them before use. The most
9551: important differences between these packages (and others) are discussed
9552: in @ref{Comparison with other object models}. All packages are written
9553: in ANS Forth and can be used with any other ANS Forth.
9554:
9555: @menu
9556: * Why object-oriented programming?::
9557: * Object-Oriented Terminology::
9558: * Objects::
9559: * OOF::
9560: * Mini-OOF::
9561: * Comparison with other object models::
9562: @end menu
9563:
9564: @c ----------------------------------------------------------------
9565: @node Why object-oriented programming?, Object-Oriented Terminology, Object-oriented Forth, Object-oriented Forth
9566: @subsection Why object-oriented programming?
9567: @cindex object-oriented programming motivation
9568: @cindex motivation for object-oriented programming
9569:
9570: Often we have to deal with several data structures (@emph{objects}),
9571: that have to be treated similarly in some respects, but differently in
9572: others. Graphical objects are the textbook example: circles, triangles,
9573: dinosaurs, icons, and others, and we may want to add more during program
9574: development. We want to apply some operations to any graphical object,
9575: e.g., @code{draw} for displaying it on the screen. However, @code{draw}
9576: has to do something different for every kind of object.
9577: @comment TODO add some other operations eg perimeter, area
9578: @comment and tie in to concrete examples later..
9579:
9580: We could implement @code{draw} as a big @code{CASE}
9581: control structure that executes the appropriate code depending on the
9582: kind of object to be drawn. This would be not be very elegant, and,
9583: moreover, we would have to change @code{draw} every time we add
9584: a new kind of graphical object (say, a spaceship).
9585:
9586: What we would rather do is: When defining spaceships, we would tell
9587: the system: ``Here's how you @code{draw} a spaceship; you figure
9588: out the rest''.
9589:
9590: This is the problem that all systems solve that (rightfully) call
9591: themselves object-oriented; the object-oriented packages presented here
9592: solve this problem (and not much else).
9593: @comment TODO ?list properties of oo systems.. oo vs o-based?
9594:
9595: @c ------------------------------------------------------------------------
9596: @node Object-Oriented Terminology, Objects, Why object-oriented programming?, Object-oriented Forth
9597: @subsection Object-Oriented Terminology
9598: @cindex object-oriented terminology
9599: @cindex terminology for object-oriented programming
9600:
9601: This section is mainly for reference, so you don't have to understand
9602: all of it right away. The terminology is mainly Smalltalk-inspired. In
9603: short:
9604:
9605: @table @emph
9606: @cindex class
9607: @item class
9608: a data structure definition with some extras.
9609:
9610: @cindex object
9611: @item object
9612: an instance of the data structure described by the class definition.
9613:
9614: @cindex instance variables
9615: @item instance variables
9616: fields of the data structure.
9617:
9618: @cindex selector
9619: @cindex method selector
9620: @cindex virtual function
9621: @item selector
9622: (or @emph{method selector}) a word (e.g.,
9623: @code{draw}) that performs an operation on a variety of data
9624: structures (classes). A selector describes @emph{what} operation to
9625: perform. In C++ terminology: a (pure) virtual function.
9626:
9627: @cindex method
9628: @item method
9629: the concrete definition that performs the operation
9630: described by the selector for a specific class. A method specifies
9631: @emph{how} the operation is performed for a specific class.
9632:
9633: @cindex selector invocation
9634: @cindex message send
9635: @cindex invoking a selector
9636: @item selector invocation
9637: a call of a selector. One argument of the call (the TOS (top-of-stack))
9638: is used for determining which method is used. In Smalltalk terminology:
9639: a message (consisting of the selector and the other arguments) is sent
9640: to the object.
9641:
9642: @cindex receiving object
9643: @item receiving object
9644: the object used for determining the method executed by a selector
9645: invocation. In the @file{objects.fs} model, it is the object that is on
9646: the TOS when the selector is invoked. (@emph{Receiving} comes from
9647: the Smalltalk @emph{message} terminology.)
9648:
9649: @cindex child class
9650: @cindex parent class
9651: @cindex inheritance
9652: @item child class
9653: a class that has (@emph{inherits}) all properties (instance variables,
9654: selectors, methods) from a @emph{parent class}. In Smalltalk
9655: terminology: The subclass inherits from the superclass. In C++
9656: terminology: The derived class inherits from the base class.
9657:
9658: @end table
9659:
9660: @c If you wonder about the message sending terminology, it comes from
9661: @c a time when each object had it's own task and objects communicated via
9662: @c message passing; eventually the Smalltalk developers realized that
9663: @c they can do most things through simple (indirect) calls. They kept the
9664: @c terminology.
9665:
9666: @c --------------------------------------------------------------
9667: @node Objects, OOF, Object-Oriented Terminology, Object-oriented Forth
9668: @subsection The @file{objects.fs} model
9669: @cindex objects
9670: @cindex object-oriented programming
9671:
9672: @cindex @file{objects.fs}
9673: @cindex @file{oof.fs}
9674:
9675: This section describes the @file{objects.fs} package. This material also
9676: has been published in @cite{Yet Another Forth Objects Package} by Anton
9677: Ertl and appeared in Forth Dimensions 19(2), pages 37--43
9678: (@uref{http://www.complang.tuwien.ac.at/forth/objects/objects.html}).
9679: @c McKewan's and Zsoter's packages
9680:
9681: This section assumes that you have read @ref{Structures}.
9682:
9683: The techniques on which this model is based have been used to implement
9684: the parser generator, Gray, and have also been used in Gforth for
9685: implementing the various flavours of word lists (hashed or not,
9686: case-sensitive or not, special-purpose word lists for locals etc.).
9687:
9688:
9689: @menu
9690: * Properties of the Objects model::
9691: * Basic Objects Usage::
9692: * The Objects base class::
9693: * Creating objects::
9694: * Object-Oriented Programming Style::
9695: * Class Binding::
9696: * Method conveniences::
9697: * Classes and Scoping::
9698: * Dividing classes::
9699: * Object Interfaces::
9700: * Objects Implementation::
9701: * Objects Glossary::
9702: @end menu
9703:
9704: Marcel Hendrix provided helpful comments on this section. Andras Zsoter
9705: and Bernd Paysan helped me with the related works section.
9706:
9707: @node Properties of the Objects model, Basic Objects Usage, Objects, Objects
9708: @subsubsection Properties of the @file{objects.fs} model
9709: @cindex @file{objects.fs} properties
9710:
9711: @itemize @bullet
9712: @item
9713: It is straightforward to pass objects on the stack. Passing
9714: selectors on the stack is a little less convenient, but possible.
9715:
9716: @item
9717: Objects are just data structures in memory, and are referenced by their
9718: address. You can create words for objects with normal defining words
9719: like @code{constant}. Likewise, there is no difference between instance
9720: variables that contain objects and those that contain other data.
9721:
9722: @item
9723: Late binding is efficient and easy to use.
9724:
9725: @item
9726: It avoids parsing, and thus avoids problems with state-smartness
9727: and reduced extensibility; for convenience there are a few parsing
9728: words, but they have non-parsing counterparts. There are also a few
9729: defining words that parse. This is hard to avoid, because all standard
9730: defining words parse (except @code{:noname}); however, such
9731: words are not as bad as many other parsing words, because they are not
9732: state-smart.
9733:
9734: @item
9735: It does not try to incorporate everything. It does a few things and does
9736: them well (IMO). In particular, this model was not designed to support
9737: information hiding (although it has features that may help); you can use
9738: a separate package for achieving this.
9739:
9740: @item
9741: It is layered; you don't have to learn and use all features to use this
9742: model. Only a few features are necessary (@xref{Basic Objects Usage},
9743: @xref{The Objects base class}, @xref{Creating objects}.), the others
9744: are optional and independent of each other.
9745:
9746: @item
9747: An implementation in ANS Forth is available.
9748:
9749: @end itemize
9750:
9751:
9752: @node Basic Objects Usage, The Objects base class, Properties of the Objects model, Objects
9753: @subsubsection Basic @file{objects.fs} Usage
9754: @cindex basic objects usage
9755: @cindex objects, basic usage
9756:
9757: You can define a class for graphical objects like this:
9758:
9759: @cindex @code{class} usage
9760: @cindex @code{end-class} usage
9761: @cindex @code{selector} usage
9762: @example
9763: object class \ "object" is the parent class
9764: selector draw ( x y graphical -- )
9765: end-class graphical
9766: @end example
9767:
9768: This code defines a class @code{graphical} with an
9769: operation @code{draw}. We can perform the operation
9770: @code{draw} on any @code{graphical} object, e.g.:
9771:
9772: @example
9773: 100 100 t-rex draw
9774: @end example
9775:
9776: @noindent
9777: where @code{t-rex} is a word (say, a constant) that produces a
9778: graphical object.
9779:
9780: @comment TODO add a 2nd operation eg perimeter.. and use for
9781: @comment a concrete example
9782:
9783: @cindex abstract class
9784: How do we create a graphical object? With the present definitions,
9785: we cannot create a useful graphical object. The class
9786: @code{graphical} describes graphical objects in general, but not
9787: any concrete graphical object type (C++ users would call it an
9788: @emph{abstract class}); e.g., there is no method for the selector
9789: @code{draw} in the class @code{graphical}.
9790:
9791: For concrete graphical objects, we define child classes of the
9792: class @code{graphical}, e.g.:
9793:
9794: @cindex @code{overrides} usage
9795: @cindex @code{field} usage in class definition
9796: @example
9797: graphical class \ "graphical" is the parent class
9798: cell% field circle-radius
9799:
9800: :noname ( x y circle -- )
9801: circle-radius @@ draw-circle ;
9802: overrides draw
9803:
9804: :noname ( n-radius circle -- )
9805: circle-radius ! ;
9806: overrides construct
9807:
9808: end-class circle
9809: @end example
9810:
9811: Here we define a class @code{circle} as a child of @code{graphical},
9812: with field @code{circle-radius} (which behaves just like a field
9813: (@pxref{Structures}); it defines (using @code{overrides}) new methods
9814: for the selectors @code{draw} and @code{construct} (@code{construct} is
9815: defined in @code{object}, the parent class of @code{graphical}).
9816:
9817: Now we can create a circle on the heap (i.e.,
9818: @code{allocate}d memory) with:
9819:
9820: @cindex @code{heap-new} usage
9821: @example
9822: 50 circle heap-new constant my-circle
9823: @end example
9824:
9825: @noindent
9826: @code{heap-new} invokes @code{construct}, thus
9827: initializing the field @code{circle-radius} with 50. We can draw
9828: this new circle at (100,100) with:
9829:
9830: @example
9831: 100 100 my-circle draw
9832: @end example
9833:
9834: @cindex selector invocation, restrictions
9835: @cindex class definition, restrictions
9836: Note: You can only invoke a selector if the object on the TOS
9837: (the receiving object) belongs to the class where the selector was
9838: defined or one of its descendents; e.g., you can invoke
9839: @code{draw} only for objects belonging to @code{graphical}
9840: or its descendents (e.g., @code{circle}). Immediately before
9841: @code{end-class}, the search order has to be the same as
9842: immediately after @code{class}.
9843:
9844: @node The Objects base class, Creating objects, Basic Objects Usage, Objects
9845: @subsubsection The @file{object.fs} base class
9846: @cindex @code{object} class
9847:
9848: When you define a class, you have to specify a parent class. So how do
9849: you start defining classes? There is one class available from the start:
9850: @code{object}. It is ancestor for all classes and so is the
9851: only class that has no parent. It has two selectors: @code{construct}
9852: and @code{print}.
9853:
9854: @node Creating objects, Object-Oriented Programming Style, The Objects base class, Objects
9855: @subsubsection Creating objects
9856: @cindex creating objects
9857: @cindex object creation
9858: @cindex object allocation options
9859:
9860: @cindex @code{heap-new} discussion
9861: @cindex @code{dict-new} discussion
9862: @cindex @code{construct} discussion
9863: You can create and initialize an object of a class on the heap with
9864: @code{heap-new} ( ... class -- object ) and in the dictionary
9865: (allocation with @code{allot}) with @code{dict-new} (
9866: ... class -- object ). Both words invoke @code{construct}, which
9867: consumes the stack items indicated by "..." above.
9868:
9869: @cindex @code{init-object} discussion
9870: @cindex @code{class-inst-size} discussion
9871: If you want to allocate memory for an object yourself, you can get its
9872: alignment and size with @code{class-inst-size 2@@} ( class --
9873: align size ). Once you have memory for an object, you can initialize
9874: it with @code{init-object} ( ... class object -- );
9875: @code{construct} does only a part of the necessary work.
9876:
9877: @node Object-Oriented Programming Style, Class Binding, Creating objects, Objects
9878: @subsubsection Object-Oriented Programming Style
9879: @cindex object-oriented programming style
9880: @cindex programming style, object-oriented
9881:
9882: This section is not exhaustive.
9883:
9884: @cindex stack effects of selectors
9885: @cindex selectors and stack effects
9886: In general, it is a good idea to ensure that all methods for the
9887: same selector have the same stack effect: when you invoke a selector,
9888: you often have no idea which method will be invoked, so, unless all
9889: methods have the same stack effect, you will not know the stack effect
9890: of the selector invocation.
9891:
9892: One exception to this rule is methods for the selector
9893: @code{construct}. We know which method is invoked, because we
9894: specify the class to be constructed at the same place. Actually, I
9895: defined @code{construct} as a selector only to give the users a
9896: convenient way to specify initialization. The way it is used, a
9897: mechanism different from selector invocation would be more natural
9898: (but probably would take more code and more space to explain).
9899:
9900: @node Class Binding, Method conveniences, Object-Oriented Programming Style, Objects
9901: @subsubsection Class Binding
9902: @cindex class binding
9903: @cindex early binding
9904:
9905: @cindex late binding
9906: Normal selector invocations determine the method at run-time depending
9907: on the class of the receiving object. This run-time selection is called
9908: @i{late binding}.
9909:
9910: Sometimes it's preferable to invoke a different method. For example,
9911: you might want to use the simple method for @code{print}ing
9912: @code{object}s instead of the possibly long-winded @code{print} method
9913: of the receiver class. You can achieve this by replacing the invocation
9914: of @code{print} with:
9915:
9916: @cindex @code{[bind]} usage
9917: @example
9918: [bind] object print
9919: @end example
9920:
9921: @noindent
9922: in compiled code or:
9923:
9924: @cindex @code{bind} usage
9925: @example
9926: bind object print
9927: @end example
9928:
9929: @cindex class binding, alternative to
9930: @noindent
9931: in interpreted code. Alternatively, you can define the method with a
9932: name (e.g., @code{print-object}), and then invoke it through the
9933: name. Class binding is just a (often more convenient) way to achieve
9934: the same effect; it avoids name clutter and allows you to invoke
9935: methods directly without naming them first.
9936:
9937: @cindex superclass binding
9938: @cindex parent class binding
9939: A frequent use of class binding is this: When we define a method
9940: for a selector, we often want the method to do what the selector does
9941: in the parent class, and a little more. There is a special word for
9942: this purpose: @code{[parent]}; @code{[parent]
9943: @emph{selector}} is equivalent to @code{[bind] @emph{parent
9944: selector}}, where @code{@emph{parent}} is the parent
9945: class of the current class. E.g., a method definition might look like:
9946:
9947: @cindex @code{[parent]} usage
9948: @example
9949: :noname
9950: dup [parent] foo \ do parent's foo on the receiving object
9951: ... \ do some more
9952: ; overrides foo
9953: @end example
9954:
9955: @cindex class binding as optimization
9956: In @cite{Object-oriented programming in ANS Forth} (Forth Dimensions,
9957: March 1997), Andrew McKewan presents class binding as an optimization
9958: technique. I recommend not using it for this purpose unless you are in
9959: an emergency. Late binding is pretty fast with this model anyway, so the
9960: benefit of using class binding is small; the cost of using class binding
9961: where it is not appropriate is reduced maintainability.
9962:
9963: While we are at programming style questions: You should bind
9964: selectors only to ancestor classes of the receiving object. E.g., say,
9965: you know that the receiving object is of class @code{foo} or its
9966: descendents; then you should bind only to @code{foo} and its
9967: ancestors.
9968:
9969: @node Method conveniences, Classes and Scoping, Class Binding, Objects
9970: @subsubsection Method conveniences
9971: @cindex method conveniences
9972:
9973: In a method you usually access the receiving object pretty often. If
9974: you define the method as a plain colon definition (e.g., with
9975: @code{:noname}), you may have to do a lot of stack
9976: gymnastics. To avoid this, you can define the method with @code{m:
9977: ... ;m}. E.g., you could define the method for
9978: @code{draw}ing a @code{circle} with
9979:
9980: @cindex @code{this} usage
9981: @cindex @code{m:} usage
9982: @cindex @code{;m} usage
9983: @example
9984: m: ( x y circle -- )
9985: ( x y ) this circle-radius @@ draw-circle ;m
9986: @end example
9987:
9988: @cindex @code{exit} in @code{m: ... ;m}
9989: @cindex @code{exitm} discussion
9990: @cindex @code{catch} in @code{m: ... ;m}
9991: When this method is executed, the receiver object is removed from the
9992: stack; you can access it with @code{this} (admittedly, in this
9993: example the use of @code{m: ... ;m} offers no advantage). Note
9994: that I specify the stack effect for the whole method (i.e. including
9995: the receiver object), not just for the code between @code{m:}
9996: and @code{;m}. You cannot use @code{exit} in
9997: @code{m:...;m}; instead, use
9998: @code{exitm}.@footnote{Moreover, for any word that calls
9999: @code{catch} and was defined before loading
10000: @code{objects.fs}, you have to redefine it like I redefined
10001: @code{catch}: @code{: catch this >r catch r> to-this ;}}
10002:
10003: @cindex @code{inst-var} usage
10004: You will frequently use sequences of the form @code{this
10005: @emph{field}} (in the example above: @code{this
10006: circle-radius}). If you use the field only in this way, you can
10007: define it with @code{inst-var} and eliminate the
10008: @code{this} before the field name. E.g., the @code{circle}
10009: class above could also be defined with:
10010:
10011: @example
10012: graphical class
10013: cell% inst-var radius
10014:
10015: m: ( x y circle -- )
10016: radius @@ draw-circle ;m
10017: overrides draw
10018:
10019: m: ( n-radius circle -- )
10020: radius ! ;m
10021: overrides construct
10022:
10023: end-class circle
10024: @end example
10025:
10026: @code{radius} can only be used in @code{circle} and its
10027: descendent classes and inside @code{m:...;m}.
10028:
10029: @cindex @code{inst-value} usage
10030: You can also define fields with @code{inst-value}, which is
10031: to @code{inst-var} what @code{value} is to
10032: @code{variable}. You can change the value of such a field with
10033: @code{[to-inst]}. E.g., we could also define the class
10034: @code{circle} like this:
10035:
10036: @example
10037: graphical class
10038: inst-value radius
10039:
10040: m: ( x y circle -- )
10041: radius draw-circle ;m
10042: overrides draw
10043:
10044: m: ( n-radius circle -- )
10045: [to-inst] radius ;m
10046: overrides construct
10047:
10048: end-class circle
10049: @end example
10050:
10051: Finally, you can define named methods with @code{:m}. One use of this
10052: feature is the definition of words that occur only in one class and are
10053: not intended to be overridden, but which still need method context
10054: (e.g., for accessing @code{inst-var}s). Another use is for methods that
10055: would be bound frequently, if defined anonymously.
10056:
10057:
10058: @node Classes and Scoping, Dividing classes, Method conveniences, Objects
10059: @subsubsection Classes and Scoping
10060: @cindex classes and scoping
10061: @cindex scoping and classes
10062:
10063: Inheritance is frequent, unlike structure extension. This exacerbates
10064: the problem with the field name convention (@pxref{Structure Naming
10065: Convention}): One always has to remember in which class the field was
10066: originally defined; changing a part of the class structure would require
10067: changes for renaming in otherwise unaffected code.
10068:
10069: @cindex @code{inst-var} visibility
10070: @cindex @code{inst-value} visibility
10071: To solve this problem, I added a scoping mechanism (which was not in my
10072: original charter): A field defined with @code{inst-var} (or
10073: @code{inst-value}) is visible only in the class where it is defined and in
10074: the descendent classes of this class. Using such fields only makes
10075: sense in @code{m:}-defined methods in these classes anyway.
10076:
10077: This scoping mechanism allows us to use the unadorned field name,
10078: because name clashes with unrelated words become much less likely.
10079:
10080: @cindex @code{protected} discussion
10081: @cindex @code{private} discussion
10082: Once we have this mechanism, we can also use it for controlling the
10083: visibility of other words: All words defined after
10084: @code{protected} are visible only in the current class and its
10085: descendents. @code{public} restores the compilation
10086: (i.e. @code{current}) word list that was in effect before. If you
10087: have several @code{protected}s without an intervening
10088: @code{public} or @code{set-current}, @code{public}
10089: will restore the compilation word list in effect before the first of
10090: these @code{protected}s.
10091:
10092: @node Dividing classes, Object Interfaces, Classes and Scoping, Objects
10093: @subsubsection Dividing classes
10094: @cindex Dividing classes
10095: @cindex @code{methods}...@code{end-methods}
10096:
10097: You may want to do the definition of methods separate from the
10098: definition of the class, its selectors, fields, and instance variables,
10099: i.e., separate the implementation from the definition. You can do this
10100: in the following way:
10101:
10102: @example
10103: graphical class
10104: inst-value radius
10105: end-class circle
10106:
10107: ... \ do some other stuff
10108:
10109: circle methods \ now we are ready
10110:
10111: m: ( x y circle -- )
10112: radius draw-circle ;m
10113: overrides draw
10114:
10115: m: ( n-radius circle -- )
10116: [to-inst] radius ;m
10117: overrides construct
10118:
10119: end-methods
10120: @end example
10121:
10122: You can use several @code{methods}...@code{end-methods} sections. The
10123: only things you can do to the class in these sections are: defining
10124: methods, and overriding the class's selectors. You must not define new
10125: selectors or fields.
10126:
10127: Note that you often have to override a selector before using it. In
10128: particular, you usually have to override @code{construct} with a new
10129: method before you can invoke @code{heap-new} and friends. E.g., you
10130: must not create a circle before the @code{overrides construct} sequence
10131: in the example above.
10132:
10133: @node Object Interfaces, Objects Implementation, Dividing classes, Objects
10134: @subsubsection Object Interfaces
10135: @cindex object interfaces
10136: @cindex interfaces for objects
10137:
10138: In this model you can only call selectors defined in the class of the
10139: receiving objects or in one of its ancestors. If you call a selector
10140: with a receiving object that is not in one of these classes, the
10141: result is undefined; if you are lucky, the program crashes
10142: immediately.
10143:
10144: @cindex selectors common to hardly-related classes
10145: Now consider the case when you want to have a selector (or several)
10146: available in two classes: You would have to add the selector to a
10147: common ancestor class, in the worst case to @code{object}. You
10148: may not want to do this, e.g., because someone else is responsible for
10149: this ancestor class.
10150:
10151: The solution for this problem is interfaces. An interface is a
10152: collection of selectors. If a class implements an interface, the
10153: selectors become available to the class and its descendents. A class
10154: can implement an unlimited number of interfaces. For the problem
10155: discussed above, we would define an interface for the selector(s), and
10156: both classes would implement the interface.
10157:
10158: As an example, consider an interface @code{storage} for
10159: writing objects to disk and getting them back, and a class
10160: @code{foo} that implements it. The code would look like this:
10161:
10162: @cindex @code{interface} usage
10163: @cindex @code{end-interface} usage
10164: @cindex @code{implementation} usage
10165: @example
10166: interface
10167: selector write ( file object -- )
10168: selector read1 ( file object -- )
10169: end-interface storage
10170:
10171: bar class
10172: storage implementation
10173:
10174: ... overrides write
10175: ... overrides read1
10176: ...
10177: end-class foo
10178: @end example
10179:
10180: @noindent
10181: (I would add a word @code{read} @i{( file -- object )} that uses
10182: @code{read1} internally, but that's beyond the point illustrated
10183: here.)
10184:
10185: Note that you cannot use @code{protected} in an interface; and
10186: of course you cannot define fields.
10187:
10188: In the Neon model, all selectors are available for all classes;
10189: therefore it does not need interfaces. The price you pay in this model
10190: is slower late binding, and therefore, added complexity to avoid late
10191: binding.
10192:
10193: @node Objects Implementation, Objects Glossary, Object Interfaces, Objects
10194: @subsubsection @file{objects.fs} Implementation
10195: @cindex @file{objects.fs} implementation
10196:
10197: @cindex @code{object-map} discussion
10198: An object is a piece of memory, like one of the data structures
10199: described with @code{struct...end-struct}. It has a field
10200: @code{object-map} that points to the method map for the object's
10201: class.
10202:
10203: @cindex method map
10204: @cindex virtual function table
10205: The @emph{method map}@footnote{This is Self terminology; in C++
10206: terminology: virtual function table.} is an array that contains the
10207: execution tokens (@i{xt}s) of the methods for the object's class. Each
10208: selector contains an offset into a method map.
10209:
10210: @cindex @code{selector} implementation, class
10211: @code{selector} is a defining word that uses
10212: @code{CREATE} and @code{DOES>}. The body of the
10213: selector contains the offset; the @code{DOES>} action for a
10214: class selector is, basically:
10215:
10216: @example
10217: ( object addr ) @@ over object-map @@ + @@ execute
10218: @end example
10219:
10220: Since @code{object-map} is the first field of the object, it
10221: does not generate any code. As you can see, calling a selector has a
10222: small, constant cost.
10223:
10224: @cindex @code{current-interface} discussion
10225: @cindex class implementation and representation
10226: A class is basically a @code{struct} combined with a method
10227: map. During the class definition the alignment and size of the class
10228: are passed on the stack, just as with @code{struct}s, so
10229: @code{field} can also be used for defining class
10230: fields. However, passing more items on the stack would be
10231: inconvenient, so @code{class} builds a data structure in memory,
10232: which is accessed through the variable
10233: @code{current-interface}. After its definition is complete, the
10234: class is represented on the stack by a pointer (e.g., as parameter for
10235: a child class definition).
10236:
10237: A new class starts off with the alignment and size of its parent,
10238: and a copy of the parent's method map. Defining new fields extends the
10239: size and alignment; likewise, defining new selectors extends the
10240: method map. @code{overrides} just stores a new @i{xt} in the method
10241: map at the offset given by the selector.
10242:
10243: @cindex class binding, implementation
10244: Class binding just gets the @i{xt} at the offset given by the selector
10245: from the class's method map and @code{compile,}s (in the case of
10246: @code{[bind]}) it.
10247:
10248: @cindex @code{this} implementation
10249: @cindex @code{catch} and @code{this}
10250: @cindex @code{this} and @code{catch}
10251: I implemented @code{this} as a @code{value}. At the
10252: start of an @code{m:...;m} method the old @code{this} is
10253: stored to the return stack and restored at the end; and the object on
10254: the TOS is stored @code{TO this}. This technique has one
10255: disadvantage: If the user does not leave the method via
10256: @code{;m}, but via @code{throw} or @code{exit},
10257: @code{this} is not restored (and @code{exit} may
10258: crash). To deal with the @code{throw} problem, I have redefined
10259: @code{catch} to save and restore @code{this}; the same
10260: should be done with any word that can catch an exception. As for
10261: @code{exit}, I simply forbid it (as a replacement, there is
10262: @code{exitm}).
10263:
10264: @cindex @code{inst-var} implementation
10265: @code{inst-var} is just the same as @code{field}, with
10266: a different @code{DOES>} action:
10267: @example
10268: @@ this +
10269: @end example
10270: Similar for @code{inst-value}.
10271:
10272: @cindex class scoping implementation
10273: Each class also has a word list that contains the words defined with
10274: @code{inst-var} and @code{inst-value}, and its protected
10275: words. It also has a pointer to its parent. @code{class} pushes
10276: the word lists of the class and all its ancestors onto the search order stack,
10277: and @code{end-class} drops them.
10278:
10279: @cindex interface implementation
10280: An interface is like a class without fields, parent and protected
10281: words; i.e., it just has a method map. If a class implements an
10282: interface, its method map contains a pointer to the method map of the
10283: interface. The positive offsets in the map are reserved for class
10284: methods, therefore interface map pointers have negative
10285: offsets. Interfaces have offsets that are unique throughout the
10286: system, unlike class selectors, whose offsets are only unique for the
10287: classes where the selector is available (invokable).
10288:
10289: This structure means that interface selectors have to perform one
10290: indirection more than class selectors to find their method. Their body
10291: contains the interface map pointer offset in the class method map, and
10292: the method offset in the interface method map. The
10293: @code{does>} action for an interface selector is, basically:
10294:
10295: @example
10296: ( object selector-body )
10297: 2dup selector-interface @@ ( object selector-body object interface-offset )
10298: swap object-map @@ + @@ ( object selector-body map )
10299: swap selector-offset @@ + @@ execute
10300: @end example
10301:
10302: where @code{object-map} and @code{selector-offset} are
10303: first fields and generate no code.
10304:
10305: As a concrete example, consider the following code:
10306:
10307: @example
10308: interface
10309: selector if1sel1
10310: selector if1sel2
10311: end-interface if1
10312:
10313: object class
10314: if1 implementation
10315: selector cl1sel1
10316: cell% inst-var cl1iv1
10317:
10318: ' m1 overrides construct
10319: ' m2 overrides if1sel1
10320: ' m3 overrides if1sel2
10321: ' m4 overrides cl1sel2
10322: end-class cl1
10323:
10324: create obj1 object dict-new drop
10325: create obj2 cl1 dict-new drop
10326: @end example
10327:
10328: The data structure created by this code (including the data structure
10329: for @code{object}) is shown in the <a
10330: href="objects-implementation.eps">figure</a>, assuming a cell size of 4.
10331: @comment TODO add this diagram..
10332:
10333: @node Objects Glossary, , Objects Implementation, Objects
10334: @subsubsection @file{objects.fs} Glossary
10335: @cindex @file{objects.fs} Glossary
10336:
10337:
10338: doc---objects-bind
10339: doc---objects-<bind>
10340: doc---objects-bind'
10341: doc---objects-[bind]
10342: doc---objects-class
10343: doc---objects-class->map
10344: doc---objects-class-inst-size
10345: doc---objects-class-override!
10346: doc---objects-construct
10347: doc---objects-current'
10348: doc---objects-[current]
10349: doc---objects-current-interface
10350: doc---objects-dict-new
10351: doc---objects-drop-order
10352: doc---objects-end-class
10353: doc---objects-end-class-noname
10354: doc---objects-end-interface
10355: doc---objects-end-interface-noname
10356: doc---objects-end-methods
10357: doc---objects-exitm
10358: doc---objects-heap-new
10359: doc---objects-implementation
10360: doc---objects-init-object
10361: doc---objects-inst-value
10362: doc---objects-inst-var
10363: doc---objects-interface
10364: doc---objects-m:
10365: doc---objects-:m
10366: doc---objects-;m
10367: doc---objects-method
10368: doc---objects-methods
10369: doc---objects-object
10370: doc---objects-overrides
10371: doc---objects-[parent]
10372: doc---objects-print
10373: doc---objects-protected
10374: doc---objects-public
10375: doc---objects-push-order
10376: doc---objects-selector
10377: doc---objects-this
10378: doc---objects-<to-inst>
10379: doc---objects-[to-inst]
10380: doc---objects-to-this
10381: doc---objects-xt-new
10382:
10383:
10384: @c -------------------------------------------------------------
10385: @node OOF, Mini-OOF, Objects, Object-oriented Forth
10386: @subsection The @file{oof.fs} model
10387: @cindex oof
10388: @cindex object-oriented programming
10389:
10390: @cindex @file{objects.fs}
10391: @cindex @file{oof.fs}
10392:
10393: This section describes the @file{oof.fs} package.
10394:
10395: The package described in this section has been used in bigFORTH since 1991, and
10396: used for two large applications: a chromatographic system used to
10397: create new medicaments, and a graphic user interface library (MINOS).
10398:
10399: You can find a description (in German) of @file{oof.fs} in @cite{Object
10400: oriented bigFORTH} by Bernd Paysan, published in @cite{Vierte Dimension}
10401: 10(2), 1994.
10402:
10403: @menu
10404: * Properties of the OOF model::
10405: * Basic OOF Usage::
10406: * The OOF base class::
10407: * Class Declaration::
10408: * Class Implementation::
10409: @end menu
10410:
10411: @node Properties of the OOF model, Basic OOF Usage, OOF, OOF
10412: @subsubsection Properties of the @file{oof.fs} model
10413: @cindex @file{oof.fs} properties
10414:
10415: @itemize @bullet
10416: @item
10417: This model combines object oriented programming with information
10418: hiding. It helps you writing large application, where scoping is
10419: necessary, because it provides class-oriented scoping.
10420:
10421: @item
10422: Named objects, object pointers, and object arrays can be created,
10423: selector invocation uses the ``object selector'' syntax. Selector invocation
10424: to objects and/or selectors on the stack is a bit less convenient, but
10425: possible.
10426:
10427: @item
10428: Selector invocation and instance variable usage of the active object is
10429: straightforward, since both make use of the active object.
10430:
10431: @item
10432: Late binding is efficient and easy to use.
10433:
10434: @item
10435: State-smart objects parse selectors. However, extensibility is provided
10436: using a (parsing) selector @code{postpone} and a selector @code{'}.
10437:
10438: @item
10439: An implementation in ANS Forth is available.
10440:
10441: @end itemize
10442:
10443:
10444: @node Basic OOF Usage, The OOF base class, Properties of the OOF model, OOF
10445: @subsubsection Basic @file{oof.fs} Usage
10446: @cindex @file{oof.fs} usage
10447:
10448: This section uses the same example as for @code{objects} (@pxref{Basic Objects Usage}).
10449:
10450: You can define a class for graphical objects like this:
10451:
10452: @cindex @code{class} usage
10453: @cindex @code{class;} usage
10454: @cindex @code{method} usage
10455: @example
10456: object class graphical \ "object" is the parent class
10457: method draw ( x y graphical -- )
10458: class;
10459: @end example
10460:
10461: This code defines a class @code{graphical} with an
10462: operation @code{draw}. We can perform the operation
10463: @code{draw} on any @code{graphical} object, e.g.:
10464:
10465: @example
10466: 100 100 t-rex draw
10467: @end example
10468:
10469: @noindent
10470: where @code{t-rex} is an object or object pointer, created with e.g.
10471: @code{graphical : t-rex}.
10472:
10473: @cindex abstract class
10474: How do we create a graphical object? With the present definitions,
10475: we cannot create a useful graphical object. The class
10476: @code{graphical} describes graphical objects in general, but not
10477: any concrete graphical object type (C++ users would call it an
10478: @emph{abstract class}); e.g., there is no method for the selector
10479: @code{draw} in the class @code{graphical}.
10480:
10481: For concrete graphical objects, we define child classes of the
10482: class @code{graphical}, e.g.:
10483:
10484: @example
10485: graphical class circle \ "graphical" is the parent class
10486: cell var circle-radius
10487: how:
10488: : draw ( x y -- )
10489: circle-radius @@ draw-circle ;
10490:
10491: : init ( n-radius -- (
10492: circle-radius ! ;
10493: class;
10494: @end example
10495:
10496: Here we define a class @code{circle} as a child of @code{graphical},
10497: with a field @code{circle-radius}; it defines new methods for the
10498: selectors @code{draw} and @code{init} (@code{init} is defined in
10499: @code{object}, the parent class of @code{graphical}).
10500:
10501: Now we can create a circle in the dictionary with:
10502:
10503: @example
10504: 50 circle : my-circle
10505: @end example
10506:
10507: @noindent
10508: @code{:} invokes @code{init}, thus initializing the field
10509: @code{circle-radius} with 50. We can draw this new circle at (100,100)
10510: with:
10511:
10512: @example
10513: 100 100 my-circle draw
10514: @end example
10515:
10516: @cindex selector invocation, restrictions
10517: @cindex class definition, restrictions
10518: Note: You can only invoke a selector if the receiving object belongs to
10519: the class where the selector was defined or one of its descendents;
10520: e.g., you can invoke @code{draw} only for objects belonging to
10521: @code{graphical} or its descendents (e.g., @code{circle}). The scoping
10522: mechanism will check if you try to invoke a selector that is not
10523: defined in this class hierarchy, so you'll get an error at compilation
10524: time.
10525:
10526:
10527: @node The OOF base class, Class Declaration, Basic OOF Usage, OOF
10528: @subsubsection The @file{oof.fs} base class
10529: @cindex @file{oof.fs} base class
10530:
10531: When you define a class, you have to specify a parent class. So how do
10532: you start defining classes? There is one class available from the start:
10533: @code{object}. You have to use it as ancestor for all classes. It is the
10534: only class that has no parent. Classes are also objects, except that
10535: they don't have instance variables; class manipulation such as
10536: inheritance or changing definitions of a class is handled through
10537: selectors of the class @code{object}.
10538:
10539: @code{object} provides a number of selectors:
10540:
10541: @itemize @bullet
10542: @item
10543: @code{class} for subclassing, @code{definitions} to add definitions
10544: later on, and @code{class?} to get type informations (is the class a
10545: subclass of the class passed on the stack?).
10546:
10547: doc---object-class
10548: doc---object-definitions
10549: doc---object-class?
10550:
10551:
10552: @item
10553: @code{init} and @code{dispose} as constructor and destructor of the
10554: object. @code{init} is invocated after the object's memory is allocated,
10555: while @code{dispose} also handles deallocation. Thus if you redefine
10556: @code{dispose}, you have to call the parent's dispose with @code{super
10557: dispose}, too.
10558:
10559: doc---object-init
10560: doc---object-dispose
10561:
10562:
10563: @item
10564: @code{new}, @code{new[]}, @code{:}, @code{ptr}, @code{asptr}, and
10565: @code{[]} to create named and unnamed objects and object arrays or
10566: object pointers.
10567:
10568: doc---object-new
10569: doc---object-new[]
10570: doc---object-:
10571: doc---object-ptr
10572: doc---object-asptr
10573: doc---object-[]
10574:
10575:
10576: @item
10577: @code{::} and @code{super} for explicit scoping. You should use explicit
10578: scoping only for super classes or classes with the same set of instance
10579: variables. Explicitly-scoped selectors use early binding.
10580:
10581: doc---object-::
10582: doc---object-super
10583:
10584:
10585: @item
10586: @code{self} to get the address of the object
10587:
10588: doc---object-self
10589:
10590:
10591: @item
10592: @code{bind}, @code{bound}, @code{link}, and @code{is} to assign object
10593: pointers and instance defers.
10594:
10595: doc---object-bind
10596: doc---object-bound
10597: doc---object-link
10598: doc---object-is
10599:
10600:
10601: @item
10602: @code{'} to obtain selector tokens, @code{send} to invocate selectors
10603: form the stack, and @code{postpone} to generate selector invocation code.
10604:
10605: doc---object-'
10606: doc---object-postpone
10607:
10608:
10609: @item
10610: @code{with} and @code{endwith} to select the active object from the
10611: stack, and enable its scope. Using @code{with} and @code{endwith}
10612: also allows you to create code using selector @code{postpone} without being
10613: trapped by the state-smart objects.
10614:
10615: doc---object-with
10616: doc---object-endwith
10617:
10618:
10619: @end itemize
10620:
10621: @node Class Declaration, Class Implementation, The OOF base class, OOF
10622: @subsubsection Class Declaration
10623: @cindex class declaration
10624:
10625: @itemize @bullet
10626: @item
10627: Instance variables
10628:
10629: doc---oof-var
10630:
10631:
10632: @item
10633: Object pointers
10634:
10635: doc---oof-ptr
10636: doc---oof-asptr
10637:
10638:
10639: @item
10640: Instance defers
10641:
10642: doc---oof-defer
10643:
10644:
10645: @item
10646: Method selectors
10647:
10648: doc---oof-early
10649: doc---oof-method
10650:
10651:
10652: @item
10653: Class-wide variables
10654:
10655: doc---oof-static
10656:
10657:
10658: @item
10659: End declaration
10660:
10661: doc---oof-how:
10662: doc---oof-class;
10663:
10664:
10665: @end itemize
10666:
10667: @c -------------------------------------------------------------
10668: @node Class Implementation, , Class Declaration, OOF
10669: @subsubsection Class Implementation
10670: @cindex class implementation
10671:
10672: @c -------------------------------------------------------------
10673: @node Mini-OOF, Comparison with other object models, OOF, Object-oriented Forth
10674: @subsection The @file{mini-oof.fs} model
10675: @cindex mini-oof
10676:
10677: Gforth's third object oriented Forth package is a 12-liner. It uses a
10678: mixture of the @file{object.fs} and the @file{oof.fs} syntax,
10679: and reduces to the bare minimum of features. This is based on a posting
10680: of Bernd Paysan in comp.arch.
10681:
10682: @menu
10683: * Basic Mini-OOF Usage::
10684: * Mini-OOF Example::
10685: * Mini-OOF Implementation::
10686: * Comparison with other object models::
10687: @end menu
10688:
10689: @c -------------------------------------------------------------
10690: @node Basic Mini-OOF Usage, Mini-OOF Example, Mini-OOF, Mini-OOF
10691: @subsubsection Basic @file{mini-oof.fs} Usage
10692: @cindex mini-oof usage
10693:
10694: There is a base class (@code{class}, which allocates one cell for the
10695: object pointer) plus seven other words: to define a method, a variable,
10696: a class; to end a class, to resolve binding, to allocate an object and
10697: to compile a class method.
10698: @comment TODO better description of the last one
10699:
10700:
10701: doc-object
10702: doc-method
10703: doc-var
10704: doc-class
10705: doc-end-class
10706: doc-defines
10707: doc-new
10708: doc-::
10709:
10710:
10711:
10712: @c -------------------------------------------------------------
10713: @node Mini-OOF Example, Mini-OOF Implementation, Basic Mini-OOF Usage, Mini-OOF
10714: @subsubsection Mini-OOF Example
10715: @cindex mini-oof example
10716:
10717: A short example shows how to use this package. This example, in slightly
10718: extended form, is supplied as @file{moof-exm.fs}
10719: @comment TODO could flesh this out with some comments from the Forthwrite article
10720:
10721: @example
10722: object class
10723: method init
10724: method draw
10725: end-class graphical
10726: @end example
10727:
10728: This code defines a class @code{graphical} with an
10729: operation @code{draw}. We can perform the operation
10730: @code{draw} on any @code{graphical} object, e.g.:
10731:
10732: @example
10733: 100 100 t-rex draw
10734: @end example
10735:
10736: where @code{t-rex} is an object or object pointer, created with e.g.
10737: @code{graphical new Constant t-rex}.
10738:
10739: For concrete graphical objects, we define child classes of the
10740: class @code{graphical}, e.g.:
10741:
10742: @example
10743: graphical class
10744: cell var circle-radius
10745: end-class circle \ "graphical" is the parent class
10746:
10747: :noname ( x y -- )
10748: circle-radius @@ draw-circle ; circle defines draw
10749: :noname ( r -- )
10750: circle-radius ! ; circle defines init
10751: @end example
10752:
10753: There is no implicit init method, so we have to define one. The creation
10754: code of the object now has to call init explicitely.
10755:
10756: @example
10757: circle new Constant my-circle
10758: 50 my-circle init
10759: @end example
10760:
10761: It is also possible to add a function to create named objects with
10762: automatic call of @code{init}, given that all objects have @code{init}
10763: on the same place:
10764:
10765: @example
10766: : new: ( .. o "name" -- )
10767: new dup Constant init ;
10768: 80 circle new: large-circle
10769: @end example
10770:
10771: We can draw this new circle at (100,100) with:
10772:
10773: @example
10774: 100 100 my-circle draw
10775: @end example
10776:
10777: @node Mini-OOF Implementation, , Mini-OOF Example, Mini-OOF
10778: @subsubsection @file{mini-oof.fs} Implementation
10779:
10780: Object-oriented systems with late binding typically use a
10781: ``vtable''-approach: the first variable in each object is a pointer to a
10782: table, which contains the methods as function pointers. The vtable
10783: may also contain other information.
10784:
10785: So first, let's declare methods:
10786:
10787: @example
10788: : method ( m v -- m' v ) Create over , swap cell+ swap
10789: DOES> ( ... o -- ... ) @ over @ + @ execute ;
10790: @end example
10791:
10792: During method declaration, the number of methods and instance
10793: variables is on the stack (in address units). @code{method} creates
10794: one method and increments the method number. To execute a method, it
10795: takes the object, fetches the vtable pointer, adds the offset, and
10796: executes the @i{xt} stored there. Each method takes the object it is
10797: invoked from as top of stack parameter. The method itself should
10798: consume that object.
10799:
10800: Now, we also have to declare instance variables
10801:
10802: @example
10803: : var ( m v size -- m v' ) Create over , +
10804: DOES> ( o -- addr ) @ + ;
10805: @end example
10806:
10807: As before, a word is created with the current offset. Instance
10808: variables can have different sizes (cells, floats, doubles, chars), so
10809: all we do is take the size and add it to the offset. If your machine
10810: has alignment restrictions, put the proper @code{aligned} or
10811: @code{faligned} before the variable, to adjust the variable
10812: offset. That's why it is on the top of stack.
10813:
10814: We need a starting point (the base object) and some syntactic sugar:
10815:
10816: @example
10817: Create object 1 cells , 2 cells ,
10818: : class ( class -- class methods vars ) dup 2@ ;
10819: @end example
10820:
10821: For inheritance, the vtable of the parent object has to be
10822: copied when a new, derived class is declared. This gives all the
10823: methods of the parent class, which can be overridden, though.
10824:
10825: @example
10826: : end-class ( class methods vars -- )
10827: Create here >r , dup , 2 cells ?DO ['] noop , 1 cells +LOOP
10828: cell+ dup cell+ r> rot @ 2 cells /string move ;
10829: @end example
10830:
10831: The first line creates the vtable, initialized with
10832: @code{noop}s. The second line is the inheritance mechanism, it
10833: copies the xts from the parent vtable.
10834:
10835: We still have no way to define new methods, let's do that now:
10836:
10837: @example
10838: : defines ( xt class -- ) ' >body @ + ! ;
10839: @end example
10840:
10841: To allocate a new object, we need a word, too:
10842:
10843: @example
10844: : new ( class -- o ) here over @ allot swap over ! ;
10845: @end example
10846:
10847: Sometimes derived classes want to access the method of the
10848: parent object. There are two ways to achieve this with Mini-OOF:
10849: first, you could use named words, and second, you could look up the
10850: vtable of the parent object.
10851:
10852: @example
10853: : :: ( class "name" -- ) ' >body @ + @ compile, ;
10854: @end example
10855:
10856:
10857: Nothing can be more confusing than a good example, so here is
10858: one. First let's declare a text object (called
10859: @code{button}), that stores text and position:
10860:
10861: @example
10862: object class
10863: cell var text
10864: cell var len
10865: cell var x
10866: cell var y
10867: method init
10868: method draw
10869: end-class button
10870: @end example
10871:
10872: @noindent
10873: Now, implement the two methods, @code{draw} and @code{init}:
10874:
10875: @example
10876: :noname ( o -- )
10877: >r r@ x @ r@ y @ at-xy r@ text @ r> len @ type ;
10878: button defines draw
10879: :noname ( addr u o -- )
10880: >r 0 r@ x ! 0 r@ y ! r@ len ! r> text ! ;
10881: button defines init
10882: @end example
10883:
10884: @noindent
10885: To demonstrate inheritance, we define a class @code{bold-button}, with no
10886: new data and no new methods:
10887:
10888: @example
10889: button class
10890: end-class bold-button
10891:
10892: : bold 27 emit ." [1m" ;
10893: : normal 27 emit ." [0m" ;
10894: @end example
10895:
10896: @noindent
10897: The class @code{bold-button} has a different draw method to
10898: @code{button}, but the new method is defined in terms of the draw method
10899: for @code{button}:
10900:
10901: @example
10902: :noname bold [ button :: draw ] normal ; bold-button defines draw
10903: @end example
10904:
10905: @noindent
10906: Finally, create two objects and apply methods:
10907:
10908: @example
10909: button new Constant foo
10910: s" thin foo" foo init
10911: page
10912: foo draw
10913: bold-button new Constant bar
10914: s" fat bar" bar init
10915: 1 bar y !
10916: bar draw
10917: @end example
10918:
10919:
10920: @node Comparison with other object models, , Mini-OOF, Object-oriented Forth
10921: @subsection Comparison with other object models
10922: @cindex comparison of object models
10923: @cindex object models, comparison
10924:
10925: Many object-oriented Forth extensions have been proposed (@cite{A survey
10926: of object-oriented Forths} (SIGPLAN Notices, April 1996) by Bradford
10927: J. Rodriguez and W. F. S. Poehlman lists 17). This section discusses the
10928: relation of the object models described here to two well-known and two
10929: closely-related (by the use of method maps) models.
10930:
10931: @cindex Neon model
10932: The most popular model currently seems to be the Neon model (see
10933: @cite{Object-oriented programming in ANS Forth} (Forth Dimensions, March
10934: 1997) by Andrew McKewan) but this model has a number of limitations
10935: @footnote{A longer version of this critique can be
10936: found in @cite{On Standardizing Object-Oriented Forth Extensions} (Forth
10937: Dimensions, May 1997) by Anton Ertl.}:
10938:
10939: @itemize @bullet
10940: @item
10941: It uses a @code{@emph{selector object}} syntax, which makes it unnatural
10942: to pass objects on the stack.
10943:
10944: @item
10945: It requires that the selector parses the input stream (at
10946: compile time); this leads to reduced extensibility and to bugs that are+
10947: hard to find.
10948:
10949: @item
10950: It allows using every selector to every object;
10951: this eliminates the need for classes, but makes it harder to create
10952: efficient implementations.
10953: @end itemize
10954:
10955: @cindex Pountain's object-oriented model
10956: Another well-known publication is @cite{Object-Oriented Forth} (Academic
10957: Press, London, 1987) by Dick Pountain. However, it is not really about
10958: object-oriented programming, because it hardly deals with late
10959: binding. Instead, it focuses on features like information hiding and
10960: overloading that are characteristic of modular languages like Ada (83).
10961:
10962: @cindex Zsoter's object-oriented model
10963: In @cite{Does late binding have to be slow?} (Forth Dimensions 18(1)
10964: 1996, pages 31-35) Andras Zsoter describes a model that makes heavy use
10965: of an active object (like @code{this} in @file{objects.fs}): The active
10966: object is not only used for accessing all fields, but also specifies the
10967: receiving object of every selector invocation; you have to change the
10968: active object explicitly with @code{@{ ... @}}, whereas in
10969: @file{objects.fs} it changes more or less implicitly at @code{m:
10970: ... ;m}. Such a change at the method entry point is unnecessary with the
10971: Zsoter's model, because the receiving object is the active object
10972: already. On the other hand, the explicit change is absolutely necessary
10973: in that model, because otherwise no one could ever change the active
10974: object. An ANS Forth implementation of this model is available at
10975: @uref{http://www.forth.org/fig/oopf.html}.
10976:
10977: @cindex @file{oof.fs}, differences to other models
10978: The @file{oof.fs} model combines information hiding and overloading
10979: resolution (by keeping names in various word lists) with object-oriented
10980: programming. It sets the active object implicitly on method entry, but
10981: also allows explicit changing (with @code{>o...o>} or with
10982: @code{with...endwith}). It uses parsing and state-smart objects and
10983: classes for resolving overloading and for early binding: the object or
10984: class parses the selector and determines the method from this. If the
10985: selector is not parsed by an object or class, it performs a call to the
10986: selector for the active object (late binding), like Zsoter's model.
10987: Fields are always accessed through the active object. The big
10988: disadvantage of this model is the parsing and the state-smartness, which
10989: reduces extensibility and increases the opportunities for subtle bugs;
10990: essentially, you are only safe if you never tick or @code{postpone} an
10991: object or class (Bernd disagrees, but I (Anton) am not convinced).
10992:
10993: @cindex @file{mini-oof.fs}, differences to other models
10994: The @file{mini-oof.fs} model is quite similar to a very stripped-down
10995: version of the @file{objects.fs} model, but syntactically it is a
10996: mixture of the @file{objects.fs} and @file{oof.fs} models.
10997:
10998: @c -------------------------------------------------------------
10999: @node Passing Commands to the OS, Keeping track of Time, Object-oriented Forth, Words
11000: @section Passing Commands to the Operating System
11001: @cindex operating system - passing commands
11002: @cindex shell commands
11003:
11004: Gforth allows you to pass an arbitrary string to the host operating
11005: system shell (if such a thing exists) for execution.
11006:
11007:
11008: doc-sh
11009: doc-system
11010: doc-$?
11011: doc-getenv
11012:
11013:
11014: @c -------------------------------------------------------------
11015: @node Keeping track of Time, Miscellaneous Words, Passing Commands to the OS, Words
11016: @section Keeping track of Time
11017: @cindex time-related words
11018:
11019: Gforth implements time-related operations by making calls to the C
11020: library function, @code{gettimeofday}.
11021:
11022: doc-ms
11023: doc-time&date
11024:
11025:
11026:
11027: @c -------------------------------------------------------------
11028: @node Miscellaneous Words, , Keeping track of Time, Words
11029: @section Miscellaneous Words
11030: @cindex miscellaneous words
11031:
11032: @comment TODO find homes for these
11033:
11034: These section lists the ANS Forth words that are not documented
11035: elsewhere in this manual. Ultimately, they all need proper homes.
11036:
11037: doc-[compile]
11038:
11039:
11040: The following ANS Forth words are not currently supported by Gforth
11041: (@pxref{ANS conformance}):
11042:
11043: @code{EDITOR}
11044: @code{EMIT?}
11045: @code{FORGET}
11046:
11047: @c ******************************************************************
11048: @node Error messages, Tools, Words, Top
11049: @chapter Error messages
11050: @cindex error messages
11051: @cindex backtrace
11052:
11053: A typical Gforth error message looks like this:
11054:
11055: @example
11056: in file included from :-1
11057: in file included from ./yyy.fs:1
11058: ./xxx.fs:4: Invalid memory address
11059: bar
11060: ^^^
11061: $400E664C @@
11062: $400E6664 foo
11063: @end example
11064:
11065: The message identifying the error is @code{Invalid memory address}. The
11066: error happened when text-interpreting line 4 of the file
11067: @file{./xxx.fs}. This line is given (it contains @code{bar}), and the
11068: word on the line where the error happened, is pointed out (with
11069: @code{^^^}).
11070:
11071: The file containing the error was included in line 1 of @file{./yyy.fs},
11072: and @file{yyy.fs} was included from a non-file (in this case, by giving
11073: @file{yyy.fs} as command-line parameter to Gforth).
11074:
11075: At the end of the error message you find a return stack dump that can be
11076: interpreted as a backtrace (possibly empty). On top you find the top of
11077: the return stack when the @code{throw} happened, and at the bottom you
11078: find the return stack entry just above the return stack of the topmost
11079: text interpreter.
11080:
11081: To the right of most return stack entries you see a guess for the word
11082: that pushed that return stack entry as its return address. This gives a
11083: backtrace. In our case we see that @code{bar} called @code{foo}, and
11084: @code{foo} called @code{@@} (and @code{@@} had an @emph{Invalid memory
11085: address} exception).
11086:
11087: Note that the backtrace is not perfect: We don't know which return stack
11088: entries are return addresses (so we may get false positives); and in
11089: some cases (e.g., for @code{abort"}) we cannot determine from the return
11090: address the word that pushed the return address, so for some return
11091: addresses you see no names in the return stack dump.
11092:
11093: @cindex @code{catch} and backtraces
11094: The return stack dump represents the return stack at the time when a
11095: specific @code{throw} was executed. In programs that make use of
11096: @code{catch}, it is not necessarily clear which @code{throw} should be
11097: used for the return stack dump (e.g., consider one @code{throw} that
11098: indicates an error, which is caught, and during recovery another error
11099: happens; which @code{throw} should be used for the stack dump?). Gforth
11100: presents the return stack dump for the first @code{throw} after the last
11101: executed (not returned-to) @code{catch}; this works well in the usual
11102: case.
11103:
11104: @cindex @code{gforth-fast} and backtraces
11105: @cindex @code{gforth-fast}, difference from @code{gforth}
11106: @cindex backtraces with @code{gforth-fast}
11107: @cindex return stack dump with @code{gforth-fast}
11108: @code{gforth} is able to do a return stack dump for throws generated
11109: from primitives (e.g., invalid memory address, stack empty etc.);
11110: @code{gforth-fast} is only able to do a return stack dump from a
11111: directly called @code{throw} (including @code{abort} etc.). This is the
11112: only difference (apart from a speed factor of between 1.15 (K6-2) and
11113: 1.6 (21164A)) between @code{gforth} and @code{gforth-fast}. Given an
11114: exception caused by a primitive in @code{gforth-fast}, you will
11115: typically see no return stack dump at all; however, if the exception is
11116: caught by @code{catch} (e.g., for restoring some state), and then
11117: @code{throw}n again, the return stack dump will be for the first such
11118: @code{throw}.
11119:
11120: @c ******************************************************************
11121: @node Tools, ANS conformance, Error messages, Top
11122: @chapter Tools
11123:
11124: @menu
11125: * ANS Report:: Report the words used, sorted by wordset.
11126: @end menu
11127:
11128: See also @ref{Emacs and Gforth}.
11129:
11130: @node ANS Report, , Tools, Tools
11131: @section @file{ans-report.fs}: Report the words used, sorted by wordset
11132: @cindex @file{ans-report.fs}
11133: @cindex report the words used in your program
11134: @cindex words used in your program
11135:
11136: If you want to label a Forth program as ANS Forth Program, you must
11137: document which wordsets the program uses; for extension wordsets, it is
11138: helpful to list the words the program requires from these wordsets
11139: (because Forth systems are allowed to provide only some words of them).
11140:
11141: The @file{ans-report.fs} tool makes it easy for you to determine which
11142: words from which wordset and which non-ANS words your application
11143: uses. You simply have to include @file{ans-report.fs} before loading the
11144: program you want to check. After loading your program, you can get the
11145: report with @code{print-ans-report}. A typical use is to run this as
11146: batch job like this:
11147: @example
11148: gforth ans-report.fs myprog.fs -e "print-ans-report bye"
11149: @end example
11150:
11151: The output looks like this (for @file{compat/control.fs}):
11152: @example
11153: The program uses the following words
11154: from CORE :
11155: : POSTPONE THEN ; immediate ?dup IF 0=
11156: from BLOCK-EXT :
11157: \
11158: from FILE :
11159: (
11160: @end example
11161:
11162: @subsection Caveats
11163:
11164: Note that @file{ans-report.fs} just checks which words are used, not whether
11165: they are used in an ANS Forth conforming way!
11166:
11167: Some words are defined in several wordsets in the
11168: standard. @file{ans-report.fs} reports them for only one of the
11169: wordsets, and not necessarily the one you expect. It depends on usage
11170: which wordset is the right one to specify. E.g., if you only use the
11171: compilation semantics of @code{S"}, it is a Core word; if you also use
11172: its interpretation semantics, it is a File word.
11173:
11174: @c ******************************************************************
11175: @node ANS conformance, Model, Tools, Top
11176: @chapter ANS conformance
11177: @cindex ANS conformance of Gforth
11178:
11179: To the best of our knowledge, Gforth is an
11180:
11181: ANS Forth System
11182: @itemize @bullet
11183: @item providing the Core Extensions word set
11184: @item providing the Block word set
11185: @item providing the Block Extensions word set
11186: @item providing the Double-Number word set
11187: @item providing the Double-Number Extensions word set
11188: @item providing the Exception word set
11189: @item providing the Exception Extensions word set
11190: @item providing the Facility word set
11191: @item providing @code{EKEY}, @code{EKEY>CHAR}, @code{EKEY?}, @code{MS} and @code{TIME&DATE} from the Facility Extensions word set
11192: @item providing the File Access word set
11193: @item providing the File Access Extensions word set
11194: @item providing the Floating-Point word set
11195: @item providing the Floating-Point Extensions word set
11196: @item providing the Locals word set
11197: @item providing the Locals Extensions word set
11198: @item providing the Memory-Allocation word set
11199: @item providing the Memory-Allocation Extensions word set (that one's easy)
11200: @item providing the Programming-Tools word set
11201: @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
11202: @item providing the Search-Order word set
11203: @item providing the Search-Order Extensions word set
11204: @item providing the String word set
11205: @item providing the String Extensions word set (another easy one)
11206: @end itemize
11207:
11208: @cindex system documentation
11209: In addition, ANS Forth systems are required to document certain
11210: implementation choices. This chapter tries to meet these
11211: requirements. In many cases it gives a way to ask the system for the
11212: information instead of providing the information directly, in
11213: particular, if the information depends on the processor, the operating
11214: system or the installation options chosen, or if they are likely to
11215: change during the maintenance of Gforth.
11216:
11217: @comment The framework for the rest has been taken from pfe.
11218:
11219: @menu
11220: * The Core Words::
11221: * The optional Block word set::
11222: * The optional Double Number word set::
11223: * The optional Exception word set::
11224: * The optional Facility word set::
11225: * The optional File-Access word set::
11226: * The optional Floating-Point word set::
11227: * The optional Locals word set::
11228: * The optional Memory-Allocation word set::
11229: * The optional Programming-Tools word set::
11230: * The optional Search-Order word set::
11231: @end menu
11232:
11233:
11234: @c =====================================================================
11235: @node The Core Words, The optional Block word set, ANS conformance, ANS conformance
11236: @comment node-name, next, previous, up
11237: @section The Core Words
11238: @c =====================================================================
11239: @cindex core words, system documentation
11240: @cindex system documentation, core words
11241:
11242: @menu
11243: * core-idef:: Implementation Defined Options
11244: * core-ambcond:: Ambiguous Conditions
11245: * core-other:: Other System Documentation
11246: @end menu
11247:
11248: @c ---------------------------------------------------------------------
11249: @node core-idef, core-ambcond, The Core Words, The Core Words
11250: @subsection Implementation Defined Options
11251: @c ---------------------------------------------------------------------
11252: @cindex core words, implementation-defined options
11253: @cindex implementation-defined options, core words
11254:
11255:
11256: @table @i
11257: @item (Cell) aligned addresses:
11258: @cindex cell-aligned addresses
11259: @cindex aligned addresses
11260: processor-dependent. Gforth's alignment words perform natural alignment
11261: (e.g., an address aligned for a datum of size 8 is divisible by
11262: 8). Unaligned accesses usually result in a @code{-23 THROW}.
11263:
11264: @item @code{EMIT} and non-graphic characters:
11265: @cindex @code{EMIT} and non-graphic characters
11266: @cindex non-graphic characters and @code{EMIT}
11267: The character is output using the C library function (actually, macro)
11268: @code{putc}.
11269:
11270: @item character editing of @code{ACCEPT} and @code{EXPECT}:
11271: @cindex character editing of @code{ACCEPT} and @code{EXPECT}
11272: @cindex editing in @code{ACCEPT} and @code{EXPECT}
11273: @cindex @code{ACCEPT}, editing
11274: @cindex @code{EXPECT}, editing
11275: This is modeled on the GNU readline library (@pxref{Readline
11276: Interaction, , Command Line Editing, readline, The GNU Readline
11277: Library}) with Emacs-like key bindings. @kbd{Tab} deviates a little by
11278: producing a full word completion every time you type it (instead of
11279: producing the common prefix of all completions). @xref{Command-line editing}.
11280:
11281: @item character set:
11282: @cindex character set
11283: The character set of your computer and display device. Gforth is
11284: 8-bit-clean (but some other component in your system may make trouble).
11285:
11286: @item Character-aligned address requirements:
11287: @cindex character-aligned address requirements
11288: installation-dependent. Currently a character is represented by a C
11289: @code{unsigned char}; in the future we might switch to @code{wchar_t}
11290: (Comments on that requested).
11291:
11292: @item character-set extensions and matching of names:
11293: @cindex character-set extensions and matching of names
11294: @cindex case-sensitivity for name lookup
11295: @cindex name lookup, case-sensitivity
11296: @cindex locale and case-sensitivity
11297: Any character except the ASCII NUL character can be used in a
11298: name. Matching is case-insensitive (except in @code{TABLE}s). The
11299: matching is performed using the C library function @code{strncasecmp}, whose
11300: function is probably influenced by the locale. E.g., the @code{C} locale
11301: does not know about accents and umlauts, so they are matched
11302: case-sensitively in that locale. For portability reasons it is best to
11303: write programs such that they work in the @code{C} locale. Then one can
11304: use libraries written by a Polish programmer (who might use words
11305: containing ISO Latin-2 encoded characters) and by a French programmer
11306: (ISO Latin-1) in the same program (of course, @code{WORDS} will produce
11307: funny results for some of the words (which ones, depends on the font you
11308: are using)). Also, the locale you prefer may not be available in other
11309: operating systems. Hopefully, Unicode will solve these problems one day.
11310:
11311: @item conditions under which control characters match a space delimiter:
11312: @cindex space delimiters
11313: @cindex control characters as delimiters
11314: If @code{WORD} is called with the space character as a delimiter, all
11315: white-space characters (as identified by the C macro @code{isspace()})
11316: are delimiters. @code{PARSE}, on the other hand, treats space like other
11317: delimiters. @code{SWORD} treats space like @code{WORD}, but behaves
11318: like @code{PARSE} otherwise. @code{(NAME)}, which is used by the outer
11319: interpreter (aka text interpreter) by default, treats all white-space
11320: characters as delimiters.
11321:
11322: @item format of the control-flow stack:
11323: @cindex control-flow stack, format
11324: The data stack is used as control-flow stack. The size of a control-flow
11325: stack item in cells is given by the constant @code{cs-item-size}. At the
11326: time of this writing, an item consists of a (pointer to a) locals list
11327: (third), an address in the code (second), and a tag for identifying the
11328: item (TOS). The following tags are used: @code{defstart},
11329: @code{live-orig}, @code{dead-orig}, @code{dest}, @code{do-dest},
11330: @code{scopestart}.
11331:
11332: @item conversion of digits > 35
11333: @cindex digits > 35
11334: The characters @code{[\]^_'} are the digits with the decimal value
11335: 36@minus{}41. There is no way to input many of the larger digits.
11336:
11337: @item display after input terminates in @code{ACCEPT} and @code{EXPECT}:
11338: @cindex @code{EXPECT}, display after end of input
11339: @cindex @code{ACCEPT}, display after end of input
11340: The cursor is moved to the end of the entered string. If the input is
11341: terminated using the @kbd{Return} key, a space is typed.
11342:
11343: @item exception abort sequence of @code{ABORT"}:
11344: @cindex exception abort sequence of @code{ABORT"}
11345: @cindex @code{ABORT"}, exception abort sequence
11346: The error string is stored into the variable @code{"error} and a
11347: @code{-2 throw} is performed.
11348:
11349: @item input line terminator:
11350: @cindex input line terminator
11351: @cindex line terminator on input
11352: @cindex newline character on input
11353: For interactive input, @kbd{C-m} (CR) and @kbd{C-j} (LF) terminate
11354: lines. One of these characters is typically produced when you type the
11355: @kbd{Enter} or @kbd{Return} key.
11356:
11357: @item maximum size of a counted string:
11358: @cindex maximum size of a counted string
11359: @cindex counted string, maximum size
11360: @code{s" /counted-string" environment? drop .}. Currently 255 characters
11361: on all ports, but this may change.
11362:
11363: @item maximum size of a parsed string:
11364: @cindex maximum size of a parsed string
11365: @cindex parsed string, maximum size
11366: Given by the constant @code{/line}. Currently 255 characters.
11367:
11368: @item maximum size of a definition name, in characters:
11369: @cindex maximum size of a definition name, in characters
11370: @cindex name, maximum length
11371: 31
11372:
11373: @item maximum string length for @code{ENVIRONMENT?}, in characters:
11374: @cindex maximum string length for @code{ENVIRONMENT?}, in characters
11375: @cindex @code{ENVIRONMENT?} string length, maximum
11376: 31
11377:
11378: @item method of selecting the user input device:
11379: @cindex user input device, method of selecting
11380: The user input device is the standard input. There is currently no way to
11381: change it from within Gforth. However, the input can typically be
11382: redirected in the command line that starts Gforth.
11383:
11384: @item method of selecting the user output device:
11385: @cindex user output device, method of selecting
11386: @code{EMIT} and @code{TYPE} output to the file-id stored in the value
11387: @code{outfile-id} (@code{stdout} by default). Gforth uses unbuffered
11388: output when the user output device is a terminal, otherwise the output
11389: is buffered.
11390:
11391: @item methods of dictionary compilation:
11392: What are we expected to document here?
11393:
11394: @item number of bits in one address unit:
11395: @cindex number of bits in one address unit
11396: @cindex address unit, size in bits
11397: @code{s" address-units-bits" environment? drop .}. 8 in all current
11398: ports.
11399:
11400: @item number representation and arithmetic:
11401: @cindex number representation and arithmetic
11402: Processor-dependent. Binary two's complement on all current ports.
11403:
11404: @item ranges for integer types:
11405: @cindex ranges for integer types
11406: @cindex integer types, ranges
11407: Installation-dependent. Make environmental queries for @code{MAX-N},
11408: @code{MAX-U}, @code{MAX-D} and @code{MAX-UD}. The lower bounds for
11409: unsigned (and positive) types is 0. The lower bound for signed types on
11410: two's complement and one's complement machines machines can be computed
11411: by adding 1 to the upper bound.
11412:
11413: @item read-only data space regions:
11414: @cindex read-only data space regions
11415: @cindex data-space, read-only regions
11416: The whole Forth data space is writable.
11417:
11418: @item size of buffer at @code{WORD}:
11419: @cindex size of buffer at @code{WORD}
11420: @cindex @code{WORD} buffer size
11421: @code{PAD HERE - .}. 104 characters on 32-bit machines. The buffer is
11422: shared with the pictured numeric output string. If overwriting
11423: @code{PAD} is acceptable, it is as large as the remaining dictionary
11424: space, although only as much can be sensibly used as fits in a counted
11425: string.
11426:
11427: @item size of one cell in address units:
11428: @cindex cell size
11429: @code{1 cells .}.
11430:
11431: @item size of one character in address units:
11432: @cindex char size
11433: @code{1 chars .}. 1 on all current ports.
11434:
11435: @item size of the keyboard terminal buffer:
11436: @cindex size of the keyboard terminal buffer
11437: @cindex terminal buffer, size
11438: Varies. You can determine the size at a specific time using @code{lp@@
11439: tib - .}. It is shared with the locals stack and TIBs of files that
11440: include the current file. You can change the amount of space for TIBs
11441: and locals stack at Gforth startup with the command line option
11442: @code{-l}.
11443:
11444: @item size of the pictured numeric output buffer:
11445: @cindex size of the pictured numeric output buffer
11446: @cindex pictured numeric output buffer, size
11447: @code{PAD HERE - .}. 104 characters on 32-bit machines. The buffer is
11448: shared with @code{WORD}.
11449:
11450: @item size of the scratch area returned by @code{PAD}:
11451: @cindex size of the scratch area returned by @code{PAD}
11452: @cindex @code{PAD} size
11453: The remainder of dictionary space. @code{unused pad here - - .}.
11454:
11455: @item system case-sensitivity characteristics:
11456: @cindex case-sensitivity characteristics
11457: Dictionary searches are case-insensitive (except in
11458: @code{TABLE}s). However, as explained above under @i{character-set
11459: extensions}, the matching for non-ASCII characters is determined by the
11460: locale you are using. In the default @code{C} locale all non-ASCII
11461: characters are matched case-sensitively.
11462:
11463: @item system prompt:
11464: @cindex system prompt
11465: @cindex prompt
11466: @code{ ok} in interpret state, @code{ compiled} in compile state.
11467:
11468: @item division rounding:
11469: @cindex division rounding
11470: installation dependent. @code{s" floored" environment? drop .}. We leave
11471: the choice to @code{gcc} (what to use for @code{/}) and to you (whether
11472: to use @code{fm/mod}, @code{sm/rem} or simply @code{/}).
11473:
11474: @item values of @code{STATE} when true:
11475: @cindex @code{STATE} values
11476: -1.
11477:
11478: @item values returned after arithmetic overflow:
11479: On two's complement machines, arithmetic is performed modulo
11480: 2**bits-per-cell for single arithmetic and 4**bits-per-cell for double
11481: arithmetic (with appropriate mapping for signed types). Division by zero
11482: typically results in a @code{-55 throw} (Floating-point unidentified
11483: fault), although a @code{-10 throw} (divide by zero) would be more
11484: appropriate.
11485:
11486: @item whether the current definition can be found after @t{DOES>}:
11487: @cindex @t{DOES>}, visibility of current definition
11488: No.
11489:
11490: @end table
11491:
11492: @c ---------------------------------------------------------------------
11493: @node core-ambcond, core-other, core-idef, The Core Words
11494: @subsection Ambiguous conditions
11495: @c ---------------------------------------------------------------------
11496: @cindex core words, ambiguous conditions
11497: @cindex ambiguous conditions, core words
11498:
11499: @table @i
11500:
11501: @item a name is neither a word nor a number:
11502: @cindex name not found
11503: @cindex undefined word
11504: @code{-13 throw} (Undefined word). Actually, @code{-13 bounce}, which
11505: preserves the data and FP stack, so you don't lose more work than
11506: necessary.
11507:
11508: @item a definition name exceeds the maximum length allowed:
11509: @cindex word name too long
11510: @code{-19 throw} (Word name too long)
11511:
11512: @item addressing a region not inside the various data spaces of the forth system:
11513: @cindex Invalid memory address
11514: The stacks, code space and header space are accessible. Machine code space is
11515: typically readable. Accessing other addresses gives results dependent on
11516: the operating system. On decent systems: @code{-9 throw} (Invalid memory
11517: address).
11518:
11519: @item argument type incompatible with parameter:
11520: @cindex argument type mismatch
11521: This is usually not caught. Some words perform checks, e.g., the control
11522: flow words, and issue a @code{ABORT"} or @code{-12 THROW} (Argument type
11523: mismatch).
11524:
11525: @item attempting to obtain the execution token of a word with undefined execution semantics:
11526: @cindex Interpreting a compile-only word, for @code{'} etc.
11527: @cindex execution token of words with undefined execution semantics
11528: @code{-14 throw} (Interpreting a compile-only word). In some cases, you
11529: get an execution token for @code{compile-only-error} (which performs a
11530: @code{-14 throw} when executed).
11531:
11532: @item dividing by zero:
11533: @cindex dividing by zero
11534: @cindex floating point unidentified fault, integer division
11535: On better platforms, this produces a @code{-10 throw} (Division by
11536: zero); on other systems, this typically results in a @code{-55 throw}
11537: (Floating-point unidentified fault).
11538:
11539: @item insufficient data stack or return stack space:
11540: @cindex insufficient data stack or return stack space
11541: @cindex stack overflow
11542: @cindex address alignment exception, stack overflow
11543: @cindex Invalid memory address, stack overflow
11544: Depending on the operating system, the installation, and the invocation
11545: of Gforth, this is either checked by the memory management hardware, or
11546: it is not checked. If it is checked, you typically get a @code{-3 throw}
11547: (Stack overflow), @code{-5 throw} (Return stack overflow), or @code{-9
11548: throw} (Invalid memory address) (depending on the platform and how you
11549: achieved the overflow) as soon as the overflow happens. If it is not
11550: checked, overflows typically result in mysterious illegal memory
11551: accesses, producing @code{-9 throw} (Invalid memory address) or
11552: @code{-23 throw} (Address alignment exception); they might also destroy
11553: the internal data structure of @code{ALLOCATE} and friends, resulting in
11554: various errors in these words.
11555:
11556: @item insufficient space for loop control parameters:
11557: @cindex insufficient space for loop control parameters
11558: like other return stack overflows.
11559:
11560: @item insufficient space in the dictionary:
11561: @cindex insufficient space in the dictionary
11562: @cindex dictionary overflow
11563: If you try to allot (either directly with @code{allot}, or indirectly
11564: with @code{,}, @code{create} etc.) more memory than available in the
11565: dictionary, you get a @code{-8 throw} (Dictionary overflow). If you try
11566: to access memory beyond the end of the dictionary, the results are
11567: similar to stack overflows.
11568:
11569: @item interpreting a word with undefined interpretation semantics:
11570: @cindex interpreting a word with undefined interpretation semantics
11571: @cindex Interpreting a compile-only word
11572: For some words, we have defined interpretation semantics. For the
11573: others: @code{-14 throw} (Interpreting a compile-only word).
11574:
11575: @item modifying the contents of the input buffer or a string literal:
11576: @cindex modifying the contents of the input buffer or a string literal
11577: These are located in writable memory and can be modified.
11578:
11579: @item overflow of the pictured numeric output string:
11580: @cindex overflow of the pictured numeric output string
11581: @cindex pictured numeric output string, overflow
11582: @code{-17 throw} (Pictured numeric ouput string overflow).
11583:
11584: @item parsed string overflow:
11585: @cindex parsed string overflow
11586: @code{PARSE} cannot overflow. @code{WORD} does not check for overflow.
11587:
11588: @item producing a result out of range:
11589: @cindex result out of range
11590: On two's complement machines, arithmetic is performed modulo
11591: 2**bits-per-cell for single arithmetic and 4**bits-per-cell for double
11592: arithmetic (with appropriate mapping for signed types). Division by zero
11593: typically results in a @code{-10 throw} (divide by zero) or @code{-55
11594: throw} (floating point unidentified fault). @code{convert} and
11595: @code{>number} currently overflow silently.
11596:
11597: @item reading from an empty data or return stack:
11598: @cindex stack empty
11599: @cindex stack underflow
11600: @cindex return stack underflow
11601: The data stack is checked by the outer (aka text) interpreter after
11602: every word executed. If it has underflowed, a @code{-4 throw} (Stack
11603: underflow) is performed. Apart from that, stacks may be checked or not,
11604: depending on operating system, installation, and invocation. If they are
11605: caught by a check, they typically result in @code{-4 throw} (Stack
11606: underflow), @code{-6 throw} (Return stack underflow) or @code{-9 throw}
11607: (Invalid memory address), depending on the platform and which stack
11608: underflows and by how much. Note that even if the system uses checking
11609: (through the MMU), your program may have to underflow by a significant
11610: number of stack items to trigger the reaction (the reason for this is
11611: that the MMU, and therefore the checking, works with a page-size
11612: granularity). If there is no checking, the symptoms resulting from an
11613: underflow are similar to those from an overflow. Unbalanced return
11614: stack errors result in a variaty of symptoms, including @code{-9 throw}
11615: (Invalid memory address) and Illegal Instruction (typically @code{-260
11616: throw}).
11617:
11618: @item unexpected end of the input buffer, resulting in an attempt to use a zero-length string as a name:
11619: @cindex unexpected end of the input buffer
11620: @cindex zero-length string as a name
11621: @cindex Attempt to use zero-length string as a name
11622: @code{Create} and its descendants perform a @code{-16 throw} (Attempt to
11623: use zero-length string as a name). Words like @code{'} probably will not
11624: find what they search. Note that it is possible to create zero-length
11625: names with @code{nextname} (should it not?).
11626:
11627: @item @code{>IN} greater than input buffer:
11628: @cindex @code{>IN} greater than input buffer
11629: The next invocation of a parsing word returns a string with length 0.
11630:
11631: @item @code{RECURSE} appears after @code{DOES>}:
11632: @cindex @code{RECURSE} appears after @code{DOES>}
11633: Compiles a recursive call to the defining word, not to the defined word.
11634:
11635: @item argument input source different than current input source for @code{RESTORE-INPUT}:
11636: @cindex argument input source different than current input source for @code{RESTORE-INPUT}
11637: @cindex argument type mismatch, @code{RESTORE-INPUT}
11638: @cindex @code{RESTORE-INPUT}, Argument type mismatch
11639: @code{-12 THROW}. Note that, once an input file is closed (e.g., because
11640: the end of the file was reached), its source-id may be
11641: reused. Therefore, restoring an input source specification referencing a
11642: closed file may lead to unpredictable results instead of a @code{-12
11643: THROW}.
11644:
11645: In the future, Gforth may be able to restore input source specifications
11646: from other than the current input source.
11647:
11648: @item data space containing definitions gets de-allocated:
11649: @cindex data space containing definitions gets de-allocated
11650: Deallocation with @code{allot} is not checked. This typically results in
11651: memory access faults or execution of illegal instructions.
11652:
11653: @item data space read/write with incorrect alignment:
11654: @cindex data space read/write with incorrect alignment
11655: @cindex alignment faults
11656: @cindex address alignment exception
11657: Processor-dependent. Typically results in a @code{-23 throw} (Address
11658: alignment exception). Under Linux-Intel on a 486 or later processor with
11659: alignment turned on, incorrect alignment results in a @code{-9 throw}
11660: (Invalid memory address). There are reportedly some processors with
11661: alignment restrictions that do not report violations.
11662:
11663: @item data space pointer not properly aligned, @code{,}, @code{C,}:
11664: @cindex data space pointer not properly aligned, @code{,}, @code{C,}
11665: Like other alignment errors.
11666:
11667: @item less than u+2 stack items (@code{PICK} and @code{ROLL}):
11668: Like other stack underflows.
11669:
11670: @item loop control parameters not available:
11671: @cindex loop control parameters not available
11672: Not checked. The counted loop words simply assume that the top of return
11673: stack items are loop control parameters and behave accordingly.
11674:
11675: @item most recent definition does not have a name (@code{IMMEDIATE}):
11676: @cindex most recent definition does not have a name (@code{IMMEDIATE})
11677: @cindex last word was headerless
11678: @code{abort" last word was headerless"}.
11679:
11680: @item name not defined by @code{VALUE} used by @code{TO}:
11681: @cindex name not defined by @code{VALUE} used by @code{TO}
11682: @cindex @code{TO} on non-@code{VALUE}s
11683: @cindex Invalid name argument, @code{TO}
11684: @code{-32 throw} (Invalid name argument) (unless name is a local or was
11685: defined by @code{CONSTANT}; in the latter case it just changes the constant).
11686:
11687: @item name not found (@code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]}):
11688: @cindex name not found (@code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]})
11689: @cindex undefined word, @code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]}
11690: @code{-13 throw} (Undefined word)
11691:
11692: @item parameters are not of the same type (@code{DO}, @code{?DO}, @code{WITHIN}):
11693: @cindex parameters are not of the same type (@code{DO}, @code{?DO}, @code{WITHIN})
11694: Gforth behaves as if they were of the same type. I.e., you can predict
11695: the behaviour by interpreting all parameters as, e.g., signed.
11696:
11697: @item @code{POSTPONE} or @code{[COMPILE]} applied to @code{TO}:
11698: @cindex @code{POSTPONE} or @code{[COMPILE]} applied to @code{TO}
11699: Assume @code{: X POSTPONE TO ; IMMEDIATE}. @code{X} performs the
11700: compilation semantics of @code{TO}.
11701:
11702: @item String longer than a counted string returned by @code{WORD}:
11703: @cindex string longer than a counted string returned by @code{WORD}
11704: @cindex @code{WORD}, string overflow
11705: Not checked. The string will be ok, but the count will, of course,
11706: contain only the least significant bits of the length.
11707:
11708: @item u greater than or equal to the number of bits in a cell (@code{LSHIFT}, @code{RSHIFT}):
11709: @cindex @code{LSHIFT}, large shift counts
11710: @cindex @code{RSHIFT}, large shift counts
11711: Processor-dependent. Typical behaviours are returning 0 and using only
11712: the low bits of the shift count.
11713:
11714: @item word not defined via @code{CREATE}:
11715: @cindex @code{>BODY} of non-@code{CREATE}d words
11716: @code{>BODY} produces the PFA of the word no matter how it was defined.
11717:
11718: @cindex @code{DOES>} of non-@code{CREATE}d words
11719: @code{DOES>} changes the execution semantics of the last defined word no
11720: matter how it was defined. E.g., @code{CONSTANT DOES>} is equivalent to
11721: @code{CREATE , DOES>}.
11722:
11723: @item words improperly used outside @code{<#} and @code{#>}:
11724: Not checked. As usual, you can expect memory faults.
11725:
11726: @end table
11727:
11728:
11729: @c ---------------------------------------------------------------------
11730: @node core-other, , core-ambcond, The Core Words
11731: @subsection Other system documentation
11732: @c ---------------------------------------------------------------------
11733: @cindex other system documentation, core words
11734: @cindex core words, other system documentation
11735:
11736: @table @i
11737: @item nonstandard words using @code{PAD}:
11738: @cindex @code{PAD} use by nonstandard words
11739: None.
11740:
11741: @item operator's terminal facilities available:
11742: @cindex operator's terminal facilities available
11743: After processing the command line, Gforth goes into interactive mode,
11744: and you can give commands to Gforth interactively. The actual facilities
11745: available depend on how you invoke Gforth.
11746:
11747: @item program data space available:
11748: @cindex program data space available
11749: @cindex data space available
11750: @code{UNUSED .} gives the remaining dictionary space. The total
11751: dictionary space can be specified with the @code{-m} switch
11752: (@pxref{Invoking Gforth}) when Gforth starts up.
11753:
11754: @item return stack space available:
11755: @cindex return stack space available
11756: You can compute the total return stack space in cells with
11757: @code{s" RETURN-STACK-CELLS" environment? drop .}. You can specify it at
11758: startup time with the @code{-r} switch (@pxref{Invoking Gforth}).
11759:
11760: @item stack space available:
11761: @cindex stack space available
11762: You can compute the total data stack space in cells with
11763: @code{s" STACK-CELLS" environment? drop .}. You can specify it at
11764: startup time with the @code{-d} switch (@pxref{Invoking Gforth}).
11765:
11766: @item system dictionary space required, in address units:
11767: @cindex system dictionary space required, in address units
11768: Type @code{here forthstart - .} after startup. At the time of this
11769: writing, this gives 80080 (bytes) on a 32-bit system.
11770: @end table
11771:
11772:
11773: @c =====================================================================
11774: @node The optional Block word set, The optional Double Number word set, The Core Words, ANS conformance
11775: @section The optional Block word set
11776: @c =====================================================================
11777: @cindex system documentation, block words
11778: @cindex block words, system documentation
11779:
11780: @menu
11781: * block-idef:: Implementation Defined Options
11782: * block-ambcond:: Ambiguous Conditions
11783: * block-other:: Other System Documentation
11784: @end menu
11785:
11786:
11787: @c ---------------------------------------------------------------------
11788: @node block-idef, block-ambcond, The optional Block word set, The optional Block word set
11789: @subsection Implementation Defined Options
11790: @c ---------------------------------------------------------------------
11791: @cindex implementation-defined options, block words
11792: @cindex block words, implementation-defined options
11793:
11794: @table @i
11795: @item the format for display by @code{LIST}:
11796: @cindex @code{LIST} display format
11797: First the screen number is displayed, then 16 lines of 64 characters,
11798: each line preceded by the line number.
11799:
11800: @item the length of a line affected by @code{\}:
11801: @cindex length of a line affected by @code{\}
11802: @cindex @code{\}, line length in blocks
11803: 64 characters.
11804: @end table
11805:
11806:
11807: @c ---------------------------------------------------------------------
11808: @node block-ambcond, block-other, block-idef, The optional Block word set
11809: @subsection Ambiguous conditions
11810: @c ---------------------------------------------------------------------
11811: @cindex block words, ambiguous conditions
11812: @cindex ambiguous conditions, block words
11813:
11814: @table @i
11815: @item correct block read was not possible:
11816: @cindex block read not possible
11817: Typically results in a @code{throw} of some OS-derived value (between
11818: -512 and -2048). If the blocks file was just not long enough, blanks are
11819: supplied for the missing portion.
11820:
11821: @item I/O exception in block transfer:
11822: @cindex I/O exception in block transfer
11823: @cindex block transfer, I/O exception
11824: Typically results in a @code{throw} of some OS-derived value (between
11825: -512 and -2048).
11826:
11827: @item invalid block number:
11828: @cindex invalid block number
11829: @cindex block number invalid
11830: @code{-35 throw} (Invalid block number)
11831:
11832: @item a program directly alters the contents of @code{BLK}:
11833: @cindex @code{BLK}, altering @code{BLK}
11834: The input stream is switched to that other block, at the same
11835: position. If the storing to @code{BLK} happens when interpreting
11836: non-block input, the system will get quite confused when the block ends.
11837:
11838: @item no current block buffer for @code{UPDATE}:
11839: @cindex @code{UPDATE}, no current block buffer
11840: @code{UPDATE} has no effect.
11841:
11842: @end table
11843:
11844: @c ---------------------------------------------------------------------
11845: @node block-other, , block-ambcond, The optional Block word set
11846: @subsection Other system documentation
11847: @c ---------------------------------------------------------------------
11848: @cindex other system documentation, block words
11849: @cindex block words, other system documentation
11850:
11851: @table @i
11852: @item any restrictions a multiprogramming system places on the use of buffer addresses:
11853: No restrictions (yet).
11854:
11855: @item the number of blocks available for source and data:
11856: depends on your disk space.
11857:
11858: @end table
11859:
11860:
11861: @c =====================================================================
11862: @node The optional Double Number word set, The optional Exception word set, The optional Block word set, ANS conformance
11863: @section The optional Double Number word set
11864: @c =====================================================================
11865: @cindex system documentation, double words
11866: @cindex double words, system documentation
11867:
11868: @menu
11869: * double-ambcond:: Ambiguous Conditions
11870: @end menu
11871:
11872:
11873: @c ---------------------------------------------------------------------
11874: @node double-ambcond, , The optional Double Number word set, The optional Double Number word set
11875: @subsection Ambiguous conditions
11876: @c ---------------------------------------------------------------------
11877: @cindex double words, ambiguous conditions
11878: @cindex ambiguous conditions, double words
11879:
11880: @table @i
11881: @item @i{d} outside of range of @i{n} in @code{D>S}:
11882: @cindex @code{D>S}, @i{d} out of range of @i{n}
11883: The least significant cell of @i{d} is produced.
11884:
11885: @end table
11886:
11887:
11888: @c =====================================================================
11889: @node The optional Exception word set, The optional Facility word set, The optional Double Number word set, ANS conformance
11890: @section The optional Exception word set
11891: @c =====================================================================
11892: @cindex system documentation, exception words
11893: @cindex exception words, system documentation
11894:
11895: @menu
11896: * exception-idef:: Implementation Defined Options
11897: @end menu
11898:
11899:
11900: @c ---------------------------------------------------------------------
11901: @node exception-idef, , The optional Exception word set, The optional Exception word set
11902: @subsection Implementation Defined Options
11903: @c ---------------------------------------------------------------------
11904: @cindex implementation-defined options, exception words
11905: @cindex exception words, implementation-defined options
11906:
11907: @table @i
11908: @item @code{THROW}-codes used in the system:
11909: @cindex @code{THROW}-codes used in the system
11910: The codes -256@minus{}-511 are used for reporting signals. The mapping
11911: from OS signal numbers to throw codes is -256@minus{}@i{signal}. The
11912: codes -512@minus{}-2047 are used for OS errors (for file and memory
11913: allocation operations). The mapping from OS error numbers to throw codes
11914: is -512@minus{}@code{errno}. One side effect of this mapping is that
11915: undefined OS errors produce a message with a strange number; e.g.,
11916: @code{-1000 THROW} results in @code{Unknown error 488} on my system.
11917: @end table
11918:
11919: @c =====================================================================
11920: @node The optional Facility word set, The optional File-Access word set, The optional Exception word set, ANS conformance
11921: @section The optional Facility word set
11922: @c =====================================================================
11923: @cindex system documentation, facility words
11924: @cindex facility words, system documentation
11925:
11926: @menu
11927: * facility-idef:: Implementation Defined Options
11928: * facility-ambcond:: Ambiguous Conditions
11929: @end menu
11930:
11931:
11932: @c ---------------------------------------------------------------------
11933: @node facility-idef, facility-ambcond, The optional Facility word set, The optional Facility word set
11934: @subsection Implementation Defined Options
11935: @c ---------------------------------------------------------------------
11936: @cindex implementation-defined options, facility words
11937: @cindex facility words, implementation-defined options
11938:
11939: @table @i
11940: @item encoding of keyboard events (@code{EKEY}):
11941: @cindex keyboard events, encoding in @code{EKEY}
11942: @cindex @code{EKEY}, encoding of keyboard events
11943: Keys corresponding to ASCII characters are encoded as ASCII characters.
11944: Other keys are encoded with the constants @code{k-left}, @code{k-right},
11945: @code{k-up}, @code{k-down}, @code{k-home}, @code{k-end}, @code{k1},
11946: @code{k2}, @code{k3}, @code{k4}, @code{k5}, @code{k6}, @code{k7},
11947: @code{k8}, @code{k9}, @code{k10}, @code{k11}, @code{k12}.
11948:
11949:
11950: @item duration of a system clock tick:
11951: @cindex duration of a system clock tick
11952: @cindex clock tick duration
11953: System dependent. With respect to @code{MS}, the time is specified in
11954: microseconds. How well the OS and the hardware implement this, is
11955: another question.
11956:
11957: @item repeatability to be expected from the execution of @code{MS}:
11958: @cindex repeatability to be expected from the execution of @code{MS}
11959: @cindex @code{MS}, repeatability to be expected
11960: System dependent. On Unix, a lot depends on load. If the system is
11961: lightly loaded, and the delay is short enough that Gforth does not get
11962: swapped out, the performance should be acceptable. Under MS-DOS and
11963: other single-tasking systems, it should be good.
11964:
11965: @end table
11966:
11967:
11968: @c ---------------------------------------------------------------------
11969: @node facility-ambcond, , facility-idef, The optional Facility word set
11970: @subsection Ambiguous conditions
11971: @c ---------------------------------------------------------------------
11972: @cindex facility words, ambiguous conditions
11973: @cindex ambiguous conditions, facility words
11974:
11975: @table @i
11976: @item @code{AT-XY} can't be performed on user output device:
11977: @cindex @code{AT-XY} can't be performed on user output device
11978: Largely terminal dependent. No range checks are done on the arguments.
11979: No errors are reported. You may see some garbage appearing, you may see
11980: simply nothing happen.
11981:
11982: @end table
11983:
11984:
11985: @c =====================================================================
11986: @node The optional File-Access word set, The optional Floating-Point word set, The optional Facility word set, ANS conformance
11987: @section The optional File-Access word set
11988: @c =====================================================================
11989: @cindex system documentation, file words
11990: @cindex file words, system documentation
11991:
11992: @menu
11993: * file-idef:: Implementation Defined Options
11994: * file-ambcond:: Ambiguous Conditions
11995: @end menu
11996:
11997: @c ---------------------------------------------------------------------
11998: @node file-idef, file-ambcond, The optional File-Access word set, The optional File-Access word set
11999: @subsection Implementation Defined Options
12000: @c ---------------------------------------------------------------------
12001: @cindex implementation-defined options, file words
12002: @cindex file words, implementation-defined options
12003:
12004: @table @i
12005: @item file access methods used:
12006: @cindex file access methods used
12007: @code{R/O}, @code{R/W} and @code{BIN} work as you would
12008: expect. @code{W/O} translates into the C file opening mode @code{w} (or
12009: @code{wb}): The file is cleared, if it exists, and created, if it does
12010: not (with both @code{open-file} and @code{create-file}). Under Unix
12011: @code{create-file} creates a file with 666 permissions modified by your
12012: umask.
12013:
12014: @item file exceptions:
12015: @cindex file exceptions
12016: The file words do not raise exceptions (except, perhaps, memory access
12017: faults when you pass illegal addresses or file-ids).
12018:
12019: @item file line terminator:
12020: @cindex file line terminator
12021: System-dependent. Gforth uses C's newline character as line
12022: terminator. What the actual character code(s) of this are is
12023: system-dependent.
12024:
12025: @item file name format:
12026: @cindex file name format
12027: System dependent. Gforth just uses the file name format of your OS.
12028:
12029: @item information returned by @code{FILE-STATUS}:
12030: @cindex @code{FILE-STATUS}, returned information
12031: @code{FILE-STATUS} returns the most powerful file access mode allowed
12032: for the file: Either @code{R/O}, @code{W/O} or @code{R/W}. If the file
12033: cannot be accessed, @code{R/O BIN} is returned. @code{BIN} is applicable
12034: along with the returned mode.
12035:
12036: @item input file state after an exception when including source:
12037: @cindex exception when including source
12038: All files that are left via the exception are closed.
12039:
12040: @item @i{ior} values and meaning:
12041: @cindex @i{ior} values and meaning
12042: The @i{ior}s returned by the file and memory allocation words are
12043: intended as throw codes. They typically are in the range
12044: -512@minus{}-2047 of OS errors. The mapping from OS error numbers to
12045: @i{ior}s is -512@minus{}@i{errno}.
12046:
12047: @item maximum depth of file input nesting:
12048: @cindex maximum depth of file input nesting
12049: @cindex file input nesting, maximum depth
12050: limited by the amount of return stack, locals/TIB stack, and the number
12051: of open files available. This should not give you troubles.
12052:
12053: @item maximum size of input line:
12054: @cindex maximum size of input line
12055: @cindex input line size, maximum
12056: @code{/line}. Currently 255.
12057:
12058: @item methods of mapping block ranges to files:
12059: @cindex mapping block ranges to files
12060: @cindex files containing blocks
12061: @cindex blocks in files
12062: By default, blocks are accessed in the file @file{blocks.fb} in the
12063: current working directory. The file can be switched with @code{USE}.
12064:
12065: @item number of string buffers provided by @code{S"}:
12066: @cindex @code{S"}, number of string buffers
12067: 1
12068:
12069: @item size of string buffer used by @code{S"}:
12070: @cindex @code{S"}, size of string buffer
12071: @code{/line}. currently 255.
12072:
12073: @end table
12074:
12075: @c ---------------------------------------------------------------------
12076: @node file-ambcond, , file-idef, The optional File-Access word set
12077: @subsection Ambiguous conditions
12078: @c ---------------------------------------------------------------------
12079: @cindex file words, ambiguous conditions
12080: @cindex ambiguous conditions, file words
12081:
12082: @table @i
12083: @item attempting to position a file outside its boundaries:
12084: @cindex @code{REPOSITION-FILE}, outside the file's boundaries
12085: @code{REPOSITION-FILE} is performed as usual: Afterwards,
12086: @code{FILE-POSITION} returns the value given to @code{REPOSITION-FILE}.
12087:
12088: @item attempting to read from file positions not yet written:
12089: @cindex reading from file positions not yet written
12090: End-of-file, i.e., zero characters are read and no error is reported.
12091:
12092: @item @i{file-id} is invalid (@code{INCLUDE-FILE}):
12093: @cindex @code{INCLUDE-FILE}, @i{file-id} is invalid
12094: An appropriate exception may be thrown, but a memory fault or other
12095: problem is more probable.
12096:
12097: @item I/O exception reading or closing @i{file-id} (@code{INCLUDE-FILE}, @code{INCLUDED}):
12098: @cindex @code{INCLUDE-FILE}, I/O exception reading or closing @i{file-id}
12099: @cindex @code{INCLUDED}, I/O exception reading or closing @i{file-id}
12100: The @i{ior} produced by the operation, that discovered the problem, is
12101: thrown.
12102:
12103: @item named file cannot be opened (@code{INCLUDED}):
12104: @cindex @code{INCLUDED}, named file cannot be opened
12105: The @i{ior} produced by @code{open-file} is thrown.
12106:
12107: @item requesting an unmapped block number:
12108: @cindex unmapped block numbers
12109: There are no unmapped legal block numbers. On some operating systems,
12110: writing a block with a large number may overflow the file system and
12111: have an error message as consequence.
12112:
12113: @item using @code{source-id} when @code{blk} is non-zero:
12114: @cindex @code{SOURCE-ID}, behaviour when @code{BLK} is non-zero
12115: @code{source-id} performs its function. Typically it will give the id of
12116: the source which loaded the block. (Better ideas?)
12117:
12118: @end table
12119:
12120:
12121: @c =====================================================================
12122: @node The optional Floating-Point word set, The optional Locals word set, The optional File-Access word set, ANS conformance
12123: @section The optional Floating-Point word set
12124: @c =====================================================================
12125: @cindex system documentation, floating-point words
12126: @cindex floating-point words, system documentation
12127:
12128: @menu
12129: * floating-idef:: Implementation Defined Options
12130: * floating-ambcond:: Ambiguous Conditions
12131: @end menu
12132:
12133:
12134: @c ---------------------------------------------------------------------
12135: @node floating-idef, floating-ambcond, The optional Floating-Point word set, The optional Floating-Point word set
12136: @subsection Implementation Defined Options
12137: @c ---------------------------------------------------------------------
12138: @cindex implementation-defined options, floating-point words
12139: @cindex floating-point words, implementation-defined options
12140:
12141: @table @i
12142: @item format and range of floating point numbers:
12143: @cindex format and range of floating point numbers
12144: @cindex floating point numbers, format and range
12145: System-dependent; the @code{double} type of C.
12146:
12147: @item results of @code{REPRESENT} when @i{float} is out of range:
12148: @cindex @code{REPRESENT}, results when @i{float} is out of range
12149: System dependent; @code{REPRESENT} is implemented using the C library
12150: function @code{ecvt()} and inherits its behaviour in this respect.
12151:
12152: @item rounding or truncation of floating-point numbers:
12153: @cindex rounding of floating-point numbers
12154: @cindex truncation of floating-point numbers
12155: @cindex floating-point numbers, rounding or truncation
12156: System dependent; the rounding behaviour is inherited from the hosting C
12157: compiler. IEEE-FP-based (i.e., most) systems by default round to
12158: nearest, and break ties by rounding to even (i.e., such that the last
12159: bit of the mantissa is 0).
12160:
12161: @item size of floating-point stack:
12162: @cindex floating-point stack size
12163: @code{s" FLOATING-STACK" environment? drop .} gives the total size of
12164: the floating-point stack (in floats). You can specify this on startup
12165: with the command-line option @code{-f} (@pxref{Invoking Gforth}).
12166:
12167: @item width of floating-point stack:
12168: @cindex floating-point stack width
12169: @code{1 floats}.
12170:
12171: @end table
12172:
12173:
12174: @c ---------------------------------------------------------------------
12175: @node floating-ambcond, , floating-idef, The optional Floating-Point word set
12176: @subsection Ambiguous conditions
12177: @c ---------------------------------------------------------------------
12178: @cindex floating-point words, ambiguous conditions
12179: @cindex ambiguous conditions, floating-point words
12180:
12181: @table @i
12182: @item @code{df@@} or @code{df!} used with an address that is not double-float aligned:
12183: @cindex @code{df@@} or @code{df!} used with an address that is not double-float aligned
12184: System-dependent. Typically results in a @code{-23 THROW} like other
12185: alignment violations.
12186:
12187: @item @code{f@@} or @code{f!} used with an address that is not float aligned:
12188: @cindex @code{f@@} used with an address that is not float aligned
12189: @cindex @code{f!} used with an address that is not float aligned
12190: System-dependent. Typically results in a @code{-23 THROW} like other
12191: alignment violations.
12192:
12193: @item floating-point result out of range:
12194: @cindex floating-point result out of range
12195: System-dependent. Can result in a @code{-55 THROW} (Floating-point
12196: unidentified fault), or can produce a special value representing, e.g.,
12197: Infinity.
12198:
12199: @item @code{sf@@} or @code{sf!} used with an address that is not single-float aligned:
12200: @cindex @code{sf@@} or @code{sf!} used with an address that is not single-float aligned
12201: System-dependent. Typically results in an alignment fault like other
12202: alignment violations.
12203:
12204: @item @code{base} is not decimal (@code{REPRESENT}, @code{F.}, @code{FE.}, @code{FS.}):
12205: @cindex @code{base} is not decimal (@code{REPRESENT}, @code{F.}, @code{FE.}, @code{FS.})
12206: The floating-point number is converted into decimal nonetheless.
12207:
12208: @item Both arguments are equal to zero (@code{FATAN2}):
12209: @cindex @code{FATAN2}, both arguments are equal to zero
12210: System-dependent. @code{FATAN2} is implemented using the C library
12211: function @code{atan2()}.
12212:
12213: @item Using @code{FTAN} on an argument @i{r1} where cos(@i{r1}) is zero:
12214: @cindex @code{FTAN} on an argument @i{r1} where cos(@i{r1}) is zero
12215: System-dependent. Anyway, typically the cos of @i{r1} will not be zero
12216: because of small errors and the tan will be a very large (or very small)
12217: but finite number.
12218:
12219: @item @i{d} cannot be presented precisely as a float in @code{D>F}:
12220: @cindex @code{D>F}, @i{d} cannot be presented precisely as a float
12221: The result is rounded to the nearest float.
12222:
12223: @item dividing by zero:
12224: @cindex dividing by zero, floating-point
12225: @cindex floating-point dividing by zero
12226: @cindex floating-point unidentified fault, FP divide-by-zero
12227: @code{-55 throw} (Floating-point unidentified fault)
12228:
12229: @item exponent too big for conversion (@code{DF!}, @code{DF@@}, @code{SF!}, @code{SF@@}):
12230: @cindex exponent too big for conversion (@code{DF!}, @code{DF@@}, @code{SF!}, @code{SF@@})
12231: System dependent. On IEEE-FP based systems the number is converted into
12232: an infinity.
12233:
12234: @item @i{float}<1 (@code{FACOSH}):
12235: @cindex @code{FACOSH}, @i{float}<1
12236: @cindex floating-point unidentified fault, @code{FACOSH}
12237: @code{-55 throw} (Floating-point unidentified fault)
12238:
12239: @item @i{float}=<-1 (@code{FLNP1}):
12240: @cindex @code{FLNP1}, @i{float}=<-1
12241: @cindex floating-point unidentified fault, @code{FLNP1}
12242: @code{-55 throw} (Floating-point unidentified fault). On IEEE-FP systems
12243: negative infinity is typically produced for @i{float}=-1.
12244:
12245: @item @i{float}=<0 (@code{FLN}, @code{FLOG}):
12246: @cindex @code{FLN}, @i{float}=<0
12247: @cindex @code{FLOG}, @i{float}=<0
12248: @cindex floating-point unidentified fault, @code{FLN} or @code{FLOG}
12249: @code{-55 throw} (Floating-point unidentified fault). On IEEE-FP systems
12250: negative infinity is typically produced for @i{float}=0.
12251:
12252: @item @i{float}<0 (@code{FASINH}, @code{FSQRT}):
12253: @cindex @code{FASINH}, @i{float}<0
12254: @cindex @code{FSQRT}, @i{float}<0
12255: @cindex floating-point unidentified fault, @code{FASINH} or @code{FSQRT}
12256: @code{-55 throw} (Floating-point unidentified fault). @code{fasinh}
12257: produces values for these inputs on my Linux box (Bug in the C library?)
12258:
12259: @item |@i{float}|>1 (@code{FACOS}, @code{FASIN}, @code{FATANH}):
12260: @cindex @code{FACOS}, |@i{float}|>1
12261: @cindex @code{FASIN}, |@i{float}|>1
12262: @cindex @code{FATANH}, |@i{float}|>1
12263: @cindex floating-point unidentified fault, @code{FACOS}, @code{FASIN} or @code{FATANH}
12264: @code{-55 throw} (Floating-point unidentified fault).
12265:
12266: @item integer part of float cannot be represented by @i{d} in @code{F>D}:
12267: @cindex @code{F>D}, integer part of float cannot be represented by @i{d}
12268: @cindex floating-point unidentified fault, @code{F>D}
12269: @code{-55 throw} (Floating-point unidentified fault).
12270:
12271: @item string larger than pictured numeric output area (@code{f.}, @code{fe.}, @code{fs.}):
12272: @cindex string larger than pictured numeric output area (@code{f.}, @code{fe.}, @code{fs.})
12273: This does not happen.
12274: @end table
12275:
12276: @c =====================================================================
12277: @node The optional Locals word set, The optional Memory-Allocation word set, The optional Floating-Point word set, ANS conformance
12278: @section The optional Locals word set
12279: @c =====================================================================
12280: @cindex system documentation, locals words
12281: @cindex locals words, system documentation
12282:
12283: @menu
12284: * locals-idef:: Implementation Defined Options
12285: * locals-ambcond:: Ambiguous Conditions
12286: @end menu
12287:
12288:
12289: @c ---------------------------------------------------------------------
12290: @node locals-idef, locals-ambcond, The optional Locals word set, The optional Locals word set
12291: @subsection Implementation Defined Options
12292: @c ---------------------------------------------------------------------
12293: @cindex implementation-defined options, locals words
12294: @cindex locals words, implementation-defined options
12295:
12296: @table @i
12297: @item maximum number of locals in a definition:
12298: @cindex maximum number of locals in a definition
12299: @cindex locals, maximum number in a definition
12300: @code{s" #locals" environment? drop .}. Currently 15. This is a lower
12301: bound, e.g., on a 32-bit machine there can be 41 locals of up to 8
12302: characters. The number of locals in a definition is bounded by the size
12303: of locals-buffer, which contains the names of the locals.
12304:
12305: @end table
12306:
12307:
12308: @c ---------------------------------------------------------------------
12309: @node locals-ambcond, , locals-idef, The optional Locals word set
12310: @subsection Ambiguous conditions
12311: @c ---------------------------------------------------------------------
12312: @cindex locals words, ambiguous conditions
12313: @cindex ambiguous conditions, locals words
12314:
12315: @table @i
12316: @item executing a named local in interpretation state:
12317: @cindex local in interpretation state
12318: @cindex Interpreting a compile-only word, for a local
12319: Locals have no interpretation semantics. If you try to perform the
12320: interpretation semantics, you will get a @code{-14 throw} somewhere
12321: (Interpreting a compile-only word). If you perform the compilation
12322: semantics, the locals access will be compiled (irrespective of state).
12323:
12324: @item @i{name} not defined by @code{VALUE} or @code{(LOCAL)} (@code{TO}):
12325: @cindex name not defined by @code{VALUE} or @code{(LOCAL)} used by @code{TO}
12326: @cindex @code{TO} on non-@code{VALUE}s and non-locals
12327: @cindex Invalid name argument, @code{TO}
12328: @code{-32 throw} (Invalid name argument)
12329:
12330: @end table
12331:
12332:
12333: @c =====================================================================
12334: @node The optional Memory-Allocation word set, The optional Programming-Tools word set, The optional Locals word set, ANS conformance
12335: @section The optional Memory-Allocation word set
12336: @c =====================================================================
12337: @cindex system documentation, memory-allocation words
12338: @cindex memory-allocation words, system documentation
12339:
12340: @menu
12341: * memory-idef:: Implementation Defined Options
12342: @end menu
12343:
12344:
12345: @c ---------------------------------------------------------------------
12346: @node memory-idef, , The optional Memory-Allocation word set, The optional Memory-Allocation word set
12347: @subsection Implementation Defined Options
12348: @c ---------------------------------------------------------------------
12349: @cindex implementation-defined options, memory-allocation words
12350: @cindex memory-allocation words, implementation-defined options
12351:
12352: @table @i
12353: @item values and meaning of @i{ior}:
12354: @cindex @i{ior} values and meaning
12355: The @i{ior}s returned by the file and memory allocation words are
12356: intended as throw codes. They typically are in the range
12357: -512@minus{}-2047 of OS errors. The mapping from OS error numbers to
12358: @i{ior}s is -512@minus{}@i{errno}.
12359:
12360: @end table
12361:
12362: @c =====================================================================
12363: @node The optional Programming-Tools word set, The optional Search-Order word set, The optional Memory-Allocation word set, ANS conformance
12364: @section The optional Programming-Tools word set
12365: @c =====================================================================
12366: @cindex system documentation, programming-tools words
12367: @cindex programming-tools words, system documentation
12368:
12369: @menu
12370: * programming-idef:: Implementation Defined Options
12371: * programming-ambcond:: Ambiguous Conditions
12372: @end menu
12373:
12374:
12375: @c ---------------------------------------------------------------------
12376: @node programming-idef, programming-ambcond, The optional Programming-Tools word set, The optional Programming-Tools word set
12377: @subsection Implementation Defined Options
12378: @c ---------------------------------------------------------------------
12379: @cindex implementation-defined options, programming-tools words
12380: @cindex programming-tools words, implementation-defined options
12381:
12382: @table @i
12383: @item ending sequence for input following @code{;CODE} and @code{CODE}:
12384: @cindex @code{;CODE} ending sequence
12385: @cindex @code{CODE} ending sequence
12386: @code{END-CODE}
12387:
12388: @item manner of processing input following @code{;CODE} and @code{CODE}:
12389: @cindex @code{;CODE}, processing input
12390: @cindex @code{CODE}, processing input
12391: The @code{ASSEMBLER} vocabulary is pushed on the search order stack, and
12392: the input is processed by the text interpreter, (starting) in interpret
12393: state.
12394:
12395: @item search order capability for @code{EDITOR} and @code{ASSEMBLER}:
12396: @cindex @code{ASSEMBLER}, search order capability
12397: The ANS Forth search order word set.
12398:
12399: @item source and format of display by @code{SEE}:
12400: @cindex @code{SEE}, source and format of output
12401: The source for @code{see} is the intermediate code used by the inner
12402: interpreter. The current @code{see} tries to output Forth source code
12403: as well as possible.
12404:
12405: @end table
12406:
12407: @c ---------------------------------------------------------------------
12408: @node programming-ambcond, , programming-idef, The optional Programming-Tools word set
12409: @subsection Ambiguous conditions
12410: @c ---------------------------------------------------------------------
12411: @cindex programming-tools words, ambiguous conditions
12412: @cindex ambiguous conditions, programming-tools words
12413:
12414: @table @i
12415:
12416: @item deleting the compilation word list (@code{FORGET}):
12417: @cindex @code{FORGET}, deleting the compilation word list
12418: Not implemented (yet).
12419:
12420: @item fewer than @i{u}+1 items on the control-flow stack (@code{CS-PICK}, @code{CS-ROLL}):
12421: @cindex @code{CS-PICK}, fewer than @i{u}+1 items on the control flow-stack
12422: @cindex @code{CS-ROLL}, fewer than @i{u}+1 items on the control flow-stack
12423: @cindex control-flow stack underflow
12424: This typically results in an @code{abort"} with a descriptive error
12425: message (may change into a @code{-22 throw} (Control structure mismatch)
12426: in the future). You may also get a memory access error. If you are
12427: unlucky, this ambiguous condition is not caught.
12428:
12429: @item @i{name} can't be found (@code{FORGET}):
12430: @cindex @code{FORGET}, @i{name} can't be found
12431: Not implemented (yet).
12432:
12433: @item @i{name} not defined via @code{CREATE}:
12434: @cindex @code{;CODE}, @i{name} not defined via @code{CREATE}
12435: @code{;CODE} behaves like @code{DOES>} in this respect, i.e., it changes
12436: the execution semantics of the last defined word no matter how it was
12437: defined.
12438:
12439: @item @code{POSTPONE} applied to @code{[IF]}:
12440: @cindex @code{POSTPONE} applied to @code{[IF]}
12441: @cindex @code{[IF]} and @code{POSTPONE}
12442: After defining @code{: X POSTPONE [IF] ; IMMEDIATE}. @code{X} is
12443: equivalent to @code{[IF]}.
12444:
12445: @item reaching the end of the input source before matching @code{[ELSE]} or @code{[THEN]}:
12446: @cindex @code{[IF]}, end of the input source before matching @code{[ELSE]} or @code{[THEN]}
12447: Continue in the same state of conditional compilation in the next outer
12448: input source. Currently there is no warning to the user about this.
12449:
12450: @item removing a needed definition (@code{FORGET}):
12451: @cindex @code{FORGET}, removing a needed definition
12452: Not implemented (yet).
12453:
12454: @end table
12455:
12456:
12457: @c =====================================================================
12458: @node The optional Search-Order word set, , The optional Programming-Tools word set, ANS conformance
12459: @section The optional Search-Order word set
12460: @c =====================================================================
12461: @cindex system documentation, search-order words
12462: @cindex search-order words, system documentation
12463:
12464: @menu
12465: * search-idef:: Implementation Defined Options
12466: * search-ambcond:: Ambiguous Conditions
12467: @end menu
12468:
12469:
12470: @c ---------------------------------------------------------------------
12471: @node search-idef, search-ambcond, The optional Search-Order word set, The optional Search-Order word set
12472: @subsection Implementation Defined Options
12473: @c ---------------------------------------------------------------------
12474: @cindex implementation-defined options, search-order words
12475: @cindex search-order words, implementation-defined options
12476:
12477: @table @i
12478: @item maximum number of word lists in search order:
12479: @cindex maximum number of word lists in search order
12480: @cindex search order, maximum depth
12481: @code{s" wordlists" environment? drop .}. Currently 16.
12482:
12483: @item minimum search order:
12484: @cindex minimum search order
12485: @cindex search order, minimum
12486: @code{root root}.
12487:
12488: @end table
12489:
12490: @c ---------------------------------------------------------------------
12491: @node search-ambcond, , search-idef, The optional Search-Order word set
12492: @subsection Ambiguous conditions
12493: @c ---------------------------------------------------------------------
12494: @cindex search-order words, ambiguous conditions
12495: @cindex ambiguous conditions, search-order words
12496:
12497: @table @i
12498: @item changing the compilation word list (during compilation):
12499: @cindex changing the compilation word list (during compilation)
12500: @cindex compilation word list, change before definition ends
12501: The word is entered into the word list that was the compilation word list
12502: at the start of the definition. Any changes to the name field (e.g.,
12503: @code{immediate}) or the code field (e.g., when executing @code{DOES>})
12504: are applied to the latest defined word (as reported by @code{last} or
12505: @code{lastxt}), if possible, irrespective of the compilation word list.
12506:
12507: @item search order empty (@code{previous}):
12508: @cindex @code{previous}, search order empty
12509: @cindex vocstack empty, @code{previous}
12510: @code{abort" Vocstack empty"}.
12511:
12512: @item too many word lists in search order (@code{also}):
12513: @cindex @code{also}, too many word lists in search order
12514: @cindex vocstack full, @code{also}
12515: @code{abort" Vocstack full"}.
12516:
12517: @end table
12518:
12519: @c ***************************************************************
12520: @node Model, Integrating Gforth, ANS conformance, Top
12521: @chapter Model
12522:
12523: This chapter has yet to be written. It will contain information, on
12524: which internal structures you can rely.
12525:
12526: @c ***************************************************************
12527: @node Integrating Gforth, Emacs and Gforth, Model, Top
12528: @chapter Integrating Gforth into C programs
12529:
12530: This is not yet implemented.
12531:
12532: Several people like to use Forth as scripting language for applications
12533: that are otherwise written in C, C++, or some other language.
12534:
12535: The Forth system ATLAST provides facilities for embedding it into
12536: applications; unfortunately it has several disadvantages: most
12537: importantly, it is not based on ANS Forth, and it is apparently dead
12538: (i.e., not developed further and not supported). The facilities
12539: provided by Gforth in this area are inspired by ATLAST's facilities, so
12540: making the switch should not be hard.
12541:
12542: We also tried to design the interface such that it can easily be
12543: implemented by other Forth systems, so that we may one day arrive at a
12544: standardized interface. Such a standard interface would allow you to
12545: replace the Forth system without having to rewrite C code.
12546:
12547: You embed the Gforth interpreter by linking with the library
12548: @code{libgforth.a} (give the compiler the option @code{-lgforth}). All
12549: global symbols in this library that belong to the interface, have the
12550: prefix @code{forth_}. (Global symbols that are used internally have the
12551: prefix @code{gforth_}).
12552:
12553: You can include the declarations of Forth types and the functions and
12554: variables of the interface with @code{#include <forth.h>}.
12555:
12556: Types.
12557:
12558: Variables.
12559:
12560: Data and FP Stack pointer. Area sizes.
12561:
12562: functions.
12563:
12564: forth_init(imagefile)
12565: forth_evaluate(string) exceptions?
12566: forth_goto(address) (or forth_execute(xt)?)
12567: forth_continue() (a corountining mechanism)
12568:
12569: Adding primitives.
12570:
12571: No checking.
12572:
12573: Signals?
12574:
12575: Accessing the Stacks
12576:
12577: @c ******************************************************************
12578: @node Emacs and Gforth, Image Files, Integrating Gforth, Top
12579: @chapter Emacs and Gforth
12580: @cindex Emacs and Gforth
12581:
12582: @cindex @file{gforth.el}
12583: @cindex @file{forth.el}
12584: @cindex Rydqvist, Goran
12585: @cindex comment editing commands
12586: @cindex @code{\}, editing with Emacs
12587: @cindex debug tracer editing commands
12588: @cindex @code{~~}, removal with Emacs
12589: @cindex Forth mode in Emacs
12590: Gforth comes with @file{gforth.el}, an improved version of
12591: @file{forth.el} by Goran Rydqvist (included in the TILE package). The
12592: improvements are:
12593:
12594: @itemize @bullet
12595: @item
12596: A better (but still not perfect) handling of indentation.
12597: @item
12598: Comment paragraph filling (@kbd{M-q})
12599: @item
12600: Commenting (@kbd{C-x \}) and uncommenting (@kbd{C-u C-x \}) of regions
12601: @item
12602: Removal of debugging tracers (@kbd{C-x ~}, @pxref{Debugging}).
12603: @item
12604: Support of the @code{info-lookup} feature for looking up the
12605: documentation of a word.
12606: @end itemize
12607:
12608: I left the stuff I do not use alone, even though some of it only makes
12609: sense for TILE. To get a description of these features, enter Forth mode
12610: and type @kbd{C-h m}.
12611:
12612: @cindex source location of error or debugging output in Emacs
12613: @cindex error output, finding the source location in Emacs
12614: @cindex debugging output, finding the source location in Emacs
12615: In addition, Gforth supports Emacs quite well: The source code locations
12616: given in error messages, debugging output (from @code{~~}) and failed
12617: assertion messages are in the right format for Emacs' compilation mode
12618: (@pxref{Compilation, , Running Compilations under Emacs, emacs, Emacs
12619: Manual}) so the source location corresponding to an error or other
12620: message is only a few keystrokes away (@kbd{C-x `} for the next error,
12621: @kbd{C-c C-c} for the error under the cursor).
12622:
12623: @cindex @file{TAGS} file
12624: @cindex @file{etags.fs}
12625: @cindex viewing the source of a word in Emacs
12626: @cindex @code{require}, placement in files
12627: @cindex @code{include}, placement in files
12628: Also, if you @code{require} @file{etags.fs}, a new @file{TAGS} file will
12629: be produced (@pxref{Tags, , Tags Tables, emacs, Emacs Manual}) that
12630: contains the definitions of all words defined afterwards. You can then
12631: find the source for a word using @kbd{M-.}. Note that emacs can use
12632: several tags files at the same time (e.g., one for the Gforth sources
12633: and one for your program, @pxref{Select Tags Table,,Selecting a Tags
12634: Table,emacs, Emacs Manual}). The TAGS file for the preloaded words is
12635: @file{$(datadir)/gforth/$(VERSION)/TAGS} (e.g.,
12636: @file{/usr/local/share/gforth/0.2.0/TAGS}). To get the best behaviour
12637: with @file{etags.fs}, you should avoid putting definitions both before
12638: and after @code{require} etc., otherwise you will see the same file
12639: visited several times by commands like @code{tags-search}.
12640:
12641: @cindex viewing the documentation of a word in Emacs
12642: @cindex context-sensitive help
12643: Moreover, for words documented in this manual, you can look up the
12644: glossary entry quickly by using @kbd{C-h TAB}
12645: (@code{info-lookup-symbol}, see @pxref{Documentation, ,Documentation
12646: Commands, emacs, Emacs Manual}). This feature requires Emacs 20.3 or
12647: later and does not work for words containing @code{:}.
12648:
12649:
12650: @cindex @file{.emacs}
12651: To get all these benefits, add the following lines to your @file{.emacs}
12652: file:
12653:
12654: @example
12655: (autoload 'forth-mode "gforth.el")
12656: (setq auto-mode-alist (cons '("\\.fs\\'" . forth-mode) auto-mode-alist))
12657: @end example
12658:
12659: @c ******************************************************************
12660: @node Image Files, Engine, Emacs and Gforth, Top
12661: @chapter Image Files
12662: @cindex image file
12663: @cindex @file{.fi} files
12664: @cindex precompiled Forth code
12665: @cindex dictionary in persistent form
12666: @cindex persistent form of dictionary
12667:
12668: An image file is a file containing an image of the Forth dictionary,
12669: i.e., compiled Forth code and data residing in the dictionary. By
12670: convention, we use the extension @code{.fi} for image files.
12671:
12672: @menu
12673: * Image Licensing Issues:: Distribution terms for images.
12674: * Image File Background:: Why have image files?
12675: * Non-Relocatable Image Files:: don't always work.
12676: * Data-Relocatable Image Files:: are better.
12677: * Fully Relocatable Image Files:: better yet.
12678: * Stack and Dictionary Sizes:: Setting the default sizes for an image.
12679: * Running Image Files:: @code{gforth -i @i{file}} or @i{file}.
12680: * Modifying the Startup Sequence:: and turnkey applications.
12681: @end menu
12682:
12683: @node Image Licensing Issues, Image File Background, Image Files, Image Files
12684: @section Image Licensing Issues
12685: @cindex license for images
12686: @cindex image license
12687:
12688: An image created with @code{gforthmi} (@pxref{gforthmi}) or
12689: @code{savesystem} (@pxref{Non-Relocatable Image Files}) includes the
12690: original image; i.e., according to copyright law it is a derived work of
12691: the original image.
12692:
12693: Since Gforth is distributed under the GNU GPL, the newly created image
12694: falls under the GNU GPL, too. In particular, this means that if you
12695: distribute the image, you have to make all of the sources for the image
12696: available, including those you wrote. For details see @ref{License, ,
12697: GNU General Public License (Section 3)}.
12698:
12699: If you create an image with @code{cross} (@pxref{cross.fs}), the image
12700: contains only code compiled from the sources you gave it; if none of
12701: these sources is under the GPL, the terms discussed above do not apply
12702: to the image. However, if your image needs an engine (a gforth binary)
12703: that is under the GPL, you should make sure that you distribute both in
12704: a way that is at most a @emph{mere aggregation}, if you don't want the
12705: terms of the GPL to apply to the image.
12706:
12707: @node Image File Background, Non-Relocatable Image Files, Image Licensing Issues, Image Files
12708: @section Image File Background
12709: @cindex image file background
12710:
12711: Our Forth system consists not only of primitives, but also of
12712: definitions written in Forth. Since the Forth compiler itself belongs to
12713: those definitions, it is not possible to start the system with the
12714: primitives and the Forth source alone. Therefore we provide the Forth
12715: code as an image file in nearly executable form. When Gforth starts up,
12716: a C routine loads the image file into memory, optionally relocates the
12717: addresses, then sets up the memory (stacks etc.) according to
12718: information in the image file, and (finally) starts executing Forth
12719: code.
12720:
12721: The image file variants represent different compromises between the
12722: goals of making it easy to generate image files and making them
12723: portable.
12724:
12725: @cindex relocation at run-time
12726: Win32Forth 3.4 and Mitch Bradley's @code{cforth} use relocation at
12727: run-time. This avoids many of the complications discussed below (image
12728: files are data relocatable without further ado), but costs performance
12729: (one addition per memory access).
12730:
12731: @cindex relocation at load-time
12732: By contrast, the Gforth loader performs relocation at image load time. The
12733: loader also has to replace tokens that represent primitive calls with the
12734: appropriate code-field addresses (or code addresses in the case of
12735: direct threading).
12736:
12737: There are three kinds of image files, with different degrees of
12738: relocatability: non-relocatable, data-relocatable, and fully relocatable
12739: image files.
12740:
12741: @cindex image file loader
12742: @cindex relocating loader
12743: @cindex loader for image files
12744: These image file variants have several restrictions in common; they are
12745: caused by the design of the image file loader:
12746:
12747: @itemize @bullet
12748: @item
12749: There is only one segment; in particular, this means, that an image file
12750: cannot represent @code{ALLOCATE}d memory chunks (and pointers to
12751: them). The contents of the stacks are not represented, either.
12752:
12753: @item
12754: The only kinds of relocation supported are: adding the same offset to
12755: all cells that represent data addresses; and replacing special tokens
12756: with code addresses or with pieces of machine code.
12757:
12758: If any complex computations involving addresses are performed, the
12759: results cannot be represented in the image file. Several applications that
12760: use such computations come to mind:
12761: @itemize @minus
12762: @item
12763: Hashing addresses (or data structures which contain addresses) for table
12764: lookup. If you use Gforth's @code{table}s or @code{wordlist}s for this
12765: purpose, you will have no problem, because the hash tables are
12766: recomputed automatically when the system is started. If you use your own
12767: hash tables, you will have to do something similar.
12768:
12769: @item
12770: There's a cute implementation of doubly-linked lists that uses
12771: @code{XOR}ed addresses. You could represent such lists as singly-linked
12772: in the image file, and restore the doubly-linked representation on
12773: startup.@footnote{In my opinion, though, you should think thrice before
12774: using a doubly-linked list (whatever implementation).}
12775:
12776: @item
12777: The code addresses of run-time routines like @code{docol:} cannot be
12778: represented in the image file (because their tokens would be replaced by
12779: machine code in direct threaded implementations). As a workaround,
12780: compute these addresses at run-time with @code{>code-address} from the
12781: executions tokens of appropriate words (see the definitions of
12782: @code{docol:} and friends in @file{kernel.fs}).
12783:
12784: @item
12785: On many architectures addresses are represented in machine code in some
12786: shifted or mangled form. You cannot put @code{CODE} words that contain
12787: absolute addresses in this form in a relocatable image file. Workarounds
12788: are representing the address in some relative form (e.g., relative to
12789: the CFA, which is present in some register), or loading the address from
12790: a place where it is stored in a non-mangled form.
12791: @end itemize
12792: @end itemize
12793:
12794: @node Non-Relocatable Image Files, Data-Relocatable Image Files, Image File Background, Image Files
12795: @section Non-Relocatable Image Files
12796: @cindex non-relocatable image files
12797: @cindex image file, non-relocatable
12798:
12799: These files are simple memory dumps of the dictionary. They are specific
12800: to the executable (i.e., @file{gforth} file) they were created
12801: with. What's worse, they are specific to the place on which the
12802: dictionary resided when the image was created. Now, there is no
12803: guarantee that the dictionary will reside at the same place the next
12804: time you start Gforth, so there's no guarantee that a non-relocatable
12805: image will work the next time (Gforth will complain instead of crashing,
12806: though).
12807:
12808: You can create a non-relocatable image file with
12809:
12810:
12811: doc-savesystem
12812:
12813:
12814: @node Data-Relocatable Image Files, Fully Relocatable Image Files, Non-Relocatable Image Files, Image Files
12815: @section Data-Relocatable Image Files
12816: @cindex data-relocatable image files
12817: @cindex image file, data-relocatable
12818:
12819: These files contain relocatable data addresses, but fixed code addresses
12820: (instead of tokens). They are specific to the executable (i.e.,
12821: @file{gforth} file) they were created with. For direct threading on some
12822: architectures (e.g., the i386), data-relocatable images do not work. You
12823: get a data-relocatable image, if you use @file{gforthmi} with a
12824: Gforth binary that is not doubly indirect threaded (@pxref{Fully
12825: Relocatable Image Files}).
12826:
12827: @node Fully Relocatable Image Files, Stack and Dictionary Sizes, Data-Relocatable Image Files, Image Files
12828: @section Fully Relocatable Image Files
12829: @cindex fully relocatable image files
12830: @cindex image file, fully relocatable
12831:
12832: @cindex @file{kern*.fi}, relocatability
12833: @cindex @file{gforth.fi}, relocatability
12834: These image files have relocatable data addresses, and tokens for code
12835: addresses. They can be used with different binaries (e.g., with and
12836: without debugging) on the same machine, and even across machines with
12837: the same data formats (byte order, cell size, floating point
12838: format). However, they are usually specific to the version of Gforth
12839: they were created with. The files @file{gforth.fi} and @file{kernl*.fi}
12840: are fully relocatable.
12841:
12842: There are two ways to create a fully relocatable image file:
12843:
12844: @menu
12845: * gforthmi:: The normal way
12846: * cross.fs:: The hard way
12847: @end menu
12848:
12849: @node gforthmi, cross.fs, Fully Relocatable Image Files, Fully Relocatable Image Files
12850: @subsection @file{gforthmi}
12851: @cindex @file{comp-i.fs}
12852: @cindex @file{gforthmi}
12853:
12854: You will usually use @file{gforthmi}. If you want to create an
12855: image @i{file} that contains everything you would load by invoking
12856: Gforth with @code{gforth @i{options}}, you simply say:
12857: @example
12858: gforthmi @i{file} @i{options}
12859: @end example
12860:
12861: E.g., if you want to create an image @file{asm.fi} that has the file
12862: @file{asm.fs} loaded in addition to the usual stuff, you could do it
12863: like this:
12864:
12865: @example
12866: gforthmi asm.fi asm.fs
12867: @end example
12868:
12869: @file{gforthmi} is implemented as a sh script and works like this: It
12870: produces two non-relocatable images for different addresses and then
12871: compares them. Its output reflects this: first you see the output (if
12872: any) of the two Gforth invocations that produce the nonrelocatable image
12873: files, then you see the output of the comparing program: It displays the
12874: offset used for data addresses and the offset used for code addresses;
12875: moreover, for each cell that cannot be represented correctly in the
12876: image files, it displays a line like this:
12877:
12878: @example
12879: 78DC BFFFFA50 BFFFFA40
12880: @end example
12881:
12882: This means that at offset $78dc from @code{forthstart}, one input image
12883: contains $bffffa50, and the other contains $bffffa40. Since these cells
12884: cannot be represented correctly in the output image, you should examine
12885: these places in the dictionary and verify that these cells are dead
12886: (i.e., not read before they are written).
12887:
12888: @cindex --application, @code{gforthmi} option
12889: If you insert the option @code{--application} in front of the image file
12890: name, you will get an image that uses the @code{--appl-image} option
12891: instead of the @code{--image-file} option (@pxref{Invoking
12892: Gforth}). When you execute such an image on Unix (by typing the image
12893: name as command), the Gforth engine will pass all options to the image
12894: instead of trying to interpret them as engine options.
12895:
12896: If you type @file{gforthmi} with no arguments, it prints some usage
12897: instructions.
12898:
12899: @cindex @code{savesystem} during @file{gforthmi}
12900: @cindex @code{bye} during @file{gforthmi}
12901: @cindex doubly indirect threaded code
12902: @cindex environment variables
12903: @cindex @code{GFORTHD} -- environment variable
12904: @cindex @code{GFORTH} -- environment variable
12905: @cindex @code{gforth-ditc}
12906: There are a few wrinkles: After processing the passed @i{options}, the
12907: words @code{savesystem} and @code{bye} must be visible. A special doubly
12908: indirect threaded version of the @file{gforth} executable is used for
12909: creating the nonrelocatable images; you can pass the exact filename of
12910: this executable through the environment variable @code{GFORTHD}
12911: (default: @file{gforth-ditc}); if you pass a version that is not doubly
12912: indirect threaded, you will not get a fully relocatable image, but a
12913: data-relocatable image (because there is no code address offset). The
12914: normal @file{gforth} executable is used for creating the relocatable
12915: image; you can pass the exact filename of this executable through the
12916: environment variable @code{GFORTH}.
12917:
12918: @node cross.fs, , gforthmi, Fully Relocatable Image Files
12919: @subsection @file{cross.fs}
12920: @cindex @file{cross.fs}
12921: @cindex cross-compiler
12922: @cindex metacompiler
12923: @cindex target compiler
12924:
12925: You can also use @code{cross}, a batch compiler that accepts a Forth-like
12926: programming language (@pxref{Cross Compiler}).
12927:
12928: @code{cross} allows you to create image files for machines with
12929: different data sizes and data formats than the one used for generating
12930: the image file. You can also use it to create an application image that
12931: does not contain a Forth compiler. These features are bought with
12932: restrictions and inconveniences in programming. E.g., addresses have to
12933: be stored in memory with special words (@code{A!}, @code{A,}, etc.) in
12934: order to make the code relocatable.
12935:
12936:
12937: @node Stack and Dictionary Sizes, Running Image Files, Fully Relocatable Image Files, Image Files
12938: @section Stack and Dictionary Sizes
12939: @cindex image file, stack and dictionary sizes
12940: @cindex dictionary size default
12941: @cindex stack size default
12942:
12943: If you invoke Gforth with a command line flag for the size
12944: (@pxref{Invoking Gforth}), the size you specify is stored in the
12945: dictionary. If you save the dictionary with @code{savesystem} or create
12946: an image with @file{gforthmi}, this size will become the default
12947: for the resulting image file. E.g., the following will create a
12948: fully relocatable version of @file{gforth.fi} with a 1MB dictionary:
12949:
12950: @example
12951: gforthmi gforth.fi -m 1M
12952: @end example
12953:
12954: In other words, if you want to set the default size for the dictionary
12955: and the stacks of an image, just invoke @file{gforthmi} with the
12956: appropriate options when creating the image.
12957:
12958: @cindex stack size, cache-friendly
12959: Note: For cache-friendly behaviour (i.e., good performance), you should
12960: make the sizes of the stacks modulo, say, 2K, somewhat different. E.g.,
12961: the default stack sizes are: data: 16k (mod 2k=0); fp: 15.5k (mod
12962: 2k=1.5k); return: 15k(mod 2k=1k); locals: 14.5k (mod 2k=0.5k).
12963:
12964: @node Running Image Files, Modifying the Startup Sequence, Stack and Dictionary Sizes, Image Files
12965: @section Running Image Files
12966: @cindex running image files
12967: @cindex invoking image files
12968: @cindex image file invocation
12969:
12970: @cindex -i, invoke image file
12971: @cindex --image file, invoke image file
12972: You can invoke Gforth with an image file @i{image} instead of the
12973: default @file{gforth.fi} with the @code{-i} flag (@pxref{Invoking Gforth}):
12974: @example
12975: gforth -i @i{image}
12976: @end example
12977:
12978: @cindex executable image file
12979: @cindex image file, executable
12980: If your operating system supports starting scripts with a line of the
12981: form @code{#! ...}, you just have to type the image file name to start
12982: Gforth with this image file (note that the file extension @code{.fi} is
12983: just a convention). I.e., to run Gforth with the image file @i{image},
12984: you can just type @i{image} instead of @code{gforth -i @i{image}}.
12985: This works because every @code{.fi} file starts with a line of this
12986: format:
12987:
12988: @example
12989: #! /usr/local/bin/gforth-0.4.0 -i
12990: @end example
12991:
12992: The file and pathname for the Gforth engine specified on this line is
12993: the specific Gforth executable that it was built against; i.e. the value
12994: of the environment variable @code{GFORTH} at the time that
12995: @file{gforthmi} was executed.
12996:
12997: You can make use of the same shell capability to make a Forth source
12998: file into an executable. For example, if you place this text in a file:
12999:
13000: @example
13001: #! /usr/local/bin/gforth
13002:
13003: ." Hello, world" CR
13004: bye
13005: @end example
13006:
13007: @noindent
13008: and then make the file executable (chmod +x in Unix), you can run it
13009: directly from the command line. The sequence @code{#!} is used in two
13010: ways; firstly, it is recognised as a ``magic sequence'' by the operating
13011: system@footnote{The Unix kernel actually recognises two types of files:
13012: executable files and files of data, where the data is processed by an
13013: interpreter that is specified on the ``interpreter line'' -- the first
13014: line of the file, starting with the sequence #!. There may be a small
13015: limit (e.g., 32) on the number of characters that may be specified on
13016: the interpreter line.} secondly it is treated as a comment character by
13017: Gforth. Because of the second usage, a space is required between
13018: @code{#!} and the path to the executable.
13019:
13020: The disadvantage of this latter technique, compared with using
13021: @file{gforthmi}, is that it is slower; the Forth source code is compiled
13022: on-the-fly, each time the program is invoked.
13023:
13024:
13025: doc-#!
13026:
13027:
13028: @node Modifying the Startup Sequence, , Running Image Files, Image Files
13029: @section Modifying the Startup Sequence
13030: @cindex startup sequence for image file
13031: @cindex image file initialization sequence
13032: @cindex initialization sequence of image file
13033:
13034: You can add your own initialization to the startup sequence through the
13035: deferred word @code{'cold}. @code{'cold} is invoked just before the
13036: image-specific command line processing (by default, loading files and
13037: evaluating (@code{-e}) strings) starts.
13038:
13039: A sequence for adding your initialization usually looks like this:
13040:
13041: @example
13042: :noname
13043: Defers 'cold \ do other initialization stuff (e.g., rehashing wordlists)
13044: ... \ your stuff
13045: ; IS 'cold
13046: @end example
13047:
13048: @cindex turnkey image files
13049: @cindex image file, turnkey applications
13050: You can make a turnkey image by letting @code{'cold} execute a word
13051: (your turnkey application) that never returns; instead, it exits Gforth
13052: via @code{bye} or @code{throw}.
13053:
13054: @cindex command-line arguments, access
13055: @cindex arguments on the command line, access
13056: You can access the (image-specific) command-line arguments through the
13057: variables @code{argc} and @code{argv}. @code{arg} provides convenient
13058: access to @code{argv}.
13059:
13060: If @code{'cold} exits normally, Gforth processes the command-line
13061: arguments as files to be loaded and strings to be evaluated. Therefore,
13062: @code{'cold} should remove the arguments it has used in this case.
13063:
13064:
13065:
13066: doc-'cold
13067: doc-argc
13068: doc-argv
13069: doc-arg
13070:
13071:
13072:
13073: @c ******************************************************************
13074: @node Engine, Binding to System Library, Image Files, Top
13075: @chapter Engine
13076: @cindex engine
13077: @cindex virtual machine
13078:
13079: Reading this chapter is not necessary for programming with Gforth. It
13080: may be helpful for finding your way in the Gforth sources.
13081:
13082: The ideas in this section have also been published in the papers
13083: @cite{ANS fig/GNU/??? Forth} (in German) by Bernd Paysan, presented at
13084: the Forth-Tagung '93 and @cite{A Portable Forth Engine} by M. Anton
13085: Ertl, presented at EuroForth '93; the latter is available at
13086: @*@uref{http://www.complang.tuwien.ac.at/papers/ertl93.ps.Z}.
13087:
13088: @menu
13089: * Portability::
13090: * Threading::
13091: * Primitives::
13092: * Performance::
13093: @end menu
13094:
13095: @node Portability, Threading, Engine, Engine
13096: @section Portability
13097: @cindex engine portability
13098:
13099: An important goal of the Gforth Project is availability across a wide
13100: range of personal machines. fig-Forth, and, to a lesser extent, F83,
13101: achieved this goal by manually coding the engine in assembly language
13102: for several then-popular processors. This approach is very
13103: labor-intensive and the results are short-lived due to progress in
13104: computer architecture.
13105:
13106: @cindex C, using C for the engine
13107: Others have avoided this problem by coding in C, e.g., Mitch Bradley
13108: (cforth), Mikael Patel (TILE) and Dirk Zoller (pfe). This approach is
13109: particularly popular for UNIX-based Forths due to the large variety of
13110: architectures of UNIX machines. Unfortunately an implementation in C
13111: does not mix well with the goals of efficiency and with using
13112: traditional techniques: Indirect or direct threading cannot be expressed
13113: in C, and switch threading, the fastest technique available in C, is
13114: significantly slower. Another problem with C is that it is very
13115: cumbersome to express double integer arithmetic.
13116:
13117: @cindex GNU C for the engine
13118: @cindex long long
13119: Fortunately, there is a portable language that does not have these
13120: limitations: GNU C, the version of C processed by the GNU C compiler
13121: (@pxref{C Extensions, , Extensions to the C Language Family, gcc.info,
13122: GNU C Manual}). Its labels as values feature (@pxref{Labels as Values, ,
13123: Labels as Values, gcc.info, GNU C Manual}) makes direct and indirect
13124: threading possible, its @code{long long} type (@pxref{Long Long, ,
13125: Double-Word Integers, gcc.info, GNU C Manual}) corresponds to Forth's
13126: double numbers@footnote{Unfortunately, long longs are not implemented
13127: properly on all machines (e.g., on alpha-osf1, long longs are only 64
13128: bits, the same size as longs (and pointers), but they should be twice as
13129: long according to @pxref{Long Long, , Double-Word Integers, gcc.info, GNU
13130: C Manual}). So, we had to implement doubles in C after all. Still, on
13131: most machines we can use long longs and achieve better performance than
13132: with the emulation package.}. GNU C is available for free on all
13133: important (and many unimportant) UNIX machines, VMS, 80386s running
13134: MS-DOS, the Amiga, and the Atari ST, so a Forth written in GNU C can run
13135: on all these machines.
13136:
13137: Writing in a portable language has the reputation of producing code that
13138: is slower than assembly. For our Forth engine we repeatedly looked at
13139: the code produced by the compiler and eliminated most compiler-induced
13140: inefficiencies by appropriate changes in the source code.
13141:
13142: @cindex explicit register declarations
13143: @cindex --enable-force-reg, configuration flag
13144: @cindex -DFORCE_REG
13145: However, register allocation cannot be portably influenced by the
13146: programmer, leading to some inefficiencies on register-starved
13147: machines. We use explicit register declarations (@pxref{Explicit Reg
13148: Vars, , Variables in Specified Registers, gcc.info, GNU C Manual}) to
13149: improve the speed on some machines. They are turned on by using the
13150: configuration flag @code{--enable-force-reg} (@code{gcc} switch
13151: @code{-DFORCE_REG}). Unfortunately, this feature not only depends on the
13152: machine, but also on the compiler version: On some machines some
13153: compiler versions produce incorrect code when certain explicit register
13154: declarations are used. So by default @code{-DFORCE_REG} is not used.
13155:
13156: @node Threading, Primitives, Portability, Engine
13157: @section Threading
13158: @cindex inner interpreter implementation
13159: @cindex threaded code implementation
13160:
13161: @cindex labels as values
13162: GNU C's labels as values extension (available since @code{gcc-2.0},
13163: @pxref{Labels as Values, , Labels as Values, gcc.info, GNU C Manual})
13164: makes it possible to take the address of @i{label} by writing
13165: @code{&&@i{label}}. This address can then be used in a statement like
13166: @code{goto *@i{address}}. I.e., @code{goto *&&x} is the same as
13167: @code{goto x}.
13168:
13169: @cindex @code{NEXT}, indirect threaded
13170: @cindex indirect threaded inner interpreter
13171: @cindex inner interpreter, indirect threaded
13172: With this feature an indirect threaded @code{NEXT} looks like:
13173: @example
13174: cfa = *ip++;
13175: ca = *cfa;
13176: goto *ca;
13177: @end example
13178: @cindex instruction pointer
13179: For those unfamiliar with the names: @code{ip} is the Forth instruction
13180: pointer; the @code{cfa} (code-field address) corresponds to ANS Forths
13181: execution token and points to the code field of the next word to be
13182: executed; The @code{ca} (code address) fetched from there points to some
13183: executable code, e.g., a primitive or the colon definition handler
13184: @code{docol}.
13185:
13186: @cindex @code{NEXT}, direct threaded
13187: @cindex direct threaded inner interpreter
13188: @cindex inner interpreter, direct threaded
13189: Direct threading is even simpler:
13190: @example
13191: ca = *ip++;
13192: goto *ca;
13193: @end example
13194:
13195: Of course we have packaged the whole thing neatly in macros called
13196: @code{NEXT} and @code{NEXT1} (the part of @code{NEXT} after fetching the cfa).
13197:
13198: @menu
13199: * Scheduling::
13200: * Direct or Indirect Threaded?::
13201: * DOES>::
13202: @end menu
13203:
13204: @node Scheduling, Direct or Indirect Threaded?, Threading, Threading
13205: @subsection Scheduling
13206: @cindex inner interpreter optimization
13207:
13208: There is a little complication: Pipelined and superscalar processors,
13209: i.e., RISC and some modern CISC machines can process independent
13210: instructions while waiting for the results of an instruction. The
13211: compiler usually reorders (schedules) the instructions in a way that
13212: achieves good usage of these delay slots. However, on our first tries
13213: the compiler did not do well on scheduling primitives. E.g., for
13214: @code{+} implemented as
13215: @example
13216: n=sp[0]+sp[1];
13217: sp++;
13218: sp[0]=n;
13219: NEXT;
13220: @end example
13221: the @code{NEXT} comes strictly after the other code, i.e., there is nearly no
13222: scheduling. After a little thought the problem becomes clear: The
13223: compiler cannot know that @code{sp} and @code{ip} point to different
13224: addresses (and the version of @code{gcc} we used would not know it even
13225: if it was possible), so it could not move the load of the cfa above the
13226: store to the TOS. Indeed the pointers could be the same, if code on or
13227: very near the top of stack were executed. In the interest of speed we
13228: chose to forbid this probably unused ``feature'' and helped the compiler
13229: in scheduling: @code{NEXT} is divided into the loading part (@code{NEXT_P1})
13230: and the goto part (@code{NEXT_P2}). @code{+} now looks like:
13231: @example
13232: n=sp[0]+sp[1];
13233: sp++;
13234: NEXT_P1;
13235: sp[0]=n;
13236: NEXT_P2;
13237: @end example
13238: This can be scheduled optimally by the compiler.
13239:
13240: This division can be turned off with the switch @code{-DCISC_NEXT}. This
13241: switch is on by default on machines that do not profit from scheduling
13242: (e.g., the 80386), in order to preserve registers.
13243:
13244: @node Direct or Indirect Threaded?, DOES>, Scheduling, Threading
13245: @subsection Direct or Indirect Threaded?
13246: @cindex threading, direct or indirect?
13247:
13248: @cindex -DDIRECT_THREADED
13249: Both! After packaging the nasty details in macro definitions we
13250: realized that we could switch between direct and indirect threading by
13251: simply setting a compilation flag (@code{-DDIRECT_THREADED}) and
13252: defining a few machine-specific macros for the direct-threading case.
13253: On the Forth level we also offer access words that hide the
13254: differences between the threading methods (@pxref{Threading Words}).
13255:
13256: Indirect threading is implemented completely machine-independently.
13257: Direct threading needs routines for creating jumps to the executable
13258: code (e.g. to @code{docol} or @code{dodoes}). These routines are inherently
13259: machine-dependent, but they do not amount to many source lines. Therefore,
13260: even porting direct threading to a new machine requires little effort.
13261:
13262: @cindex --enable-indirect-threaded, configuration flag
13263: @cindex --enable-direct-threaded, configuration flag
13264: The default threading method is machine-dependent. You can enforce a
13265: specific threading method when building Gforth with the configuration
13266: flag @code{--enable-direct-threaded} or
13267: @code{--enable-indirect-threaded}. Note that direct threading is not
13268: supported on all machines.
13269:
13270: @node DOES>, , Direct or Indirect Threaded?, Threading
13271: @subsection DOES>
13272: @cindex @code{DOES>} implementation
13273:
13274: @cindex @code{dodoes} routine
13275: @cindex @code{DOES>}-code
13276: One of the most complex parts of a Forth engine is @code{dodoes}, i.e.,
13277: the chunk of code executed by every word defined by a
13278: @code{CREATE}...@code{DOES>} pair. The main problem here is: How to find
13279: the Forth code to be executed, i.e. the code after the
13280: @code{DOES>} (the @code{DOES>}-code)? There are two solutions:
13281:
13282: In fig-Forth the code field points directly to the @code{dodoes} and the
13283: @code{DOES>}-code address is stored in the cell after the code address (i.e. at
13284: @code{@i{CFA} cell+}). It may seem that this solution is illegal in
13285: the Forth-79 and all later standards, because in fig-Forth this address
13286: lies in the body (which is illegal in these standards). However, by
13287: making the code field larger for all words this solution becomes legal
13288: again. We use this approach for the indirect threaded version and for
13289: direct threading on some machines. Leaving a cell unused in most words
13290: is a bit wasteful, but on the machines we are targeting this is hardly a
13291: problem. The other reason for having a code field size of two cells is
13292: to avoid having different image files for direct and indirect threaded
13293: systems (direct threaded systems require two-cell code fields on many
13294: machines).
13295:
13296: @cindex @code{DOES>}-handler
13297: The other approach is that the code field points or jumps to the cell
13298: after @code{DOES>}. In this variant there is a jump to @code{dodoes} at
13299: this address (the @code{DOES>}-handler). @code{dodoes} can then get the
13300: @code{DOES>}-code address by computing the code address, i.e., the address of
13301: the jump to @code{dodoes}, and add the length of that jump field. A variant of
13302: this is to have a call to @code{dodoes} after the @code{DOES>}; then the
13303: return address (which can be found in the return register on RISCs) is
13304: the @code{DOES>}-code address. Since the two cells available in the code field
13305: are used up by the jump to the code address in direct threading on many
13306: architectures, we use this approach for direct threading on these
13307: architectures. We did not want to add another cell to the code field.
13308:
13309: @node Primitives, Performance, Threading, Engine
13310: @section Primitives
13311: @cindex primitives, implementation
13312: @cindex virtual machine instructions, implementation
13313:
13314: @menu
13315: * Automatic Generation::
13316: * TOS Optimization::
13317: * Produced code::
13318: @end menu
13319:
13320: @node Automatic Generation, TOS Optimization, Primitives, Primitives
13321: @subsection Automatic Generation
13322: @cindex primitives, automatic generation
13323:
13324: @cindex @file{prims2x.fs}
13325: Since the primitives are implemented in a portable language, there is no
13326: longer any need to minimize the number of primitives. On the contrary,
13327: having many primitives has an advantage: speed. In order to reduce the
13328: number of errors in primitives and to make programming them easier, we
13329: provide a tool, the primitive generator (@file{prims2x.fs}), that
13330: automatically generates most (and sometimes all) of the C code for a
13331: primitive from the stack effect notation. The source for a primitive
13332: has the following form:
13333:
13334: @cindex primitive source format
13335: @format
13336: @i{Forth-name} @i{stack-effect} @i{category} [@i{pronounc.}]
13337: [@code{""}@i{glossary entry}@code{""}]
13338: @i{C code}
13339: [@code{:}
13340: @i{Forth code}]
13341: @end format
13342:
13343: The items in brackets are optional. The category and glossary fields
13344: are there for generating the documentation, the Forth code is there
13345: for manual implementations on machines without GNU C. E.g., the source
13346: for the primitive @code{+} is:
13347: @example
13348: + n1 n2 -- n core plus
13349: n = n1+n2;
13350: @end example
13351:
13352: This looks like a specification, but in fact @code{n = n1+n2} is C
13353: code. Our primitive generation tool extracts a lot of information from
13354: the stack effect notations@footnote{We use a one-stack notation, even
13355: though we have separate data and floating-point stacks; The separate
13356: notation can be generated easily from the unified notation.}: The number
13357: of items popped from and pushed on the stack, their type, and by what
13358: name they are referred to in the C code. It then generates a C code
13359: prelude and postlude for each primitive. The final C code for @code{+}
13360: looks like this:
13361:
13362: @example
13363: I_plus: /* + ( n1 n2 -- n ) */ /* label, stack effect */
13364: /* */ /* documentation */
13365: @{
13366: DEF_CA /* definition of variable ca (indirect threading) */
13367: Cell n1; /* definitions of variables */
13368: Cell n2;
13369: Cell n;
13370: n1 = (Cell) sp[1]; /* input */
13371: n2 = (Cell) TOS;
13372: sp += 1; /* stack adjustment */
13373: NAME("+") /* debugging output (with -DDEBUG) */
13374: @{
13375: n = n1+n2; /* C code taken from the source */
13376: @}
13377: NEXT_P1; /* NEXT part 1 */
13378: TOS = (Cell)n; /* output */
13379: NEXT_P2; /* NEXT part 2 */
13380: @}
13381: @end example
13382:
13383: This looks long and inefficient, but the GNU C compiler optimizes quite
13384: well and produces optimal code for @code{+} on, e.g., the R3000 and the
13385: HP RISC machines: Defining the @code{n}s does not produce any code, and
13386: using them as intermediate storage also adds no cost.
13387:
13388: There are also other optimizations that are not illustrated by this
13389: example: assignments between simple variables are usually for free (copy
13390: propagation). If one of the stack items is not used by the primitive
13391: (e.g. in @code{drop}), the compiler eliminates the load from the stack
13392: (dead code elimination). On the other hand, there are some things that
13393: the compiler does not do, therefore they are performed by
13394: @file{prims2x.fs}: The compiler does not optimize code away that stores
13395: a stack item to the place where it just came from (e.g., @code{over}).
13396:
13397: While programming a primitive is usually easy, there are a few cases
13398: where the programmer has to take the actions of the generator into
13399: account, most notably @code{?dup}, but also words that do not (always)
13400: fall through to @code{NEXT}.
13401:
13402: @node TOS Optimization, Produced code, Automatic Generation, Primitives
13403: @subsection TOS Optimization
13404: @cindex TOS optimization for primitives
13405: @cindex primitives, keeping the TOS in a register
13406:
13407: An important optimization for stack machine emulators, e.g., Forth
13408: engines, is keeping one or more of the top stack items in
13409: registers. If a word has the stack effect @i{in1}...@i{inx} @code{--}
13410: @i{out1}...@i{outy}, keeping the top @i{n} items in registers
13411: @itemize @bullet
13412: @item
13413: is better than keeping @i{n-1} items, if @i{x>=n} and @i{y>=n},
13414: due to fewer loads from and stores to the stack.
13415: @item is slower than keeping @i{n-1} items, if @i{x<>y} and @i{x<n} and
13416: @i{y<n}, due to additional moves between registers.
13417: @end itemize
13418:
13419: @cindex -DUSE_TOS
13420: @cindex -DUSE_NO_TOS
13421: In particular, keeping one item in a register is never a disadvantage,
13422: if there are enough registers. Keeping two items in registers is a
13423: disadvantage for frequent words like @code{?branch}, constants,
13424: variables, literals and @code{i}. Therefore our generator only produces
13425: code that keeps zero or one items in registers. The generated C code
13426: covers both cases; the selection between these alternatives is made at
13427: C-compile time using the switch @code{-DUSE_TOS}. @code{TOS} in the C
13428: code for @code{+} is just a simple variable name in the one-item case,
13429: otherwise it is a macro that expands into @code{sp[0]}. Note that the
13430: GNU C compiler tries to keep simple variables like @code{TOS} in
13431: registers, and it usually succeeds, if there are enough registers.
13432:
13433: @cindex -DUSE_FTOS
13434: @cindex -DUSE_NO_FTOS
13435: The primitive generator performs the TOS optimization for the
13436: floating-point stack, too (@code{-DUSE_FTOS}). For floating-point
13437: operations the benefit of this optimization is even larger:
13438: floating-point operations take quite long on most processors, but can be
13439: performed in parallel with other operations as long as their results are
13440: not used. If the FP-TOS is kept in a register, this works. If
13441: it is kept on the stack, i.e., in memory, the store into memory has to
13442: wait for the result of the floating-point operation, lengthening the
13443: execution time of the primitive considerably.
13444:
13445: The TOS optimization makes the automatic generation of primitives a
13446: bit more complicated. Just replacing all occurrences of @code{sp[0]} by
13447: @code{TOS} is not sufficient. There are some special cases to
13448: consider:
13449: @itemize @bullet
13450: @item In the case of @code{dup ( w -- w w )} the generator must not
13451: eliminate the store to the original location of the item on the stack,
13452: if the TOS optimization is turned on.
13453: @item Primitives with stack effects of the form @code{--}
13454: @i{out1}...@i{outy} must store the TOS to the stack at the start.
13455: Likewise, primitives with the stack effect @i{in1}...@i{inx} @code{--}
13456: must load the TOS from the stack at the end. But for the null stack
13457: effect @code{--} no stores or loads should be generated.
13458: @end itemize
13459:
13460: @node Produced code, , TOS Optimization, Primitives
13461: @subsection Produced code
13462: @cindex primitives, assembly code listing
13463:
13464: @cindex @file{engine.s}
13465: To see what assembly code is produced for the primitives on your machine
13466: with your compiler and your flag settings, type @code{make engine.s} and
13467: look at the resulting file @file{engine.s}.
13468:
13469: @node Performance, , Primitives, Engine
13470: @section Performance
13471: @cindex performance of some Forth interpreters
13472: @cindex engine performance
13473: @cindex benchmarking Forth systems
13474: @cindex Gforth performance
13475:
13476: On RISCs the Gforth engine is very close to optimal; i.e., it is usually
13477: impossible to write a significantly faster engine.
13478:
13479: On register-starved machines like the 386 architecture processors
13480: improvements are possible, because @code{gcc} does not utilize the
13481: registers as well as a human, even with explicit register declarations;
13482: e.g., Bernd Beuster wrote a Forth system fragment in assembly language
13483: and hand-tuned it for the 486; this system is 1.19 times faster on the
13484: Sieve benchmark on a 486DX2/66 than Gforth compiled with
13485: @code{gcc-2.6.3} with @code{-DFORCE_REG}. The situation has improved
13486: with gcc-2.95 and gforth-0.4.9; now the most important virtual machine
13487: registers fit in real registers (and we can even afford to use the TOS
13488: optimization), resulting in a speedup of 1.14 on the sieve over the
13489: earlier results.
13490:
13491: @cindex Win32Forth performance
13492: @cindex NT Forth performance
13493: @cindex eforth performance
13494: @cindex ThisForth performance
13495: @cindex PFE performance
13496: @cindex TILE performance
13497: The potential advantage of assembly language implementations
13498: is not necessarily realized in complete Forth systems: We compared
13499: Gforth-0.4.9 (direct threaded, compiled with @code{gcc-2.95.1} and
13500: @code{-DFORCE_REG}) with Win32Forth 1.2093, LMI's NT Forth (Beta, May
13501: 1994) and Eforth (with and without peephole (aka pinhole) optimization
13502: of the threaded code); all these systems were written in assembly
13503: language. We also compared Gforth with three systems written in C:
13504: PFE-0.9.14 (compiled with @code{gcc-2.6.3} with the default
13505: configuration for Linux: @code{-O2 -fomit-frame-pointer -DUSE_REGS
13506: -DUNROLL_NEXT}), ThisForth Beta (compiled with @code{gcc-2.6.3 -O3
13507: -fomit-frame-pointer}; ThisForth employs peephole optimization of the
13508: threaded code) and TILE (compiled with @code{make opt}). We benchmarked
13509: Gforth, PFE, ThisForth and TILE on a 486DX2/66 under Linux. Kenneth
13510: O'Heskin kindly provided the results for Win32Forth and NT Forth on a
13511: 486DX2/66 with similar memory performance under Windows NT. Marcel
13512: Hendrix ported Eforth to Linux, then extended it to run the benchmarks,
13513: added the peephole optimizer, ran the benchmarks and reported the
13514: results.
13515:
13516: We used four small benchmarks: the ubiquitous Sieve; bubble-sorting and
13517: matrix multiplication come from the Stanford integer benchmarks and have
13518: been translated into Forth by Martin Fraeman; we used the versions
13519: included in the TILE Forth package, but with bigger data set sizes; and
13520: a recursive Fibonacci number computation for benchmarking calling
13521: performance. The following table shows the time taken for the benchmarks
13522: scaled by the time taken by Gforth (in other words, it shows the speedup
13523: factor that Gforth achieved over the other systems).
13524:
13525: @example
13526: relative Win32- NT eforth This-
13527: time Gforth Forth Forth eforth +opt PFE Forth TILE
13528: sieve 1.00 1.58 1.30 1.58 0.97 1.80 3.63 9.79
13529: bubble 1.00 1.55 1.67 1.75 1.04 1.78 4.59
13530: matmul 1.00 1.67 1.53 1.66 0.84 1.79 4.63
13531: fib 1.00 1.75 1.53 1.40 0.99 1.99 3.43 4.93
13532: @end example
13533:
13534: You may be quite surprised by the good performance of Gforth when
13535: compared with systems written in assembly language. One important reason
13536: for the disappointing performance of these other systems is probably
13537: that they are not written optimally for the 486 (e.g., they use the
13538: @code{lods} instruction). In addition, Win32Forth uses a comfortable,
13539: but costly method for relocating the Forth image: like @code{cforth}, it
13540: computes the actual addresses at run time, resulting in two address
13541: computations per @code{NEXT} (@pxref{Image File Background}).
13542:
13543: Only Eforth with the peephole optimizer performs comparable to
13544: Gforth. The speedups achieved with peephole optimization of threaded
13545: code are quite remarkable. Adding a peephole optimizer to Gforth should
13546: cause similar speedups.
13547:
13548: The speedup of Gforth over PFE, ThisForth and TILE can be easily
13549: explained with the self-imposed restriction of the latter systems to
13550: standard C, which makes efficient threading impossible (however, the
13551: measured implementation of PFE uses a GNU C extension: @pxref{Global Reg
13552: Vars, , Defining Global Register Variables, gcc.info, GNU C Manual}).
13553: Moreover, current C compilers have a hard time optimizing other aspects
13554: of the ThisForth and the TILE source.
13555:
13556: The performance of Gforth on 386 architecture processors varies widely
13557: with the version of @code{gcc} used. E.g., @code{gcc-2.5.8} failed to
13558: allocate any of the virtual machine registers into real machine
13559: registers by itself and would not work correctly with explicit register
13560: declarations, giving a 1.5 times slower engine (on a 486DX2/66 running
13561: the Sieve) than the one measured above.
13562:
13563: Note that there have been several releases of Win32Forth since the
13564: release presented here, so the results presented above may have little
13565: predictive value for the performance of Win32Forth today (results for
13566: the current release on an i486DX2/66 are welcome).
13567:
13568: @cindex @file{Benchres}
13569: In @cite{Translating Forth to Efficient C} by M. Anton Ertl and Martin
13570: Maierhofer (presented at EuroForth '95), an indirect threaded version of
13571: Gforth is compared with Win32Forth, NT Forth, PFE, and ThisForth; that
13572: version of Gforth is slower on a 486 than the direct threaded version
13573: used here. The paper available at
13574: @*@uref{http://www.complang.tuwien.ac.at/papers/ertl&maierhofer95.ps.gz};
13575: it also contains numbers for some native code systems. You can find a
13576: newer version of these measurements at
13577: @uref{http://www.complang.tuwien.ac.at/forth/performance.html}. You can
13578: find numbers for Gforth on various machines in @file{Benchres}.
13579:
13580: @c ******************************************************************
13581: @node Binding to System Library, Cross Compiler, Engine, Top
13582: @chapter Binding to System Library
13583:
13584: @node Cross Compiler, Bugs, Binding to System Library, Top
13585: @chapter Cross Compiler
13586: @cindex @file{cross.fs}
13587: @cindex cross-compiler
13588: @cindex metacompiler
13589: @cindex target compiler
13590:
13591: The cross compiler is used to bootstrap a Forth kernel. Since Gforth is
13592: mostly written in Forth, including crucial parts like the outer
13593: interpreter and compiler, it needs compiled Forth code to get
13594: started. The cross compiler allows to create new images for other
13595: architectures, even running under another Forth system.
13596:
13597: @menu
13598: * Using the Cross Compiler::
13599: * How the Cross Compiler Works::
13600: @end menu
13601:
13602: @node Using the Cross Compiler, How the Cross Compiler Works, Cross Compiler, Cross Compiler
13603: @section Using the Cross Compiler
13604:
13605: The cross compiler uses a language that resembles Forth, but isn't. The
13606: main difference is that you can execute Forth code after definition,
13607: while you usually can't execute the code compiled by cross, because the
13608: code you are compiling is typically for a different computer than the
13609: one you are compiling on.
13610:
13611: The Makefile is already set up to allow you to create kernels for new
13612: architectures with a simple make command. The generic kernels using the
13613: GCC compiled virtual machine are created in the normal build process
13614: with @code{make}. To create a embedded Gforth executable for e.g. the
13615: 8086 processor (running on a DOS machine), type
13616:
13617: @example
13618: make kernl-8086.fi
13619: @end example
13620:
13621: This will use the machine description from the @file{arch/8086}
13622: directory to create a new kernel. A machine file may look like that:
13623:
13624: @example
13625: \ Parameter for target systems 06oct92py
13626:
13627: 4 Constant cell \ cell size in bytes
13628: 2 Constant cell<< \ cell shift to bytes
13629: 5 Constant cell>bit \ cell shift to bits
13630: 8 Constant bits/char \ bits per character
13631: 8 Constant bits/byte \ bits per byte [default: 8]
13632: 8 Constant float \ bytes per float
13633: 8 Constant /maxalign \ maximum alignment in bytes
13634: false Constant bigendian \ byte order
13635: ( true=big, false=little )
13636:
13637: include machpc.fs \ feature list
13638: @end example
13639:
13640: This part is obligatory for the cross compiler itself, the feature list
13641: is used by the kernel to conditionally compile some features in and out,
13642: depending on whether the target supports these features.
13643:
13644: There are some optional features, if you define your own primitives,
13645: have an assembler, or need special, nonstandard preparation to make the
13646: boot process work. @code{asm-include} include an assembler,
13647: @code{prims-include} includes primitives, and @code{>boot} prepares for
13648: booting.
13649:
13650: @example
13651: : asm-include ." Include assembler" cr
13652: s" arch/8086/asm.fs" included ;
13653:
13654: : prims-include ." Include primitives" cr
13655: s" arch/8086/prim.fs" included ;
13656:
13657: : >boot ." Prepare booting" cr
13658: s" ' boot >body into-forth 1+ !" evaluate ;
13659: @end example
13660:
13661: These words are used as sort of macro during the cross compilation in
13662: the file @file{kernel/main.fs}. Instead of using this macros, it would
13663: be possible --- but more complicated --- to write a new kernel project
13664: file, too.
13665:
13666: @file{kernel/main.fs} expects the machine description file name on the
13667: stack; the cross compiler itself (@file{cross.fs}) assumes that either
13668: @code{mach-file} leaves a counted string on the stack, or
13669: @code{machine-file} leaves an address, count pair of the filename on the
13670: stack.
13671:
13672: The feature list is typically controlled using @code{SetValue}, generic
13673: files that are used by several projects can use @code{DefaultValue}
13674: instead. Both functions work like @code{Value}, when the value isn't
13675: defined, but @code{SetValue} works like @code{to} if the value is
13676: defined, and @code{DefaultValue} doesn't set anything, if the value is
13677: defined.
13678:
13679: @example
13680: \ generic mach file for pc gforth 03sep97jaw
13681:
13682: true DefaultValue NIL \ relocating
13683:
13684: >ENVIRON
13685:
13686: true DefaultValue file \ controls the presence of the
13687: \ file access wordset
13688: true DefaultValue OS \ flag to indicate a operating system
13689:
13690: true DefaultValue prims \ true: primitives are c-code
13691:
13692: true DefaultValue floating \ floating point wordset is present
13693:
13694: true DefaultValue glocals \ gforth locals are present
13695: \ will be loaded
13696: true DefaultValue dcomps \ double number comparisons
13697:
13698: true DefaultValue hash \ hashing primitives are loaded/present
13699:
13700: true DefaultValue xconds \ used together with glocals,
13701: \ special conditionals supporting gforths'
13702: \ local variables
13703: true DefaultValue header \ save a header information
13704:
13705: true DefaultValue backtrace \ enables backtrace code
13706:
13707: false DefaultValue ec
13708: false DefaultValue crlf
13709:
13710: cell 2 = [IF] &32 [ELSE] &256 [THEN] KB DefaultValue kernel-size
13711:
13712: &16 KB DefaultValue stack-size
13713: &15 KB &512 + DefaultValue fstack-size
13714: &15 KB DefaultValue rstack-size
13715: &14 KB &512 + DefaultValue lstack-size
13716: @end example
13717:
13718: @node How the Cross Compiler Works, , Using the Cross Compiler, Cross Compiler
13719: @section How the Cross Compiler Works
13720:
13721: @node Bugs, Origin, Cross Compiler, Top
13722: @appendix Bugs
13723: @cindex bug reporting
13724:
13725: Known bugs are described in the file @file{BUGS} in the Gforth distribution.
13726:
13727: If you find a bug, please send a bug report to
13728: @email{bug-gforth@@gnu.org}. A bug report should include this
13729: information:
13730:
13731: @itemize @bullet
13732: @item
13733: The Gforth version used (it is announced at the start of an
13734: interactive Gforth session).
13735: @item
13736: The machine and operating system (on Unix
13737: systems @code{uname -a} will report this information).
13738: @item
13739: The installation options (send the file @file{config.status}).
13740: @item
13741: A complete list of changes (if any) you (or your installer) have made to the
13742: Gforth sources.
13743: @item
13744: A program (or a sequence of keyboard commands) that reproduces the bug.
13745: @item
13746: A description of what you think constitutes the buggy behaviour.
13747: @end itemize
13748:
13749: For a thorough guide on reporting bugs read @ref{Bug Reporting, , How
13750: to Report Bugs, gcc.info, GNU C Manual}.
13751:
13752:
13753: @node Origin, Forth-related information, Bugs, Top
13754: @appendix Authors and Ancestors of Gforth
13755:
13756: @section Authors and Contributors
13757: @cindex authors of Gforth
13758: @cindex contributors to Gforth
13759:
13760: The Gforth project was started in mid-1992 by Bernd Paysan and Anton
13761: Ertl. The third major author was Jens Wilke. Lennart Benschop (who was
13762: one of Gforth's first users, in mid-1993) and Stuart Ramsden inspired us
13763: with their continuous feedback. Lennart Benshop contributed
13764: @file{glosgen.fs}, while Stuart Ramsden has been working on automatic
13765: support for calling C libraries. Helpful comments also came from Paul
13766: Kleinrubatscher, Christian Pirker, Dirk Zoller, Marcel Hendrix, John
13767: Wavrik, Barrie Stott, Marc de Groot, and Jorge Acerada. Since the
13768: release of Gforth-0.2.1 there were also helpful comments from many
13769: others; thank you all, sorry for not listing you here (but digging
13770: through my mailbox to extract your names is on my to-do list). Since the
13771: release of Gforth-0.4.0 Neal Crook worked on the manual.
13772:
13773: Gforth also owes a lot to the authors of the tools we used (GCC, CVS,
13774: and autoconf, among others), and to the creators of the Internet: Gforth
13775: was developed across the Internet, and its authors did not meet
13776: physically for the first 4 years of development.
13777:
13778: @section Pedigree
13779: @cindex pedigree of Gforth
13780:
13781: Gforth descends from bigFORTH (1993) and fig-Forth. Gforth and PFE (by
13782: Dirk Zoller) will cross-fertilize each other. Of course, a significant
13783: part of the design of Gforth was prescribed by ANS Forth.
13784:
13785: Bernd Paysan wrote bigFORTH, a descendent from TurboForth, an unreleased
13786: 32 bit native code version of VolksForth for the Atari ST, written
13787: mostly by Dietrich Weineck.
13788:
13789: VolksForth descends from F83. It was written by Klaus Schleisiek, Bernd
13790: Pennemann, Georg Rehfeld and Dietrich Weineck for the C64 (called
13791: UltraForth there) in the mid-80s and ported to the Atari ST in 1986.
13792:
13793: Henry Laxen and Mike Perry wrote F83 as a model implementation of the
13794: Forth-83 standard. !! Pedigree? When?
13795:
13796: A team led by Bill Ragsdale implemented fig-Forth on many processors in
13797: 1979. Robert Selzer and Bill Ragsdale developed the original
13798: implementation of fig-Forth for the 6502 based on microForth.
13799:
13800: The principal architect of microForth was Dean Sanderson. microForth was
13801: FORTH, Inc.'s first off-the-shelf product. It was developed in 1976 for
13802: the 1802, and subsequently implemented on the 8080, the 6800 and the
13803: Z80.
13804:
13805: All earlier Forth systems were custom-made, usually by Charles Moore,
13806: who discovered (as he puts it) Forth during the late 60s. The first full
13807: Forth existed in 1971.
13808:
13809: A part of the information in this section comes from @cite{The Evolution
13810: of Forth} by Elizabeth D. Rather, Donald R. Colburn and Charles
13811: H. Moore, presented at the HOPL-II conference and preprinted in SIGPLAN
13812: Notices 28(3), 1993. You can find more historical and genealogical
13813: information about Forth there.
13814:
13815: @node Forth-related information, Word Index, Origin, Top
13816: @appendix Other Forth-related information
13817: @cindex Forth-related information
13818:
13819: @menu
13820: * Internet resources::
13821: * Books::
13822: * The Forth Interest Group::
13823: * Conferences::
13824: @end menu
13825:
13826:
13827: @node Internet resources, Books, Forth-related information, Forth-related information
13828: @section Internet resources
13829: @cindex internet resources
13830:
13831: @cindex comp.lang.forth
13832: @cindex frequently asked questions
13833: There is an active news group (comp.lang.forth) discussing Forth and
13834: Forth-related issues. A frequently-asked-questions (FAQ) list
13835: is posted to the news group regularly, and archived at these sites:
13836:
13837: @itemize @bullet
13838: @item
13839: @uref{ftp://rtfm.mit.edu/pub/usenet-by-group/comp.lang.forth/}
13840: @item
13841: @uref{ftp://ftp.forth.org/pub/Forth/FAQ/}
13842: @end itemize
13843:
13844: The FAQ list should be considered mandatory reading before posting to
13845: the news group.
13846:
13847: Here are some other web sites holding Forth-related material:
13848:
13849: @itemize @bullet
13850: @item
13851: @uref{http://www.taygeta.com/forth.html} -- Skip Carter's Forth pages.
13852: @item
13853: @uref{http://www.jwdt.com/~paysan/gforth.html} -- the Gforth home page.
13854: @item
13855: @uref{http://www.minerva.com/uathena.htm} -- home of ANS Forth Standard.
13856: @item
13857: @uref{http://dec.bournemouth.ac.uk/forth/index.html} -- the Forth
13858: Research page, including links to the Journal of Forth Application and
13859: Research (JFAR) and a searchable Forth bibliography.
13860: @end itemize
13861:
13862:
13863: @node Books, The Forth Interest Group, Internet resources, Forth-related information
13864: @section Books
13865: @cindex books on Forth
13866:
13867: As the Standard is relatively new, there are not many books out yet. It
13868: is not recommended to learn Forth by using Gforth and a book that is not
13869: written for ANS Forth, as you will not know your mistakes from the
13870: deviations of the book. However, books based on the Forth-83 standard
13871: should be ok, because ANS Forth is primarily an extension of Forth-83.
13872: Refer to the Forth FAQ for details of Forth-related books.
13873:
13874: @cindex standard document for ANS Forth
13875: @cindex ANS Forth document
13876: The definite reference if you want to write ANS Forth programs is, of
13877: course, the ANS Forth document. It is available in printed form from the
13878: National Standards Institute Sales Department (Tel.: USA (212) 642-4900;
13879: Fax.: USA (212) 302-1286) as document @cite{X3.215-1994} for about
13880: $200. You can also get it from Global Engineering Documents (Tel.: USA
13881: (800) 854-7179; Fax.: (303) 843-9880) for about $300.
13882:
13883: @cite{dpANS6}, the last draft of the standard, which was then submitted
13884: to ANSI for publication is available electronically and for free in some
13885: MS Word format, and it has been converted to HTML
13886: (@uref{http://www.taygeta.com/forth/dpans.html}; this HTML version also
13887: includes the answers to Requests for Interpretation (RFIs). Some
13888: pointers to these versions can be found through
13889: @*@uref{http://www.complang.tuwien.ac.at/projects/forth.html}.
13890:
13891:
13892: @node The Forth Interest Group, Conferences, Books, Forth-related information
13893: @section The Forth Interest Group
13894: @cindex Forth interest group (FIG)
13895:
13896: The Forth Interest Group (FIG) is a world-wide, non-profit,
13897: member-supported organisation. It publishes a regular magazine,
13898: @var{FORTH Dimensions}, and offers other benefits of membership. You can
13899: contact the FIG through their office email address:
13900: @email{office@@forth.org} or by visiting their web site at
13901: @uref{http://www.forth.org/}. This web site also includes links to FIG
13902: chapters in other countries and American cities
13903: (@uref{http://www.forth.org/chapters.html}).
13904:
13905: @node Conferences, , The Forth Interest Group, Forth-related information
13906: @section Conferences
13907: @cindex Conferences
13908:
13909: There are several regular conferences related to Forth. They are all
13910: well-publicised in @var{FORTH Dimensions} and on the comp.lang.forth
13911: news group:
13912:
13913: @itemize @bullet
13914: @item
13915: FORML -- the Forth modification laboratory convenes every year near
13916: Monterey, California.
13917: @item
13918: The Rochester Forth Conference -- an annual conference traditionally
13919: held in Rochester, New York.
13920: @item
13921: EuroForth -- this European conference takes place annually.
13922: @end itemize
13923:
13924:
13925: @node Word Index, Name Index, Forth-related information, Top
13926: @unnumbered Word Index
13927:
13928: This index is a list of Forth words that have ``glossary'' entries
13929: within this manual. Each word is listed with its stack effect and
13930: wordset.
13931:
13932: @printindex fn
13933:
13934: @node Name Index, Concept Index, Word Index, Top
13935: @unnumbered Name Index
13936:
13937: This index is a list of Forth words that have ``glossary'' entries
13938: within this manual.
13939:
13940: @printindex ky
13941:
13942: @node Concept Index, , Name Index, Top
13943: @unnumbered Concept and Word Index
13944:
13945: Not all entries listed in this index are present verbatim in the
13946: text. This index also duplicates, in abbreviated form, all of the words
13947: listed in the Word Index (only the names are listed for the words here).
13948:
13949: @printindex cp
13950:
13951: @contents
13952: @bye
13953:
FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>
![[BACK]](/cvsweb/icons/back.gif){kind=icon}
Return to gforth.ds CVS log ![[TXT]](/cvsweb/icons/text.gif){kind=icon}
![[DIR]](/cvsweb/icons/dir.gif){kind=icon}
Up to [gforth] / gforth / doc