File:
[gforth] /
gforth /
doc /
gforth.ds
Revision
1.70:
download - view:
text,
annotated -
select for diffs
Mon Aug 14 19:15:55 2000 UTC (23 years, 8 months ago) by
pazsan
Branches:
MAIN
CVS tags:
HEAD
Added conditions to the new primitives (floating)
Fixed prims2x.fs to accept comments after the last primitive
Fixed newline Forth definition
Small docs fixes
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: @c POSTPONE, COMPILE, [COMPILE], LITERAL should have their own section
18:
19: @comment %**start of header (This is for running Texinfo on a region.)
20: @setfilename gforth.info
21: @settitle Gforth Manual
22: @dircategory GNU programming tools
23: @direntry
24: * Gforth: (gforth). A fast interpreter for the Forth language.
25: @end direntry
26: @c The Texinfo manual also recommends doing this, but for Gforth it may
27: @c not make much sense
28: @c @dircategory Individual utilities
29: @c @direntry
30: @c * Gforth: (gforth)Invoking Gforth. gforth, gforth-fast, gforthmi
31: @c @end direntry
32:
33: @comment @setchapternewpage odd
34: @comment TODO this gets left in by HTML converter
35: @macro progstyle {}
36: Programming style note:
37: @end macro
38:
39: @macro assignment {}
40: @table @i
41: @item Assignment:
42: @end macro
43: @macro endassignment {}
44: @end table
45: @end macro
46:
47: @comment %**end of header (This is for running Texinfo on a region.)
48:
49:
50: @comment ----------------------------------------------------------
51: @comment macros for beautifying glossary entries
52: @comment if these are used, need to strip them out for HTML converter
53: @comment else they get repeated verbatim in HTML output.
54: @comment .. not working yet.
55:
56: @macro GLOSS-START {}
57: @iftex
58: @ninerm
59: @end iftex
60: @end macro
61:
62: @macro GLOSS-END {}
63: @iftex
64: @rm
65: @end iftex
66: @end macro
67:
68: @comment ----------------------------------------------------------
69:
70:
71: @include version.texi
72:
73: @ifnottex
74: This file documents Gforth @value{VERSION}
75:
76: Copyright @copyright{} 1995--2000 Free Software Foundation, Inc.
77:
78: Permission is granted to make and distribute verbatim copies of
79: this manual provided the copyright notice and this permission notice
80: are preserved on all copies.
81:
82: @ignore
83: Permission is granted to process this file through TeX and print the
84: results, provided the printed document carries a copying permission
85: notice identical to this one except for the removal of this paragraph
86: (this paragraph not being relevant to the printed manual).
87:
88: @end ignore
89: Permission is granted to copy and distribute modified versions of this
90: manual under the conditions for verbatim copying, provided also that the
91: sections entitled "Distribution" and "General Public License" are
92: included exactly as in the original, and provided that the entire
93: resulting derived work is distributed under the terms of a permission
94: notice identical to this one.
95:
96: Permission is granted to copy and distribute translations of this manual
97: into another language, under the above conditions for modified versions,
98: except that the sections entitled "Distribution" and "General Public
99: License" may be included in a translation approved by the author instead
100: of in the original English.
101: @end ifnottex
102:
103: @finalout
104: @titlepage
105: @sp 10
106: @center @titlefont{Gforth Manual}
107: @sp 2
108: @center for version @value{VERSION}
109: @sp 2
110: @center Neal Crook
111: @center Anton Ertl
112: @center Bernd Paysan
113: @center Jens Wilke
114: @sp 3
115: @center This manual is permanently under construction and was last updated on 15-Mar-2000
116:
117: @comment The following two commands start the copyright page.
118: @page
119: @vskip 0pt plus 1filll
120: Copyright @copyright{} 1995--2000 Free Software Foundation, Inc.
121:
122: @comment !! Published by ... or You can get a copy of this manual ...
123:
124: Permission is granted to make and distribute verbatim copies of
125: this manual provided the copyright notice and this permission notice
126: are preserved on all copies.
127:
128: Permission is granted to copy and distribute modified versions of this
129: manual under the conditions for verbatim copying, provided also that the
130: sections entitled "Distribution" and "General Public License" are
131: included exactly as in the original, and provided that the entire
132: resulting derived work is distributed under the terms of a permission
133: notice identical to this one.
134:
135: Permission is granted to copy and distribute translations of this manual
136: into another language, under the above conditions for modified versions,
137: except that the sections entitled "Distribution" and "General Public
138: License" may be included in a translation approved by the author instead
139: of in the original English.
140: @end titlepage
141:
142: @node Top, License, (dir), (dir)
143: @ifnottex
144: Gforth is a free implementation of ANS Forth available on many
145: personal machines. This manual corresponds to version @value{VERSION}.
146: @end ifnottex
147:
148: @menu
149: * License:: The GPL
150: * Goals:: About the Gforth Project
151: * Gforth Environment:: Starting (and exiting) Gforth
152: * Tutorial:: Hands-on Forth Tutorial
153: * Introduction:: An introduction to ANS Forth
154: * Words:: Forth words available in Gforth
155: * Error messages:: How to interpret them
156: * Tools:: Programming tools
157: * ANS conformance:: Implementation-defined options etc.
158: * Standard vs Extensions:: Should I use extensions?
159: * Model:: The abstract machine of Gforth
160: * Integrating Gforth:: Forth as scripting language for applications
161: * Emacs and Gforth:: The Gforth Mode
162: * Image Files:: @code{.fi} files contain compiled code
163: * Engine:: The inner interpreter and the primitives
164: * Binding to System Library::
165: * Cross Compiler:: The Cross Compiler
166: * Bugs:: How to report them
167: * Origin:: Authors and ancestors of Gforth
168: * Forth-related information:: Books and places to look on the WWW
169: * Word Index:: An item for each Forth word
170: * Name Index:: Forth words, only names listed
171: * Concept Index:: A menu covering many topics
172:
173: @detailmenu --- The Detailed Node Listing ---
174:
175: Gforth Environment
176:
177: * Invoking Gforth:: Getting in
178: * Leaving Gforth:: Getting out
179: * Command-line editing::
180: * Environment variables:: that affect how Gforth starts up
181: * Gforth Files:: What gets installed and where
182: * Startup speed:: When 35ms is not fast enough ...
183:
184: Forth Tutorial
185:
186: * Starting Gforth Tutorial::
187: * Syntax Tutorial::
188: * Crash Course Tutorial::
189: * Stack Tutorial::
190: * Arithmetics Tutorial::
191: * Stack Manipulation Tutorial::
192: * Using files for Forth code Tutorial::
193: * Comments Tutorial::
194: * Colon Definitions Tutorial::
195: * Decompilation Tutorial::
196: * Stack-Effect Comments Tutorial::
197: * Types Tutorial::
198: * Factoring Tutorial::
199: * Designing the stack effect Tutorial::
200: * Local Variables Tutorial::
201: * Conditional execution Tutorial::
202: * Flags and Comparisons Tutorial::
203: * General Loops Tutorial::
204: * Counted loops Tutorial::
205: * Recursion Tutorial::
206: * Leaving definitions or loops Tutorial::
207: * Return Stack Tutorial::
208: * Memory Tutorial::
209: * Characters and Strings Tutorial::
210: * Alignment Tutorial::
211: * Interpretation and Compilation Semantics and Immediacy Tutorial::
212: * Execution Tokens Tutorial::
213: * Exceptions Tutorial::
214: * Defining Words Tutorial::
215: * Arrays and Records Tutorial::
216: * POSTPONE Tutorial::
217: * Literal Tutorial::
218: * Advanced macros Tutorial::
219: * Compilation Tokens Tutorial::
220: * Wordlists and Search Order Tutorial::
221:
222: An Introduction to ANS Forth
223:
224: * Introducing the Text Interpreter::
225: * Stacks and Postfix notation::
226: * Your first definition::
227: * How does that work?::
228: * Forth is written in Forth::
229: * Review - elements of a Forth system::
230: * Where to go next::
231: * Exercises::
232:
233: Forth Words
234:
235: * Notation::
236: * Case insensitivity::
237: * Comments::
238: * Boolean Flags::
239: * Arithmetic::
240: * Stack Manipulation::
241: * Memory::
242: * Control Structures::
243: * Defining Words::
244: * Interpretation and Compilation Semantics::
245: * Tokens for Words::
246: * The Text Interpreter::
247: * Word Lists::
248: * Environmental Queries::
249: * Files::
250: * Blocks::
251: * Other I/O::
252: * Programming Tools::
253: * Assembler and Code Words::
254: * Threading Words::
255: * Locals::
256: * Structures::
257: * Object-oriented Forth::
258: * Passing Commands to the OS::
259: * Keeping track of Time::
260: * Miscellaneous Words::
261:
262: Arithmetic
263:
264: * Single precision::
265: * Double precision:: Double-cell integer arithmetic
266: * Bitwise operations::
267: * Numeric comparison::
268: * Mixed precision:: Operations with single and double-cell integers
269: * Floating Point::
270:
271: Stack Manipulation
272:
273: * Data stack::
274: * Floating point stack::
275: * Return stack::
276: * Locals stack::
277: * Stack pointer manipulation::
278:
279: Memory
280:
281: * Memory model::
282: * Dictionary allocation::
283: * Heap Allocation::
284: * Memory Access::
285: * Address arithmetic::
286: * Memory Blocks::
287:
288: Control Structures
289:
290: * Selection:: IF ... ELSE ... ENDIF
291: * Simple Loops:: BEGIN ...
292: * Counted Loops:: DO
293: * Arbitrary control structures::
294: * Calls and returns::
295: * Exception Handling::
296:
297: Defining Words
298:
299: * CREATE::
300: * Variables:: Variables and user variables
301: * Constants::
302: * Values:: Initialised variables
303: * Colon Definitions::
304: * Anonymous Definitions:: Definitions without names
305: * Supplying names::
306: * User-defined Defining Words::
307: * Deferred words:: Allow forward references
308: * Aliases::
309:
310: User-defined Defining Words
311:
312: * CREATE..DOES> applications::
313: * CREATE..DOES> details::
314: * Advanced does> usage example::
315:
316: Interpretation and Compilation Semantics
317:
318: * Combined words::
319:
320: The Text Interpreter
321:
322: * Input Sources::
323: * Number Conversion::
324: * Interpret/Compile states::
325: * Literals::
326: * Interpreter Directives::
327:
328: Word Lists
329:
330: * Why use word lists?::
331: * Word list examples::
332:
333: Files
334:
335: * Forth source files::
336: * General files::
337: * Search Paths::
338:
339: Search Paths
340:
341: * Forth Search Paths::
342: * General Search Paths::
343:
344: Other I/O
345:
346: * Simple numeric output:: Predefined formats
347: * Formatted numeric output:: Formatted (pictured) output
348: * String Formats:: How Forth stores strings in memory
349: * Displaying characters and strings:: Other stuff
350: * Input:: Input
351:
352: Programming Tools
353:
354: * Debugging:: Simple and quick.
355: * Assertions:: Making your programs self-checking.
356: * Singlestep Debugger:: Executing your program word by word.
357:
358: Assembler and Code Words
359:
360: * Code and ;code::
361: * Common Assembler:: Assembler Syntax
362: * Common Disassembler::
363: * 386 Assembler:: Deviations and special cases
364: * Alpha Assembler:: Deviations and special cases
365: * MIPS assembler:: Deviations and special cases
366: * Other assemblers:: How to write them
367:
368: Locals
369:
370: * Gforth locals::
371: * ANS Forth locals::
372:
373: Gforth locals
374:
375: * Where are locals visible by name?::
376: * How long do locals live?::
377: * Programming Style::
378: * Implementation::
379:
380: Structures
381:
382: * Why explicit structure support?::
383: * Structure Usage::
384: * Structure Naming Convention::
385: * Structure Implementation::
386: * Structure Glossary::
387:
388: Object-oriented Forth
389:
390: * Why object-oriented programming?::
391: * Object-Oriented Terminology::
392: * Objects::
393: * OOF::
394: * Mini-OOF::
395: * Comparison with other object models::
396:
397: The @file{objects.fs} model
398:
399: * Properties of the Objects model::
400: * Basic Objects Usage::
401: * The Objects base class::
402: * Creating objects::
403: * Object-Oriented Programming Style::
404: * Class Binding::
405: * Method conveniences::
406: * Classes and Scoping::
407: * Dividing classes::
408: * Object Interfaces::
409: * Objects Implementation::
410: * Objects Glossary::
411:
412: The @file{oof.fs} model
413:
414: * Properties of the OOF model::
415: * Basic OOF Usage::
416: * The OOF base class::
417: * Class Declaration::
418: * Class Implementation::
419:
420: The @file{mini-oof.fs} model
421:
422: * Basic Mini-OOF Usage::
423: * Mini-OOF Example::
424: * Mini-OOF Implementation::
425:
426: Tools
427:
428: * ANS Report:: Report the words used, sorted by wordset.
429:
430: ANS conformance
431:
432: * The Core Words::
433: * The optional Block word set::
434: * The optional Double Number word set::
435: * The optional Exception word set::
436: * The optional Facility word set::
437: * The optional File-Access word set::
438: * The optional Floating-Point word set::
439: * The optional Locals word set::
440: * The optional Memory-Allocation word set::
441: * The optional Programming-Tools word set::
442: * The optional Search-Order word set::
443:
444: The Core Words
445:
446: * core-idef:: Implementation Defined Options
447: * core-ambcond:: Ambiguous Conditions
448: * core-other:: Other System Documentation
449:
450: The optional Block word set
451:
452: * block-idef:: Implementation Defined Options
453: * block-ambcond:: Ambiguous Conditions
454: * block-other:: Other System Documentation
455:
456: The optional Double Number word set
457:
458: * double-ambcond:: Ambiguous Conditions
459:
460: The optional Exception word set
461:
462: * exception-idef:: Implementation Defined Options
463:
464: The optional Facility word set
465:
466: * facility-idef:: Implementation Defined Options
467: * facility-ambcond:: Ambiguous Conditions
468:
469: The optional File-Access word set
470:
471: * file-idef:: Implementation Defined Options
472: * file-ambcond:: Ambiguous Conditions
473:
474: The optional Floating-Point word set
475:
476: * floating-idef:: Implementation Defined Options
477: * floating-ambcond:: Ambiguous Conditions
478:
479: The optional Locals word set
480:
481: * locals-idef:: Implementation Defined Options
482: * locals-ambcond:: Ambiguous Conditions
483:
484: The optional Memory-Allocation word set
485:
486: * memory-idef:: Implementation Defined Options
487:
488: The optional Programming-Tools word set
489:
490: * programming-idef:: Implementation Defined Options
491: * programming-ambcond:: Ambiguous Conditions
492:
493: The optional Search-Order word set
494:
495: * search-idef:: Implementation Defined Options
496: * search-ambcond:: Ambiguous Conditions
497:
498: Image Files
499:
500: * Image Licensing Issues:: Distribution terms for images.
501: * Image File Background:: Why have image files?
502: * Non-Relocatable Image Files:: don't always work.
503: * Data-Relocatable Image Files:: are better.
504: * Fully Relocatable Image Files:: better yet.
505: * Stack and Dictionary Sizes:: Setting the default sizes for an image.
506: * Running Image Files:: @code{gforth -i @i{file}} or @i{file}.
507: * Modifying the Startup Sequence:: and turnkey applications.
508:
509: Fully Relocatable Image Files
510:
511: * gforthmi:: The normal way
512: * cross.fs:: The hard way
513:
514: Engine
515:
516: * Portability::
517: * Threading::
518: * Primitives::
519: * Performance::
520:
521: Threading
522:
523: * Scheduling::
524: * Direct or Indirect Threaded?::
525: * DOES>::
526:
527: Primitives
528:
529: * Automatic Generation::
530: * TOS Optimization::
531: * Produced code::
532:
533: Cross Compiler
534:
535: * Using the Cross Compiler::
536: * How the Cross Compiler Works::
537:
538: Other Forth-related information
539:
540: * Internet resources::
541: * Books::
542: * The Forth Interest Group::
543: * Conferences::
544:
545: @end detailmenu
546: @end menu
547:
548: @node License, Goals, Top, Top
549: @unnumbered GNU GENERAL PUBLIC LICENSE
550: @center Version 2, June 1991
551:
552: @display
553: Copyright @copyright{} 1989, 1991 Free Software Foundation, Inc.
554: 675 Mass Ave, Cambridge, MA 02139, USA
555:
556: Everyone is permitted to copy and distribute verbatim copies
557: of this license document, but changing it is not allowed.
558: @end display
559:
560: @unnumberedsec Preamble
561:
562: The licenses for most software are designed to take away your
563: freedom to share and change it. By contrast, the GNU General Public
564: License is intended to guarantee your freedom to share and change free
565: software---to make sure the software is free for all its users. This
566: General Public License applies to most of the Free Software
567: Foundation's software and to any other program whose authors commit to
568: using it. (Some other Free Software Foundation software is covered by
569: the GNU Library General Public License instead.) You can apply it to
570: your programs, too.
571:
572: When we speak of free software, we are referring to freedom, not
573: price. Our General Public Licenses are designed to make sure that you
574: have the freedom to distribute copies of free software (and charge for
575: this service if you wish), that you receive source code or can get it
576: if you want it, that you can change the software or use pieces of it
577: in new free programs; and that you know you can do these things.
578:
579: To protect your rights, we need to make restrictions that forbid
580: anyone to deny you these rights or to ask you to surrender the rights.
581: These restrictions translate to certain responsibilities for you if you
582: distribute copies of the software, or if you modify it.
583:
584: For example, if you distribute copies of such a program, whether
585: gratis or for a fee, you must give the recipients all the rights that
586: you have. You must make sure that they, too, receive or can get the
587: source code. And you must show them these terms so they know their
588: rights.
589:
590: We protect your rights with two steps: (1) copyright the software, and
591: (2) offer you this license which gives you legal permission to copy,
592: distribute and/or modify the software.
593:
594: Also, for each author's protection and ours, we want to make certain
595: that everyone understands that there is no warranty for this free
596: software. If the software is modified by someone else and passed on, we
597: want its recipients to know that what they have is not the original, so
598: that any problems introduced by others will not reflect on the original
599: authors' reputations.
600:
601: Finally, any free program is threatened constantly by software
602: patents. We wish to avoid the danger that redistributors of a free
603: program will individually obtain patent licenses, in effect making the
604: program proprietary. To prevent this, we have made it clear that any
605: patent must be licensed for everyone's free use or not licensed at all.
606:
607: The precise terms and conditions for copying, distribution and
608: modification follow.
609:
610: @iftex
611: @unnumberedsec TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION
612: @end iftex
613: @ifnottex
614: @center TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION
615: @end ifnottex
616:
617: @enumerate 0
618: @item
619: This License applies to any program or other work which contains
620: a notice placed by the copyright holder saying it may be distributed
621: under the terms of this General Public License. The ``Program'', below,
622: refers to any such program or work, and a ``work based on the Program''
623: means either the Program or any derivative work under copyright law:
624: that is to say, a work containing the Program or a portion of it,
625: either verbatim or with modifications and/or translated into another
626: language. (Hereinafter, translation is included without limitation in
627: the term ``modification''.) Each licensee is addressed as ``you''.
628:
629: Activities other than copying, distribution and modification are not
630: covered by this License; they are outside its scope. The act of
631: running the Program is not restricted, and the output from the Program
632: is covered only if its contents constitute a work based on the
633: Program (independent of having been made by running the Program).
634: Whether that is true depends on what the Program does.
635:
636: @item
637: You may copy and distribute verbatim copies of the Program's
638: source code as you receive it, in any medium, provided that you
639: conspicuously and appropriately publish on each copy an appropriate
640: copyright notice and disclaimer of warranty; keep intact all the
641: notices that refer to this License and to the absence of any warranty;
642: and give any other recipients of the Program a copy of this License
643: along with the Program.
644:
645: You may charge a fee for the physical act of transferring a copy, and
646: you may at your option offer warranty protection in exchange for a fee.
647:
648: @item
649: You may modify your copy or copies of the Program or any portion
650: of it, thus forming a work based on the Program, and copy and
651: distribute such modifications or work under the terms of Section 1
652: above, provided that you also meet all of these conditions:
653:
654: @enumerate a
655: @item
656: You must cause the modified files to carry prominent notices
657: stating that you changed the files and the date of any change.
658:
659: @item
660: You must cause any work that you distribute or publish, that in
661: whole or in part contains or is derived from the Program or any
662: part thereof, to be licensed as a whole at no charge to all third
663: parties under the terms of this License.
664:
665: @item
666: If the modified program normally reads commands interactively
667: when run, you must cause it, when started running for such
668: interactive use in the most ordinary way, to print or display an
669: announcement including an appropriate copyright notice and a
670: notice that there is no warranty (or else, saying that you provide
671: a warranty) and that users may redistribute the program under
672: these conditions, and telling the user how to view a copy of this
673: License. (Exception: if the Program itself is interactive but
674: does not normally print such an announcement, your work based on
675: the Program is not required to print an announcement.)
676: @end enumerate
677:
678: These requirements apply to the modified work as a whole. If
679: identifiable sections of that work are not derived from the Program,
680: and can be reasonably considered independent and separate works in
681: themselves, then this License, and its terms, do not apply to those
682: sections when you distribute them as separate works. But when you
683: distribute the same sections as part of a whole which is a work based
684: on the Program, the distribution of the whole must be on the terms of
685: this License, whose permissions for other licensees extend to the
686: entire whole, and thus to each and every part regardless of who wrote it.
687:
688: Thus, it is not the intent of this section to claim rights or contest
689: your rights to work written entirely by you; rather, the intent is to
690: exercise the right to control the distribution of derivative or
691: collective works based on the Program.
692:
693: In addition, mere aggregation of another work not based on the Program
694: with the Program (or with a work based on the Program) on a volume of
695: a storage or distribution medium does not bring the other work under
696: the scope of this License.
697:
698: @item
699: You may copy and distribute the Program (or a work based on it,
700: under Section 2) in object code or executable form under the terms of
701: Sections 1 and 2 above provided that you also do one of the following:
702:
703: @enumerate a
704: @item
705: Accompany it with the complete corresponding machine-readable
706: source code, which must be distributed under the terms of Sections
707: 1 and 2 above on a medium customarily used for software interchange; or,
708:
709: @item
710: Accompany it with a written offer, valid for at least three
711: years, to give any third party, for a charge no more than your
712: cost of physically performing source distribution, a complete
713: machine-readable copy of the corresponding source code, to be
714: distributed under the terms of Sections 1 and 2 above on a medium
715: customarily used for software interchange; or,
716:
717: @item
718: Accompany it with the information you received as to the offer
719: to distribute corresponding source code. (This alternative is
720: allowed only for noncommercial distribution and only if you
721: received the program in object code or executable form with such
722: an offer, in accord with Subsection b above.)
723: @end enumerate
724:
725: The source code for a work means the preferred form of the work for
726: making modifications to it. For an executable work, complete source
727: code means all the source code for all modules it contains, plus any
728: associated interface definition files, plus the scripts used to
729: control compilation and installation of the executable. However, as a
730: special exception, the source code distributed need not include
731: anything that is normally distributed (in either source or binary
732: form) with the major components (compiler, kernel, and so on) of the
733: operating system on which the executable runs, unless that component
734: itself accompanies the executable.
735:
736: If distribution of executable or object code is made by offering
737: access to copy from a designated place, then offering equivalent
738: access to copy the source code from the same place counts as
739: distribution of the source code, even though third parties are not
740: compelled to copy the source along with the object code.
741:
742: @item
743: You may not copy, modify, sublicense, or distribute the Program
744: except as expressly provided under this License. Any attempt
745: otherwise to copy, modify, sublicense or distribute the Program is
746: void, and will automatically terminate your rights under this License.
747: However, parties who have received copies, or rights, from you under
748: this License will not have their licenses terminated so long as such
749: parties remain in full compliance.
750:
751: @item
752: You are not required to accept this License, since you have not
753: signed it. However, nothing else grants you permission to modify or
754: distribute the Program or its derivative works. These actions are
755: prohibited by law if you do not accept this License. Therefore, by
756: modifying or distributing the Program (or any work based on the
757: Program), you indicate your acceptance of this License to do so, and
758: all its terms and conditions for copying, distributing or modifying
759: the Program or works based on it.
760:
761: @item
762: Each time you redistribute the Program (or any work based on the
763: Program), the recipient automatically receives a license from the
764: original licensor to copy, distribute or modify the Program subject to
765: these terms and conditions. You may not impose any further
766: restrictions on the recipients' exercise of the rights granted herein.
767: You are not responsible for enforcing compliance by third parties to
768: this License.
769:
770: @item
771: If, as a consequence of a court judgment or allegation of patent
772: infringement or for any other reason (not limited to patent issues),
773: conditions are imposed on you (whether by court order, agreement or
774: otherwise) that contradict the conditions of this License, they do not
775: excuse you from the conditions of this License. If you cannot
776: distribute so as to satisfy simultaneously your obligations under this
777: License and any other pertinent obligations, then as a consequence you
778: may not distribute the Program at all. For example, if a patent
779: license would not permit royalty-free redistribution of the Program by
780: all those who receive copies directly or indirectly through you, then
781: the only way you could satisfy both it and this License would be to
782: refrain entirely from distribution of the Program.
783:
784: If any portion of this section is held invalid or unenforceable under
785: any particular circumstance, the balance of the section is intended to
786: apply and the section as a whole is intended to apply in other
787: circumstances.
788:
789: It is not the purpose of this section to induce you to infringe any
790: patents or other property right claims or to contest validity of any
791: such claims; this section has the sole purpose of protecting the
792: integrity of the free software distribution system, which is
793: implemented by public license practices. Many people have made
794: generous contributions to the wide range of software distributed
795: through that system in reliance on consistent application of that
796: system; it is up to the author/donor to decide if he or she is willing
797: to distribute software through any other system and a licensee cannot
798: impose that choice.
799:
800: This section is intended to make thoroughly clear what is believed to
801: be a consequence of the rest of this License.
802:
803: @item
804: If the distribution and/or use of the Program is restricted in
805: certain countries either by patents or by copyrighted interfaces, the
806: original copyright holder who places the Program under this License
807: may add an explicit geographical distribution limitation excluding
808: those countries, so that distribution is permitted only in or among
809: countries not thus excluded. In such case, this License incorporates
810: the limitation as if written in the body of this License.
811:
812: @item
813: The Free Software Foundation may publish revised and/or new versions
814: of the General Public License from time to time. Such new versions will
815: be similar in spirit to the present version, but may differ in detail to
816: address new problems or concerns.
817:
818: Each version is given a distinguishing version number. If the Program
819: specifies a version number of this License which applies to it and ``any
820: later version'', you have the option of following the terms and conditions
821: either of that version or of any later version published by the Free
822: Software Foundation. If the Program does not specify a version number of
823: this License, you may choose any version ever published by the Free Software
824: Foundation.
825:
826: @item
827: If you wish to incorporate parts of the Program into other free
828: programs whose distribution conditions are different, write to the author
829: to ask for permission. For software which is copyrighted by the Free
830: Software Foundation, write to the Free Software Foundation; we sometimes
831: make exceptions for this. Our decision will be guided by the two goals
832: of preserving the free status of all derivatives of our free software and
833: of promoting the sharing and reuse of software generally.
834:
835: @iftex
836: @heading NO WARRANTY
837: @end iftex
838: @ifnottex
839: @center NO WARRANTY
840: @end ifnottex
841:
842: @item
843: BECAUSE THE PROGRAM IS LICENSED FREE OF CHARGE, THERE IS NO WARRANTY
844: FOR THE PROGRAM, TO THE EXTENT PERMITTED BY APPLICABLE LAW. EXCEPT WHEN
845: OTHERWISE STATED IN WRITING THE COPYRIGHT HOLDERS AND/OR OTHER PARTIES
846: PROVIDE THE PROGRAM ``AS IS'' WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESSED
847: OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
848: MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. THE ENTIRE RISK AS
849: TO THE QUALITY AND PERFORMANCE OF THE PROGRAM IS WITH YOU. SHOULD THE
850: PROGRAM PROVE DEFECTIVE, YOU ASSUME THE COST OF ALL NECESSARY SERVICING,
851: REPAIR OR CORRECTION.
852:
853: @item
854: IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING
855: WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MAY MODIFY AND/OR
856: REDISTRIBUTE THE PROGRAM AS PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES,
857: INCLUDING ANY GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING
858: OUT OF THE USE OR INABILITY TO USE THE PROGRAM (INCLUDING BUT NOT LIMITED
859: TO LOSS OF DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY
860: YOU OR THIRD PARTIES OR A FAILURE OF THE PROGRAM TO OPERATE WITH ANY OTHER
861: PROGRAMS), EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE
862: POSSIBILITY OF SUCH DAMAGES.
863: @end enumerate
864:
865: @iftex
866: @heading END OF TERMS AND CONDITIONS
867: @end iftex
868: @ifnottex
869: @center END OF TERMS AND CONDITIONS
870: @end ifnottex
871:
872: @page
873: @unnumberedsec How to Apply These Terms to Your New Programs
874:
875: If you develop a new program, and you want it to be of the greatest
876: possible use to the public, the best way to achieve this is to make it
877: free software which everyone can redistribute and change under these terms.
878:
879: To do so, attach the following notices to the program. It is safest
880: to attach them to the start of each source file to most effectively
881: convey the exclusion of warranty; and each file should have at least
882: the ``copyright'' line and a pointer to where the full notice is found.
883:
884: @smallexample
885: @var{one line to give the program's name and a brief idea of what it does.}
886: Copyright (C) 19@var{yy} @var{name of author}
887:
888: This program is free software; you can redistribute it and/or modify
889: it under the terms of the GNU General Public License as published by
890: the Free Software Foundation; either version 2 of the License, or
891: (at your option) any later version.
892:
893: This program is distributed in the hope that it will be useful,
894: but WITHOUT ANY WARRANTY; without even the implied warranty of
895: MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
896: GNU General Public License for more details.
897:
898: You should have received a copy of the GNU General Public License
899: along with this program; if not, write to the Free Software
900: Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
901: @end smallexample
902:
903: Also add information on how to contact you by electronic and paper mail.
904:
905: If the program is interactive, make it output a short notice like this
906: when it starts in an interactive mode:
907:
908: @smallexample
909: Gnomovision version 69, Copyright (C) 19@var{yy} @var{name of author}
910: Gnomovision comes with ABSOLUTELY NO WARRANTY; for details
911: type `show w'.
912: This is free software, and you are welcome to redistribute it
913: under certain conditions; type `show c' for details.
914: @end smallexample
915:
916: The hypothetical commands @samp{show w} and @samp{show c} should show
917: the appropriate parts of the General Public License. Of course, the
918: commands you use may be called something other than @samp{show w} and
919: @samp{show c}; they could even be mouse-clicks or menu items---whatever
920: suits your program.
921:
922: You should also get your employer (if you work as a programmer) or your
923: school, if any, to sign a ``copyright disclaimer'' for the program, if
924: necessary. Here is a sample; alter the names:
925:
926: @smallexample
927: Yoyodyne, Inc., hereby disclaims all copyright interest in the program
928: `Gnomovision' (which makes passes at compilers) written by James Hacker.
929:
930: @var{signature of Ty Coon}, 1 April 1989
931: Ty Coon, President of Vice
932: @end smallexample
933:
934: This General Public License does not permit incorporating your program into
935: proprietary programs. If your program is a subroutine library, you may
936: consider it more useful to permit linking proprietary applications with the
937: library. If this is what you want to do, use the GNU Library General
938: Public License instead of this License.
939:
940: @iftex
941: @unnumbered Preface
942: @cindex Preface
943: This manual documents Gforth. Some introductory material is provided for
944: readers who are unfamiliar with Forth or who are migrating to Gforth
945: from other Forth compilers. However, this manual is primarily a
946: reference manual.
947: @end iftex
948:
949: @comment TODO much more blurb here.
950:
951: @c ******************************************************************
952: @node Goals, Gforth Environment, License, Top
953: @comment node-name, next, previous, up
954: @chapter Goals of Gforth
955: @cindex goals of the Gforth project
956: The goal of the Gforth Project is to develop a standard model for
957: ANS Forth. This can be split into several subgoals:
958:
959: @itemize @bullet
960: @item
961: Gforth should conform to the ANS Forth Standard.
962: @item
963: It should be a model, i.e. it should define all the
964: implementation-dependent things.
965: @item
966: It should become standard, i.e. widely accepted and used. This goal
967: is the most difficult one.
968: @end itemize
969:
970: To achieve these goals Gforth should be
971: @itemize @bullet
972: @item
973: Similar to previous models (fig-Forth, F83)
974: @item
975: Powerful. It should provide for all the things that are considered
976: necessary today and even some that are not yet considered necessary.
977: @item
978: Efficient. It should not get the reputation of being exceptionally
979: slow.
980: @item
981: Free.
982: @item
983: Available on many machines/easy to port.
984: @end itemize
985:
986: Have we achieved these goals? Gforth conforms to the ANS Forth
987: standard. It may be considered a model, but we have not yet documented
988: which parts of the model are stable and which parts we are likely to
989: change. It certainly has not yet become a de facto standard, but it
990: appears to be quite popular. It has some similarities to and some
991: differences from previous models. It has some powerful features, but not
992: yet everything that we envisioned. We certainly have achieved our
993: execution speed goals (@pxref{Performance})@footnote{However, in 1998
994: the bar was raised when the major commercial Forth vendors switched to
995: native code compilers.}. It is free and available on many machines.
996:
997: @c ******************************************************************
998: @node Gforth Environment, Tutorial, Goals, Top
999: @chapter Gforth Environment
1000: @cindex Gforth environment
1001:
1002: Note: ultimately, the Gforth man page will be auto-generated from the
1003: material in this chapter.
1004:
1005: @menu
1006: * Invoking Gforth:: Getting in
1007: * Leaving Gforth:: Getting out
1008: * Command-line editing::
1009: * Environment variables:: that affect how Gforth starts up
1010: * Gforth Files:: What gets installed and where
1011: * Startup speed:: When 35ms is not fast enough ...
1012: @end menu
1013:
1014: For related information about the creation of images see @ref{Image Files}.
1015:
1016: @comment ----------------------------------------------
1017: @node Invoking Gforth, Leaving Gforth, Gforth Environment, Gforth Environment
1018: @section Invoking Gforth
1019: @cindex invoking Gforth
1020: @cindex running Gforth
1021: @cindex command-line options
1022: @cindex options on the command line
1023: @cindex flags on the command line
1024:
1025: Gforth is made up of two parts; an executable ``engine'' (named
1026: @file{gforth} or @file{gforth-fast}) and an image file. To start it, you
1027: will usually just say @code{gforth} -- this automatically loads the
1028: default image file @file{gforth.fi}. In many other cases the default
1029: Gforth image will be invoked like this:
1030: @example
1031: gforth [file | -e forth-code] ...
1032: @end example
1033: @noindent
1034: This interprets the contents of the files and the Forth code in the order they
1035: are given.
1036:
1037: In addition to the @file{gforth} engine, there is also an engine called
1038: @file{gforth-fast}, which is faster, but gives less informative error
1039: messages (@pxref{Error messages}).
1040:
1041: In general, the command line looks like this:
1042:
1043: @example
1044: gforth[-fast] [engine options] [image options]
1045: @end example
1046:
1047: The engine options must come before the rest of the command
1048: line. They are:
1049:
1050: @table @code
1051: @cindex -i, command-line option
1052: @cindex --image-file, command-line option
1053: @item --image-file @i{file}
1054: @itemx -i @i{file}
1055: Loads the Forth image @i{file} instead of the default
1056: @file{gforth.fi} (@pxref{Image Files}).
1057:
1058: @cindex --appl-image, command-line option
1059: @item --appl-image @i{file}
1060: Loads the image @i{file} and leaves all further command-line arguments
1061: to the image (instead of processing them as engine options). This is
1062: useful for building executable application images on Unix, built with
1063: @code{gforthmi --application ...}.
1064:
1065: @cindex --path, command-line option
1066: @cindex -p, command-line option
1067: @item --path @i{path}
1068: @itemx -p @i{path}
1069: Uses @i{path} for searching the image file and Forth source code files
1070: instead of the default in the environment variable @code{GFORTHPATH} or
1071: the path specified at installation time (e.g.,
1072: @file{/usr/local/share/gforth/0.2.0:.}). A path is given as a list of
1073: directories, separated by @samp{:} (on Unix) or @samp{;} (on other OSs).
1074:
1075: @cindex --dictionary-size, command-line option
1076: @cindex -m, command-line option
1077: @cindex @i{size} parameters for command-line options
1078: @cindex size of the dictionary and the stacks
1079: @item --dictionary-size @i{size}
1080: @itemx -m @i{size}
1081: Allocate @i{size} space for the Forth dictionary space instead of
1082: using the default specified in the image (typically 256K). The
1083: @i{size} specification for this and subsequent options consists of
1084: an integer and a unit (e.g.,
1085: @code{4M}). The unit can be one of @code{b} (bytes), @code{e} (element
1086: size, in this case Cells), @code{k} (kilobytes), @code{M} (Megabytes),
1087: @code{G} (Gigabytes), and @code{T} (Terabytes). If no unit is specified,
1088: @code{e} is used.
1089:
1090: @cindex --data-stack-size, command-line option
1091: @cindex -d, command-line option
1092: @item --data-stack-size @i{size}
1093: @itemx -d @i{size}
1094: Allocate @i{size} space for the data stack instead of using the
1095: default specified in the image (typically 16K).
1096:
1097: @cindex --return-stack-size, command-line option
1098: @cindex -r, command-line option
1099: @item --return-stack-size @i{size}
1100: @itemx -r @i{size}
1101: Allocate @i{size} space for the return stack instead of using the
1102: default specified in the image (typically 15K).
1103:
1104: @cindex --fp-stack-size, command-line option
1105: @cindex -f, command-line option
1106: @item --fp-stack-size @i{size}
1107: @itemx -f @i{size}
1108: Allocate @i{size} space for the floating point stack instead of
1109: using the default specified in the image (typically 15.5K). In this case
1110: the unit specifier @code{e} refers to floating point numbers.
1111:
1112: @cindex --locals-stack-size, command-line option
1113: @cindex -l, command-line option
1114: @item --locals-stack-size @i{size}
1115: @itemx -l @i{size}
1116: Allocate @i{size} space for the locals stack instead of using the
1117: default specified in the image (typically 14.5K).
1118:
1119: @cindex -h, command-line option
1120: @cindex --help, command-line option
1121: @item --help
1122: @itemx -h
1123: Print a message about the command-line options
1124:
1125: @cindex -v, command-line option
1126: @cindex --version, command-line option
1127: @item --version
1128: @itemx -v
1129: Print version and exit
1130:
1131: @cindex --debug, command-line option
1132: @item --debug
1133: Print some information useful for debugging on startup.
1134:
1135: @cindex --offset-image, command-line option
1136: @item --offset-image
1137: Start the dictionary at a slightly different position than would be used
1138: otherwise (useful for creating data-relocatable images,
1139: @pxref{Data-Relocatable Image Files}).
1140:
1141: @cindex --no-offset-im, command-line option
1142: @item --no-offset-im
1143: Start the dictionary at the normal position.
1144:
1145: @cindex --clear-dictionary, command-line option
1146: @item --clear-dictionary
1147: Initialize all bytes in the dictionary to 0 before loading the image
1148: (@pxref{Data-Relocatable Image Files}).
1149:
1150: @cindex --die-on-signal, command-line-option
1151: @item --die-on-signal
1152: Normally Gforth handles most signals (e.g., the user interrupt SIGINT,
1153: or the segmentation violation SIGSEGV) by translating it into a Forth
1154: @code{THROW}. With this option, Gforth exits if it receives such a
1155: signal. This option is useful when the engine and/or the image might be
1156: severely broken (such that it causes another signal before recovering
1157: from the first); this option avoids endless loops in such cases.
1158: @end table
1159:
1160: @cindex loading files at startup
1161: @cindex executing code on startup
1162: @cindex batch processing with Gforth
1163: As explained above, the image-specific command-line arguments for the
1164: default image @file{gforth.fi} consist of a sequence of filenames and
1165: @code{-e @var{forth-code}} options that are interpreted in the sequence
1166: in which they are given. The @code{-e @var{forth-code}} or
1167: @code{--evaluate @var{forth-code}} option evaluates the Forth
1168: code. This option takes only one argument; if you want to evaluate more
1169: Forth words, you have to quote them or use @code{-e} several times. To exit
1170: after processing the command line (instead of entering interactive mode)
1171: append @code{-e bye} to the command line.
1172:
1173: @cindex versions, invoking other versions of Gforth
1174: If you have several versions of Gforth installed, @code{gforth} will
1175: invoke the version that was installed last. @code{gforth-@i{version}}
1176: invokes a specific version. If your environment contains the variable
1177: @code{GFORTHPATH}, you may want to override it by using the
1178: @code{--path} option.
1179:
1180: Not yet implemented:
1181: On startup the system first executes the system initialization file
1182: (unless the option @code{--no-init-file} is given; note that the system
1183: resulting from using this option may not be ANS Forth conformant). Then
1184: the user initialization file @file{.gforth.fs} is executed, unless the
1185: option @code{--no-rc} is given; this file is searched for in @file{.},
1186: then in @file{~}, then in the normal path (see above).
1187:
1188:
1189:
1190: @comment ----------------------------------------------
1191: @node Leaving Gforth, Command-line editing, Invoking Gforth, Gforth Environment
1192: @section Leaving Gforth
1193: @cindex Gforth - leaving
1194: @cindex leaving Gforth
1195:
1196: You can leave Gforth by typing @code{bye} or @kbd{Ctrl-d} (at the start
1197: of a line) or (if you invoked Gforth with the @code{--die-on-signal}
1198: option) @kbd{Ctrl-c}. When you leave Gforth, all of your definitions and
1199: data are discarded. For ways of saving the state of the system before
1200: leaving Gforth see @ref{Image Files}.
1201:
1202: doc-bye
1203:
1204:
1205: @comment ----------------------------------------------
1206: @node Command-line editing, Environment variables, Leaving Gforth, Gforth Environment
1207: @section Command-line editing
1208: @cindex command-line editing
1209:
1210: Gforth maintains a history file that records every line that you type to
1211: the text interpreter. This file is preserved between sessions, and is
1212: used to provide a command-line recall facility; if you type @kbd{Ctrl-P}
1213: repeatedly you can recall successively older commands from this (or
1214: previous) session(s). The full list of command-line editing facilities is:
1215:
1216: @itemize @bullet
1217: @item
1218: @kbd{Ctrl-p} (``previous'') (or up-arrow) to recall successively older
1219: commands from the history buffer.
1220: @item
1221: @kbd{Ctrl-n} (``next'') (or down-arrow) to recall successively newer commands
1222: from the history buffer.
1223: @item
1224: @kbd{Ctrl-f} (or right-arrow) to move the cursor right, non-destructively.
1225: @item
1226: @kbd{Ctrl-b} (or left-arrow) to move the cursor left, non-destructively.
1227: @item
1228: @kbd{Ctrl-h} (backspace) to delete the character to the left of the cursor,
1229: closing up the line.
1230: @item
1231: @kbd{Ctrl-k} to delete (``kill'') from the cursor to the end of the line.
1232: @item
1233: @kbd{Ctrl-a} to move the cursor to the start of the line.
1234: @item
1235: @kbd{Ctrl-e} to move the cursor to the end of the line.
1236: @item
1237: @key{RET} (@kbd{Ctrl-m}) or @key{LFD} (@kbd{Ctrl-j}) to submit the current
1238: line.
1239: @item
1240: @key{TAB} to step through all possible full-word completions of the word
1241: currently being typed.
1242: @item
1243: @kbd{Ctrl-d} on an empty line line to terminate Gforth (gracefully,
1244: using @code{bye}).
1245: @item
1246: @kbd{Ctrl-x} (or @code{Ctrl-d} on a non-empty line) to delete the
1247: character under the cursor.
1248: @end itemize
1249:
1250: When editing, displayable characters are inserted to the left of the
1251: cursor position; the line is always in ``insert'' (as opposed to
1252: ``overstrike'') mode.
1253:
1254: @cindex history file
1255: @cindex @file{.gforth-history}
1256: On Unix systems, the history file is @file{~/.gforth-history} by
1257: default@footnote{i.e. it is stored in the user's home directory.}. You
1258: can find out the name and location of your history file using:
1259:
1260: @example
1261: history-file type \ Unix-class systems
1262:
1263: history-file type \ Other systems
1264: history-dir type
1265: @end example
1266:
1267: If you enter long definitions by hand, you can use a text editor to
1268: paste them out of the history file into a Forth source file for reuse at
1269: a later time.
1270:
1271: Gforth never trims the size of the history file, so you should do this
1272: periodically, if necessary.
1273:
1274: @comment this is all defined in history.fs
1275: @comment NAC TODO the ctrl-D behaviour can either do a bye or a beep.. how is that option
1276: @comment chosen?
1277:
1278:
1279: @comment ----------------------------------------------
1280: @node Environment variables, Gforth Files, Command-line editing, Gforth Environment
1281: @section Environment variables
1282: @cindex environment variables
1283:
1284: Gforth uses these environment variables:
1285:
1286: @itemize @bullet
1287: @item
1288: @cindex @code{GFORTHHIST} -- environment variable
1289: @code{GFORTHHIST} -- (Unix systems only) specifies the directory in which to
1290: open/create the history file, @file{.gforth-history}. Default:
1291: @code{$HOME}.
1292:
1293: @item
1294: @cindex @code{GFORTHPATH} -- environment variable
1295: @code{GFORTHPATH} -- specifies the path used when searching for the gforth image file and
1296: for Forth source-code files.
1297:
1298: @item
1299: @cindex @code{GFORTH} -- environment variable
1300: @code{GFORTH} -- used by @file{gforthmi}, @xref{gforthmi}.
1301:
1302: @item
1303: @cindex @code{GFORTHD} -- environment variable
1304: @code{GFORTHD} -- used by @file{gforthmi}, @xref{gforthmi}.
1305:
1306: @item
1307: @cindex @code{TMP}, @code{TEMP} - environment variable
1308: @code{TMP}, @code{TEMP} - (non-Unix systems only) used as a potential
1309: location for the history file.
1310: @end itemize
1311:
1312: @comment also POSIXELY_CORRECT LINES COLUMNS HOME but no interest in
1313: @comment mentioning these.
1314:
1315: All the Gforth environment variables default to sensible values if they
1316: are not set.
1317:
1318:
1319: @comment ----------------------------------------------
1320: @node Gforth Files, Startup speed, Environment variables, Gforth Environment
1321: @section Gforth files
1322: @cindex Gforth files
1323:
1324: When you install Gforth on a Unix system, it installs files in these
1325: locations by default:
1326:
1327: @itemize @bullet
1328: @item
1329: @file{/usr/local/bin/gforth}
1330: @item
1331: @file{/usr/local/bin/gforthmi}
1332: @item
1333: @file{/usr/local/man/man1/gforth.1} - man page.
1334: @item
1335: @file{/usr/local/info} - the Info version of this manual.
1336: @item
1337: @file{/usr/local/lib/gforth/<version>/...} - Gforth @file{.fi} files.
1338: @item
1339: @file{/usr/local/share/gforth/<version>/TAGS} - Emacs TAGS file.
1340: @item
1341: @file{/usr/local/share/gforth/<version>/...} - Gforth source files.
1342: @item
1343: @file{.../emacs/site-lisp/gforth.el} - Emacs gforth mode.
1344: @end itemize
1345:
1346: You can select different places for installation by using
1347: @code{configure} options (listed with @code{configure --help}).
1348:
1349: @comment ----------------------------------------------
1350: @node Startup speed, , Gforth Files, Gforth Environment
1351: @section Startup speed
1352: @cindex Startup speed
1353: @cindex speed, startup
1354:
1355: If Gforth is used for CGI scripts or in shell scripts, its startup
1356: speed may become a problem. On a 300MHz 21064a under Linux-2.2.13 with
1357: glibc-2.0.7, @code{gforth -e bye} takes about 24.6ms user and 11.3ms
1358: system time.
1359:
1360: If startup speed is a problem, you may consider the following ways to
1361: improve it; or you may consider ways to reduce the number of startups
1362: (for example, by using Fast-CGI).
1363:
1364: The first step to improve startup speed is to statically link Gforth, by
1365: building it with @code{XLDFLAGS=-static}. This requires more memory for
1366: the code and will therefore slow down the first invocation, but
1367: subsequent invocations avoid the dynamic linking overhead. Another
1368: disadvantage is that Gforth won't profit from library upgrades. As a
1369: result, @code{gforth-static -e bye} takes about 17.1ms user and
1370: 8.2ms system time.
1371:
1372: The next step to improve startup speed is to use a non-relocatable image
1373: (@pxref{Non-Relocatable Image Files}). You can create this image with
1374: @code{gforth -e "savesystem gforthnr.fi bye"} and later use it with
1375: @code{gforth -i gforthnr.fi ...}. This avoids the relocation overhead
1376: and a part of the copy-on-write overhead. The disadvantage is that the
1377: non-relocatable image does not work if the OS gives Gforth a different
1378: address for the dictionary, for whatever reason; so you better provide a
1379: fallback on a relocatable image. @code{gforth-static -i gforthnr.fi -e
1380: bye} takes about 15.3ms user and 7.5ms system time.
1381:
1382: The final step is to disable dictionary hashing in Gforth. Gforth
1383: builds the hash table on startup, which takes much of the startup
1384: overhead. You can do this by commenting out the @code{include hash.fs}
1385: in @file{startup.fs} and everything that requires @file{hash.fs} (at the
1386: moment @file{table.fs} and @file{ekey.fs}) and then doing @code{make}.
1387: The disadvantages are that functionality like @code{table} and
1388: @code{ekey} is missing and that text interpretation (e.g., compiling)
1389: now takes much longer. So, you should only use this method if there is
1390: no significant text interpretation to perform (the script should be
1391: compiled into the image, amongst other things). @code{gforth-static -i
1392: gforthnrnh.fi -e bye} takes about 2.1ms user and 6.1ms system time.
1393:
1394: @c ******************************************************************
1395: @node Tutorial, Introduction, Gforth Environment, Top
1396: @chapter Forth Tutorial
1397: @cindex Tutorial
1398: @cindex Forth Tutorial
1399:
1400: @c Topics from nac's Introduction that could be mentioned:
1401: @c press <ret> after each line
1402: @c Prompt
1403: @c numbers vs. words in dictionary on text interpretation
1404: @c what happens on redefinition
1405: @c parsing words (in particular, defining words)
1406:
1407: This tutorial can be used with any ANS-compliant Forth; any
1408: Gforth-specific features are marked as such and you can skip them if you
1409: work with another Forth. This tutorial does not explain all features of
1410: Forth, just enough to get you started and give you some ideas about the
1411: facilities available in Forth. Read the rest of the manual and the
1412: standard when you are through this.
1413:
1414: The intended way to use this tutorial is that you work through it while
1415: sitting in front of the console, take a look at the examples and predict
1416: what they will do, then try them out; if the outcome is not as expected,
1417: find out why (e.g., by trying out variations of the example), so you
1418: understand what's going on. There are also some assignments that you
1419: should solve.
1420:
1421: This tutorial assumes that you have programmed before and know what,
1422: e.g., a loop is.
1423:
1424: @c !! explain compat library
1425:
1426: @menu
1427: * Starting Gforth Tutorial::
1428: * Syntax Tutorial::
1429: * Crash Course Tutorial::
1430: * Stack Tutorial::
1431: * Arithmetics Tutorial::
1432: * Stack Manipulation Tutorial::
1433: * Using files for Forth code Tutorial::
1434: * Comments Tutorial::
1435: * Colon Definitions Tutorial::
1436: * Decompilation Tutorial::
1437: * Stack-Effect Comments Tutorial::
1438: * Types Tutorial::
1439: * Factoring Tutorial::
1440: * Designing the stack effect Tutorial::
1441: * Local Variables Tutorial::
1442: * Conditional execution Tutorial::
1443: * Flags and Comparisons Tutorial::
1444: * General Loops Tutorial::
1445: * Counted loops Tutorial::
1446: * Recursion Tutorial::
1447: * Leaving definitions or loops Tutorial::
1448: * Return Stack Tutorial::
1449: * Memory Tutorial::
1450: * Characters and Strings Tutorial::
1451: * Alignment Tutorial::
1452: * Interpretation and Compilation Semantics and Immediacy Tutorial::
1453: * Execution Tokens Tutorial::
1454: * Exceptions Tutorial::
1455: * Defining Words Tutorial::
1456: * Arrays and Records Tutorial::
1457: * POSTPONE Tutorial::
1458: * Literal Tutorial::
1459: * Advanced macros Tutorial::
1460: * Compilation Tokens Tutorial::
1461: * Wordlists and Search Order Tutorial::
1462: @end menu
1463:
1464: @node Starting Gforth Tutorial, Syntax Tutorial, Tutorial, Tutorial
1465: @section Starting Gforth
1466: @cindex starting Gforth tutorial
1467: You can start Gforth by typing its name:
1468:
1469: @example
1470: gforth
1471: @end example
1472:
1473: That puts you into interactive mode; you can leave Gforth by typing
1474: @code{bye}. While in Gforth, you can edit the command line and access
1475: the command line history with cursor keys, similar to bash.
1476:
1477:
1478: @node Syntax Tutorial, Crash Course Tutorial, Starting Gforth Tutorial, Tutorial
1479: @section Syntax
1480: @cindex syntax tutorial
1481:
1482: A @dfn{word} is a sequence of arbitrary characters (expcept white
1483: space). Words are separated by white space. E.g., each of the
1484: following lines contains exactly one word:
1485:
1486: @example
1487: word
1488: !@@#$%^&*()
1489: 1234567890
1490: 5!a
1491: @end example
1492:
1493: A frequent beginner's error is to leave away necessary white space,
1494: resulting in an error like @samp{Undefined word}; so if you see such an
1495: error, check if you have put spaces wherever necessary.
1496:
1497: @example
1498: ." hello, world" \ correct
1499: ."hello, world" \ gives an "Undefined word" error
1500: @end example
1501:
1502: Gforth and most other Forth systems ignore differences in case (they are
1503: case-insensitive), i.e., @samp{word} is the same as @samp{Word}. If
1504: your system is case-sensitive, you may have to type all the examples
1505: given here in upper case.
1506:
1507:
1508: @node Crash Course Tutorial, Stack Tutorial, Syntax Tutorial, Tutorial
1509: @section Crash Course
1510:
1511: Type
1512:
1513: @example
1514: 0 0 !
1515: here execute
1516: ' catch >body 20 erase abort
1517: ' (quit) >body 20 erase
1518: @end example
1519:
1520: The last two examples are guaranteed to destroy parts of Gforth (and
1521: most other systems), so you better leave Gforth afterwards (if it has
1522: not finished by itself). On some systems you may have to kill gforth
1523: from outside (e.g., in Unix with @code{kill}).
1524:
1525: Now that you know how to produce crashes (and that there's not much to
1526: them), let's learn how to produce meaningful programs.
1527:
1528:
1529: @node Stack Tutorial, Arithmetics Tutorial, Crash Course Tutorial, Tutorial
1530: @section Stack
1531: @cindex stack tutorial
1532:
1533: The most obvious feature of Forth is the stack. When you type in a
1534: number, it is pushed on the stack. You can display the content of the
1535: stack with @code{.s}.
1536:
1537: @example
1538: 1 2 .s
1539: 3 .s
1540: @end example
1541:
1542: @code{.s} displays the top-of-stack to the right, i.e., the numbers
1543: appear in @code{.s} output as they appeared in the input.
1544:
1545: You can print the top of stack element with @code{.}.
1546:
1547: @example
1548: 1 2 3 . . .
1549: @end example
1550:
1551: In general, words consume their stack arguments (@code{.s} is an
1552: exception).
1553:
1554: @assignment
1555: What does the stack contain after @code{5 6 7 .}?
1556: @endassignment
1557:
1558:
1559: @node Arithmetics Tutorial, Stack Manipulation Tutorial, Stack Tutorial, Tutorial
1560: @section Arithmetics
1561: @cindex arithmetics tutorial
1562:
1563: The words @code{+}, @code{-}, @code{*}, @code{/}, and @code{mod} always
1564: operate on the top two stack items:
1565:
1566: @example
1567: 2 2 .s
1568: + .s
1569: .
1570: 2 1 - .
1571: 7 3 mod .
1572: @end example
1573:
1574: The operands of @code{-}, @code{/}, and @code{mod} are in the same order
1575: as in the corresponding infix expression (this is generally the case in
1576: Forth).
1577:
1578: Parentheses are superfluous (and not available), because the order of
1579: the words unambiguously determines the order of evaluation and the
1580: operands:
1581:
1582: @example
1583: 3 4 + 5 * .
1584: 3 4 5 * + .
1585: @end example
1586:
1587: @assignment
1588: What are the infix expressions corresponding to the Forth code above?
1589: Write @code{6-7*8+9} in Forth notation@footnote{This notation is also
1590: known as Postfix or RPN (Reverse Polish Notation).}.
1591: @endassignment
1592:
1593: To change the sign, use @code{negate}:
1594:
1595: @example
1596: 2 negate .
1597: @end example
1598:
1599: @assignment
1600: Convert -(-3)*4-5 to Forth.
1601: @endassignment
1602:
1603: @code{/mod} performs both @code{/} and @code{mod}.
1604:
1605: @example
1606: 7 3 /mod . .
1607: @end example
1608:
1609: Reference: @ref{Arithmetic}.
1610:
1611:
1612: @node Stack Manipulation Tutorial, Using files for Forth code Tutorial, Arithmetics Tutorial, Tutorial
1613: @section Stack Manipulation
1614: @cindex stack manipulation tutorial
1615:
1616: Stack manipulation words rearrange the data on the stack.
1617:
1618: @example
1619: 1 .s drop .s
1620: 1 .s dup .s drop drop .s
1621: 1 2 .s over .s drop drop drop
1622: 1 2 .s swap .s drop drop
1623: 1 2 3 .s rot .s drop drop drop
1624: @end example
1625:
1626: These are the most important stack manipulation words. There are also
1627: variants that manipulate twice as many stack items:
1628:
1629: @example
1630: 1 2 3 4 .s 2swap .s 2drop 2drop
1631: @end example
1632:
1633: Two more stack manipulation words are:
1634:
1635: @example
1636: 1 2 .s nip .s drop
1637: 1 2 .s tuck .s 2drop drop
1638: @end example
1639:
1640: @assignment
1641: Replace @code{nip} and @code{tuck} with combinations of other stack
1642: manipulation words.
1643:
1644: @example
1645: Given: How do you get:
1646: 1 2 3 3 2 1
1647: 1 2 3 1 2 3 2
1648: 1 2 3 1 2 3 3
1649: 1 2 3 1 3 3
1650: 1 2 3 2 1 3
1651: 1 2 3 4 4 3 2 1
1652: 1 2 3 1 2 3 1 2 3
1653: 1 2 3 4 1 2 3 4 1 2
1654: 1 2 3
1655: 1 2 3 1 2 3 4
1656: 1 2 3 1 3
1657: @end example
1658: @endassignment
1659:
1660: @example
1661: 5 dup * .
1662: @end example
1663:
1664: @assignment
1665: Write 17^3 and 17^4 in Forth, without writing @code{17} more than once.
1666: Write a piece of Forth code that expects two numbers on the stack
1667: (@var{a} and @var{b}, with @var{b} on top) and computes
1668: @code{(a-b)(a+1)}.
1669: @endassignment
1670:
1671: Reference: @ref{Stack Manipulation}.
1672:
1673:
1674: @node Using files for Forth code Tutorial, Comments Tutorial, Stack Manipulation Tutorial, Tutorial
1675: @section Using files for Forth code
1676: @cindex loading Forth code, tutorial
1677: @cindex files containing Forth code, tutorial
1678:
1679: While working at the Forth command line is convenient for one-line
1680: examples and short one-off code, you probably want to store your source
1681: code in files for convenient editing and persistence. You can use your
1682: favourite editor (Gforth includes Emacs support, @pxref{Emacs and
1683: Gforth}) to create @var{file} and use
1684:
1685: @example
1686: s" @var{file}" included
1687: @end example
1688:
1689: to load it into your Forth system. The file name extension I use for
1690: Forth files is @samp{.fs}.
1691:
1692: You can easily start Gforth with some files loaded like this:
1693:
1694: @example
1695: gforth @var{file1} @var{file2}
1696: @end example
1697:
1698: If an error occurs during loading these files, Gforth terminates,
1699: whereas an error during @code{INCLUDED} within Gforth usually gives you
1700: a Gforth command line. Starting the Forth system every time gives you a
1701: clean start every time, without interference from the results of earlier
1702: tries.
1703:
1704: I often put all the tests in a file, then load the code and run the
1705: tests with
1706:
1707: @example
1708: gforth @var{code} @var{tests} -e bye
1709: @end example
1710:
1711: (often by performing this command with @kbd{C-x C-e} in Emacs). The
1712: @code{-e bye} ensures that Gforth terminates afterwards so that I can
1713: restart this command without ado.
1714:
1715: The advantage of this approach is that the tests can be repeated easily
1716: every time the program ist changed, making it easy to catch bugs
1717: introduced by the change.
1718:
1719: Reference: @ref{Forth source files}.
1720:
1721:
1722: @node Comments Tutorial, Colon Definitions Tutorial, Using files for Forth code Tutorial, Tutorial
1723: @section Comments
1724: @cindex comments tutorial
1725:
1726: @example
1727: \ That's a comment; it ends at the end of the line
1728: ( Another comment; it ends here: ) .s
1729: @end example
1730:
1731: @code{\} and @code{(} are ordinary Forth words and therefore have to be
1732: separated with white space from the following text.
1733:
1734: @example
1735: \This gives an "Undefined word" error
1736: @end example
1737:
1738: The first @code{)} ends a comment started with @code{(}, so you cannot
1739: nest @code{(}-comments; and you cannot comment out text containing a
1740: @code{)} with @code{( ... )}@footnote{therefore it's a good idea to
1741: avoid @code{)} in word names.}.
1742:
1743: I use @code{\}-comments for descriptive text and for commenting out code
1744: of one or more line; I use @code{(}-comments for describing the stack
1745: effect, the stack contents, or for commenting out sub-line pieces of
1746: code.
1747:
1748: The Emacs mode @file{gforth.el} (@pxref{Emacs and Gforth}) supports
1749: these uses by commenting out a region with @kbd{C-x \}, uncommenting a
1750: region with @kbd{C-u C-x \}, and filling a @code{\}-commented region
1751: with @kbd{M-q}.
1752:
1753: Reference: @ref{Comments}.
1754:
1755:
1756: @node Colon Definitions Tutorial, Decompilation Tutorial, Comments Tutorial, Tutorial
1757: @section Colon Definitions
1758: @cindex colon definitions, tutorial
1759: @cindex definitions, tutorial
1760: @cindex procedures, tutorial
1761: @cindex functions, tutorial
1762:
1763: are similar to procedures and functions in other programming languages.
1764:
1765: @example
1766: : squared ( n -- n^2 )
1767: dup * ;
1768: 5 squared .
1769: 7 squared .
1770: @end example
1771:
1772: @code{:} starts the colon definition; its name is @code{squared}. The
1773: following comment describes its stack effect. The words @code{dup *}
1774: are not executed, but compiled into the definition. @code{;} ends the
1775: colon definition.
1776:
1777: The newly-defined word can be used like any other word, including using
1778: it in other definitions:
1779:
1780: @example
1781: : cubed ( n -- n^3 )
1782: dup squared * ;
1783: -5 cubed .
1784: : fourth-power ( n -- n^4 )
1785: squared squared ;
1786: 3 fourth-power .
1787: @end example
1788:
1789: @assignment
1790: Write colon definitions for @code{nip}, @code{tuck}, @code{negate}, and
1791: @code{/mod} in terms of other Forth words, and check if they work (hint:
1792: test your tests on the originals first). Don't let the
1793: @samp{redefined}-Messages spook you, they are just warnings.
1794: @endassignment
1795:
1796: Reference: @ref{Colon Definitions}.
1797:
1798:
1799: @node Decompilation Tutorial, Stack-Effect Comments Tutorial, Colon Definitions Tutorial, Tutorial
1800: @section Decompilation
1801: @cindex decompilation tutorial
1802: @cindex see tutorial
1803:
1804: You can decompile colon definitions with @code{see}:
1805:
1806: @example
1807: see squared
1808: see cubed
1809: @end example
1810:
1811: In Gforth @code{see} shows you a reconstruction of the source code from
1812: the executable code. Informations that were present in the source, but
1813: not in the executable code, are lost (e.g., comments).
1814:
1815: You can also decompile the predefined words:
1816:
1817: @example
1818: see .
1819: see +
1820: @end example
1821:
1822:
1823: @node Stack-Effect Comments Tutorial, Types Tutorial, Decompilation Tutorial, Tutorial
1824: @section Stack-Effect Comments
1825: @cindex stack-effect comments, tutorial
1826: @cindex --, tutorial
1827: By convention the comment after the name of a definition describes the
1828: stack effect: The part in from of the @samp{--} describes the state of
1829: the stack before the execution of the definition, i.e., the parameters
1830: that are passed into the colon definition; the part behind the @samp{--}
1831: is the state of the stack after the execution of the definition, i.e.,
1832: the results of the definition. The stack comment only shows the top
1833: stack items that the definition accesses and/or changes.
1834:
1835: You should put a correct stack effect on every definition, even if it is
1836: just @code{( -- )}. You should also add some descriptive comment to
1837: more complicated words (I usually do this in the lines following
1838: @code{:}). If you don't do this, your code becomes unreadable (because
1839: you have to work through every definition before you can undertsand
1840: any).
1841:
1842: @assignment
1843: The stack effect of @code{swap} can be written like this: @code{x1 x2 --
1844: x2 x1}. Describe the stack effect of @code{-}, @code{drop}, @code{dup},
1845: @code{over}, @code{rot}, @code{nip}, and @code{tuck}. Hint: When you
1846: are done, you can compare your stack effects to those in this manual
1847: (@pxref{Word Index}).
1848: @endassignment
1849:
1850: Sometimes programmers put comments at various places in colon
1851: definitions that describe the contents of the stack at that place (stack
1852: comments); i.e., they are like the first part of a stack-effect
1853: comment. E.g.,
1854:
1855: @example
1856: : cubed ( n -- n^3 )
1857: dup squared ( n n^2 ) * ;
1858: @end example
1859:
1860: In this case the stack comment is pretty superfluous, because the word
1861: is simple enough. If you think it would be a good idea to add such a
1862: comment to increase readability, you should also consider factoring the
1863: word into several simpler words (@pxref{Factoring Tutorial,,
1864: Factoring}), which typically eliminates the need for the stack comment;
1865: however, if you decide not to refactor it, then having such a comment is
1866: better than not having it.
1867:
1868: The names of the stack items in stack-effect and stack comments in the
1869: standard, in this manual, and in many programs specify the type through
1870: a type prefix, similar to Fortran and Hungarian notation. The most
1871: frequent prefixes are:
1872:
1873: @table @code
1874: @item n
1875: signed integer
1876: @item u
1877: unsigned integer
1878: @item c
1879: character
1880: @item f
1881: Boolean flags, i.e. @code{false} or @code{true}.
1882: @item a-addr,a-
1883: Cell-aligned address
1884: @item c-addr,c-
1885: Char-aligned address (note that a Char may have two bytes in Windows NT)
1886: @item xt
1887: Execution token, same size as Cell
1888: @item w,x
1889: Cell, can contain an integer or an address. It usually takes 32, 64 or
1890: 16 bits (depending on your platform and Forth system). A cell is more
1891: commonly known as machine word, but the term @emph{word} already means
1892: something different in Forth.
1893: @item d
1894: signed double-cell integer
1895: @item ud
1896: unsigned double-cell integer
1897: @item r
1898: Float (on the FP stack)
1899: @end table
1900:
1901: You can find a more complete list in @ref{Notation}.
1902:
1903: @assignment
1904: Write stack-effect comments for all definitions you have written up to
1905: now.
1906: @endassignment
1907:
1908:
1909: @node Types Tutorial, Factoring Tutorial, Stack-Effect Comments Tutorial, Tutorial
1910: @section Types
1911: @cindex types tutorial
1912:
1913: In Forth the names of the operations are not overloaded; so similar
1914: operations on different types need different names; e.g., @code{+} adds
1915: integers, and you have to use @code{f+} to add floating-point numbers.
1916: The following prefixes are often used for related operations on
1917: different types:
1918:
1919: @table @code
1920: @item (none)
1921: signed integer
1922: @item u
1923: unsigned integer
1924: @item c
1925: character
1926: @item d
1927: signed double-cell integer
1928: @item ud, du
1929: unsigned double-cell integer
1930: @item 2
1931: two cells (not-necessarily double-cell numbers)
1932: @item m, um
1933: mixed single-cell and double-cell operations
1934: @item f
1935: floating-point (note that in stack comments @samp{f} represents flags,
1936: and @samp{r} represents FP numbers).
1937: @end table
1938:
1939: If there are no differences between the signed and the unsigned variant
1940: (e.g., for @code{+}), there is only the prefix-less variant.
1941:
1942: Forth does not perform type checking, neither at compile time, nor at
1943: run time. If you use the wrong oeration, the data are interpreted
1944: incorrectly:
1945:
1946: @example
1947: -1 u.
1948: @end example
1949:
1950: If you have only experience with type-checked languages until now, and
1951: have heard how important type-checking is, don't panic! In my
1952: experience (and that of other Forthers), type errors in Forth code are
1953: usually easy to find (once you get used to it), the increased vigilance
1954: of the programmer tends to catch some harder errors in addition to most
1955: type errors, and you never have to work around the type system, so in
1956: most situations the lack of type-checking seems to be a win (projects to
1957: add type checking to Forth have not caught on).
1958:
1959:
1960: @node Factoring Tutorial, Designing the stack effect Tutorial, Types Tutorial, Tutorial
1961: @section Factoring
1962: @cindex factoring tutorial
1963:
1964: If you try to write longer definitions, you will soon find it hard to
1965: keep track of the stack contents. Therefore, good Forth programmers
1966: tend to write only short definitions (e.g., three lines). The art of
1967: finding meaningful short definitions is known as factoring (as in
1968: factoring polynomials).
1969:
1970: Well-factored programs offer additional advantages: smaller, more
1971: general words, are easier to test and debug and can be reused more and
1972: better than larger, specialized words.
1973:
1974: So, if you run into difficulties with stack management, when writing
1975: code, try to define meaningful factors for the word, and define the word
1976: in terms of those. Even if a factor contains only two words, it is
1977: often helpful.
1978:
1979: Good factoring is not easy, and it takes some practice to get the knack
1980: for it; but even experienced Forth programmers often don't find the
1981: right solution right away, but only when rewriting the program. So, if
1982: you don't come up with a good solution immediately, keep trying, don't
1983: despair.
1984:
1985: @c example !!
1986:
1987:
1988: @node Designing the stack effect Tutorial, Local Variables Tutorial, Factoring Tutorial, Tutorial
1989: @section Designing the stack effect
1990: @cindex Stack effect design, tutorial
1991: @cindex design of stack effects, tutorial
1992:
1993: In other languages you can use an arbitrary order of parameters for a
1994: function; and since there is only one result, you don't have to deal with
1995: the order of results, either.
1996:
1997: In Forth (and other stack-based languages, e.g., Postscript) the
1998: parameter and result order of a definition is important and should be
1999: designed well. The general guideline is to design the stack effect such
2000: that the word is simple to use in most cases, even if that complicates
2001: the implementation of the word. Some concrete rules are:
2002:
2003: @itemize @bullet
2004:
2005: @item
2006: Words consume all of their parameters (e.g., @code{.}).
2007:
2008: @item
2009: If there is a convention on the order of parameters (e.g., from
2010: mathematics or another programming language), stick with it (e.g.,
2011: @code{-}).
2012:
2013: @item
2014: If one parameter usually requires only a short computation (e.g., it is
2015: a constant), pass it on the top of the stack. Conversely, parameters
2016: that usually require a long sequence of code to compute should be passed
2017: as the bottom (i.e., first) parameter. This makes the code easier to
2018: read, because reader does not need to keep track of the bottom item
2019: through a long sequence of code (or, alternatively, through stack
2020: manipulations). E.g., @code{!} (store, @pxref{Memory}) expects the
2021: address on top of the stack because it is usually simpler to compute
2022: than the stored value (often the address is just a variable).
2023:
2024: @item
2025: Similarly, results that are usually consumed quickly should be returned
2026: on the top of stack, whereas a result that is often used in long
2027: computations should be passed as bottom result. E.g., the file words
2028: like @code{open-file} return the error code on the top of stack, because
2029: it is usually consumed quickly by @code{throw}; moreover, the error code
2030: has to be checked before doing anything with the other results.
2031:
2032: @end itemize
2033:
2034: These rules are just general guidelines, don't lose sight of the overall
2035: goal to make the words easy to use. E.g., if the convention rule
2036: conflicts with the computation-length rule, you might decide in favour
2037: of the convention if the word will be used rarely, and in favour of the
2038: computation-length rule if the word will be used frequently (because
2039: with frequent use the cost of breaking the computation-length rule would
2040: be quite high, and frequent use makes it easier to remember an
2041: unconventional order).
2042:
2043: @c example !! structure package
2044:
2045:
2046: @node Local Variables Tutorial, Conditional execution Tutorial, Designing the stack effect Tutorial, Tutorial
2047: @section Local Variables
2048: @cindex local variables, tutorial
2049:
2050: You can define local variables (@emph{locals}) in a colon definition:
2051:
2052: @example
2053: : swap @{ a b -- b a @}
2054: b a ;
2055: 1 2 swap .s 2drop
2056: @end example
2057:
2058: (If your Forth system does not support this syntax, include
2059: @file{compat/anslocals.fs} first).
2060:
2061: In this example @code{@{ a b -- b a @}} is the locals definition; it
2062: takes two cells from the stack, puts the top of stack in @code{b} and
2063: the next stack element in @code{a}. @code{--} starts a comment ending
2064: with @code{@}}. After the locals definition, using the name of the
2065: local will push its value on the stack. You can leave the comment
2066: part (@code{-- b a}) away:
2067:
2068: @example
2069: : swap ( x1 x2 -- x2 x1 )
2070: @{ a b @} b a ;
2071: @end example
2072:
2073: In Gforth you can have several locals definitions, anywhere in a colon
2074: definition; in contrast, in a standard program you can have only one
2075: locals definition per colon definition, and that locals definition must
2076: be outside any controll structure.
2077:
2078: With locals you can write slightly longer definitions without running
2079: into stack trouble. However, I recommend trying to write colon
2080: definitions without locals for exercise purposes to help you gain the
2081: essential factoring skills.
2082:
2083: @assignment
2084: Rewrite your definitions until now with locals
2085: @endassignment
2086:
2087: Reference: @ref{Locals}.
2088:
2089:
2090: @node Conditional execution Tutorial, Flags and Comparisons Tutorial, Local Variables Tutorial, Tutorial
2091: @section Conditional execution
2092: @cindex conditionals, tutorial
2093: @cindex if, tutorial
2094:
2095: In Forth you can use control structures only inside colon definitions.
2096: An @code{if}-structure looks like this:
2097:
2098: @example
2099: : abs ( n1 -- +n2 )
2100: dup 0 < if
2101: negate
2102: endif ;
2103: 5 abs .
2104: -5 abs .
2105: @end example
2106:
2107: @code{if} takes a flag from the stack. If the flag is non-zero (true),
2108: the following code is performed, otherwise execution continues after the
2109: @code{endif} (or @code{else}). @code{<} compares the top two stack
2110: elements and prioduces a flag:
2111:
2112: @example
2113: 1 2 < .
2114: 2 1 < .
2115: 1 1 < .
2116: @end example
2117:
2118: Actually the standard name for @code{endif} is @code{then}. This
2119: tutorial presents the examples using @code{endif}, because this is often
2120: less confusing for people familiar with other programming languages
2121: where @code{then} has a different meaning. If your system does not have
2122: @code{endif}, define it with
2123:
2124: @example
2125: : endif postpone then ; immediate
2126: @end example
2127:
2128: You can optionally use an @code{else}-part:
2129:
2130: @example
2131: : min ( n1 n2 -- n )
2132: 2dup < if
2133: drop
2134: else
2135: nip
2136: endif ;
2137: 2 3 min .
2138: 3 2 min .
2139: @end example
2140:
2141: @assignment
2142: Write @code{min} without @code{else}-part (hint: what's the definition
2143: of @code{nip}?).
2144: @endassignment
2145:
2146: Reference: @ref{Selection}.
2147:
2148:
2149: @node Flags and Comparisons Tutorial, General Loops Tutorial, Conditional execution Tutorial, Tutorial
2150: @section Flags and Comparisons
2151: @cindex flags tutorial
2152: @cindex comparison tutorial
2153:
2154: In a false-flag all bits are clear (0 when interpreted as integer). In
2155: a canonical true-flag all bits are set (-1 as a twos-complement signed
2156: integer); in many contexts (e.g., @code{if}) any non-zero value is
2157: treated as true flag.
2158:
2159: @example
2160: false .
2161: true .
2162: true hex u. decimal
2163: @end example
2164:
2165: Comparison words produce canonical flags:
2166:
2167: @example
2168: 1 1 = .
2169: 1 0= .
2170: 0 1 < .
2171: 0 0 < .
2172: -1 1 u< . \ type error, u< interprets -1 as large unsigned number
2173: -1 1 < .
2174: @end example
2175:
2176: Gforth supports all combinations of the prefixes @code{0 u d d0 du f f0}
2177: (or none) and the comparisons @code{= <> < > <= >=}. Only a part of
2178: these combinations are standard (for details see the standard,
2179: @ref{Numeric comparison}, @ref{Floating Point} or @ref{Word Index}).
2180:
2181: You can use @code{and or xor invert} can be used as operations on
2182: canonical flags. Actually they are bitwise operations:
2183:
2184: @example
2185: 1 2 and .
2186: 1 2 or .
2187: 1 3 xor .
2188: 1 invert .
2189: @end example
2190:
2191: You can convert a zero/non-zero flag into a canonical flag with
2192: @code{0<>} (and complement it on the way with @code{0=}).
2193:
2194: @example
2195: 1 0= .
2196: 1 0<> .
2197: @end example
2198:
2199: You can use the all-bits-set feature of canonical flags and the bitwise
2200: operation of the Boolean operations to avoid @code{if}s:
2201:
2202: @example
2203: : foo ( n1 -- n2 )
2204: 0= if
2205: 14
2206: else
2207: 0
2208: endif ;
2209: 0 foo .
2210: 1 foo .
2211:
2212: : foo ( n1 -- n2 )
2213: 0= 14 and ;
2214: 0 foo .
2215: 1 foo .
2216: @end example
2217:
2218: @assignment
2219: Write @code{min} without @code{if}.
2220: @endassignment
2221:
2222: For reference, see @ref{Boolean Flags}, @ref{Numeric comparison}, and
2223: @ref{Bitwise operations}.
2224:
2225:
2226: @node General Loops Tutorial, Counted loops Tutorial, Flags and Comparisons Tutorial, Tutorial
2227: @section General Loops
2228: @cindex loops, indefinite, tutorial
2229:
2230: The endless loop is the most simple one:
2231:
2232: @example
2233: : endless ( -- )
2234: 0 begin
2235: dup . 1+
2236: again ;
2237: endless
2238: @end example
2239:
2240: Terminate this loop by pressing @kbd{Ctrl-C} (in Gforth). @code{begin}
2241: does nothing at run-time, @code{again} jumps back to @code{begin}.
2242:
2243: A loop with one exit at any place looks like this:
2244:
2245: @example
2246: : log2 ( +n1 -- n2 )
2247: \ logarithmus dualis of n1>0, rounded down to the next integer
2248: assert( dup 0> )
2249: 2/ 0 begin
2250: over 0> while
2251: 1+ swap 2/ swap
2252: repeat
2253: nip ;
2254: 7 log2 .
2255: 8 log2 .
2256: @end example
2257:
2258: At run-time @code{while} consumes a flag; if it is 0, execution
2259: continues behind the @code{repeat}; if the flag is non-zero, execution
2260: continues behind the @code{while}. @code{Repeat} jumps back to
2261: @code{begin}, just like @code{again}.
2262:
2263: In Forth there are many combinations/abbreviations, like @code{1+}.
2264: However, @code{2/} is not one of them; it shifts it's argument right by
2265: one bit (arithmetic shift right):
2266:
2267: @example
2268: -5 2 / .
2269: -5 2/ .
2270: @end example
2271:
2272: @code{assert(} is no standard word, but you can get it on systems other
2273: then Gforth by including @file{compat/assert.fs}. You can see what it
2274: does by trying
2275:
2276: @example
2277: 0 log2 .
2278: @end example
2279:
2280: Here's a loop with an exit at the end:
2281:
2282: @example
2283: : log2 ( +n1 -- n2 )
2284: \ logarithmus dualis of n1>0, rounded down to the next integer
2285: assert( dup 0 > )
2286: -1 begin
2287: 1+ swap 2/ swap
2288: over 0 <=
2289: until
2290: nip ;
2291: @end example
2292:
2293: @code{Until} consumes a flag; if it is non-zero, execution continues at
2294: the @code{begin}, otherwise after the @code{until}.
2295:
2296: @assignment
2297: Write a definition for computing the greatest common divisor.
2298: @endassignment
2299:
2300: Reference: @ref{Simple Loops}.
2301:
2302:
2303: @node Counted loops Tutorial, Recursion Tutorial, General Loops Tutorial, Tutorial
2304: @section Counted loops
2305: @cindex loops, counted, tutorial
2306:
2307: @example
2308: : ^ ( n1 u -- n )
2309: \ n = the uth power of u1
2310: 1 swap 0 u+do
2311: over *
2312: loop
2313: nip ;
2314: 3 2 ^ .
2315: 4 3 ^ .
2316: @end example
2317:
2318: @code{U+do} (from @file{compat/loops.fs}, if your Forth system doesn't
2319: have it) takes two numbers of the stack @code{( u3 u4 -- )}, and then
2320: performs the code between @code{u+do} and @code{loop} for @code{u3-u4}
2321: times (or not at all, if @code{u3-u4<0}).
2322:
2323: You can see the stack effect design rules at work in the stack effect of
2324: the loop start words: Since the start value of the loop is more
2325: frequently constant than the end value, the start value is passed on
2326: the top-of-stack.
2327:
2328: You can access the counter of a counted loop with @code{i}:
2329:
2330: @example
2331: : fac ( u -- u! )
2332: 1 swap 1+ 1 u+do
2333: i *
2334: loop ;
2335: 5 fac .
2336: 7 fac .
2337: @end example
2338:
2339: There is also @code{+do}, which expects signed numbers (important for
2340: deciding whether to enter the loop).
2341:
2342: @assignment
2343: Write a definition for computing the nth Fibonacci number.
2344: @endassignment
2345:
2346: You can also use increments other than 1:
2347:
2348: @example
2349: : up2 ( n1 n2 -- )
2350: +do
2351: i .
2352: 2 +loop ;
2353: 10 0 up2
2354:
2355: : down2 ( n1 n2 -- )
2356: -do
2357: i .
2358: 2 -loop ;
2359: 0 10 down2
2360: @end example
2361:
2362: Reference: @ref{Counted Loops}.
2363:
2364:
2365: @node Recursion Tutorial, Leaving definitions or loops Tutorial, Counted loops Tutorial, Tutorial
2366: @section Recursion
2367: @cindex recursion tutorial
2368:
2369: Usually the name of a definition is not visible in the definition; but
2370: earlier definitions are usually visible:
2371:
2372: @example
2373: 1 0 / . \ "Floating-point unidentified fault" in Gforth on most platforms
2374: : / ( n1 n2 -- n )
2375: dup 0= if
2376: -10 throw \ report division by zero
2377: endif
2378: / \ old version
2379: ;
2380: 1 0 /
2381: @end example
2382:
2383: For recursive definitions you can use @code{recursive} (non-standard) or
2384: @code{recurse}:
2385:
2386: @example
2387: : fac1 ( n -- n! ) recursive
2388: dup 0> if
2389: dup 1- fac1 *
2390: else
2391: drop 1
2392: endif ;
2393: 7 fac1 .
2394:
2395: : fac2 ( n -- n! )
2396: dup 0> if
2397: dup 1- recurse *
2398: else
2399: drop 1
2400: endif ;
2401: 8 fac2 .
2402: @end example
2403:
2404: @assignment
2405: Write a recursive definition for computing the nth Fibonacci number.
2406: @endassignment
2407:
2408: Reference (including indirect recursion): @xref{Calls and returns}.
2409:
2410:
2411: @node Leaving definitions or loops Tutorial, Return Stack Tutorial, Recursion Tutorial, Tutorial
2412: @section Leaving definitions or loops
2413: @cindex leaving definitions, tutorial
2414: @cindex leaving loops, tutorial
2415:
2416: @code{EXIT} exits the current definition right away. For every counted
2417: loop that is left in this way, an @code{UNLOOP} has to be performed
2418: before the @code{EXIT}:
2419:
2420: @c !! real examples
2421: @example
2422: : ...
2423: ... u+do
2424: ... if
2425: ... unloop exit
2426: endif
2427: ...
2428: loop
2429: ... ;
2430: @end example
2431:
2432: @code{LEAVE} leaves the innermost counted loop right away:
2433:
2434: @example
2435: : ...
2436: ... u+do
2437: ... if
2438: ... leave
2439: endif
2440: ...
2441: loop
2442: ... ;
2443: @end example
2444:
2445: @c !! example
2446:
2447: Reference: @ref{Calls and returns}, @ref{Counted Loops}.
2448:
2449:
2450: @node Return Stack Tutorial, Memory Tutorial, Leaving definitions or loops Tutorial, Tutorial
2451: @section Return Stack
2452: @cindex return stack tutorial
2453:
2454: In addition to the data stack Forth also has a second stack, the return
2455: stack; most Forth systems store the return addresses of procedure calls
2456: there (thus its name). Programmers can also use this stack:
2457:
2458: @example
2459: : foo ( n1 n2 -- )
2460: .s
2461: >r .s
2462: r@@ .
2463: >r .s
2464: r@@ .
2465: r> .
2466: r@@ .
2467: r> . ;
2468: 1 2 foo
2469: @end example
2470:
2471: @code{>r} takes an element from the data stack and pushes it onto the
2472: return stack; conversely, @code{r>} moves an elementm from the return to
2473: the data stack; @code{r@@} pushes a copy of the top of the return stack
2474: on the return stack.
2475:
2476: Forth programmers usually use the return stack for storing data
2477: temporarily, if using the data stack alone would be too complex, and
2478: factoring and locals are not an option:
2479:
2480: @example
2481: : 2swap ( x1 x2 x3 x4 -- x3 x4 x1 x2 )
2482: rot >r rot r> ;
2483: @end example
2484:
2485: The return address of the definition and the loop control parameters of
2486: counted loops usually reside on the return stack, so you have to take
2487: all items, that you have pushed on the return stack in a colon
2488: definition or counted loop, from the return stack before the definition
2489: or loop ends. You cannot access items that you pushed on the return
2490: stack outside some definition or loop within the definition of loop.
2491:
2492: If you miscount the return stack items, this usually ends in a crash:
2493:
2494: @example
2495: : crash ( n -- )
2496: >r ;
2497: 5 crash
2498: @end example
2499:
2500: You cannot mix using locals and using the return stack (according to the
2501: standard; Gforth has no problem). However, they solve the same
2502: problems, so this shouldn't be an issue.
2503:
2504: @assignment
2505: Can you rewrite any of the definitions you wrote until now in a better
2506: way using the return stack?
2507: @endassignment
2508:
2509: Reference: @ref{Return stack}.
2510:
2511:
2512: @node Memory Tutorial, Characters and Strings Tutorial, Return Stack Tutorial, Tutorial
2513: @section Memory
2514: @cindex memory access/allocation tutorial
2515:
2516: You can create a global variable @code{v} with
2517:
2518: @example
2519: variable v ( -- addr )
2520: @end example
2521:
2522: @code{v} pushes the address of a cell in memory on the stack. This cell
2523: was reserved by @code{variable}. You can use @code{!} (store) to store
2524: values into this cell and @code{@@} (fetch) to load the value from the
2525: stack into memory:
2526:
2527: @example
2528: v .
2529: 5 v ! .s
2530: v @@ .
2531: @end example
2532:
2533: You can see a raw dump of memory with @code{dump}:
2534:
2535: @example
2536: v 1 cells .s dump
2537: @end example
2538:
2539: @code{Cells ( n1 -- n2 )} gives you the number of bytes (or, more
2540: generally, address units (aus)) that @code{n1 cells} occupy. You can
2541: also reserve more memory:
2542:
2543: @example
2544: create v2 20 cells allot
2545: v2 20 cells dump
2546: @end example
2547:
2548: creates a word @code{v2} and reserves 20 uninitialized cells; the
2549: address pushed by @code{v2} points to the start of these 20 cells. You
2550: can use address arithmetic to access these cells:
2551:
2552: @example
2553: 3 v2 5 cells + !
2554: v2 20 cells dump
2555: @end example
2556:
2557: You can reserve and initialize memory with @code{,}:
2558:
2559: @example
2560: create v3
2561: 5 , 4 , 3 , 2 , 1 ,
2562: v3 @@ .
2563: v3 cell+ @@ .
2564: v3 2 cells + @@ .
2565: v3 5 cells dump
2566: @end example
2567:
2568: @assignment
2569: Write a definition @code{vsum ( addr u -- n )} that computes the sum of
2570: @code{u} cells, with the first of these cells at @code{addr}, the next
2571: one at @code{addr cell+} etc.
2572: @endassignment
2573:
2574: You can also reserve memory without creating a new word:
2575:
2576: @example
2577: here 10 cells allot .
2578: here .
2579: @end example
2580:
2581: @code{Here} pushes the start address of the memory area. You should
2582: store it somewhere, or you will have a hard time finding the memory area
2583: again.
2584:
2585: @code{Allot} manages dictionary memory. The dictionary memory contains
2586: the system's data structures for words etc. on Gforth and most other
2587: Forth systems. It is managed like a stack: You can free the memory that
2588: you have just @code{allot}ed with
2589:
2590: @example
2591: -10 cells allot
2592: here .
2593: @end example
2594:
2595: Note that you cannot do this if you have created a new word in the
2596: meantime (because then your @code{allot}ed memory is no longer on the
2597: top of the dictionary ``stack'').
2598:
2599: Alternatively, you can use @code{allocate} and @code{free} which allow
2600: freeing memory in any order:
2601:
2602: @example
2603: 10 cells allocate throw .s
2604: 20 cells allocate throw .s
2605: swap
2606: free throw
2607: free throw
2608: @end example
2609:
2610: The @code{throw}s deal with errors (e.g., out of memory).
2611:
2612: And there is also a
2613: @uref{http://www.complang.tuwien.ac.at/forth/garbage-collection.zip,
2614: garbage collector}, which eliminates the need to @code{free} memory
2615: explicitly.
2616:
2617: Reference: @ref{Memory}.
2618:
2619:
2620: @node Characters and Strings Tutorial, Alignment Tutorial, Memory Tutorial, Tutorial
2621: @section Characters and Strings
2622: @cindex strings tutorial
2623: @cindex characters tutorial
2624:
2625: On the stack characters take up a cell, like numbers. In memory they
2626: have their own size (one 8-bit byte on most systems), and therefore
2627: require their own words for memory access:
2628:
2629: @example
2630: create v4
2631: 104 c, 97 c, 108 c, 108 c, 111 c,
2632: v4 4 chars + c@@ .
2633: v4 5 chars dump
2634: @end example
2635:
2636: The preferred representation of strings on the stack is @code{addr
2637: u-count}, where @code{addr} is the address of the first character and
2638: @code{u-count} is the number of characters in the string.
2639:
2640: @example
2641: v4 5 type
2642: @end example
2643:
2644: You get a string constant with
2645:
2646: @example
2647: s" hello, world" .s
2648: type
2649: @end example
2650:
2651: Make sure you have a space between @code{s"} and the string; @code{s"}
2652: is a normal Forth word and must be delimited with white space (try what
2653: happens when you remove the space).
2654:
2655: However, this interpretive use of @code{s"} is quite restricted: the
2656: string exists only until the next call of @code{s"} (some Forth systems
2657: keep more than one of these strings, but usually they still have a
2658: limited lifetime).
2659:
2660: @example
2661: s" hello," s" world" .s
2662: type
2663: type
2664: @end example
2665:
2666: You can also use @code{s"} in a definition, and the resulting
2667: strings then live forever (well, for as long as the definition):
2668:
2669: @example
2670: : foo s" hello," s" world" ;
2671: foo .s
2672: type
2673: type
2674: @end example
2675:
2676: @assignment
2677: @code{Emit ( c -- )} types @code{c} as character (not a number).
2678: Implement @code{type ( addr u -- )}.
2679: @endassignment
2680:
2681: Reference: @ref{Memory Blocks}.
2682:
2683:
2684: @node Alignment Tutorial, Interpretation and Compilation Semantics and Immediacy Tutorial, Characters and Strings Tutorial, Tutorial
2685: @section Alignment
2686: @cindex alignment tutorial
2687: @cindex memory alignment tutorial
2688:
2689: On many processors cells have to be aligned in memory, if you want to
2690: access them with @code{@@} and @code{!} (and even if the processor does
2691: not require alignment, access to aligned cells is faster).
2692:
2693: @code{Create} aligns @code{here} (i.e., the place where the next
2694: allocation will occur, and that the @code{create}d word points to).
2695: Likewise, the memory produced by @code{allocate} starts at an aligned
2696: address. Adding a number of @code{cells} to an aligned address produces
2697: another aligned address.
2698:
2699: However, address arithmetic involving @code{char+} and @code{chars} can
2700: create an address that is not cell-aligned. @code{Aligned ( addr --
2701: a-addr )} produces the next aligned address:
2702:
2703: @example
2704: v3 char+ aligned .s @@ .
2705: v3 char+ .s @@ .
2706: @end example
2707:
2708: Similarly, @code{align} advances @code{here} to the next aligned
2709: address:
2710:
2711: @example
2712: create v5 97 c,
2713: here .
2714: align here .
2715: 1000 ,
2716: @end example
2717:
2718: Note that you should use aligned addresses even if your processor does
2719: not require them, if you want your program to be portable.
2720:
2721: Reference: @ref{Address arithmetic}.
2722:
2723:
2724: @node Interpretation and Compilation Semantics and Immediacy Tutorial, Execution Tokens Tutorial, Alignment Tutorial, Tutorial
2725: @section Interpretation and Compilation Semantics and Immediacy
2726: @cindex semantics tutorial
2727: @cindex interpretation semantics tutorial
2728: @cindex compilation semantics tutorial
2729: @cindex immediate, tutorial
2730:
2731: When a word is compiled, it behaves differently from being interpreted.
2732: E.g., consider @code{+}:
2733:
2734: @example
2735: 1 2 + .
2736: : foo + ;
2737: @end example
2738:
2739: These two behaviours are known as compilation and interpretation
2740: semantics. For normal words (e.g., @code{+}), the compilation semantics
2741: is to append the interpretation semantics to the currently defined word
2742: (@code{foo} in the example above). I.e., when @code{foo} is executed
2743: later, the interpretation semantics of @code{+} (i.e., adding two
2744: numbers) will be performed.
2745:
2746: However, there are words with non-default compilation semantics, e.g.,
2747: the control-flow words like @code{if}. You can use @code{immediate} to
2748: change the compilation semantics of the last defined word to be equal to
2749: the interpretation semantics:
2750:
2751: @example
2752: : [FOO] ( -- )
2753: 5 . ; immediate
2754:
2755: [FOO]
2756: : bar ( -- )
2757: [FOO] ;
2758: bar
2759: see bar
2760: @end example
2761:
2762: Two conventions to mark words with non-default compilation semnatics are
2763: names with brackets (more frequently used) and to write them all in
2764: upper case (less frequently used).
2765:
2766: In Gforth (and many other systems) you can also remove the
2767: interpretation semantics with @code{compile-only} (the compilation
2768: semantics is derived from the original interpretation semantics):
2769:
2770: @example
2771: : flip ( -- )
2772: 6 . ; compile-only \ but not immediate
2773: flip
2774:
2775: : flop ( -- )
2776: flip ;
2777: flop
2778: @end example
2779:
2780: In this example the interpretation semantics of @code{flop} is equal to
2781: the original interpretation semantics of @code{flip}.
2782:
2783: The text interpreter has two states: in interpret state, it performs the
2784: interpretation semantics of words it encounters; in compile state, it
2785: performs the compilation semantics of these words.
2786:
2787: Among other things, @code{:} switches into compile state, and @code{;}
2788: switches back to interpret state. They contain the factors @code{]}
2789: (switch to compile state) and @code{[} (switch to interpret state), that
2790: do nothing but switch the state.
2791:
2792: @example
2793: : xxx ( -- )
2794: [ 5 . ]
2795: ;
2796:
2797: xxx
2798: see xxx
2799: @end example
2800:
2801: These brackets are also the source of the naming convention mentioned
2802: above.
2803:
2804: Reference: @ref{Interpretation and Compilation Semantics}.
2805:
2806:
2807: @node Execution Tokens Tutorial, Exceptions Tutorial, Interpretation and Compilation Semantics and Immediacy Tutorial, Tutorial
2808: @section Execution Tokens
2809: @cindex execution tokens tutorial
2810: @cindex XT tutorial
2811:
2812: @code{' word} gives you the execution token (XT) of a word. The XT is a
2813: cell representing the interpretation semantics of a word. You can
2814: execute this semantics with @code{execute}:
2815:
2816: @example
2817: ' + .s
2818: 1 2 rot execute .
2819: @end example
2820:
2821: The XT is similar to a function pointer in C. However, parameter
2822: passing through the stack makes it a little more flexible:
2823:
2824: @example
2825: : map-array ( ... addr u xt -- ... )
2826: \ executes xt ( ... x -- ... ) for every element of the array starting
2827: \ at addr and containing u elements
2828: @{ xt @}
2829: cells over + swap ?do
2830: i @@ xt execute
2831: 1 cells +loop ;
2832:
2833: create a 3 , 4 , 2 , -1 , 4 ,
2834: a 5 ' . map-array .s
2835: 0 a 5 ' + map-array .
2836: s" max-n" environment? drop .s
2837: a 5 ' min map-array .
2838: @end example
2839:
2840: You can use map-array with the XTs of words that consume one element
2841: more than they produce. In theory you can also use it with other XTs,
2842: but the stack effect then depends on the size of the array, which is
2843: hard to understand.
2844:
2845: Since XTs are cell-sized, you can store them in memory and manipulate
2846: them on the stack like other cells. You can also compile the XT into a
2847: word with @code{compile,}:
2848:
2849: @example
2850: : foo1 ( n1 n2 -- n )
2851: [ ' + compile, ] ;
2852: see foo
2853: @end example
2854:
2855: This is non-standard, because @code{compile,} has no compilation
2856: semantics in the standard, but it works in good Forth systems. For the
2857: broken ones, use
2858:
2859: @example
2860: : [compile,] compile, ; immediate
2861:
2862: : foo1 ( n1 n2 -- n )
2863: [ ' + ] [compile,] ;
2864: see foo
2865: @end example
2866:
2867: @code{'} is a word with default compilation semantics; it parses the
2868: next word when its interpretation semantics are executed, not during
2869: compilation:
2870:
2871: @example
2872: : foo ( -- xt )
2873: ' ;
2874: see foo
2875: : bar ( ... "word" -- ... )
2876: ' execute ;
2877: see bar
2878: 1 2 bar + .
2879: @end example
2880:
2881: You often want to parse a word during compilation and compile its XT so
2882: it will be pushed on the stack at run-time. @code{[']} does this:
2883:
2884: @example
2885: : xt-+ ( -- xt )
2886: ['] + ;
2887: see xt-+
2888: 1 2 xt-+ execute .
2889: @end example
2890:
2891: Many programmers tend to see @code{'} and the word it parses as one
2892: unit, and expect it to behave like @code{[']} when compiled, and are
2893: confused by the actual behaviour. If you are, just remember that the
2894: Forth system just takes @code{'} as one unit and has no idea that it is
2895: a parsing word (attempts to convenience programmers in this issue have
2896: usually resulted in even worse pitfalls, see
2897: @uref{http://www.complang.tuwien.ac.at/papers/ertl98.ps.gz,
2898: @code{State}-smartness---Why it is evil and How to Exorcise it}).
2899:
2900: Note that the state of the interpreter does not come into play when
2901: creating and executing XTs. I.e., even when you execute @code{'} in
2902: compile state, it still gives you the interpretation semantics. And
2903: whatever that state is, @code{execute} performs the semantics
2904: represented by the XT (i.e., for XTs produced with @code{'} the
2905: interpretation semantics).
2906:
2907: Reference: @ref{Tokens for Words}.
2908:
2909:
2910: @node Exceptions Tutorial, Defining Words Tutorial, Execution Tokens Tutorial, Tutorial
2911: @section Exceptions
2912: @cindex exceptions tutorial
2913:
2914: @code{throw ( n -- )} causes an exception unless n is zero.
2915:
2916: @example
2917: 100 throw .s
2918: 0 throw .s
2919: @end example
2920:
2921: @code{catch ( ... xt -- ... n )} behaves similar to @code{execute}, but
2922: it catches exceptions and pushes the number of the exception on the
2923: stack (or 0, if the xt executed without exception). If there was an
2924: exception, the stacks have the same depth as when entering @code{catch}:
2925:
2926: @example
2927: .s
2928: 3 0 ' / catch .s
2929: 3 2 ' / catch .s
2930: @end example
2931:
2932: @assignment
2933: Try the same with @code{execute} instead of @code{catch}.
2934: @endassignment
2935:
2936: @code{Throw} always jumps to the dynamically next enclosing
2937: @code{catch}, even if it has to leave several call levels to achieve
2938: this:
2939:
2940: @example
2941: : foo 100 throw ;
2942: : foo1 foo ." after foo" ;
2943: : bar ['] foo1 catch ;
2944: bar .
2945: @end example
2946:
2947: It is often important to restore a value upon leaving a definition, even
2948: if the definition is left through an exception. You can ensure this
2949: like this:
2950:
2951: @example
2952: : ...
2953: save-x
2954: ['] word-changing-x catch ( ... n )
2955: restore-x
2956: ( ... n ) throw ;
2957: @end example
2958:
2959: Gforth provides an alternative syntax in addition to @code{catch}:
2960: @code{try ... recover ... endtry}. If the code between @code{try} and
2961: @code{recover} has an exception, the stack depths are restored, the
2962: exception number is pushed on the stack, and the code between
2963: @code{recover} and @code{endtry} is performed. E.g., the definition for
2964: @code{catch} is
2965:
2966: @example
2967: : catch ( x1 .. xn xt -- y1 .. ym 0 / z1 .. zn error ) \ exception
2968: try
2969: execute 0
2970: recover
2971: nip
2972: endtry ;
2973: @end example
2974:
2975: The equivalent to the restoration code above is
2976:
2977: @example
2978: : ...
2979: save-x
2980: try
2981: word-changing-x
2982: end-try
2983: restore-x
2984: throw ;
2985: @end example
2986:
2987: As you can see, the @code{recover} part is optional.
2988:
2989: Reference: @ref{Exception Handling}.
2990:
2991:
2992: @node Defining Words Tutorial, Arrays and Records Tutorial, Exceptions Tutorial, Tutorial
2993: @section Defining Words
2994: @cindex defining words tutorial
2995: @cindex does> tutorial
2996: @cindex create...does> tutorial
2997:
2998: @c before semantics?
2999:
3000: @code{:}, @code{create}, and @code{variable} are definition words: They
3001: define other words. @code{Constant} is another definition word:
3002:
3003: @example
3004: 5 constant foo
3005: foo .
3006: @end example
3007:
3008: You can also use the prefixes @code{2} (double-cell) and @code{f}
3009: (floating point) with @code{variable} and @code{constant}.
3010:
3011: You can also define your own defining words. E.g.:
3012:
3013: @example
3014: : variable ( "name" -- )
3015: create 0 , ;
3016: @end example
3017:
3018: You can also define defining words that create words that do something
3019: other than just producing their address:
3020:
3021: @example
3022: : constant ( n "name" -- )
3023: create ,
3024: does> ( -- n )
3025: ( addr ) @@ ;
3026:
3027: 5 constant foo
3028: foo .
3029: @end example
3030:
3031: The definition of @code{constant} above ends at the @code{does>}; i.e.,
3032: @code{does>} replaces @code{;}, but it also does something else: It
3033: changes the last defined word such that it pushes the address of the
3034: body of the word and then performs the code after the @code{does>}
3035: whenever it is called.
3036:
3037: In the example above, @code{constant} uses @code{,} to store 5 into the
3038: body of @code{foo}. When @code{foo} executes, it pushes the address of
3039: the body onto the stack, then (in the code after the @code{does>})
3040: fetches the 5 from there.
3041:
3042: The stack comment near the @code{does>} reflects the stack effect of the
3043: defined word, not the stack effect of the code after the @code{does>}
3044: (the difference is that the code expects the address of the body that
3045: the stack comment does not show).
3046:
3047: You can use these definition words to do factoring in cases that involve
3048: (other) definition words. E.g., a field offset is always added to an
3049: address. Instead of defining
3050:
3051: @example
3052: 2 cells constant offset-field1
3053: @end example
3054:
3055: and using this like
3056:
3057: @example
3058: ( addr ) offset-field1 +
3059: @end example
3060:
3061: you can define a definition word
3062:
3063: @example
3064: : simple-field ( n "name" -- )
3065: create ,
3066: does> ( n1 -- n1+n )
3067: ( addr ) @@ + ;
3068: @end example
3069:
3070: Definition and use of field offsets now look like this:
3071:
3072: @example
3073: 2 cells simple-field field1
3074: create mystruct 4 cells allot
3075: mystruct .s field1 .s drop
3076: @end example
3077:
3078: If you want to do something with the word without performing the code
3079: after the @code{does>}, you can access the body of a @code{create}d word
3080: with @code{>body ( xt -- addr )}:
3081:
3082: @example
3083: : value ( n "name" -- )
3084: create ,
3085: does> ( -- n1 )
3086: @@ ;
3087: : to ( n "name" -- )
3088: ' >body ! ;
3089:
3090: 5 value foo
3091: foo .
3092: 7 to foo
3093: foo .
3094: @end example
3095:
3096: @assignment
3097: Define @code{defer ( "name" -- )}, which creates a word that stores an
3098: XT (at the start the XT of @code{abort}), and upon execution
3099: @code{execute}s the XT. Define @code{is ( xt "name" -- )} that stores
3100: @code{xt} into @code{name}, a word defined with @code{defer}. Indirect
3101: recursion is one application of @code{defer}.
3102: @endassignment
3103:
3104: Reference: @ref{User-defined Defining Words}.
3105:
3106:
3107: @node Arrays and Records Tutorial, POSTPONE Tutorial, Defining Words Tutorial, Tutorial
3108: @section Arrays and Records
3109: @cindex arrays tutorial
3110: @cindex records tutorial
3111: @cindex structs tutorial
3112:
3113: Forth has no standard words for defining data structures such as arrays
3114: and records (structs in C terminology), but you can build them yourself
3115: based on address arithmetic. You can also define words for defining
3116: arrays and records (@pxref{Defining Words Tutorial,, Defining Words}).
3117:
3118: One of the first projects a Forth newcomer sets out upon when learning
3119: about defining words is an array defining word (possibly for
3120: n-dimensional arrays). Go ahead and do it, I did it, too; you will
3121: learn something from it. However, don't be disappointed when you later
3122: learn that you have little use for these words (inappropriate use would
3123: be even worse). I have not yet found a set of useful array words yet;
3124: the needs are just too diverse, and named, global arrays (the result of
3125: naive use of defining words) are often not flexible enough (e.g.,
3126: consider how to pass them as parameters). Another such project is a set
3127: of words to help dealing with strings.
3128:
3129: On the other hand, there is a useful set of record words, and it has
3130: been defined in @file{compat/struct.fs}; these words are predefined in
3131: Gforth. They are explained in depth elsewhere in this manual (see
3132: @pxref{Structures}). The @code{simple-field} example above is
3133: simplified variant of fields in this package.
3134:
3135:
3136: @node POSTPONE Tutorial, Literal Tutorial, Arrays and Records Tutorial, Tutorial
3137: @section @code{POSTPONE}
3138: @cindex postpone tutorial
3139:
3140: You can compile the compilation semantics (instead of compiling the
3141: interpretation semantics) of a word with @code{POSTPONE}:
3142:
3143: @example
3144: : MY-+ ( Compilation: -- ; Run-time of compiled code: n1 n2 -- n )
3145: POSTPONE + ; immediate
3146: : foo ( n1 n2 -- n )
3147: MY-+ ;
3148: 1 2 foo .
3149: see foo
3150: @end example
3151:
3152: During the definition of @code{foo} the text interpreter performs the
3153: compilation semantics of @code{MY-+}, which performs the compilation
3154: semantics of @code{+}, i.e., it compiles @code{+} into @code{foo}.
3155:
3156: This example also displays separate stack comments for the compilation
3157: semantics and for the stack effect of the compiled code. For words with
3158: default compilation semantics these stack effects are usually not
3159: displayed; the stack effect of the compilation semantics is always
3160: @code{( -- )} for these words, the stack effect for the compiled code is
3161: the stack effect of the interpretation semantics.
3162:
3163: Note that the state of the interpreter does not come into play when
3164: performing the compilation semantics in this way. You can also perform
3165: it interpretively, e.g.:
3166:
3167: @example
3168: : foo2 ( n1 n2 -- n )
3169: [ MY-+ ] ;
3170: 1 2 foo .
3171: see foo
3172: @end example
3173:
3174: However, there are some broken Forth systems where this does not always
3175: work, and therefore this practice was been declared non-standard in
3176: 1999.
3177: @c !! repair.fs
3178:
3179: Here is another example for using @code{POSTPONE}:
3180:
3181: @example
3182: : MY-- ( Compilation: -- ; Run-time of compiled code: n1 n2 -- n )
3183: POSTPONE negate POSTPONE + ; immediate compile-only
3184: : bar ( n1 n2 -- n )
3185: MY-- ;
3186: 2 1 bar .
3187: see bar
3188: @end example
3189:
3190: You can define @code{ENDIF} in this way:
3191:
3192: @example
3193: : ENDIF ( Compilation: orig -- )
3194: POSTPONE then ; immediate
3195: @end example
3196:
3197: @assignment
3198: Write @code{MY-2DUP} that has compilation semantics equivalent to
3199: @code{2dup}, but compiles @code{over over}.
3200: @endassignment
3201:
3202: @c !! @xref{Macros} for reference
3203:
3204:
3205: @node Literal Tutorial, Advanced macros Tutorial, POSTPONE Tutorial, Tutorial
3206: @section @code{Literal}
3207: @cindex literal tutorial
3208:
3209: You cannot @code{POSTPONE} numbers:
3210:
3211: @example
3212: : [FOO] POSTPONE 500 ; immediate
3213: @end example
3214:
3215: Instead, you can use @code{LITERAL (compilation: n --; run-time: -- n )}:
3216:
3217: @example
3218: : [FOO] ( compilation: --; run-time: -- n )
3219: 500 POSTPONE literal ; immediate
3220:
3221: : flip [FOO] ;
3222: flip .
3223: see flip
3224: @end example
3225:
3226: @code{LITERAL} consumes a number at compile-time (when it's compilation
3227: semantics are executed) and pushes it at run-time (when the code it
3228: compiled is executed). A frequent use of @code{LITERAL} is to compile a
3229: number computed at compile time into the current word:
3230:
3231: @example
3232: : bar ( -- n )
3233: [ 2 2 + ] literal ;
3234: see bar
3235: @end example
3236:
3237: @assignment
3238: Write @code{]L} which allows writing the example above as @code{: bar (
3239: -- n ) [ 2 2 + ]L ;}
3240: @endassignment
3241:
3242: @c !! @xref{Macros} for reference
3243:
3244:
3245: @node Advanced macros Tutorial, Compilation Tokens Tutorial, Literal Tutorial, Tutorial
3246: @section Advanced macros
3247: @cindex macros, advanced tutorial
3248: @cindex run-time code generation, tutorial
3249:
3250: Reconsider @code{map-array} from @ref{Execution Tokens Tutorial,,
3251: Execution Tokens}. It frequently performs @code{execute}, a relatively
3252: expensive operation in some Forth implementations. You can use
3253: @code{compile,} and @code{POSTPONE} to eliminate these @code{execute}s
3254: and produce a word that contains the word to be performed directly:
3255:
3256: @c use ]] ... [[
3257: @example
3258: : compile-map-array ( compilation: xt -- ; run-time: ... addr u -- ... )
3259: \ at run-time, execute xt ( ... x -- ... ) for each element of the
3260: \ array beginning at addr and containing u elements
3261: @{ xt @}
3262: POSTPONE cells POSTPONE over POSTPONE + POSTPONE swap POSTPONE ?do
3263: POSTPONE i POSTPONE @@ xt compile,
3264: 1 cells POSTPONE literal POSTPONE +loop ;
3265:
3266: : sum-array ( addr u -- n )
3267: 0 rot rot [ ' + compile-map-array ] ;
3268: see sum-array
3269: a 5 sum-array .
3270: @end example
3271:
3272: You can use the full power of Forth for generating the code; here's an
3273: example where the code is generated in a loop:
3274:
3275: @example
3276: : compile-vmul-step ( compilation: n --; run-time: n1 addr1 -- n2 addr2 )
3277: \ n2=n1+(addr1)*n, addr2=addr1+cell
3278: POSTPONE tuck POSTPONE @@
3279: POSTPONE literal POSTPONE * POSTPONE +
3280: POSTPONE swap POSTPONE cell+ ;
3281:
3282: : compile-vmul ( compilation: addr1 u -- ; run-time: addr2 -- n )
3283: \ n=v1*v2 (inner product), where the v_i are represented as addr_i u
3284: 0 postpone literal postpone swap
3285: [ ' compile-vmul-step compile-map-array ]
3286: postpone drop ;
3287: see compile-vmul
3288:
3289: : a-vmul ( addr -- n )
3290: \ n=a*v, where v is a vector that's as long as a and starts at addr
3291: [ a 5 compile-vmul ] ;
3292: see a-vmul
3293: a a-vmul .
3294: @end example
3295:
3296: This example uses @code{compile-map-array} to show off, but you could
3297: also use @code{map-array} instead (try it now!).
3298:
3299: You can use this technique for efficient multiplication of large
3300: matrices. In matrix multiplication, you multiply every line of one
3301: matrix with every column of the other matrix. You can generate the code
3302: for one line once, and use it for every column. The only downside of
3303: this technique is that it is cumbersome to recover the memory consumed
3304: by the generated code when you are done (and in more complicated cases
3305: it is not possible portably).
3306:
3307: @c !! @xref{Macros} for reference
3308:
3309:
3310: @node Compilation Tokens Tutorial, Wordlists and Search Order Tutorial, Advanced macros Tutorial, Tutorial
3311: @section Compilation Tokens
3312: @cindex compilation tokens, tutorial
3313: @cindex CT, tutorial
3314:
3315: This section is Gforth-specific. You can skip it.
3316:
3317: @code{' word compile,} compiles the interpretation semantics. For words
3318: with default compilation semantics this is the same as performing the
3319: compilation semantics. To represent the compilation semantics of other
3320: words (e.g., words like @code{if} that have no interpretation
3321: semantics), Gforth has the concept of a compilation token (CT,
3322: consisting of two cells), and words @code{comp'} and @code{[comp']}.
3323: You can perform the compilation semantics represented by a CT with
3324: @code{execute}:
3325:
3326: @example
3327: : foo2 ( n1 n2 -- n )
3328: [ comp' + execute ] ;
3329: see foo
3330: @end example
3331:
3332: You can compile the compilation semantics represented by a CT with
3333: @code{postpone,}:
3334:
3335: @example
3336: : foo3 ( -- )
3337: [ comp' + postpone, ] ;
3338: see foo3
3339: @end example
3340:
3341: @code{[ comp' word postpone, ]} is equivalent to @code{POSTPONE word}.
3342: @code{comp'} is particularly useful for words that have no
3343: interpretation semantics:
3344:
3345: @example
3346: ' if
3347: comp' if .s 2drop
3348: @end example
3349:
3350: Reference: @ref{Tokens for Words}.
3351:
3352:
3353: @node Wordlists and Search Order Tutorial, , Compilation Tokens Tutorial, Tutorial
3354: @section Wordlists and Search Order
3355: @cindex wordlists tutorial
3356: @cindex search order, tutorial
3357:
3358: The dictionary is not just a memory area that allows you to allocate
3359: memory with @code{allot}, it also contains the Forth words, arranged in
3360: several wordlists. When searching for a word in a wordlist,
3361: conceptually you start searching at the youngest and proceed towards
3362: older words (in reality most systems nowadays use hash-tables); i.e., if
3363: you define a word with the same name as an older word, the new word
3364: shadows the older word.
3365:
3366: Which wordlists are searched in which order is determined by the search
3367: order. You can display the search order with @code{order}. It displays
3368: first the search order, starting with the wordlist searched first, then
3369: it displays the wordlist that will contain newly defined words.
3370:
3371: You can create a new, empty wordlist with @code{wordlist ( -- wid )}:
3372:
3373: @example
3374: wordlist constant mywords
3375: @end example
3376:
3377: @code{Set-current ( wid -- )} sets the wordlist that will contain newly
3378: defined words (the @emph{current} wordlist):
3379:
3380: @example
3381: mywords set-current
3382: order
3383: @end example
3384:
3385: Gforth does not display a name for the wordlist in @code{mywords}
3386: because this wordlist was created anonymously with @code{wordlist}.
3387:
3388: You can get the current wordlist with @code{get-current ( -- wid)}. If
3389: you want to put something into a specific wordlist without overall
3390: effect on the current wordlist, this typically looks like this:
3391:
3392: @example
3393: get-current mywords set-current ( wid )
3394: create someword
3395: ( wid ) set-current
3396: @end example
3397:
3398: You can write the search order with @code{set-order ( wid1 .. widn n --
3399: )} and read it with @code{get-order ( -- wid1 .. widn n )}. The first
3400: searched wordlist is topmost.
3401:
3402: @example
3403: get-order mywords swap 1+ set-order
3404: order
3405: @end example
3406:
3407: Yes, the order of wordlists in the output of @code{order} is reversed
3408: from stack comments and the output of @code{.s} and thus unintuitive.
3409:
3410: @assignment
3411: Define @code{>order ( wid -- )} with adds @code{wid} as first searched
3412: wordlist to the search order. Define @code{previous ( -- )}, which
3413: removes the first searched wordlist from the search order. Experiment
3414: with boundary conditions (you will see some crashes or situations that
3415: are hard or impossible to leave).
3416: @endassignment
3417:
3418: The search order is a powerful foundation for providing features similar
3419: to Modula-2 modules and C++ namespaces. However, trying to modularize
3420: programs in this way has disadvantages for debugging and reuse/factoring
3421: that overcome the advantages in my experience (I don't do huge projects,
3422: though). These disadvantages are not so clear in other
3423: languages/programming environments, because these langauges are not so
3424: strong in debugging and reuse.
3425:
3426: @c !! example
3427:
3428: Reference: @ref{Word Lists}.
3429:
3430: @c ******************************************************************
3431: @node Introduction, Words, Tutorial, Top
3432: @comment node-name, next, previous, up
3433: @chapter An Introduction to ANS Forth
3434: @cindex Forth - an introduction
3435:
3436: The primary purpose of this manual is to document Gforth. However, since
3437: Forth is not a widely-known language and there is a lack of up-to-date
3438: teaching material, it seems worthwhile to provide some introductory
3439: material. For other sources of Forth-related
3440: information, see @ref{Forth-related information}.
3441:
3442: The examples in this section should work on any ANS Forth; the
3443: output shown was produced using Gforth. Each example attempts to
3444: reproduce the exact output that Gforth produces. If you try out the
3445: examples (and you should), what you should type is shown @kbd{like this}
3446: and Gforth's response is shown @code{like this}. The single exception is
3447: that, where the example shows @key{RET} it means that you should
3448: press the ``carriage return'' key. Unfortunately, some output formats for
3449: this manual cannot show the difference between @kbd{this} and
3450: @code{this} which will make trying out the examples harder (but not
3451: impossible).
3452:
3453: Forth is an unusual language. It provides an interactive development
3454: environment which includes both an interpreter and compiler. Forth
3455: programming style encourages you to break a problem down into many
3456: @cindex factoring
3457: small fragments (@dfn{factoring}), and then to develop and test each
3458: fragment interactively. Forth advocates assert that breaking the
3459: edit-compile-test cycle used by conventional programming languages can
3460: lead to great productivity improvements.
3461:
3462: @menu
3463: * Introducing the Text Interpreter::
3464: * Stacks and Postfix notation::
3465: * Your first definition::
3466: * How does that work?::
3467: * Forth is written in Forth::
3468: * Review - elements of a Forth system::
3469: * Where to go next::
3470: * Exercises::
3471: @end menu
3472:
3473: @comment ----------------------------------------------
3474: @node Introducing the Text Interpreter, Stacks and Postfix notation, Introduction, Introduction
3475: @section Introducing the Text Interpreter
3476: @cindex text interpreter
3477: @cindex outer interpreter
3478:
3479: @c IMO this is too detailed and the pace is too slow for
3480: @c an introduction. If you know German, take a look at
3481: @c http://www.complang.tuwien.ac.at/anton/lvas/skriptum-stack.html
3482: @c to see how I do it - anton
3483:
3484: @c nac-> Where I have accepted your comments 100% and modified the text
3485: @c accordingly, I have deleted your comments. Elsewhere I have added a
3486: @c response like this to attempt to rationalise what I have done. Of
3487: @c course, this is a very clumsy mechanism for something that would be
3488: @c done far more efficiently over a beer. Please delete any dialogue
3489: @c you consider closed.
3490:
3491: When you invoke the Forth image, you will see a startup banner printed
3492: and nothing else (if you have Gforth installed on your system, try
3493: invoking it now, by typing @kbd{gforth@key{RET}}). Forth is now running
3494: its command line interpreter, which is called the @dfn{Text Interpreter}
3495: (also known as the @dfn{Outer Interpreter}). (You will learn a lot
3496: about the text interpreter as you read through this chapter, for more
3497: detail @pxref{The Text Interpreter}).
3498:
3499: Although it's not obvious, Forth is actually waiting for your
3500: input. Type a number and press the @key{RET} key:
3501:
3502: @example
3503: @kbd{45@key{RET}} ok
3504: @end example
3505:
3506: Rather than give you a prompt to invite you to input something, the text
3507: interpreter prints a status message @i{after} it has processed a line
3508: of input. The status message in this case (``@code{ ok}'' followed by
3509: carriage-return) indicates that the text interpreter was able to process
3510: all of your input successfully. Now type something illegal:
3511:
3512: @example
3513: @kbd{qwer341@key{RET}}
3514: :1: Undefined word
3515: qwer341
3516: ^^^^^^^
3517: $400D2BA8 Bounce
3518: $400DBDA8 no.extensions
3519: @end example
3520:
3521: The exact text, other than the ``Undefined word'' may differ slightly on
3522: your system, but the effect is the same; when the text interpreter
3523: detects an error, it discards any remaining text on a line, resets
3524: certain internal state and prints an error message. For a detailed description of error messages see @ref{Error
3525: messages}.
3526:
3527: The text interpreter waits for you to press carriage-return, and then
3528: processes your input line. Starting at the beginning of the line, it
3529: breaks the line into groups of characters separated by spaces. For each
3530: group of characters in turn, it makes two attempts to do something:
3531:
3532: @itemize @bullet
3533: @item
3534: @cindex name dictionary
3535: It tries to treat it as a command. It does this by searching a @dfn{name
3536: dictionary}. If the group of characters matches an entry in the name
3537: dictionary, the name dictionary provides the text interpreter with
3538: information that allows the text interpreter perform some actions. In
3539: Forth jargon, we say that the group
3540: @cindex word
3541: @cindex definition
3542: @cindex execution token
3543: @cindex xt
3544: of characters names a @dfn{word}, that the dictionary search returns an
3545: @dfn{execution token (xt)} corresponding to the @dfn{definition} of the
3546: word, and that the text interpreter executes the xt. Often, the terms
3547: @dfn{word} and @dfn{definition} are used interchangeably.
3548: @item
3549: If the text interpreter fails to find a match in the name dictionary, it
3550: tries to treat the group of characters as a number in the current number
3551: base (when you start up Forth, the current number base is base 10). If
3552: the group of characters legitimately represents a number, the text
3553: interpreter pushes the number onto a stack (we'll learn more about that
3554: in the next section).
3555: @end itemize
3556:
3557: If the text interpreter is unable to do either of these things with any
3558: group of characters, it discards the group of characters and the rest of
3559: the line, then prints an error message. If the text interpreter reaches
3560: the end of the line without error, it prints the status message ``@code{ ok}''
3561: followed by carriage-return.
3562:
3563: This is the simplest command we can give to the text interpreter:
3564:
3565: @example
3566: @key{RET} ok
3567: @end example
3568:
3569: The text interpreter did everything we asked it to do (nothing) without
3570: an error, so it said that everything is ``@code{ ok}''. Try a slightly longer
3571: command:
3572:
3573: @example
3574: @kbd{12 dup fred dup@key{RET}}
3575: :1: Undefined word
3576: 12 dup fred dup
3577: ^^^^
3578: $400D2BA8 Bounce
3579: $400DBDA8 no.extensions
3580: @end example
3581:
3582: When you press the carriage-return key, the text interpreter starts to
3583: work its way along the line:
3584:
3585: @itemize @bullet
3586: @item
3587: When it gets to the space after the @code{2}, it takes the group of
3588: characters @code{12} and looks them up in the name
3589: dictionary@footnote{We can't tell if it found them or not, but assume
3590: for now that it did not}. There is no match for this group of characters
3591: in the name dictionary, so it tries to treat them as a number. It is
3592: able to do this successfully, so it puts the number, 12, ``on the stack''
3593: (whatever that means).
3594: @item
3595: The text interpreter resumes scanning the line and gets the next group
3596: of characters, @code{dup}. It looks it up in the name dictionary and
3597: (you'll have to take my word for this) finds it, and executes the word
3598: @code{dup} (whatever that means).
3599: @item
3600: Once again, the text interpreter resumes scanning the line and gets the
3601: group of characters @code{fred}. It looks them up in the name
3602: dictionary, but can't find them. It tries to treat them as a number, but
3603: they don't represent any legal number.
3604: @end itemize
3605:
3606: At this point, the text interpreter gives up and prints an error
3607: message. The error message shows exactly how far the text interpreter
3608: got in processing the line. In particular, it shows that the text
3609: interpreter made no attempt to do anything with the final character
3610: group, @code{dup}, even though we have good reason to believe that the
3611: text interpreter would have no problem looking that word up and
3612: executing it a second time.
3613:
3614:
3615: @comment ----------------------------------------------
3616: @node Stacks and Postfix notation, Your first definition, Introducing the Text Interpreter, Introduction
3617: @section Stacks, postfix notation and parameter passing
3618: @cindex text interpreter
3619: @cindex outer interpreter
3620:
3621: In procedural programming languages (like C and Pascal), the
3622: building-block of programs is the @dfn{function} or @dfn{procedure}. These
3623: functions or procedures are called with @dfn{explicit parameters}. For
3624: example, in C we might write:
3625:
3626: @example
3627: total = total + new_volume(length,height,depth);
3628: @end example
3629:
3630: @noindent
3631: where new_volume is a function-call to another piece of code, and total,
3632: length, height and depth are all variables. length, height and depth are
3633: parameters to the function-call.
3634:
3635: In Forth, the equivalent of the function or procedure is the
3636: @dfn{definition} and parameters are implicitly passed between
3637: definitions using a shared stack that is visible to the
3638: programmer. Although Forth does support variables, the existence of the
3639: stack means that they are used far less often than in most other
3640: programming languages. When the text interpreter encounters a number, it
3641: will place (@dfn{push}) it on the stack. There are several stacks (the
3642: actual number is implementation-dependent ...) and the particular stack
3643: used for any operation is implied unambiguously by the operation being
3644: performed. The stack used for all integer operations is called the @dfn{data
3645: stack} and, since this is the stack used most commonly, references to
3646: ``the data stack'' are often abbreviated to ``the stack''.
3647:
3648: The stacks have a last-in, first-out (LIFO) organisation. If you type:
3649:
3650: @example
3651: @kbd{1 2 3@key{RET}} ok
3652: @end example
3653:
3654: Then this instructs the text interpreter to placed three numbers on the
3655: (data) stack. An analogy for the behaviour of the stack is to take a
3656: pack of playing cards and deal out the ace (1), 2 and 3 into a pile on
3657: the table. The 3 was the last card onto the pile (``last-in'') and if
3658: you take a card off the pile then, unless you're prepared to fiddle a
3659: bit, the card that you take off will be the 3 (``first-out''). The
3660: number that will be first-out of the stack is called the @dfn{top of
3661: stack}, which
3662: @cindex TOS definition
3663: is often abbreviated to @dfn{TOS}.
3664:
3665: To understand how parameters are passed in Forth, consider the
3666: behaviour of the definition @code{+} (pronounced ``plus''). You will not
3667: be surprised to learn that this definition performs addition. More
3668: precisely, it adds two number together and produces a result. Where does
3669: it get the two numbers from? It takes the top two numbers off the
3670: stack. Where does it place the result? On the stack. You can act-out the
3671: behaviour of @code{+} with your playing cards like this:
3672:
3673: @itemize @bullet
3674: @item
3675: Pick up two cards from the stack on the table
3676: @item
3677: Stare at them intently and ask yourself ``what @i{is} the sum of these two
3678: numbers''
3679: @item
3680: Decide that the answer is 5
3681: @item
3682: Shuffle the two cards back into the pack and find a 5
3683: @item
3684: Put a 5 on the remaining ace that's on the table.
3685: @end itemize
3686:
3687: If you don't have a pack of cards handy but you do have Forth running,
3688: you can use the definition @code{.s} to show the current state of the stack,
3689: without affecting the stack. Type:
3690:
3691: @example
3692: @kbd{clearstack 1 2 3@key{RET}} ok
3693: @kbd{.s@key{RET}} <3> 1 2 3 ok
3694: @end example
3695:
3696: The text interpreter looks up the word @code{clearstack} and executes
3697: it; it tidies up the stack and removes any entries that may have been
3698: left on it by earlier examples. The text interpreter pushes each of the
3699: three numbers in turn onto the stack. Finally, the text interpreter
3700: looks up the word @code{.s} and executes it. The effect of executing
3701: @code{.s} is to print the ``<3>'' (the total number of items on the stack)
3702: followed by a list of all the items on the stack; the item on the far
3703: right-hand side is the TOS.
3704:
3705: You can now type:
3706:
3707: @example
3708: @kbd{+ .s@key{RET}} <2> 1 5 ok
3709: @end example
3710:
3711: @noindent
3712: which is correct; there are now 2 items on the stack and the result of
3713: the addition is 5.
3714:
3715: If you're playing with cards, try doing a second addition: pick up the
3716: two cards, work out that their sum is 6, shuffle them into the pack,
3717: look for a 6 and place that on the table. You now have just one item on
3718: the stack. What happens if you try to do a third addition? Pick up the
3719: first card, pick up the second card -- ah! There is no second card. This
3720: is called a @dfn{stack underflow} and consitutes an error. If you try to
3721: do the same thing with Forth it will report an error (probably a Stack
3722: Underflow or an Invalid Memory Address error).
3723:
3724: The opposite situation to a stack underflow is a @dfn{stack overflow},
3725: which simply accepts that there is a finite amount of storage space
3726: reserved for the stack. To stretch the playing card analogy, if you had
3727: enough packs of cards and you piled the cards up on the table, you would
3728: eventually be unable to add another card; you'd hit the ceiling. Gforth
3729: allows you to set the maximum size of the stacks. In general, the only
3730: time that you will get a stack overflow is because a definition has a
3731: bug in it and is generating data on the stack uncontrollably.
3732:
3733: There's one final use for the playing card analogy. If you model your
3734: stack using a pack of playing cards, the maximum number of items on
3735: your stack will be 52 (I assume you didn't use the Joker). The maximum
3736: @i{value} of any item on the stack is 13 (the King). In fact, the only
3737: possible numbers are positive integer numbers 1 through 13; you can't
3738: have (for example) 0 or 27 or 3.52 or -2. If you change the way you
3739: think about some of the cards, you can accommodate different
3740: numbers. For example, you could think of the Jack as representing 0,
3741: the Queen as representing -1 and the King as representing -2. Your
3742: @i{range} remains unchanged (you can still only represent a total of 13
3743: numbers) but the numbers that you can represent are -2 through 10.
3744:
3745: In that analogy, the limit was the amount of information that a single
3746: stack entry could hold, and Forth has a similar limit. In Forth, the
3747: size of a stack entry is called a @dfn{cell}. The actual size of a cell is
3748: implementation dependent and affects the maximum value that a stack
3749: entry can hold. A Standard Forth provides a cell size of at least
3750: 16-bits, and most desktop systems use a cell size of 32-bits.
3751:
3752: Forth does not do any type checking for you, so you are free to
3753: manipulate and combine stack items in any way you wish. A convenient way
3754: of treating stack items is as 2's complement signed integers, and that
3755: is what Standard words like @code{+} do. Therefore you can type:
3756:
3757: @example
3758: @kbd{-5 12 + .s@key{RET}} <1> 7 ok
3759: @end example
3760:
3761: If you use numbers and definitions like @code{+} in order to turn Forth
3762: into a great big pocket calculator, you will realise that it's rather
3763: different from a normal calculator. Rather than typing 2 + 3 = you had
3764: to type 2 3 + (ignore the fact that you had to use @code{.s} to see the
3765: result). The terminology used to describe this difference is to say that
3766: your calculator uses @dfn{Infix Notation} (parameters and operators are
3767: mixed) whilst Forth uses @dfn{Postfix Notation} (parameters and
3768: operators are separate), also called @dfn{Reverse Polish Notation}.
3769:
3770: Whilst postfix notation might look confusing to begin with, it has
3771: several important advantages:
3772:
3773: @itemize @bullet
3774: @item
3775: it is unambiguous
3776: @item
3777: it is more concise
3778: @item
3779: it fits naturally with a stack-based system
3780: @end itemize
3781:
3782: To examine these claims in more detail, consider these sums:
3783:
3784: @example
3785: 6 + 5 * 4 =
3786: 4 * 5 + 6 =
3787: @end example
3788:
3789: If you're just learning maths or your maths is very rusty, you will
3790: probably come up with the answer 44 for the first and 26 for the
3791: second. If you are a bit of a whizz at maths you will remember the
3792: @i{convention} that multiplication takes precendence over addition, and
3793: you'd come up with the answer 26 both times. To explain the answer 26
3794: to someone who got the answer 44, you'd probably rewrite the first sum
3795: like this:
3796:
3797: @example
3798: 6 + (5 * 4) =
3799: @end example
3800:
3801: If what you really wanted was to perform the addition before the
3802: multiplication, you would have to use parentheses to force it.
3803:
3804: If you did the first two sums on a pocket calculator you would probably
3805: get the right answers, unless you were very cautious and entered them using
3806: these keystroke sequences:
3807:
3808: 6 + 5 = * 4 =
3809: 4 * 5 = + 6 =
3810:
3811: Postfix notation is unambiguous because the order that the operators
3812: are applied is always explicit; that also means that parentheses are
3813: never required. The operators are @i{active} (the act of quoting the
3814: operator makes the operation occur) which removes the need for ``=''.
3815:
3816: The sum 6 + 5 * 4 can be written (in postfix notation) in two
3817: equivalent ways:
3818:
3819: @example
3820: 6 5 4 * + or:
3821: 5 4 * 6 +
3822: @end example
3823:
3824: An important thing that you should notice about this notation is that
3825: the @i{order} of the numbers does not change; if you want to subtract
3826: 2 from 10 you type @code{10 2 -}.
3827:
3828: The reason that Forth uses postfix notation is very simple to explain: it
3829: makes the implementation extremely simple, and it follows naturally from
3830: using the stack as a mechanism for passing parameters. Another way of
3831: thinking about this is to realise that all Forth definitions are
3832: @i{active}; they execute as they are encountered by the text
3833: interpreter. The result of this is that the syntax of Forth is trivially
3834: simple.
3835:
3836:
3837:
3838: @comment ----------------------------------------------
3839: @node Your first definition, How does that work?, Stacks and Postfix notation, Introduction
3840: @section Your first Forth definition
3841: @cindex first definition
3842:
3843: Until now, the examples we've seen have been trivial; we've just been
3844: using Forth as a bigger-than-pocket calculator. Also, each calculation
3845: we've shown has been a ``one-off'' -- to repeat it we'd need to type it in
3846: again@footnote{That's not quite true. If you press the up-arrow key on
3847: your keyboard you should be able to scroll back to any earlier command,
3848: edit it and re-enter it.} In this section we'll see how to add new
3849: words to Forth's vocabulary.
3850:
3851: The easiest way to create a new word is to use a @dfn{colon
3852: definition}. We'll define a few and try them out before worrying too
3853: much about how they work. Try typing in these examples; be careful to
3854: copy the spaces accurately:
3855:
3856: @example
3857: : add-two 2 + . ;
3858: : greet ." Hello and welcome" ;
3859: : demo 5 add-two ;
3860: @end example
3861:
3862: @noindent
3863: Now try them out:
3864:
3865: @example
3866: @kbd{greet@key{RET}} Hello and welcome ok
3867: @kbd{greet greet@key{RET}} Hello and welcomeHello and welcome ok
3868: @kbd{4 add-two@key{RET}} 6 ok
3869: @kbd{demo@key{RET}} 7 ok
3870: @kbd{9 greet demo add-two@key{RET}} Hello and welcome7 11 ok
3871: @end example
3872:
3873: The first new thing that we've introduced here is the pair of words
3874: @code{:} and @code{;}. These are used to start and terminate a new
3875: definition, respectively. The first word after the @code{:} is the name
3876: for the new definition.
3877:
3878: As you can see from the examples, a definition is built up of words that
3879: have already been defined; Forth makes no distinction between
3880: definitions that existed when you started the system up, and those that
3881: you define yourself.
3882:
3883: The examples also introduce the words @code{.} (dot), @code{."}
3884: (dot-quote) and @code{dup} (dewp). Dot takes the value from the top of
3885: the stack and displays it. It's like @code{.s} except that it only
3886: displays the top item of the stack and it is destructive; after it has
3887: executed, the number is no longer on the stack. There is always one
3888: space printed after the number, and no spaces before it. Dot-quote
3889: defines a string (a sequence of characters) that will be printed when
3890: the word is executed. The string can contain any printable characters
3891: except @code{"}. A @code{"} has a special function; it is not a Forth
3892: word but it acts as a delimiter (the way that delimiters work is
3893: described in the next section). Finally, @code{dup} duplicates the value
3894: at the top of the stack. Try typing @code{5 dup .s} to see what it does.
3895:
3896: We already know that the text interpreter searches through the
3897: dictionary to locate names. If you've followed the examples earlier, you
3898: will already have a definition called @code{add-two}. Lets try modifying
3899: it by typing in a new definition:
3900:
3901: @example
3902: @kbd{: add-two dup . ." + 2 =" 2 + . ;@key{RET}} redefined add-two ok
3903: @end example
3904:
3905: Forth recognised that we were defining a word that already exists, and
3906: printed a message to warn us of that fact. Let's try out the new
3907: definition:
3908:
3909: @example
3910: @kbd{9 add-two@key{RET}} 9 + 2 =11 ok
3911: @end example
3912:
3913: @noindent
3914: All that we've actually done here, though, is to create a new
3915: definition, with a particular name. The fact that there was already a
3916: definition with the same name did not make any difference to the way
3917: that the new definition was created (except that Forth printed a warning
3918: message). The old definition of add-two still exists (try @code{demo}
3919: again to see that this is true). Any new definition will use the new
3920: definition of @code{add-two}, but old definitions continue to use the
3921: version that already existed at the time that they were @code{compiled}.
3922:
3923: Before you go on to the next section, try defining and redefining some
3924: words of your own.
3925:
3926: @comment ----------------------------------------------
3927: @node How does that work?, Forth is written in Forth, Your first definition, Introduction
3928: @section How does that work?
3929: @cindex parsing words
3930:
3931: @c That's pretty deep (IMO way too deep) for an introduction. - anton
3932:
3933: @c Is it a good idea to talk about the interpretation semantics of a
3934: @c number? We don't have an xt to go along with it. - anton
3935:
3936: @c Now that I have eliminated execution semantics, I wonder if it would not
3937: @c be better to keep them (or add run-time semantics), to make it easier to
3938: @c explain what compilation semantics usually does. - anton
3939:
3940: @c nac-> I removed the term ``default compilation sematics'' from the
3941: @c introductory chapter. Removing ``execution semantics'' was making
3942: @c everything simpler to explain, then I think the use of this term made
3943: @c everything more complex again. I replaced it with ``default
3944: @c semantics'' (which is used elsewhere in the manual) by which I mean
3945: @c ``a definition that has neither the immediate nor the compile-only
3946: @c flag set''. I reworded big chunks of the ``how does that work''
3947: @c section (and, unusually for me, I think I even made it shorter!). See
3948: @c what you think -- I know I have not addressed your primary concern
3949: @c that it is too heavy-going for an introduction. From what I understood
3950: @c of your course notes it looks as though they might be a good framework.
3951: @c Things that I've tried to capture here are some things that came as a
3952: @c great revelation here when I first understood them. Also, I like the
3953: @c fact that a very simple code example shows up almost all of the issues
3954: @c that you need to understand to see how Forth works. That's unique and
3955: @c worthwhile to emphasise.
3956:
3957: Now we're going to take another look at the definition of @code{add-two}
3958: from the previous section. From our knowledge of the way that the text
3959: interpreter works, we would have expected this result when we tried to
3960: define @code{add-two}:
3961:
3962: @example
3963: @kbd{: add-two 2 + . ;@key{RET}}
3964: ^^^^^^^
3965: Error: Undefined word
3966: @end example
3967:
3968: The reason that this didn't happen is bound up in the way that @code{:}
3969: works. The word @code{:} does two special things. The first special
3970: thing that it does prevents the text interpreter from ever seeing the
3971: characters @code{add-two}. The text interpreter uses a variable called
3972: @cindex modifying >IN
3973: @code{>IN} (pronounced ``to-in'') to keep track of where it is in the
3974: input line. When it encounters the word @code{:} it behaves in exactly
3975: the same way as it does for any other word; it looks it up in the name
3976: dictionary, finds its xt and executes it. When @code{:} executes, it
3977: looks at the input buffer, finds the word @code{add-two} and advances the
3978: value of @code{>IN} to point past it. It then does some other stuff
3979: associated with creating the new definition (including creating an entry
3980: for @code{add-two} in the name dictionary). When the execution of @code{:}
3981: completes, control returns to the text interpreter, which is oblivious
3982: to the fact that it has been tricked into ignoring part of the input
3983: line.
3984:
3985: @cindex parsing words
3986: Words like @code{:} -- words that advance the value of @code{>IN} and so
3987: prevent the text interpreter from acting on the whole of the input line
3988: -- are called @dfn{parsing words}.
3989:
3990: @cindex @code{state} - effect on the text interpreter
3991: @cindex text interpreter - effect of state
3992: The second special thing that @code{:} does is change the value of a
3993: variable called @code{state}, which affects the way that the text
3994: interpreter behaves. When Gforth starts up, @code{state} has the value
3995: 0, and the text interpreter is said to be @dfn{interpreting}. During a
3996: colon definition (started with @code{:}), @code{state} is set to -1 and
3997: the text interpreter is said to be @dfn{compiling}.
3998:
3999: In this example, the text interpreter is compiling when it processes the
4000: string ``@code{2 + . ;}''. It still breaks the string down into
4001: character sequences in the same way. However, instead of pushing the
4002: number @code{2} onto the stack, it lays down (@dfn{compiles}) some magic
4003: into the definition of @code{add-two} that will make the number @code{2} get
4004: pushed onto the stack when @code{add-two} is @dfn{executed}. Similarly,
4005: the behaviours of @code{+} and @code{.} are also compiled into the
4006: definition.
4007:
4008: One category of words don't get compiled. These so-called @dfn{immediate
4009: words} get executed (performed @i{now}) regardless of whether the text
4010: interpreter is interpreting or compiling. The word @code{;} is an
4011: immediate word. Rather than being compiled into the definition, it
4012: executes. Its effect is to terminate the current definition, which
4013: includes changing the value of @code{state} back to 0.
4014:
4015: When you execute @code{add-two}, it has a @dfn{run-time effect} that is
4016: exactly the same as if you had typed @code{2 + . @key{RET}} outside of a
4017: definition.
4018:
4019: In Forth, every word or number can be described in terms of two
4020: properties:
4021:
4022: @itemize @bullet
4023: @item
4024: @cindex interpretation semantics
4025: Its @dfn{interpretation semantics} describe how it will behave when the
4026: text interpreter encounters it in @dfn{interpret} state. The
4027: interpretation semantics of a word are represented by an @dfn{execution
4028: token}.
4029: @item
4030: @cindex compilation semantics
4031: Its @dfn{compilation semantics} describe how it will behave when the
4032: text interpreter encounters it in @dfn{compile} state. The compilation
4033: semantics of a word are represented in an implementation-dependent way;
4034: Gforth uses a @dfn{compilation token}.
4035: @end itemize
4036:
4037: @noindent
4038: Numbers are always treated in a fixed way:
4039:
4040: @itemize @bullet
4041: @item
4042: When the number is @dfn{interpreted}, its behaviour is to push the
4043: number onto the stack.
4044: @item
4045: When the number is @dfn{compiled}, a piece of code is appended to the
4046: current definition that pushes the number when it runs. (In other words,
4047: the compilation semantics of a number are to postpone its interpretation
4048: semantics until the run-time of the definition that it is being compiled
4049: into.)
4050: @end itemize
4051:
4052: Words don't behave in such a regular way, but most have @i{default
4053: semantics} which means that they behave like this:
4054:
4055: @itemize @bullet
4056: @item
4057: The @dfn{interpretation semantics} of the word are to do something useful.
4058: @item
4059: The @dfn{compilation semantics} of the word are to append its
4060: @dfn{interpretation semantics} to the current definition (so that its
4061: run-time behaviour is to do something useful).
4062: @end itemize
4063:
4064: @cindex immediate words
4065: The actual behaviour of any particular word can be controlled by using
4066: the words @code{immediate} and @code{compile-only} when the word is
4067: defined. These words set flags in the name dictionary entry of the most
4068: recently defined word, and these flags are retrieved by the text
4069: interpreter when it finds the word in the name dictionary.
4070:
4071: A word that is marked as @dfn{immediate} has compilation semantics that
4072: are identical to its interpretation semantics. In other words, it
4073: behaves like this:
4074:
4075: @itemize @bullet
4076: @item
4077: The @dfn{interpretation semantics} of the word are to do something useful.
4078: @item
4079: The @dfn{compilation semantics} of the word are to do something useful
4080: (and actually the same thing); i.e., it is executed during compilation.
4081: @end itemize
4082:
4083: Marking a word as @dfn{compile-only} prohibits the text interpreter from
4084: performing the interpretation semantics of the word directly; an attempt
4085: to do so will generate an error. It is never necessary to use
4086: @code{compile-only} (and it is not even part of ANS Forth, though it is
4087: provided by many implementations) but it is good etiquette to apply it
4088: to a word that will not behave correctly (and might have unexpected
4089: side-effects) in interpret state. For example, it is only legal to use
4090: the conditional word @code{IF} within a definition. If you forget this
4091: and try to use it elsewhere, the fact that (in Gforth) it is marked as
4092: @code{compile-only} allows the text interpreter to generate a helpful
4093: error message rather than subjecting you to the consequences of your
4094: folly.
4095:
4096: This example shows the difference between an immediate and a
4097: non-immediate word:
4098:
4099: @example
4100: : show-state state @@ . ;
4101: : show-state-now show-state ; immediate
4102: : word1 show-state ;
4103: : word2 show-state-now ;
4104: @end example
4105:
4106: The word @code{immediate} after the definition of @code{show-state-now}
4107: makes that word an immediate word. These definitions introduce a new
4108: word: @code{@@} (pronounced ``fetch''). This word fetches the value of a
4109: variable, and leaves it on the stack. Therefore, the behaviour of
4110: @code{show-state} is to print a number that represents the current value
4111: of @code{state}.
4112:
4113: When you execute @code{word1}, it prints the number 0, indicating that
4114: the system is interpreting. When the text interpreter compiled the
4115: definition of @code{word1}, it encountered @code{show-state} whose
4116: compilation semantics are to append its interpretation semantics to the
4117: current definition. When you execute @code{word1}, it performs the
4118: interpretation semantics of @code{show-state}. At the time that @code{word1}
4119: (and therefore @code{show-state}) are executed, the system is
4120: interpreting.
4121:
4122: When you pressed @key{RET} after entering the definition of @code{word2},
4123: you should have seen the number -1 printed, followed by ``@code{
4124: ok}''. When the text interpreter compiled the definition of
4125: @code{word2}, it encountered @code{show-state-now}, an immediate word,
4126: whose compilation semantics are therefore to perform its interpretation
4127: semantics. It is executed straight away (even before the text
4128: interpreter has moved on to process another group of characters; the
4129: @code{;} in this example). The effect of executing it are to display the
4130: value of @code{state} @i{at the time that the definition of}
4131: @code{word2} @i{is being defined}. Printing -1 demonstrates that the
4132: system is compiling at this time. If you execute @code{word2} it does
4133: nothing at all.
4134:
4135: @cindex @code{."}, how it works
4136: Before leaving the subject of immediate words, consider the behaviour of
4137: @code{."} in the definition of @code{greet}, in the previous
4138: section. This word is both a parsing word and an immediate word. Notice
4139: that there is a space between @code{."} and the start of the text
4140: @code{Hello and welcome}, but that there is no space between the last
4141: letter of @code{welcome} and the @code{"} character. The reason for this
4142: is that @code{."} is a Forth word; it must have a space after it so that
4143: the text interpreter can identify it. The @code{"} is not a Forth word;
4144: it is a @dfn{delimiter}. The examples earlier show that, when the string
4145: is displayed, there is neither a space before the @code{H} nor after the
4146: @code{e}. Since @code{."} is an immediate word, it executes at the time
4147: that @code{greet} is defined. When it executes, its behaviour is to
4148: search forward in the input line looking for the delimiter. When it
4149: finds the delimiter, it updates @code{>IN} to point past the
4150: delimiter. It also compiles some magic code into the definition of
4151: @code{greet}; the xt of a run-time routine that prints a text string. It
4152: compiles the string @code{Hello and welcome} into memory so that it is
4153: available to be printed later. When the text interpreter gains control,
4154: the next word it finds in the input stream is @code{;} and so it
4155: terminates the definition of @code{greet}.
4156:
4157:
4158: @comment ----------------------------------------------
4159: @node Forth is written in Forth, Review - elements of a Forth system, How does that work?, Introduction
4160: @section Forth is written in Forth
4161: @cindex structure of Forth programs
4162:
4163: When you start up a Forth compiler, a large number of definitions
4164: already exist. In Forth, you develop a new application using bottom-up
4165: programming techniques to create new definitions that are defined in
4166: terms of existing definitions. As you create each definition you can
4167: test and debug it interactively.
4168:
4169: If you have tried out the examples in this section, you will probably
4170: have typed them in by hand; when you leave Gforth, your definitions will
4171: be lost. You can avoid this by using a text editor to enter Forth source
4172: code into a file, and then loading code from the file using
4173: @code{include} (@pxref{Forth source files}). A Forth source file is
4174: processed by the text interpreter, just as though you had typed it in by
4175: hand@footnote{Actually, there are some subtle differences -- see
4176: @ref{The Text Interpreter}.}.
4177:
4178: Gforth also supports the traditional Forth alternative to using text
4179: files for program entry (@pxref{Blocks}).
4180:
4181: In common with many, if not most, Forth compilers, most of Gforth is
4182: actually written in Forth. All of the @file{.fs} files in the
4183: installation directory@footnote{For example,
4184: @file{/usr/local/share/gforth...}} are Forth source files, which you can
4185: study to see examples of Forth programming.
4186:
4187: Gforth maintains a history file that records every line that you type to
4188: the text interpreter. This file is preserved between sessions, and is
4189: used to provide a command-line recall facility. If you enter long
4190: definitions by hand, you can use a text editor to paste them out of the
4191: history file into a Forth source file for reuse at a later time
4192: (for more information @pxref{Command-line editing}).
4193:
4194:
4195: @comment ----------------------------------------------
4196: @node Review - elements of a Forth system, Where to go next, Forth is written in Forth, Introduction
4197: @section Review - elements of a Forth system
4198: @cindex elements of a Forth system
4199:
4200: To summarise this chapter:
4201:
4202: @itemize @bullet
4203: @item
4204: Forth programs use @dfn{factoring} to break a problem down into small
4205: fragments called @dfn{words} or @dfn{definitions}.
4206: @item
4207: Forth program development is an interactive process.
4208: @item
4209: The main command loop that accepts input, and controls both
4210: interpretation and compilation, is called the @dfn{text interpreter}
4211: (also known as the @dfn{outer interpreter}).
4212: @item
4213: Forth has a very simple syntax, consisting of words and numbers
4214: separated by spaces or carriage-return characters. Any additional syntax
4215: is imposed by @dfn{parsing words}.
4216: @item
4217: Forth uses a stack to pass parameters between words. As a result, it
4218: uses postfix notation.
4219: @item
4220: To use a word that has previously been defined, the text interpreter
4221: searches for the word in the @dfn{name dictionary}.
4222: @item
4223: Words have @dfn{interpretation semantics} and @dfn{compilation semantics}.
4224: @item
4225: The text interpreter uses the value of @code{state} to select between
4226: the use of the @dfn{interpretation semantics} and the @dfn{compilation
4227: semantics} of a word that it encounters.
4228: @item
4229: The relationship between the @dfn{interpretation semantics} and
4230: @dfn{compilation semantics} for a word
4231: depend upon the way in which the word was defined (for example, whether
4232: it is an @dfn{immediate} word).
4233: @item
4234: Forth definitions can be implemented in Forth (called @dfn{high-level
4235: definitions}) or in some other way (usually a lower-level language and
4236: as a result often called @dfn{low-level definitions}, @dfn{code
4237: definitions} or @dfn{primitives}).
4238: @item
4239: Many Forth systems are implemented mainly in Forth.
4240: @end itemize
4241:
4242:
4243: @comment ----------------------------------------------
4244: @node Where to go next, Exercises, Review - elements of a Forth system, Introduction
4245: @section Where To Go Next
4246: @cindex where to go next
4247:
4248: Amazing as it may seem, if you have read (and understood) this far, you
4249: know almost all the fundamentals about the inner workings of a Forth
4250: system. You certainly know enough to be able to read and understand the
4251: rest of this manual and the ANS Forth document, to learn more about the
4252: facilities that Forth in general and Gforth in particular provide. Even
4253: scarier, you know almost enough to implement your own Forth system.
4254: However, that's not a good idea just yet... better to try writing some
4255: programs in Gforth.
4256:
4257: Forth has such a rich vocabulary that it can be hard to know where to
4258: start in learning it. This section suggests a few sets of words that are
4259: enough to write small but useful programs. Use the word index in this
4260: document to learn more about each word, then try it out and try to write
4261: small definitions using it. Start by experimenting with these words:
4262:
4263: @itemize @bullet
4264: @item
4265: Arithmetic: @code{+ - * / /MOD */ ABS INVERT}
4266: @item
4267: Comparison: @code{MIN MAX =}
4268: @item
4269: Logic: @code{AND OR XOR NOT}
4270: @item
4271: Stack manipulation: @code{DUP DROP SWAP OVER}
4272: @item
4273: Loops and decisions: @code{IF ELSE ENDIF ?DO I LOOP}
4274: @item
4275: Input/Output: @code{. ." EMIT CR KEY}
4276: @item
4277: Defining words: @code{: ; CREATE}
4278: @item
4279: Memory allocation words: @code{ALLOT ,}
4280: @item
4281: Tools: @code{SEE WORDS .S MARKER}
4282: @end itemize
4283:
4284: When you have mastered those, go on to:
4285:
4286: @itemize @bullet
4287: @item
4288: More defining words: @code{VARIABLE CONSTANT VALUE TO CREATE DOES>}
4289: @item
4290: Memory access: @code{@@ !}
4291: @end itemize
4292:
4293: When you have mastered these, there's nothing for it but to read through
4294: the whole of this manual and find out what you've missed.
4295:
4296: @comment ----------------------------------------------
4297: @node Exercises, , Where to go next, Introduction
4298: @section Exercises
4299: @cindex exercises
4300:
4301: TODO: provide a set of programming excercises linked into the stuff done
4302: already and into other sections of the manual. Provide solutions to all
4303: the exercises in a .fs file in the distribution.
4304:
4305: @c Get some inspiration from Starting Forth and Kelly&Spies.
4306:
4307: @c excercises:
4308: @c 1. take inches and convert to feet and inches.
4309: @c 2. take temperature and convert from fahrenheight to celcius;
4310: @c may need to care about symmetric vs floored??
4311: @c 3. take input line and do character substitution
4312: @c to encipher or decipher
4313: @c 4. as above but work on a file for in and out
4314: @c 5. take input line and convert to pig-latin
4315: @c
4316: @c thing of sets of things to exercise then come up with
4317: @c problems that need those things.
4318:
4319:
4320: @c ******************************************************************
4321: @node Words, Error messages, Introduction, Top
4322: @chapter Forth Words
4323: @cindex words
4324:
4325: @menu
4326: * Notation::
4327: * Case insensitivity::
4328: * Comments::
4329: * Boolean Flags::
4330: * Arithmetic::
4331: * Stack Manipulation::
4332: * Memory::
4333: * Control Structures::
4334: * Defining Words::
4335: * Interpretation and Compilation Semantics::
4336: * Tokens for Words::
4337: * The Text Interpreter::
4338: * Word Lists::
4339: * Environmental Queries::
4340: * Files::
4341: * Blocks::
4342: * Other I/O::
4343: * Programming Tools::
4344: * Assembler and Code Words::
4345: * Threading Words::
4346: * Locals::
4347: * Structures::
4348: * Object-oriented Forth::
4349: * Passing Commands to the OS::
4350: * Keeping track of Time::
4351: * Miscellaneous Words::
4352: @end menu
4353:
4354: @node Notation, Case insensitivity, Words, Words
4355: @section Notation
4356: @cindex notation of glossary entries
4357: @cindex format of glossary entries
4358: @cindex glossary notation format
4359: @cindex word glossary entry format
4360:
4361: The Forth words are described in this section in the glossary notation
4362: that has become a de-facto standard for Forth texts:
4363:
4364: @format
4365: @i{word} @i{Stack effect} @i{wordset} @i{pronunciation}
4366: @end format
4367: @i{Description}
4368:
4369: @table @var
4370: @item word
4371: The name of the word.
4372:
4373: @item Stack effect
4374: @cindex stack effect
4375: The stack effect is written in the notation @code{@i{before} --
4376: @i{after}}, where @i{before} and @i{after} describe the top of
4377: stack entries before and after the execution of the word. The rest of
4378: the stack is not touched by the word. The top of stack is rightmost,
4379: i.e., a stack sequence is written as it is typed in. Note that Gforth
4380: uses a separate floating point stack, but a unified stack
4381: notation. Also, return stack effects are not shown in @i{stack
4382: effect}, but in @i{Description}. The name of a stack item describes
4383: the type and/or the function of the item. See below for a discussion of
4384: the types.
4385:
4386: All words have two stack effects: A compile-time stack effect and a
4387: run-time stack effect. The compile-time stack-effect of most words is
4388: @i{ -- }. If the compile-time stack-effect of a word deviates from
4389: this standard behaviour, or the word does other unusual things at
4390: compile time, both stack effects are shown; otherwise only the run-time
4391: stack effect is shown.
4392:
4393: @cindex pronounciation of words
4394: @item pronunciation
4395: How the word is pronounced.
4396:
4397: @cindex wordset
4398: @cindex environment wordset
4399: @item wordset
4400: The ANS Forth standard is divided into several word sets. A standard
4401: system need not support all of them. Therefore, in theory, the fewer
4402: word sets your program uses the more portable it will be. However, we
4403: suspect that most ANS Forth systems on personal machines will feature
4404: all word sets. Words that are not defined in ANS Forth have
4405: @code{gforth} or @code{gforth-internal} as word set. @code{gforth}
4406: describes words that will work in future releases of Gforth;
4407: @code{gforth-internal} words are more volatile. Environmental query
4408: strings are also displayed like words; you can recognize them by the
4409: @code{environment} in the word set field.
4410:
4411: @item Description
4412: A description of the behaviour of the word.
4413: @end table
4414:
4415: @cindex types of stack items
4416: @cindex stack item types
4417: The type of a stack item is specified by the character(s) the name
4418: starts with:
4419:
4420: @table @code
4421: @item f
4422: @cindex @code{f}, stack item type
4423: Boolean flags, i.e. @code{false} or @code{true}.
4424: @item c
4425: @cindex @code{c}, stack item type
4426: Char
4427: @item w
4428: @cindex @code{w}, stack item type
4429: Cell, can contain an integer or an address
4430: @item n
4431: @cindex @code{n}, stack item type
4432: signed integer
4433: @item u
4434: @cindex @code{u}, stack item type
4435: unsigned integer
4436: @item d
4437: @cindex @code{d}, stack item type
4438: double sized signed integer
4439: @item ud
4440: @cindex @code{ud}, stack item type
4441: double sized unsigned integer
4442: @item r
4443: @cindex @code{r}, stack item type
4444: Float (on the FP stack)
4445: @item a-
4446: @cindex @code{a_}, stack item type
4447: Cell-aligned address
4448: @item c-
4449: @cindex @code{c_}, stack item type
4450: Char-aligned address (note that a Char may have two bytes in Windows NT)
4451: @item f-
4452: @cindex @code{f_}, stack item type
4453: Float-aligned address
4454: @item df-
4455: @cindex @code{df_}, stack item type
4456: Address aligned for IEEE double precision float
4457: @item sf-
4458: @cindex @code{sf_}, stack item type
4459: Address aligned for IEEE single precision float
4460: @item xt
4461: @cindex @code{xt}, stack item type
4462: Execution token, same size as Cell
4463: @item wid
4464: @cindex @code{wid}, stack item type
4465: Word list ID, same size as Cell
4466: @item ior, wior
4467: @cindex ior type description
4468: @cindex wior type description
4469: I/O result code, cell-sized. In Gforth, you can @code{throw} iors.
4470: @item f83name
4471: @cindex @code{f83name}, stack item type
4472: Pointer to a name structure
4473: @item "
4474: @cindex @code{"}, stack item type
4475: string in the input stream (not on the stack). The terminating character
4476: is a blank by default. If it is not a blank, it is shown in @code{<>}
4477: quotes.
4478: @end table
4479:
4480: @comment ----------------------------------------------
4481: @node Case insensitivity, Comments, Notation, Words
4482: @section Case insensitivity
4483: @cindex case sensitivity
4484: @cindex upper and lower case
4485:
4486: Gforth is case-insensitive; you can enter definitions and invoke
4487: Standard words using upper, lower or mixed case (however,
4488: @pxref{core-idef, Implementation-defined options, Implementation-defined
4489: options}).
4490:
4491: ANS Forth only @i{requires} implementations to recognise Standard words
4492: when they are typed entirely in upper case. Therefore, a Standard
4493: program must use upper case for all Standard words. You can use whatever
4494: case you like for words that you define, but in a Standard program you
4495: have to use the words in the same case that you defined them.
4496:
4497: Gforth supports case sensitivity through @code{table}s (case-sensitive
4498: wordlists, @pxref{Word Lists}).
4499:
4500: Two people have asked how to convert Gforth to be case-sensitive; while
4501: we think this is a bad idea, you can change all wordlists into tables
4502: like this:
4503:
4504: @example
4505: ' table-find forth-wordlist wordlist-map @ !
4506: @end example
4507:
4508: Note that you now have to type the predefined words in the same case
4509: that we defined them, which are varying. You may want to convert them
4510: to your favourite case before doing this operation (I won't explain how,
4511: because if you are even contemplating doing this, you'd better have
4512: enough knowledge of Forth systems to know this already).
4513:
4514: @node Comments, Boolean Flags, Case insensitivity, Words
4515: @section Comments
4516: @cindex comments
4517:
4518: Forth supports two styles of comment; the traditional @i{in-line} comment,
4519: @code{(} and its modern cousin, the @i{comment to end of line}; @code{\}.
4520:
4521:
4522: doc-(
4523: doc-\
4524: doc-\G
4525:
4526:
4527: @node Boolean Flags, Arithmetic, Comments, Words
4528: @section Boolean Flags
4529: @cindex Boolean flags
4530:
4531: A Boolean flag is cell-sized. A cell with all bits clear represents the
4532: flag @code{false} and a flag with all bits set represents the flag
4533: @code{true}. Words that check a flag (for example, @code{IF}) will treat
4534: a cell that has @i{any} bit set as @code{true}.
4535: @c on and off to Memory?
4536: @c true and false to "Bitwise operations" or "Numeric comparison"?
4537:
4538: doc-true
4539: doc-false
4540: doc-on
4541: doc-off
4542:
4543:
4544: @node Arithmetic, Stack Manipulation, Boolean Flags, Words
4545: @section Arithmetic
4546: @cindex arithmetic words
4547:
4548: @cindex division with potentially negative operands
4549: Forth arithmetic is not checked, i.e., you will not hear about integer
4550: overflow on addition or multiplication, you may hear about division by
4551: zero if you are lucky. The operator is written after the operands, but
4552: the operands are still in the original order. I.e., the infix @code{2-1}
4553: corresponds to @code{2 1 -}. Forth offers a variety of division
4554: operators. If you perform division with potentially negative operands,
4555: you do not want to use @code{/} or @code{/mod} with its undefined
4556: behaviour, but rather @code{fm/mod} or @code{sm/mod} (probably the
4557: former, @pxref{Mixed precision}).
4558: @comment TODO discuss the different division forms and the std approach
4559:
4560: @menu
4561: * Single precision::
4562: * Double precision:: Double-cell integer arithmetic
4563: * Bitwise operations::
4564: * Numeric comparison::
4565: * Mixed precision:: Operations with single and double-cell integers
4566: * Floating Point::
4567: @end menu
4568:
4569: @node Single precision, Double precision, Arithmetic, Arithmetic
4570: @subsection Single precision
4571: @cindex single precision arithmetic words
4572:
4573: @c !! cell undefined
4574:
4575: By default, numbers in Forth are single-precision integers that are one
4576: cell in size. They can be signed or unsigned, depending upon how you
4577: treat them. For the rules used by the text interpreter for recognising
4578: single-precision integers see @ref{Number Conversion}.
4579:
4580: These words are all defined for signed operands, but some of them also
4581: work for unsigned numbers: @code{+}, @code{1+}, @code{-}, @code{1-},
4582: @code{*}.
4583:
4584: doc-+
4585: doc-1+
4586: doc--
4587: doc-1-
4588: doc-*
4589: doc-/
4590: doc-mod
4591: doc-/mod
4592: doc-negate
4593: doc-abs
4594: doc-min
4595: doc-max
4596: doc-floored
4597:
4598:
4599: @node Double precision, Bitwise operations, Single precision, Arithmetic
4600: @subsection Double precision
4601: @cindex double precision arithmetic words
4602:
4603: For the rules used by the text interpreter for
4604: recognising double-precision integers, see @ref{Number Conversion}.
4605:
4606: A double precision number is represented by a cell pair, with the most
4607: significant cell at the TOS. It is trivial to convert an unsigned single
4608: to a double: simply push a @code{0} onto the TOS. Since numbers are
4609: represented by Gforth using 2's complement arithmetic, converting a
4610: signed single to a (signed) double requires sign-extension across the
4611: most significant cell. This can be achieved using @code{s>d}. The moral
4612: of the story is that you cannot convert a number without knowing whether
4613: it represents an unsigned or a signed number.
4614:
4615: These words are all defined for signed operands, but some of them also
4616: work for unsigned numbers: @code{d+}, @code{d-}.
4617:
4618: doc-s>d
4619: doc-d>s
4620: doc-d+
4621: doc-d-
4622: doc-dnegate
4623: doc-dabs
4624: doc-dmin
4625: doc-dmax
4626:
4627:
4628: @node Bitwise operations, Numeric comparison, Double precision, Arithmetic
4629: @subsection Bitwise operations
4630: @cindex bitwise operation words
4631:
4632:
4633: doc-and
4634: doc-or
4635: doc-xor
4636: doc-invert
4637: doc-lshift
4638: doc-rshift
4639: doc-2*
4640: doc-d2*
4641: doc-2/
4642: doc-d2/
4643:
4644:
4645: @node Numeric comparison, Mixed precision, Bitwise operations, Arithmetic
4646: @subsection Numeric comparison
4647: @cindex numeric comparison words
4648:
4649: Note that the words that compare for equality (@code{= <> 0= 0<> d= d<>
4650: d0= d0<>}) work for for both signed and unsigned numbers.
4651:
4652: doc-<
4653: doc-<=
4654: doc-<>
4655: doc-=
4656: doc->
4657: doc->=
4658:
4659: doc-0<
4660: doc-0<=
4661: doc-0<>
4662: doc-0=
4663: doc-0>
4664: doc-0>=
4665:
4666: doc-u<
4667: doc-u<=
4668: @c u<> and u= exist but are the same as <> and =
4669: @c doc-u<>
4670: @c doc-u=
4671: doc-u>
4672: doc-u>=
4673:
4674: doc-within
4675:
4676: doc-d<
4677: doc-d<=
4678: doc-d<>
4679: doc-d=
4680: doc-d>
4681: doc-d>=
4682:
4683: doc-d0<
4684: doc-d0<=
4685: doc-d0<>
4686: doc-d0=
4687: doc-d0>
4688: doc-d0>=
4689:
4690: doc-du<
4691: doc-du<=
4692: @c du<> and du= exist but are the same as d<> and d=
4693: @c doc-du<>
4694: @c doc-du=
4695: doc-du>
4696: doc-du>=
4697:
4698:
4699: @node Mixed precision, Floating Point, Numeric comparison, Arithmetic
4700: @subsection Mixed precision
4701: @cindex mixed precision arithmetic words
4702:
4703:
4704: doc-m+
4705: doc-*/
4706: doc-*/mod
4707: doc-m*
4708: doc-um*
4709: doc-m*/
4710: doc-um/mod
4711: doc-fm/mod
4712: doc-sm/rem
4713:
4714:
4715: @node Floating Point, , Mixed precision, Arithmetic
4716: @subsection Floating Point
4717: @cindex floating point arithmetic words
4718:
4719: For the rules used by the text interpreter for
4720: recognising floating-point numbers see @ref{Number Conversion}.
4721:
4722: Gforth has a separate floating point stack, but the documentation uses
4723: the unified notation.@footnote{It's easy to generate the separate
4724: notation from that by just separating the floating-point numbers out:
4725: e.g. @code{( n r1 u r2 -- r3 )} becomes @code{( n u -- ) ( F: r1 r2 --
4726: r3 )}.}
4727:
4728: @cindex floating-point arithmetic, pitfalls
4729: Floating point numbers have a number of unpleasant surprises for the
4730: unwary (e.g., floating point addition is not associative) and even a few
4731: for the wary. You should not use them unless you know what you are doing
4732: or you don't care that the results you get are totally bogus. If you
4733: want to learn about the problems of floating point numbers (and how to
4734: avoid them), you might start with @cite{David Goldberg,
4735: @uref{http://www.validgh.com/goldberg/paper.ps,What Every Computer
4736: Scientist Should Know About Floating-Point Arithmetic}, ACM Computing
4737: Surveys 23(1):5@minus{}48, March 1991}.
4738:
4739:
4740: doc-d>f
4741: doc-f>d
4742: doc-f+
4743: doc-f-
4744: doc-f*
4745: doc-f/
4746: doc-fnegate
4747: doc-fabs
4748: doc-fmax
4749: doc-fmin
4750: doc-floor
4751: doc-fround
4752: doc-f**
4753: doc-fsqrt
4754: doc-fexp
4755: doc-fexpm1
4756: doc-fln
4757: doc-flnp1
4758: doc-flog
4759: doc-falog
4760: doc-f2*
4761: doc-f2/
4762: doc-1/f
4763: doc-precision
4764: doc-set-precision
4765:
4766: @cindex angles in trigonometric operations
4767: @cindex trigonometric operations
4768: Angles in floating point operations are given in radians (a full circle
4769: has 2 pi radians).
4770:
4771: doc-fsin
4772: doc-fcos
4773: doc-fsincos
4774: doc-ftan
4775: doc-fasin
4776: doc-facos
4777: doc-fatan
4778: doc-fatan2
4779: doc-fsinh
4780: doc-fcosh
4781: doc-ftanh
4782: doc-fasinh
4783: doc-facosh
4784: doc-fatanh
4785: doc-pi
4786:
4787: @cindex equality of floats
4788: @cindex floating-point comparisons
4789: One particular problem with floating-point arithmetic is that comparison
4790: for equality often fails when you would expect it to succeed. For this
4791: reason approximate equality is often preferred (but you still have to
4792: know what you are doing). Also note that IEEE NaNs may compare
4793: differently from what you might expect. The comparison words are:
4794:
4795: doc-f~rel
4796: doc-f~abs
4797: doc-f~
4798: doc-f=
4799: doc-f<>
4800:
4801: doc-f<
4802: doc-f<=
4803: doc-f>
4804: doc-f>=
4805:
4806: doc-f0<
4807: doc-f0<=
4808: doc-f0<>
4809: doc-f0=
4810: doc-f0>
4811: doc-f0>=
4812:
4813:
4814: @node Stack Manipulation, Memory, Arithmetic, Words
4815: @section Stack Manipulation
4816: @cindex stack manipulation words
4817:
4818: @cindex floating-point stack in the standard
4819: Gforth maintains a number of separate stacks:
4820:
4821: @cindex data stack
4822: @cindex parameter stack
4823: @itemize @bullet
4824: @item
4825: A data stack (also known as the @dfn{parameter stack}) -- for
4826: characters, cells, addresses, and double cells.
4827:
4828: @cindex floating-point stack
4829: @item
4830: A floating point stack -- for holding floating point (FP) numbers.
4831:
4832: @cindex return stack
4833: @item
4834: A return stack -- for holding the return addresses of colon
4835: definitions and other (non-FP) data.
4836:
4837: @cindex locals stack
4838: @item
4839: A locals stack -- for holding local variables.
4840: @end itemize
4841:
4842: @menu
4843: * Data stack::
4844: * Floating point stack::
4845: * Return stack::
4846: * Locals stack::
4847: * Stack pointer manipulation::
4848: @end menu
4849:
4850: @node Data stack, Floating point stack, Stack Manipulation, Stack Manipulation
4851: @subsection Data stack
4852: @cindex data stack manipulation words
4853: @cindex stack manipulations words, data stack
4854:
4855:
4856: doc-drop
4857: doc-nip
4858: doc-dup
4859: doc-over
4860: doc-tuck
4861: doc-swap
4862: doc-pick
4863: doc-rot
4864: doc--rot
4865: doc-?dup
4866: doc-roll
4867: doc-2drop
4868: doc-2nip
4869: doc-2dup
4870: doc-2over
4871: doc-2tuck
4872: doc-2swap
4873: doc-2rot
4874:
4875:
4876: @node Floating point stack, Return stack, Data stack, Stack Manipulation
4877: @subsection Floating point stack
4878: @cindex floating-point stack manipulation words
4879: @cindex stack manipulation words, floating-point stack
4880:
4881: Whilst every sane Forth has a separate floating-point stack, it is not
4882: strictly required; an ANS Forth system could theoretically keep
4883: floating-point numbers on the data stack. As an additional difficulty,
4884: you don't know how many cells a floating-point number takes. It is
4885: reportedly possible to write words in a way that they work also for a
4886: unified stack model, but we do not recommend trying it. Instead, just
4887: say that your program has an environmental dependency on a separate
4888: floating-point stack.
4889:
4890: doc-floating-stack
4891:
4892: doc-fdrop
4893: doc-fnip
4894: doc-fdup
4895: doc-fover
4896: doc-ftuck
4897: doc-fswap
4898: doc-fpick
4899: doc-frot
4900:
4901:
4902: @node Return stack, Locals stack, Floating point stack, Stack Manipulation
4903: @subsection Return stack
4904: @cindex return stack manipulation words
4905: @cindex stack manipulation words, return stack
4906:
4907: @cindex return stack and locals
4908: @cindex locals and return stack
4909: A Forth system is allowed to keep local variables on the
4910: return stack. This is reasonable, as local variables usually eliminate
4911: the need to use the return stack explicitly. So, if you want to produce
4912: a standard compliant program and you are using local variables in a
4913: word, forget about return stack manipulations in that word (refer to the
4914: standard document for the exact rules).
4915:
4916: doc->r
4917: doc-r>
4918: doc-r@
4919: doc-rdrop
4920: doc-2>r
4921: doc-2r>
4922: doc-2r@
4923: doc-2rdrop
4924:
4925:
4926: @node Locals stack, Stack pointer manipulation, Return stack, Stack Manipulation
4927: @subsection Locals stack
4928:
4929: Gforth uses an extra locals stack. It is described, along with the
4930: reasons for its existence, in @ref{Implementation,Implementation of locals}.
4931:
4932: @node Stack pointer manipulation, , Locals stack, Stack Manipulation
4933: @subsection Stack pointer manipulation
4934: @cindex stack pointer manipulation words
4935:
4936: @c removed s0 r0 l0 -- they are obsolete aliases for sp0 rp0 lp0
4937: doc-sp0
4938: doc-sp@
4939: doc-sp!
4940: doc-fp0
4941: doc-fp@
4942: doc-fp!
4943: doc-rp0
4944: doc-rp@
4945: doc-rp!
4946: doc-lp0
4947: doc-lp@
4948: doc-lp!
4949:
4950:
4951: @node Memory, Control Structures, Stack Manipulation, Words
4952: @section Memory
4953: @cindex memory words
4954:
4955: @menu
4956: * Memory model::
4957: * Dictionary allocation::
4958: * Heap Allocation::
4959: * Memory Access::
4960: * Address arithmetic::
4961: * Memory Blocks::
4962: @end menu
4963:
4964: In addition to the standard Forth memory allocation words, there is also
4965: a @uref{http://www.complang.tuwien.ac.at/forth/garbage-collection.zip,
4966: garbage collector}.
4967:
4968: @node Memory model, Dictionary allocation, Memory, Memory
4969: @subsection ANS Forth and Gforth memory models
4970:
4971: @c The ANS Forth description is a mess (e.g., is the heap part of
4972: @c the dictionary?), so let's not stick to closely with it.
4973:
4974: ANS Forth considers a Forth system as consisting of several address
4975: spaces, of which only @dfn{data space} is managed and accessible with
4976: the memory words. Memory not necessarily in data space includes the
4977: stacks, the code (called code space) and the headers (called name
4978: space). In Gforth everything is in data space, but the code for the
4979: primitives is usually read-only.
4980:
4981: Data space is divided into a number of areas: The (data space portion of
4982: the) dictionary@footnote{Sometimes, the term @dfn{dictionary} is used to
4983: refer to the search data structure embodied in word lists and headers,
4984: because it is used for looking up names, just as you would in a
4985: conventional dictionary.}, the heap, and a number of system-allocated
4986: buffers.
4987:
4988: @cindex address arithmetic restrictions, ANS vs. Gforth
4989: @cindex contiguous regions, ANS vs. Gforth
4990: In ANS Forth data space is also divided into contiguous regions. You
4991: can only use address arithmetic within a contiguous region, not between
4992: them. Usually each allocation gives you one contiguous region, but the
4993: dictionary allocation words have additional rules (@pxref{Dictionary
4994: allocation}).
4995:
4996: Gforth provides one big address space, and address arithmetic can be
4997: performed between any addresses. However, in the dictionary headers or
4998: code are interleaved with data, so almost the only contiguous data space
4999: regions there are those described by ANS Forth as contiguous; but you
5000: can be sure that the dictionary is allocated towards increasing
5001: addresses even between contiguous regions. The memory order of
5002: allocations in the heap is platform-dependent (and possibly different
5003: from one run to the next).
5004:
5005:
5006: @node Dictionary allocation, Heap Allocation, Memory model, Memory
5007: @subsection Dictionary allocation
5008: @cindex reserving data space
5009: @cindex data space - reserving some
5010:
5011: Dictionary allocation is a stack-oriented allocation scheme, i.e., if
5012: you want to deallocate X, you also deallocate everything
5013: allocated after X.
5014:
5015: @cindex contiguous regions in dictionary allocation
5016: The allocations using the words below are contiguous and grow the region
5017: towards increasing addresses. Other words that allocate dictionary
5018: memory of any kind (i.e., defining words including @code{:noname}) end
5019: the contiguous region and start a new one.
5020:
5021: In ANS Forth only @code{create}d words are guaranteed to produce an
5022: address that is the start of the following contiguous region. In
5023: particular, the cell allocated by @code{variable} is not guaranteed to
5024: be contiguous with following @code{allot}ed memory.
5025:
5026: You can deallocate memory by using @code{allot} with a negative argument
5027: (with some restrictions, see @code{allot}). For larger deallocations use
5028: @code{marker}.
5029:
5030:
5031: doc-here
5032: doc-unused
5033: doc-allot
5034: doc-c,
5035: doc-f,
5036: doc-,
5037: doc-2,
5038:
5039: Memory accesses have to be aligned (@pxref{Address arithmetic}). So of
5040: course you should allocate memory in an aligned way, too. I.e., before
5041: allocating allocating a cell, @code{here} must be cell-aligned, etc.
5042: The words below align @code{here} if it is not already. Basically it is
5043: only already aligned for a type, if the last allocation was a multiple
5044: of the size of this type and if @code{here} was aligned for this type
5045: before.
5046:
5047: After freshly @code{create}ing a word, @code{here} is @code{align}ed in
5048: ANS Forth (@code{maxalign}ed in Gforth).
5049:
5050: doc-align
5051: doc-falign
5052: doc-sfalign
5053: doc-dfalign
5054: doc-maxalign
5055: doc-cfalign
5056:
5057:
5058: @node Heap Allocation, Memory Access, Dictionary allocation, Memory
5059: @subsection Heap allocation
5060: @cindex heap allocation
5061: @cindex dynamic allocation of memory
5062: @cindex memory-allocation word set
5063:
5064: @cindex contiguous regions and heap allocation
5065: Heap allocation supports deallocation of allocated memory in any
5066: order. Dictionary allocation is not affected by it (i.e., it does not
5067: end a contiguous region). In Gforth, these words are implemented using
5068: the standard C library calls malloc(), free() and resize().
5069:
5070: The memory region produced by one invocation of @code{allocate} or
5071: @code{resize} is internally contiguous. There is no contiguity between
5072: such a region and any other region (including others allocated from the
5073: heap).
5074:
5075: doc-allocate
5076: doc-free
5077: doc-resize
5078:
5079:
5080: @node Memory Access, Address arithmetic, Heap Allocation, Memory
5081: @subsection Memory Access
5082: @cindex memory access words
5083:
5084: doc-@
5085: doc-!
5086: doc-+!
5087: doc-c@
5088: doc-c!
5089: doc-2@
5090: doc-2!
5091: doc-f@
5092: doc-f!
5093: doc-sf@
5094: doc-sf!
5095: doc-df@
5096: doc-df!
5097:
5098:
5099: @node Address arithmetic, Memory Blocks, Memory Access, Memory
5100: @subsection Address arithmetic
5101: @cindex address arithmetic words
5102:
5103: Address arithmetic is the foundation on which you can build data
5104: structures like arrays, records (@pxref{Structures}) and objects
5105: (@pxref{Object-oriented Forth}).
5106:
5107: @cindex address unit
5108: @cindex au (address unit)
5109: ANS Forth does not specify the sizes of the data types. Instead, it
5110: offers a number of words for computing sizes and doing address
5111: arithmetic. Address arithmetic is performed in terms of address units
5112: (aus); on most systems the address unit is one byte. Note that a
5113: character may have more than one au, so @code{chars} is no noop (on
5114: platforms where it is a noop, it compiles to nothing).
5115:
5116: The basic address arithmetic words are @code{+} and @code{-}. E.g., if
5117: you have the address of a cell, perform @code{1 cells +}, and you will
5118: have the address of the next cell.
5119:
5120: @cindex contiguous regions and address arithmetic
5121: In ANS Forth you can perform address arithmetic only within a contiguous
5122: region, i.e., if you have an address into one region, you can only add
5123: and subtract such that the result is still within the region; you can
5124: only subtract or compare addresses from within the same contiguous
5125: region. Reasons: several contiguous regions can be arranged in memory
5126: in any way; on segmented systems addresses may have unusual
5127: representations, such that address arithmetic only works within a
5128: region. Gforth provides a few more guarantees (linear address space,
5129: dictionary grows upwards), but in general I have found it easy to stay
5130: within contiguous regions (exception: computing and comparing to the
5131: address just beyond the end of an array).
5132:
5133: @cindex alignment of addresses for types
5134: ANS Forth also defines words for aligning addresses for specific
5135: types. Many computers require that accesses to specific data types
5136: must only occur at specific addresses; e.g., that cells may only be
5137: accessed at addresses divisible by 4. Even if a machine allows unaligned
5138: accesses, it can usually perform aligned accesses faster.
5139:
5140: For the performance-conscious: alignment operations are usually only
5141: necessary during the definition of a data structure, not during the
5142: (more frequent) accesses to it.
5143:
5144: ANS Forth defines no words for character-aligning addresses. This is not
5145: an oversight, but reflects the fact that addresses that are not
5146: char-aligned have no use in the standard and therefore will not be
5147: created.
5148:
5149: @cindex @code{CREATE} and alignment
5150: ANS Forth guarantees that addresses returned by @code{CREATE}d words
5151: are cell-aligned; in addition, Gforth guarantees that these addresses
5152: are aligned for all purposes.
5153:
5154: Note that the ANS Forth word @code{char} has nothing to do with address
5155: arithmetic.
5156:
5157:
5158: doc-chars
5159: doc-char+
5160: doc-cells
5161: doc-cell+
5162: doc-cell
5163: doc-aligned
5164: doc-floats
5165: doc-float+
5166: doc-float
5167: doc-faligned
5168: doc-sfloats
5169: doc-sfloat+
5170: doc-sfaligned
5171: doc-dfloats
5172: doc-dfloat+
5173: doc-dfaligned
5174: doc-maxaligned
5175: doc-cfaligned
5176: doc-address-unit-bits
5177:
5178:
5179: @node Memory Blocks, , Address arithmetic, Memory
5180: @subsection Memory Blocks
5181: @cindex memory block words
5182: @cindex character strings - moving and copying
5183:
5184: Memory blocks often represent character strings; For ways of storing
5185: character strings in memory see @ref{String Formats}. For other
5186: string-processing words see @ref{Displaying characters and strings}.
5187:
5188: A few of these words work on address unit blocks. In that case, you
5189: usually have to insert @code{CHARS} before the word when working on
5190: character strings. Most words work on character blocks, and expect a
5191: char-aligned address.
5192:
5193: When copying characters between overlapping memory regions, use
5194: @code{chars move} or choose carefully between @code{cmove} and
5195: @code{cmove>}.
5196:
5197: doc-move
5198: doc-erase
5199: doc-cmove
5200: doc-cmove>
5201: doc-fill
5202: doc-blank
5203: doc-compare
5204: doc-search
5205: doc--trailing
5206: doc-/string
5207:
5208:
5209: @comment TODO examples
5210:
5211:
5212: @node Control Structures, Defining Words, Memory, Words
5213: @section Control Structures
5214: @cindex control structures
5215:
5216: Control structures in Forth cannot be used interpretively, only in a
5217: colon definition@footnote{To be precise, they have no interpretation
5218: semantics (@pxref{Interpretation and Compilation Semantics}).}. We do
5219: not like this limitation, but have not seen a satisfying way around it
5220: yet, although many schemes have been proposed.
5221:
5222: @menu
5223: * Selection:: IF ... ELSE ... ENDIF
5224: * Simple Loops:: BEGIN ...
5225: * Counted Loops:: DO
5226: * Arbitrary control structures::
5227: * Calls and returns::
5228: * Exception Handling::
5229: @end menu
5230:
5231: @node Selection, Simple Loops, Control Structures, Control Structures
5232: @subsection Selection
5233: @cindex selection control structures
5234: @cindex control structures for selection
5235:
5236: @cindex @code{IF} control structure
5237: @example
5238: @i{flag}
5239: IF
5240: @i{code}
5241: ENDIF
5242: @end example
5243: @noindent
5244:
5245: If @i{flag} is non-zero (as far as @code{IF} etc. are concerned, a cell
5246: with any bit set represents truth) @i{code} is executed.
5247:
5248: @example
5249: @i{flag}
5250: IF
5251: @i{code1}
5252: ELSE
5253: @i{code2}
5254: ENDIF
5255: @end example
5256:
5257: If @var{flag} is true, @i{code1} is executed, otherwise @i{code2} is
5258: executed.
5259:
5260: You can use @code{THEN} instead of @code{ENDIF}. Indeed, @code{THEN} is
5261: standard, and @code{ENDIF} is not, although it is quite popular. We
5262: recommend using @code{ENDIF}, because it is less confusing for people
5263: who also know other languages (and is not prone to reinforcing negative
5264: prejudices against Forth in these people). Adding @code{ENDIF} to a
5265: system that only supplies @code{THEN} is simple:
5266: @example
5267: : ENDIF POSTPONE THEN ; immediate
5268: @end example
5269:
5270: [According to @cite{Webster's New Encyclopedic Dictionary}, @dfn{then
5271: (adv.)} has the following meanings:
5272: @quotation
5273: ... 2b: following next after in order ... 3d: as a necessary consequence
5274: (if you were there, then you saw them).
5275: @end quotation
5276: Forth's @code{THEN} has the meaning 2b, whereas @code{THEN} in Pascal
5277: and many other programming languages has the meaning 3d.]
5278:
5279: Gforth also provides the words @code{?DUP-IF} and @code{?DUP-0=-IF}, so
5280: you can avoid using @code{?dup}. Using these alternatives is also more
5281: efficient than using @code{?dup}. Definitions in ANS Forth
5282: for @code{ENDIF}, @code{?DUP-IF} and @code{?DUP-0=-IF} are provided in
5283: @file{compat/control.fs}.
5284:
5285: @cindex @code{CASE} control structure
5286: @example
5287: @i{n}
5288: CASE
5289: @i{n1} OF @i{code1} ENDOF
5290: @i{n2} OF @i{code2} ENDOF
5291: @dots{}
5292: ( n ) @i{default-code} ( n )
5293: ENDCASE
5294: @end example
5295:
5296: Executes the first @i{codei}, where the @i{ni} is equal to @i{n}. If no
5297: @i{ni} matches, the optional @i{default-code} is executed. The optional
5298: default case can be added by simply writing the code after the last
5299: @code{ENDOF}. It may use @i{n}, which is on top of the stack, but must
5300: not consume it.
5301:
5302: @progstyle
5303: To keep the code understandable, you should ensure that on all paths
5304: through a selection construct the stack is changed in the same way
5305: (wrt. number and types of stack items consumed and pushed).
5306:
5307: @node Simple Loops, Counted Loops, Selection, Control Structures
5308: @subsection Simple Loops
5309: @cindex simple loops
5310: @cindex loops without count
5311:
5312: @cindex @code{WHILE} loop
5313: @example
5314: BEGIN
5315: @i{code1}
5316: @i{flag}
5317: WHILE
5318: @i{code2}
5319: REPEAT
5320: @end example
5321:
5322: @i{code1} is executed and @i{flag} is computed. If it is true,
5323: @i{code2} is executed and the loop is restarted; If @i{flag} is
5324: false, execution continues after the @code{REPEAT}.
5325:
5326: @cindex @code{UNTIL} loop
5327: @example
5328: BEGIN
5329: @i{code}
5330: @i{flag}
5331: UNTIL
5332: @end example
5333:
5334: @i{code} is executed. The loop is restarted if @code{flag} is false.
5335:
5336: @progstyle
5337: To keep the code understandable, a complete iteration of the loop should
5338: not change the number and types of the items on the stacks.
5339:
5340: @cindex endless loop
5341: @cindex loops, endless
5342: @example
5343: BEGIN
5344: @i{code}
5345: AGAIN
5346: @end example
5347:
5348: This is an endless loop.
5349:
5350: @node Counted Loops, Arbitrary control structures, Simple Loops, Control Structures
5351: @subsection Counted Loops
5352: @cindex counted loops
5353: @cindex loops, counted
5354: @cindex @code{DO} loops
5355:
5356: The basic counted loop is:
5357: @example
5358: @i{limit} @i{start}
5359: ?DO
5360: @i{body}
5361: LOOP
5362: @end example
5363:
5364: This performs one iteration for every integer, starting from @i{start}
5365: and up to, but excluding @i{limit}. The counter, or @i{index}, can be
5366: accessed with @code{i}. For example, the loop:
5367: @example
5368: 10 0 ?DO
5369: i .
5370: LOOP
5371: @end example
5372: @noindent
5373: prints @code{0 1 2 3 4 5 6 7 8 9}
5374:
5375: The index of the innermost loop can be accessed with @code{i}, the index
5376: of the next loop with @code{j}, and the index of the third loop with
5377: @code{k}.
5378:
5379:
5380: doc-i
5381: doc-j
5382: doc-k
5383:
5384:
5385: The loop control data are kept on the return stack, so there are some
5386: restrictions on mixing return stack accesses and counted loop words. In
5387: particuler, if you put values on the return stack outside the loop, you
5388: cannot read them inside the loop@footnote{well, not in a way that is
5389: portable.}. If you put values on the return stack within a loop, you
5390: have to remove them before the end of the loop and before accessing the
5391: index of the loop.
5392:
5393: There are several variations on the counted loop:
5394:
5395: @itemize @bullet
5396: @item
5397: @code{LEAVE} leaves the innermost counted loop immediately; execution
5398: continues after the associated @code{LOOP} or @code{NEXT}. For example:
5399:
5400: @example
5401: 10 0 ?DO i DUP . 3 = IF LEAVE THEN LOOP
5402: @end example
5403: prints @code{0 1 2 3}
5404:
5405:
5406: @item
5407: @code{UNLOOP} prepares for an abnormal loop exit, e.g., via
5408: @code{EXIT}. @code{UNLOOP} removes the loop control parameters from the
5409: return stack so @code{EXIT} can get to its return address. For example:
5410:
5411: @example
5412: : demo 10 0 ?DO i DUP . 3 = IF UNLOOP EXIT THEN LOOP ." Done" ;
5413: @end example
5414: prints @code{0 1 2 3}
5415:
5416:
5417: @item
5418: If @i{start} is greater than @i{limit}, a @code{?DO} loop is entered
5419: (and @code{LOOP} iterates until they become equal by wrap-around
5420: arithmetic). This behaviour is usually not what you want. Therefore,
5421: Gforth offers @code{+DO} and @code{U+DO} (as replacements for
5422: @code{?DO}), which do not enter the loop if @i{start} is greater than
5423: @i{limit}; @code{+DO} is for signed loop parameters, @code{U+DO} for
5424: unsigned loop parameters.
5425:
5426: @item
5427: @code{?DO} can be replaced by @code{DO}. @code{DO} always enters
5428: the loop, independent of the loop parameters. Do not use @code{DO}, even
5429: if you know that the loop is entered in any case. Such knowledge tends
5430: to become invalid during maintenance of a program, and then the
5431: @code{DO} will make trouble.
5432:
5433: @item
5434: @code{LOOP} can be replaced with @code{@i{n} +LOOP}; this updates the
5435: index by @i{n} instead of by 1. The loop is terminated when the border
5436: between @i{limit-1} and @i{limit} is crossed. E.g.:
5437:
5438: @example
5439: 4 0 +DO i . 2 +LOOP
5440: @end example
5441: @noindent
5442: prints @code{0 2}
5443:
5444: @example
5445: 4 1 +DO i . 2 +LOOP
5446: @end example
5447: @noindent
5448: prints @code{1 3}
5449:
5450: @item
5451: @cindex negative increment for counted loops
5452: @cindex counted loops with negative increment
5453: The behaviour of @code{@i{n} +LOOP} is peculiar when @i{n} is negative:
5454:
5455: @example
5456: -1 0 ?DO i . -1 +LOOP
5457: @end example
5458: @noindent
5459: prints @code{0 -1}
5460:
5461: @example
5462: 0 0 ?DO i . -1 +LOOP
5463: @end example
5464: prints nothing.
5465:
5466: Therefore we recommend avoiding @code{@i{n} +LOOP} with negative
5467: @i{n}. One alternative is @code{@i{u} -LOOP}, which reduces the
5468: index by @i{u} each iteration. The loop is terminated when the border
5469: between @i{limit+1} and @i{limit} is crossed. Gforth also provides
5470: @code{-DO} and @code{U-DO} for down-counting loops. E.g.:
5471:
5472: @example
5473: -2 0 -DO i . 1 -LOOP
5474: @end example
5475: @noindent
5476: prints @code{0 -1}
5477:
5478: @example
5479: -1 0 -DO i . 1 -LOOP
5480: @end example
5481: @noindent
5482: prints @code{0}
5483:
5484: @example
5485: 0 0 -DO i . 1 -LOOP
5486: @end example
5487: @noindent
5488: prints nothing.
5489:
5490: @end itemize
5491:
5492: Unfortunately, @code{+DO}, @code{U+DO}, @code{-DO}, @code{U-DO} and
5493: @code{-LOOP} are not defined in ANS Forth. However, an implementation
5494: for these words that uses only standard words is provided in
5495: @file{compat/loops.fs}.
5496:
5497:
5498: @cindex @code{FOR} loops
5499: Another counted loop is:
5500: @example
5501: @i{n}
5502: FOR
5503: @i{body}
5504: NEXT
5505: @end example
5506: This is the preferred loop of native code compiler writers who are too
5507: lazy to optimize @code{?DO} loops properly. This loop structure is not
5508: defined in ANS Forth. In Gforth, this loop iterates @i{n+1} times;
5509: @code{i} produces values starting with @i{n} and ending with 0. Other
5510: Forth systems may behave differently, even if they support @code{FOR}
5511: loops. To avoid problems, don't use @code{FOR} loops.
5512:
5513: @node Arbitrary control structures, Calls and returns, Counted Loops, Control Structures
5514: @subsection Arbitrary control structures
5515: @cindex control structures, user-defined
5516:
5517: @cindex control-flow stack
5518: ANS Forth permits and supports using control structures in a non-nested
5519: way. Information about incomplete control structures is stored on the
5520: control-flow stack. This stack may be implemented on the Forth data
5521: stack, and this is what we have done in Gforth.
5522:
5523: @cindex @code{orig}, control-flow stack item
5524: @cindex @code{dest}, control-flow stack item
5525: An @i{orig} entry represents an unresolved forward branch, a @i{dest}
5526: entry represents a backward branch target. A few words are the basis for
5527: building any control structure possible (except control structures that
5528: need storage, like calls, coroutines, and backtracking).
5529:
5530:
5531: doc-if
5532: doc-ahead
5533: doc-then
5534: doc-begin
5535: doc-until
5536: doc-again
5537: doc-cs-pick
5538: doc-cs-roll
5539:
5540:
5541: The Standard words @code{CS-PICK} and @code{CS-ROLL} allow you to
5542: manipulate the control-flow stack in a portable way. Without them, you
5543: would need to know how many stack items are occupied by a control-flow
5544: entry (many systems use one cell. In Gforth they currently take three,
5545: but this may change in the future).
5546:
5547: Some standard control structure words are built from these words:
5548:
5549:
5550: doc-else
5551: doc-while
5552: doc-repeat
5553:
5554:
5555: @noindent
5556: Gforth adds some more control-structure words:
5557:
5558:
5559: doc-endif
5560: doc-?dup-if
5561: doc-?dup-0=-if
5562:
5563:
5564: @noindent
5565: Counted loop words constitute a separate group of words:
5566:
5567:
5568: doc-?do
5569: doc-+do
5570: doc-u+do
5571: doc--do
5572: doc-u-do
5573: doc-do
5574: doc-for
5575: doc-loop
5576: doc-+loop
5577: doc--loop
5578: doc-next
5579: doc-leave
5580: doc-?leave
5581: doc-unloop
5582: doc-done
5583:
5584:
5585: The standard does not allow using @code{CS-PICK} and @code{CS-ROLL} on
5586: @i{do-sys}. Gforth allows it, but it's your job to ensure that for
5587: every @code{?DO} etc. there is exactly one @code{UNLOOP} on any path
5588: through the definition (@code{LOOP} etc. compile an @code{UNLOOP} on the
5589: fall-through path). Also, you have to ensure that all @code{LEAVE}s are
5590: resolved (by using one of the loop-ending words or @code{DONE}).
5591:
5592: @noindent
5593: Another group of control structure words are:
5594:
5595:
5596: doc-case
5597: doc-endcase
5598: doc-of
5599: doc-endof
5600:
5601:
5602: @i{case-sys} and @i{of-sys} cannot be processed using @code{CS-PICK} and
5603: @code{CS-ROLL}.
5604:
5605: @subsubsection Programming Style
5606: @cindex control structures programming style
5607: @cindex programming style, arbitrary control structures
5608:
5609: In order to ensure readability we recommend that you do not create
5610: arbitrary control structures directly, but define new control structure
5611: words for the control structure you want and use these words in your
5612: program. For example, instead of writing:
5613:
5614: @example
5615: BEGIN
5616: ...
5617: IF [ 1 CS-ROLL ]
5618: ...
5619: AGAIN THEN
5620: @end example
5621:
5622: @noindent
5623: we recommend defining control structure words, e.g.,
5624:
5625: @example
5626: : WHILE ( DEST -- ORIG DEST )
5627: POSTPONE IF
5628: 1 CS-ROLL ; immediate
5629:
5630: : REPEAT ( orig dest -- )
5631: POSTPONE AGAIN
5632: POSTPONE THEN ; immediate
5633: @end example
5634:
5635: @noindent
5636: and then using these to create the control structure:
5637:
5638: @example
5639: BEGIN
5640: ...
5641: WHILE
5642: ...
5643: REPEAT
5644: @end example
5645:
5646: That's much easier to read, isn't it? Of course, @code{REPEAT} and
5647: @code{WHILE} are predefined, so in this example it would not be
5648: necessary to define them.
5649:
5650: @node Calls and returns, Exception Handling, Arbitrary control structures, Control Structures
5651: @subsection Calls and returns
5652: @cindex calling a definition
5653: @cindex returning from a definition
5654:
5655: @cindex recursive definitions
5656: A definition can be called simply be writing the name of the definition
5657: to be called. Normally a definition is invisible during its own
5658: definition. If you want to write a directly recursive definition, you
5659: can use @code{recursive} to make the current definition visible, or
5660: @code{recurse} to call the current definition directly.
5661:
5662:
5663: doc-recursive
5664: doc-recurse
5665:
5666:
5667: @comment TODO add example of the two recursion methods
5668: @quotation
5669: @progstyle
5670: I prefer using @code{recursive} to @code{recurse}, because calling the
5671: definition by name is more descriptive (if the name is well-chosen) than
5672: the somewhat cryptic @code{recurse}. E.g., in a quicksort
5673: implementation, it is much better to read (and think) ``now sort the
5674: partitions'' than to read ``now do a recursive call''.
5675: @end quotation
5676:
5677: For mutual recursion, use @code{Defer}red words, like this:
5678:
5679: @example
5680: Defer foo
5681:
5682: : bar ( ... -- ... )
5683: ... foo ... ;
5684:
5685: :noname ( ... -- ... )
5686: ... bar ... ;
5687: IS foo
5688: @end example
5689:
5690: Deferred words are discussed in more detail in @ref{Deferred words}.
5691:
5692: The current definition returns control to the calling definition when
5693: the end of the definition is reached or @code{EXIT} is encountered.
5694:
5695: doc-exit
5696: doc-;s
5697:
5698:
5699: @node Exception Handling, , Calls and returns, Control Structures
5700: @subsection Exception Handling
5701: @cindex exceptions
5702:
5703: @c quit is a very bad idea for error handling,
5704: @c because it does not translate into a THROW
5705: @c it also does not belong into this chapter
5706:
5707: If a word detects an error condition that it cannot handle, it can
5708: @code{throw} an exception. In the simplest case, this will terminate
5709: your program, and report an appropriate error.
5710:
5711: doc-throw
5712:
5713: @code{Throw} consumes a cell-sized error number on the stack. There are
5714: some predefined error numbers in ANS Forth (see @file{errors.fs}). In
5715: Gforth (and most other systems) you can use the iors produced by various
5716: words as error numbers (e.g., a typical use of @code{allocate} is
5717: @code{allocate throw}). Gforth also provides the word @code{exception}
5718: to define your own error numbers (with decent error reporting); an ANS
5719: Forth version of this word (but without the error messages) is available
5720: in @code{compat/except.fs}. And finally, you can use your own error
5721: numbers (anything outside the range -4095..0), but won't get nice error
5722: messages, only numbers. For example, try:
5723:
5724: @example
5725: -10 throw \ ANS defined
5726: -267 throw \ system defined
5727: s" my error" exception throw \ user defined
5728: 7 throw \ arbitrary number
5729: @end example
5730:
5731: doc---exception-exception
5732:
5733: A common idiom to @code{THROW} a specific error if a flag is true is
5734: this:
5735:
5736: @example
5737: @code{( flag ) 0<> @i{errno} and throw}
5738: @end example
5739:
5740: Your program can provide exception handlers to catch exceptions. An
5741: exception handler can be used to correct the problem, or to clean up
5742: some data structures and just throw the exception to the next exception
5743: handler. Note that @code{throw} jumps to the dynamically innermost
5744: exception handler. The system's exception handler is outermost, and just
5745: prints an error and restarts command-line interpretation (or, in batch
5746: mode (i.e., while processing the shell command line), leaves Gforth).
5747:
5748: The ANS Forth way to catch exceptions is @code{catch}:
5749:
5750: doc-catch
5751:
5752: The most common use of exception handlers is to clean up the state when
5753: an error happens. E.g.,
5754:
5755: @example
5756: base @ >r hex \ actually the hex should be inside foo, or we h
5757: ['] foo catch ( nerror|0 )
5758: r> base !
5759: ( nerror|0 ) throw \ pass it on
5760: @end example
5761:
5762: A use of @code{catch} for handling the error @code{myerror} might look
5763: like this:
5764:
5765: @example
5766: ['] foo catch
5767: CASE
5768: myerror OF ... ( do something about it ) ENDOF
5769: dup throw \ default: pass other errors on, do nothing on non-errors
5770: ENDCASE
5771: @end example
5772:
5773: Having to wrap the code into a separate word is often cumbersome,
5774: therefore Gforth provides an alternative syntax:
5775:
5776: @example
5777: TRY
5778: @i{code1}
5779: RECOVER \ optional
5780: @i{code2} \ optional
5781: ENDTRY
5782: @end example
5783:
5784: This performs @i{Code1}. If @i{code1} completes normally, execution
5785: continues after the @code{endtry}. If @i{Code1} throws, the stacks are
5786: reset to the state during @code{try}, the throw value is pushed on the
5787: data stack, and execution constinues at @i{code2}, and finally falls
5788: through the @code{endtry} into the following code. If there is no
5789: @code{recover} clause, this works like an empty recover clause.
5790:
5791: doc-try
5792: doc-recover
5793: doc-endtry
5794:
5795: The cleanup example from above in this syntax:
5796:
5797: @example
5798: base @ >r TRY
5799: hex foo \ now the hex is placed correctly
5800: 0 \ value for throw
5801: ENDTRY
5802: r> base ! throw
5803: @end example
5804:
5805: And here's the error handling example:
5806:
5807: @example
5808: TRY
5809: foo
5810: RECOVER
5811: CASE
5812: myerror OF ... ( do something about it ) ENDOF
5813: throw \ pass other errors on
5814: ENDCASE
5815: ENDTRY
5816: @end example
5817:
5818: @progstyle
5819: As usual, you should ensure that the stack depth is statically known at
5820: the end: either after the @code{throw} for passing on errors, or after
5821: the @code{ENDTRY} (or, if you use @code{catch}, after the end of the
5822: selection construct for handling the error).
5823:
5824: There are two alternatives to @code{throw}: @code{Abort"} is conditional
5825: and you can provide an error message. @code{Abort} just produces an
5826: ``Aborted'' error.
5827:
5828: The problem with these words is that exception handlers cannot
5829: differentiate between different @code{abort"}s; they just look like
5830: @code{-2 throw} to them (the error message cannot be accessed by
5831: standard programs). Similar @code{abort} looks like @code{-1 throw} to
5832: exception handlers.
5833:
5834: doc-abort"
5835: doc-abort
5836:
5837:
5838:
5839: @c -------------------------------------------------------------
5840: @node Defining Words, Interpretation and Compilation Semantics, Control Structures, Words
5841: @section Defining Words
5842: @cindex defining words
5843:
5844: Defining words are used to extend Forth by creating new entries in the dictionary.
5845:
5846: @menu
5847: * CREATE::
5848: * Variables:: Variables and user variables
5849: * Constants::
5850: * Values:: Initialised variables
5851: * Colon Definitions::
5852: * Anonymous Definitions:: Definitions without names
5853: * Supplying names:: Passing definition names as strings
5854: * User-defined Defining Words::
5855: * Deferred words:: Allow forward references
5856: * Aliases::
5857: @end menu
5858:
5859: @node CREATE, Variables, Defining Words, Defining Words
5860: @subsection @code{CREATE}
5861: @cindex simple defining words
5862: @cindex defining words, simple
5863:
5864: Defining words are used to create new entries in the dictionary. The
5865: simplest defining word is @code{CREATE}. @code{CREATE} is used like
5866: this:
5867:
5868: @example
5869: CREATE new-word1
5870: @end example
5871:
5872: @code{CREATE} is a parsing word, i.e., it takes an argument from the
5873: input stream (@code{new-word1} in our example). It generates a
5874: dictionary entry for @code{new-word1}. When @code{new-word1} is
5875: executed, all that it does is leave an address on the stack. The address
5876: represents the value of the data space pointer (@code{HERE}) at the time
5877: that @code{new-word1} was defined. Therefore, @code{CREATE} is a way of
5878: associating a name with the address of a region of memory.
5879:
5880: doc-create
5881:
5882: Note that in ANS Forth guarantees only for @code{create} that its body
5883: is in dictionary data space (i.e., where @code{here}, @code{allot}
5884: etc. work, @pxref{Dictionary allocation}). Also, in ANS Forth only
5885: @code{create}d words can be modified with @code{does>}
5886: (@pxref{User-defined Defining Words}). And in ANS Forth @code{>body}
5887: can only be applied to @code{create}d words.
5888:
5889: By extending this example to reserve some memory in data space, we end
5890: up with something like a @i{variable}. Here are two different ways to do
5891: it:
5892:
5893: @example
5894: CREATE new-word2 1 cells allot \ reserve 1 cell - initial value undefined
5895: CREATE new-word3 4 , \ reserve 1 cell and initialise it (to 4)
5896: @end example
5897:
5898: The variable can be examined and modified using @code{@@} (``fetch'') and
5899: @code{!} (``store'') like this:
5900:
5901: @example
5902: new-word2 @@ . \ get address, fetch from it and display
5903: 1234 new-word2 ! \ new value, get address, store to it
5904: @end example
5905:
5906: @cindex arrays
5907: A similar mechanism can be used to create arrays. For example, an
5908: 80-character text input buffer:
5909:
5910: @example
5911: CREATE text-buf 80 chars allot
5912:
5913: text-buf 0 chars c@@ \ the 1st character (offset 0)
5914: text-buf 3 chars c@@ \ the 4th character (offset 3)
5915: @end example
5916:
5917: You can build arbitrarily complex data structures by allocating
5918: appropriate areas of memory. For further discussions of this, and to
5919: learn about some Gforth tools that make it easier,
5920: @xref{Structures}.
5921:
5922:
5923: @node Variables, Constants, CREATE, Defining Words
5924: @subsection Variables
5925: @cindex variables
5926:
5927: The previous section showed how a sequence of commands could be used to
5928: generate a variable. As a final refinement, the whole code sequence can
5929: be wrapped up in a defining word (pre-empting the subject of the next
5930: section), making it easier to create new variables:
5931:
5932: @example
5933: : myvariableX ( "name" -- a-addr ) CREATE 1 cells allot ;
5934: : myvariable0 ( "name" -- a-addr ) CREATE 0 , ;
5935:
5936: myvariableX foo \ variable foo starts off with an unknown value
5937: myvariable0 joe \ whilst joe is initialised to 0
5938:
5939: 45 3 * foo ! \ set foo to 135
5940: 1234 joe ! \ set joe to 1234
5941: 3 joe +! \ increment joe by 3.. to 1237
5942: @end example
5943:
5944: Not surprisingly, there is no need to define @code{myvariable}, since
5945: Forth already has a definition @code{Variable}. ANS Forth does not
5946: guarantee that a @code{Variable} is initialised when it is created
5947: (i.e., it may behave like @code{myvariableX}). In contrast, Gforth's
5948: @code{Variable} initialises the variable to 0 (i.e., it behaves exactly
5949: like @code{myvariable0}). Forth also provides @code{2Variable} and
5950: @code{fvariable} for double and floating-point variables, respectively
5951: -- they are initialised to 0. and 0e in Gforth. If you use a @code{Variable} to
5952: store a boolean, you can use @code{on} and @code{off} to toggle its
5953: state.
5954:
5955: doc-variable
5956: doc-2variable
5957: doc-fvariable
5958:
5959: @cindex user variables
5960: @cindex user space
5961: The defining word @code{User} behaves in the same way as @code{Variable}.
5962: The difference is that it reserves space in @i{user (data) space} rather
5963: than normal data space. In a Forth system that has a multi-tasker, each
5964: task has its own set of user variables.
5965:
5966: doc-user
5967: @c doc-udp
5968: @c doc-uallot
5969:
5970: @comment TODO is that stuff about user variables strictly correct? Is it
5971: @comment just terminal tasks that have user variables?
5972: @comment should document tasker.fs (with some examples) elsewhere
5973: @comment in this manual, then expand on user space and user variables.
5974:
5975: @node Constants, Values, Variables, Defining Words
5976: @subsection Constants
5977: @cindex constants
5978:
5979: @code{Constant} allows you to declare a fixed value and refer to it by
5980: name. For example:
5981:
5982: @example
5983: 12 Constant INCHES-PER-FOOT
5984: 3E+08 fconstant SPEED-O-LIGHT
5985: @end example
5986:
5987: A @code{Variable} can be both read and written, so its run-time
5988: behaviour is to supply an address through which its current value can be
5989: manipulated. In contrast, the value of a @code{Constant} cannot be
5990: changed once it has been declared@footnote{Well, often it can be -- but
5991: not in a Standard, portable way. It's safer to use a @code{Value} (read
5992: on).} so it's not necessary to supply the address -- it is more
5993: efficient to return the value of the constant directly. That's exactly
5994: what happens; the run-time effect of a constant is to put its value on
5995: the top of the stack (You can find one
5996: way of implementing @code{Constant} in @ref{User-defined Defining Words}).
5997:
5998: Forth also provides @code{2Constant} and @code{fconstant} for defining
5999: double and floating-point constants, respectively.
6000:
6001: doc-constant
6002: doc-2constant
6003: doc-fconstant
6004:
6005: @c that's too deep, and it's not necessarily true for all ANS Forths. - anton
6006: @c nac-> How could that not be true in an ANS Forth? You can't define a
6007: @c constant, use it and then delete the definition of the constant..
6008:
6009: @c anton->An ANS Forth system can compile a constant to a literal; On
6010: @c decompilation you would see only the number, just as if it had been used
6011: @c in the first place. The word will stay, of course, but it will only be
6012: @c used by the text interpreter (no run-time duties, except when it is
6013: @c POSTPONEd or somesuch).
6014:
6015: @c nac:
6016: @c I agree that it's rather deep, but IMO it is an important difference
6017: @c relative to other programming languages.. often it's annoying: it
6018: @c certainly changes my programming style relative to C.
6019:
6020: @c anton: In what way?
6021:
6022: Constants in Forth behave differently from their equivalents in other
6023: programming languages. In other languages, a constant (such as an EQU in
6024: assembler or a #define in C) only exists at compile-time; in the
6025: executable program the constant has been translated into an absolute
6026: number and, unless you are using a symbolic debugger, it's impossible to
6027: know what abstract thing that number represents. In Forth a constant has
6028: an entry in the header space and remains there after the code that uses
6029: it has been defined. In fact, it must remain in the dictionary since it
6030: has run-time duties to perform. For example:
6031:
6032: @example
6033: 12 Constant INCHES-PER-FOOT
6034: : FEET-TO-INCHES ( n1 -- n2 ) INCHES-PER-FOOT * ;
6035: @end example
6036:
6037: @cindex in-lining of constants
6038: When @code{FEET-TO-INCHES} is executed, it will in turn execute the xt
6039: associated with the constant @code{INCHES-PER-FOOT}. If you use
6040: @code{see} to decompile the definition of @code{FEET-TO-INCHES}, you can
6041: see that it makes a call to @code{INCHES-PER-FOOT}. Some Forth compilers
6042: attempt to optimise constants by in-lining them where they are used. You
6043: can force Gforth to in-line a constant like this:
6044:
6045: @example
6046: : FEET-TO-INCHES ( n1 -- n2 ) [ INCHES-PER-FOOT ] LITERAL * ;
6047: @end example
6048:
6049: If you use @code{see} to decompile @i{this} version of
6050: @code{FEET-TO-INCHES}, you can see that @code{INCHES-PER-FOOT} is no
6051: longer present. To understand how this works, read
6052: @ref{Interpret/Compile states}, and @ref{Literals}.
6053:
6054: In-lining constants in this way might improve execution time
6055: fractionally, and can ensure that a constant is now only referenced at
6056: compile-time. However, the definition of the constant still remains in
6057: the dictionary. Some Forth compilers provide a mechanism for controlling
6058: a second dictionary for holding transient words such that this second
6059: dictionary can be deleted later in order to recover memory
6060: space. However, there is no standard way of doing this.
6061:
6062:
6063: @node Values, Colon Definitions, Constants, Defining Words
6064: @subsection Values
6065: @cindex values
6066:
6067: A @code{Value} behaves like a @code{Constant}, but it can be changed.
6068: @code{TO} is a parsing word that changes a @code{Values}. In Gforth
6069: (not in ANS Forth) you can access (and change) a @code{value} also with
6070: @code{>body}.
6071:
6072: Here are some
6073: examples:
6074:
6075: @example
6076: 12 Value APPLES \ Define APPLES with an initial value of 12
6077: 34 TO APPLES \ Change the value of APPLES. TO is a parsing word
6078: 1 ' APPLES >body +! \ Increment APPLES. Non-standard usage.
6079: APPLES \ puts 35 on the top of the stack.
6080: @end example
6081:
6082: doc-value
6083: doc-to
6084:
6085:
6086:
6087: @node Colon Definitions, Anonymous Definitions, Values, Defining Words
6088: @subsection Colon Definitions
6089: @cindex colon definitions
6090:
6091: @example
6092: : name ( ... -- ... )
6093: word1 word2 word3 ;
6094: @end example
6095:
6096: @noindent
6097: Creates a word called @code{name} that, upon execution, executes
6098: @code{word1 word2 word3}. @code{name} is a @dfn{(colon) definition}.
6099:
6100: The explanation above is somewhat superficial. For simple examples of
6101: colon definitions see @ref{Your first definition}. For an in-depth
6102: discussion of some of the issues involved, @xref{Interpretation and
6103: Compilation Semantics}.
6104:
6105: doc-:
6106: doc-;
6107:
6108:
6109: @node Anonymous Definitions, Supplying names, Colon Definitions, Defining Words
6110: @subsection Anonymous Definitions
6111: @cindex colon definitions
6112: @cindex defining words without name
6113:
6114: Sometimes you want to define an @dfn{anonymous word}; a word without a
6115: name. You can do this with:
6116:
6117: doc-:noname
6118:
6119: This leaves the execution token for the word on the stack after the
6120: closing @code{;}. Here's an example in which a deferred word is
6121: initialised with an @code{xt} from an anonymous colon definition:
6122:
6123: @example
6124: Defer deferred
6125: :noname ( ... -- ... )
6126: ... ;
6127: IS deferred
6128: @end example
6129:
6130: @noindent
6131: Gforth provides an alternative way of doing this, using two separate
6132: words:
6133:
6134: doc-noname
6135: @cindex execution token of last defined word
6136: doc-lastxt
6137:
6138: @noindent
6139: The previous example can be rewritten using @code{noname} and
6140: @code{lastxt}:
6141:
6142: @example
6143: Defer deferred
6144: noname : ( ... -- ... )
6145: ... ;
6146: lastxt IS deferred
6147: @end example
6148:
6149: @noindent
6150: @code{noname} works with any defining word, not just @code{:}.
6151:
6152: @code{lastxt} also works when the last word was not defined as
6153: @code{noname}. It also has the useful property that is is valid as soon
6154: as the header for a definition has been built. Thus:
6155:
6156: @example
6157: lastxt . : foo [ lastxt . ] ; ' foo .
6158: @end example
6159:
6160: @noindent
6161: prints 3 numbers; the last two are the same.
6162:
6163: @node Supplying names, User-defined Defining Words, Anonymous Definitions, Defining Words
6164: @subsection Supplying the name of a defined word
6165: @cindex names for defined words
6166: @cindex defining words, name given in a string
6167:
6168: By default, a defining word takes the name for the defined word from the
6169: input stream. Sometimes you want to supply the name from a string. You
6170: can do this with:
6171:
6172: doc-nextname
6173:
6174: For example:
6175:
6176: @example
6177: s" foo" nextname create
6178: @end example
6179:
6180: @noindent
6181: is equivalent to:
6182:
6183: @example
6184: create foo
6185: @end example
6186:
6187: @noindent
6188: @code{nextname} works with any defining word.
6189:
6190:
6191: @node User-defined Defining Words, Deferred words, Supplying names, Defining Words
6192: @subsection User-defined Defining Words
6193: @cindex user-defined defining words
6194: @cindex defining words, user-defined
6195:
6196: You can create a new defining word by wrapping defining-time code around
6197: an existing defining word and putting the sequence in a colon
6198: definition.
6199:
6200: @c anton: This example is very complex and leads in a quite different
6201: @c direction from the CREATE-DOES> stuff that follows. It should probably
6202: @c be done elsewhere, or as a subsubsection of this subsection (or as a
6203: @c subsection of Defining Words)
6204:
6205: For example, suppose that you have a word @code{stats} that
6206: gathers statistics about colon definitions given the @i{xt} of the
6207: definition, and you want every colon definition in your application to
6208: make a call to @code{stats}. You can define and use a new version of
6209: @code{:} like this:
6210:
6211: @example
6212: : stats ( xt -- ) DUP ." (Gathering statistics for " . ." )"
6213: ... ; \ other code
6214:
6215: : my: : lastxt postpone literal ['] stats compile, ;
6216:
6217: my: foo + - ;
6218: @end example
6219:
6220: When @code{foo} is defined using @code{my:} these steps occur:
6221:
6222: @itemize @bullet
6223: @item
6224: @code{my:} is executed.
6225: @item
6226: The @code{:} within the definition (the one between @code{my:} and
6227: @code{lastxt}) is executed, and does just what it always does; it parses
6228: the input stream for a name, builds a dictionary header for the name
6229: @code{foo} and switches @code{state} from interpret to compile.
6230: @item
6231: The word @code{lastxt} is executed. It puts the @i{xt} for the word that is
6232: being defined -- @code{foo} -- onto the stack.
6233: @item
6234: The code that was produced by @code{postpone literal} is executed; this
6235: causes the value on the stack to be compiled as a literal in the code
6236: area of @code{foo}.
6237: @item
6238: The code @code{['] stats} compiles a literal into the definition of
6239: @code{my:}. When @code{compile,} is executed, that literal -- the
6240: execution token for @code{stats} -- is layed down in the code area of
6241: @code{foo} , following the literal@footnote{Strictly speaking, the
6242: mechanism that @code{compile,} uses to convert an @i{xt} into something
6243: in the code area is implementation-dependent. A threaded implementation
6244: might spit out the execution token directly whilst another
6245: implementation might spit out a native code sequence.}.
6246: @item
6247: At this point, the execution of @code{my:} is complete, and control
6248: returns to the text interpreter. The text interpreter is in compile
6249: state, so subsequent text @code{+ -} is compiled into the definition of
6250: @code{foo} and the @code{;} terminates the definition as always.
6251: @end itemize
6252:
6253: You can use @code{see} to decompile a word that was defined using
6254: @code{my:} and see how it is different from a normal @code{:}
6255: definition. For example:
6256:
6257: @example
6258: : bar + - ; \ like foo but using : rather than my:
6259: see bar
6260: : bar
6261: + - ;
6262: see foo
6263: : foo
6264: 107645672 stats + - ;
6265:
6266: \ use ' stats . to show that 107645672 is the xt for stats
6267: @end example
6268:
6269: You can use techniques like this to make new defining words in terms of
6270: @i{any} existing defining word.
6271:
6272:
6273: @cindex defining defining words
6274: @cindex @code{CREATE} ... @code{DOES>}
6275: If you want the words defined with your defining words to behave
6276: differently from words defined with standard defining words, you can
6277: write your defining word like this:
6278:
6279: @example
6280: : def-word ( "name" -- )
6281: CREATE @i{code1}
6282: DOES> ( ... -- ... )
6283: @i{code2} ;
6284:
6285: def-word name
6286: @end example
6287:
6288: @cindex child words
6289: This fragment defines a @dfn{defining word} @code{def-word} and then
6290: executes it. When @code{def-word} executes, it @code{CREATE}s a new
6291: word, @code{name}, and executes the code @i{code1}. The code @i{code2}
6292: is not executed at this time. The word @code{name} is sometimes called a
6293: @dfn{child} of @code{def-word}.
6294:
6295: When you execute @code{name}, the address of the body of @code{name} is
6296: put on the data stack and @i{code2} is executed (the address of the body
6297: of @code{name} is the address @code{HERE} returns immediately after the
6298: @code{CREATE}, i.e., the address a @code{create}d word returns by
6299: default).
6300:
6301: @c anton:
6302: @c www.dictionary.com says:
6303: @c at·a·vism: 1.The reappearance of a characteristic in an organism after
6304: @c several generations of absence, usually caused by the chance
6305: @c recombination of genes. 2.An individual or a part that exhibits
6306: @c atavism. Also called throwback. 3.The return of a trait or recurrence
6307: @c of previous behavior after a period of absence.
6308: @c
6309: @c Doesn't seem to fit.
6310:
6311: @c @cindex atavism in child words
6312: You can use @code{def-word} to define a set of child words that behave
6313: similarly; they all have a common run-time behaviour determined by
6314: @i{code2}. Typically, the @i{code1} sequence builds a data area in the
6315: body of the child word. The structure of the data is common to all
6316: children of @code{def-word}, but the data values are specific -- and
6317: private -- to each child word. When a child word is executed, the
6318: address of its private data area is passed as a parameter on TOS to be
6319: used and manipulated@footnote{It is legitimate both to read and write to
6320: this data area.} by @i{code2}.
6321:
6322: The two fragments of code that make up the defining words act (are
6323: executed) at two completely separate times:
6324:
6325: @itemize @bullet
6326: @item
6327: At @i{define time}, the defining word executes @i{code1} to generate a
6328: child word
6329: @item
6330: At @i{child execution time}, when a child word is invoked, @i{code2}
6331: is executed, using parameters (data) that are private and specific to
6332: the child word.
6333: @end itemize
6334:
6335: Another way of understanding the behaviour of @code{def-word} and
6336: @code{name} is to say that, if you make the following definitions:
6337: @example
6338: : def-word1 ( "name" -- )
6339: CREATE @i{code1} ;
6340:
6341: : action1 ( ... -- ... )
6342: @i{code2} ;
6343:
6344: def-word1 name1
6345: @end example
6346:
6347: @noindent
6348: Then using @code{name1 action1} is equivalent to using @code{name}.
6349:
6350: The classic example is that you can define @code{CONSTANT} in this way:
6351:
6352: @example
6353: : CONSTANT ( w "name" -- )
6354: CREATE ,
6355: DOES> ( -- w )
6356: @@ ;
6357: @end example
6358:
6359: @comment There is a beautiful description of how this works and what
6360: @comment it does in the Forthwrite 100th edition.. as well as an elegant
6361: @comment commentary on the Counting Fruits problem.
6362:
6363: When you create a constant with @code{5 CONSTANT five}, a set of
6364: define-time actions take place; first a new word @code{five} is created,
6365: then the value 5 is laid down in the body of @code{five} with
6366: @code{,}. When @code{five} is executed, the address of the body is put on
6367: the stack, and @code{@@} retrieves the value 5. The word @code{five} has
6368: no code of its own; it simply contains a data field and a pointer to the
6369: code that follows @code{DOES>} in its defining word. That makes words
6370: created in this way very compact.
6371:
6372: The final example in this section is intended to remind you that space
6373: reserved in @code{CREATE}d words is @i{data} space and therefore can be
6374: both read and written by a Standard program@footnote{Exercise: use this
6375: example as a starting point for your own implementation of @code{Value}
6376: and @code{TO} -- if you get stuck, investigate the behaviour of @code{'} and
6377: @code{[']}.}:
6378:
6379: @example
6380: : foo ( "name" -- )
6381: CREATE -1 ,
6382: DOES> ( -- )
6383: @@ . ;
6384:
6385: foo first-word
6386: foo second-word
6387:
6388: 123 ' first-word >BODY !
6389: @end example
6390:
6391: If @code{first-word} had been a @code{CREATE}d word, we could simply
6392: have executed it to get the address of its data field. However, since it
6393: was defined to have @code{DOES>} actions, its execution semantics are to
6394: perform those @code{DOES>} actions. To get the address of its data field
6395: it's necessary to use @code{'} to get its xt, then @code{>BODY} to
6396: translate the xt into the address of the data field. When you execute
6397: @code{first-word}, it will display @code{123}. When you execute
6398: @code{second-word} it will display @code{-1}.
6399:
6400: @cindex stack effect of @code{DOES>}-parts
6401: @cindex @code{DOES>}-parts, stack effect
6402: In the examples above the stack comment after the @code{DOES>} specifies
6403: the stack effect of the defined words, not the stack effect of the
6404: following code (the following code expects the address of the body on
6405: the top of stack, which is not reflected in the stack comment). This is
6406: the convention that I use and recommend (it clashes a bit with using
6407: locals declarations for stack effect specification, though).
6408:
6409: @menu
6410: * CREATE..DOES> applications::
6411: * CREATE..DOES> details::
6412: * Advanced does> usage example::
6413: @end menu
6414:
6415: @node CREATE..DOES> applications, CREATE..DOES> details, User-defined Defining Words, User-defined Defining Words
6416: @subsubsection Applications of @code{CREATE..DOES>}
6417: @cindex @code{CREATE} ... @code{DOES>}, applications
6418:
6419: You may wonder how to use this feature. Here are some usage patterns:
6420:
6421: @cindex factoring similar colon definitions
6422: When you see a sequence of code occurring several times, and you can
6423: identify a meaning, you will factor it out as a colon definition. When
6424: you see similar colon definitions, you can factor them using
6425: @code{CREATE..DOES>}. E.g., an assembler usually defines several words
6426: that look very similar:
6427: @example
6428: : ori, ( reg-target reg-source n -- )
6429: 0 asm-reg-reg-imm ;
6430: : andi, ( reg-target reg-source n -- )
6431: 1 asm-reg-reg-imm ;
6432: @end example
6433:
6434: @noindent
6435: This could be factored with:
6436: @example
6437: : reg-reg-imm ( op-code -- )
6438: CREATE ,
6439: DOES> ( reg-target reg-source n -- )
6440: @@ asm-reg-reg-imm ;
6441:
6442: 0 reg-reg-imm ori,
6443: 1 reg-reg-imm andi,
6444: @end example
6445:
6446: @cindex currying
6447: Another view of @code{CREATE..DOES>} is to consider it as a crude way to
6448: supply a part of the parameters for a word (known as @dfn{currying} in
6449: the functional language community). E.g., @code{+} needs two
6450: parameters. Creating versions of @code{+} with one parameter fixed can
6451: be done like this:
6452: @example
6453: : curry+ ( n1 -- )
6454: CREATE ,
6455: DOES> ( n2 -- n1+n2 )
6456: @@ + ;
6457:
6458: 3 curry+ 3+
6459: -2 curry+ 2-
6460: @end example
6461:
6462: @node CREATE..DOES> details, Advanced does> usage example, CREATE..DOES> applications, User-defined Defining Words
6463: @subsubsection The gory details of @code{CREATE..DOES>}
6464: @cindex @code{CREATE} ... @code{DOES>}, details
6465:
6466: doc-does>
6467:
6468: @cindex @code{DOES>} in a separate definition
6469: This means that you need not use @code{CREATE} and @code{DOES>} in the
6470: same definition; you can put the @code{DOES>}-part in a separate
6471: definition. This allows us to, e.g., select among different @code{DOES>}-parts:
6472: @example
6473: : does1
6474: DOES> ( ... -- ... )
6475: ... ;
6476:
6477: : does2
6478: DOES> ( ... -- ... )
6479: ... ;
6480:
6481: : def-word ( ... -- ... )
6482: create ...
6483: IF
6484: does1
6485: ELSE
6486: does2
6487: ENDIF ;
6488: @end example
6489:
6490: In this example, the selection of whether to use @code{does1} or
6491: @code{does2} is made at definition-time; at the time that the child word is
6492: @code{CREATE}d.
6493:
6494: @cindex @code{DOES>} in interpretation state
6495: In a standard program you can apply a @code{DOES>}-part only if the last
6496: word was defined with @code{CREATE}. In Gforth, the @code{DOES>}-part
6497: will override the behaviour of the last word defined in any case. In a
6498: standard program, you can use @code{DOES>} only in a colon
6499: definition. In Gforth, you can also use it in interpretation state, in a
6500: kind of one-shot mode; for example:
6501: @example
6502: CREATE name ( ... -- ... )
6503: @i{initialization}
6504: DOES>
6505: @i{code} ;
6506: @end example
6507:
6508: @noindent
6509: is equivalent to the standard:
6510: @example
6511: :noname
6512: DOES>
6513: @i{code} ;
6514: CREATE name EXECUTE ( ... -- ... )
6515: @i{initialization}
6516: @end example
6517:
6518: doc->body
6519:
6520: @node Advanced does> usage example, , CREATE..DOES> details, User-defined Defining Words
6521: @subsubsection Advanced does> usage example
6522:
6523: The MIPS disassembler (@file{arch/mips/disasm.fs}) contains many words
6524: for disassembling instructions, that follow a very repetetive scheme:
6525:
6526: @example
6527: :noname @var{disasm-operands} s" @var{inst-name}" type ;
6528: @var{entry-num} cells @var{table} + !
6529: @end example
6530:
6531: Of course, this inspires the idea to factor out the commonalities to
6532: allow a definition like
6533:
6534: @example
6535: @var{disasm-operands} @var{entry-num} @var{table} define-inst @var{inst-name}
6536: @end example
6537:
6538: The parameters @var{disasm-operands} and @var{table} are usually
6539: correlated. Moreover, before I wrote the disassembler, there already
6540: existed code that defines instructions like this:
6541:
6542: @example
6543: @var{entry-num} @var{inst-format} @var{inst-name}
6544: @end example
6545:
6546: This code comes from the assembler and resides in
6547: @file{arch/mips/insts.fs}.
6548:
6549: So I had to define the @var{inst-format} words that performed the scheme
6550: above when executed. At first I chose to use run-time code-generation:
6551:
6552: @example
6553: : @var{inst-format} ( entry-num "name" -- ; compiled code: addr w -- )
6554: :noname Postpone @var{disasm-operands}
6555: name Postpone sliteral Postpone type Postpone ;
6556: swap cells @var{table} + ! ;
6557: @end example
6558:
6559: Note that this supplies the other two parameters of the scheme above.
6560:
6561: An alternative would have been to write this using
6562: @code{create}/@code{does>}:
6563:
6564: @example
6565: : @var{inst-format} ( entry-num "name" -- )
6566: here name string, ( entry-num c-addr ) \ parse and save "name"
6567: noname create , ( entry-num )
6568: lastxt swap cells @var{table} + !
6569: does> ( addr w -- )
6570: \ disassemble instruction w at addr
6571: @@ >r
6572: @var{disasm-operands}
6573: r> count type ;
6574: @end example
6575:
6576: Somehow the first solution is simpler, mainly because it's simpler to
6577: shift a string from definition-time to use-time with @code{sliteral}
6578: than with @code{string,} and friends.
6579:
6580: I wrote a lot of words following this scheme and soon thought about
6581: factoring out the commonalities among them. Note that this uses a
6582: two-level defining word, i.e., a word that defines ordinary defining
6583: words.
6584:
6585: This time a solution involving @code{postpone} and friends seemed more
6586: difficult (try it as an exercise), so I decided to use a
6587: @code{create}/@code{does>} word; since I was already at it, I also used
6588: @code{create}/@code{does>} for the lower level (try using
6589: @code{postpone} etc. as an exercise), resulting in the following
6590: definition:
6591:
6592: @example
6593: : define-format ( disasm-xt table-xt -- )
6594: \ define an instruction format that uses disasm-xt for
6595: \ disassembling and enters the defined instructions into table
6596: \ table-xt
6597: create 2,
6598: does> ( u "inst" -- )
6599: \ defines an anonymous word for disassembling instruction inst,
6600: \ and enters it as u-th entry into table-xt
6601: 2@@ swap here name string, ( u table-xt disasm-xt c-addr ) \ remember string
6602: noname create 2, \ define anonymous word
6603: execute lastxt swap ! \ enter xt of defined word into table-xt
6604: does> ( addr w -- )
6605: \ disassemble instruction w at addr
6606: 2@@ >r ( addr w disasm-xt R: c-addr )
6607: execute ( R: c-addr ) \ disassemble operands
6608: r> count type ; \ print name
6609: @end example
6610:
6611: Note that the tables here (in contrast to above) do the @code{cells +}
6612: by themselves (that's why you have to pass an xt). This word is used in
6613: the following way:
6614:
6615: @example
6616: ' @var{disasm-operands} ' @var{table} define-format @var{inst-format}
6617: @end example
6618:
6619: In terms of currying, this kind of two-level defining word provides the
6620: parameters in three stages: first @var{disasm-operands} and @var{table},
6621: then @var{entry-num} and @var{inst-name}, finally @code{addr w}, i.e.,
6622: the instruction to be disassembled.
6623:
6624: Of course this did not quite fit all the instruction format names used
6625: in @file{insts.fs}, so I had to define a few wrappers that conditioned
6626: the parameters into the right form.
6627:
6628: If you have trouble following this section, don't worry. First, this is
6629: involved and takes time (and probably some playing around) to
6630: understand; second, this is the first two-level
6631: @code{create}/@code{does>} word I have written in seventeen years of
6632: Forth; and if I did not have @file{insts.fs} to start with, I may well
6633: have elected to use just a one-level defining word (with some repeating
6634: of parameters when using the defining word). So it is not necessary to
6635: understand this, but it may improve your understanding of Forth.
6636:
6637:
6638: @node Deferred words, Aliases, User-defined Defining Words, Defining Words
6639: @subsection Deferred words
6640: @cindex deferred words
6641:
6642: The defining word @code{Defer} allows you to define a word by name
6643: without defining its behaviour; the definition of its behaviour is
6644: deferred. Here are two situation where this can be useful:
6645:
6646: @itemize @bullet
6647: @item
6648: Where you want to allow the behaviour of a word to be altered later, and
6649: for all precompiled references to the word to change when its behaviour
6650: is changed.
6651: @item
6652: For mutual recursion; @xref{Calls and returns}.
6653: @end itemize
6654:
6655: In the following example, @code{foo} always invokes the version of
6656: @code{greet} that prints ``@code{Good morning}'' whilst @code{bar}
6657: always invokes the version that prints ``@code{Hello}''. There is no way
6658: of getting @code{foo} to use the later version without re-ordering the
6659: source code and recompiling it.
6660:
6661: @example
6662: : greet ." Good morning" ;
6663: : foo ... greet ... ;
6664: : greet ." Hello" ;
6665: : bar ... greet ... ;
6666: @end example
6667:
6668: This problem can be solved by defining @code{greet} as a @code{Defer}red
6669: word. The behaviour of a @code{Defer}red word can be defined and
6670: redefined at any time by using @code{IS} to associate the xt of a
6671: previously-defined word with it. The previous example becomes:
6672:
6673: @example
6674: Defer greet ( -- )
6675: : foo ... greet ... ;
6676: : bar ... greet ... ;
6677: : greet1 ( -- ) ." Good morning" ;
6678: : greet2 ( -- ) ." Hello" ;
6679: ' greet2 <IS> greet \ make greet behave like greet2
6680: @end example
6681:
6682: @progstyle
6683: You should write a stack comment for every deferred word, and put only
6684: XTs into deferred words that conform to this stack effect. Otherwise
6685: it's too difficult to use the deferred word.
6686:
6687: A deferred word can be used to improve the statistics-gathering example
6688: from @ref{User-defined Defining Words}; rather than edit the
6689: application's source code to change every @code{:} to a @code{my:}, do
6690: this:
6691:
6692: @example
6693: : real: : ; \ retain access to the original
6694: defer : \ redefine as a deferred word
6695: ' my: <IS> : \ use special version of :
6696: \
6697: \ load application here
6698: \
6699: ' real: <IS> : \ go back to the original
6700: @end example
6701:
6702:
6703: One thing to note is that @code{<IS>} consumes its name when it is
6704: executed. If you want to specify the name at compile time, use
6705: @code{[IS]}:
6706:
6707: @example
6708: : set-greet ( xt -- )
6709: [IS] greet ;
6710:
6711: ' greet1 set-greet
6712: @end example
6713:
6714: A deferred word can only inherit execution semantics from the xt
6715: (because that is all that an xt can represent -- for more discussion of
6716: this @pxref{Tokens for Words}); by default it will have default
6717: interpretation and compilation semantics deriving from this execution
6718: semantics. However, you can change the interpretation and compilation
6719: semantics of the deferred word in the usual ways:
6720:
6721: @example
6722: : bar .... ; compile-only
6723: Defer fred immediate
6724: Defer jim
6725:
6726: ' bar <IS> jim \ jim has default semantics
6727: ' bar <IS> fred \ fred is immediate
6728: @end example
6729:
6730: doc-defer
6731: doc-<is>
6732: doc-[is]
6733: doc-is
6734: @comment TODO document these: what's defers [is]
6735: doc-what's
6736: doc-defers
6737:
6738: @c Use @code{words-deferred} to see a list of deferred words.
6739:
6740: Definitions in ANS Forth for @code{defer}, @code{<is>} and @code{[is]}
6741: are provided in @file{compat/defer.fs}.
6742:
6743:
6744: @node Aliases, , Deferred words, Defining Words
6745: @subsection Aliases
6746: @cindex aliases
6747:
6748: The defining word @code{Alias} allows you to define a word by name that
6749: has the same behaviour as some other word. Here are two situation where
6750: this can be useful:
6751:
6752: @itemize @bullet
6753: @item
6754: When you want access to a word's definition from a different word list
6755: (for an example of this, see the definition of the @code{Root} word list
6756: in the Gforth source).
6757: @item
6758: When you want to create a synonym; a definition that can be known by
6759: either of two names (for example, @code{THEN} and @code{ENDIF} are
6760: aliases).
6761: @end itemize
6762:
6763: Like deferred words, an alias has default compilation and interpretation
6764: semantics at the beginning (not the modifications of the other word),
6765: but you can change them in the usual ways (@code{immediate},
6766: @code{compile-only}). For example:
6767:
6768: @example
6769: : foo ... ; immediate
6770:
6771: ' foo Alias bar \ bar is not an immediate word
6772: ' foo Alias fooby immediate \ fooby is an immediate word
6773: @end example
6774:
6775: Words that are aliases have the same xt, different headers in the
6776: dictionary, and consequently different name tokens (@pxref{Tokens for
6777: Words}) and possibly different immediate flags. An alias can only have
6778: default or immediate compilation semantics; you can define aliases for
6779: combined words with @code{interpret/compile:} -- see @ref{Combined words}.
6780:
6781: doc-alias
6782:
6783:
6784: @node Interpretation and Compilation Semantics, Tokens for Words, Defining Words, Words
6785: @section Interpretation and Compilation Semantics
6786: @cindex semantics, interpretation and compilation
6787:
6788: @cindex interpretation semantics
6789: The @dfn{interpretation semantics} of a word are what the text
6790: interpreter does when it encounters the word in interpret state. It also
6791: appears in some other contexts, e.g., the execution token returned by
6792: @code{' @i{word}} identifies the interpretation semantics of
6793: @i{word} (in other words, @code{' @i{word} execute} is equivalent to
6794: interpret-state text interpretation of @code{@i{word}}).
6795:
6796: @cindex compilation semantics
6797: The @dfn{compilation semantics} of a word are what the text interpreter
6798: does when it encounters the word in compile state. It also appears in
6799: other contexts, e.g, @code{POSTPONE @i{word}} compiles@footnote{In
6800: standard terminology, ``appends to the current definition''.} the
6801: compilation semantics of @i{word}.
6802:
6803: @cindex execution semantics
6804: The standard also talks about @dfn{execution semantics}. They are used
6805: only for defining the interpretation and compilation semantics of many
6806: words. By default, the interpretation semantics of a word are to
6807: @code{execute} its execution semantics, and the compilation semantics of
6808: a word are to @code{compile,} its execution semantics.@footnote{In
6809: standard terminology: The default interpretation semantics are its
6810: execution semantics; the default compilation semantics are to append its
6811: execution semantics to the execution semantics of the current
6812: definition.}
6813:
6814: @comment TODO expand, make it co-operate with new sections on text interpreter.
6815:
6816: @cindex immediate words
6817: @cindex compile-only words
6818: You can change the semantics of the most-recently defined word:
6819:
6820:
6821: doc-immediate
6822: doc-compile-only
6823: doc-restrict
6824:
6825:
6826: Note that ticking (@code{'}) a compile-only word gives an error
6827: (``Interpreting a compile-only word'').
6828:
6829: @menu
6830: * Combined words::
6831: @end menu
6832:
6833: @node Combined words, , Interpretation and Compilation Semantics, Interpretation and Compilation Semantics
6834: @subsection Combined Words
6835: @cindex combined words
6836:
6837: Gforth allows you to define @dfn{combined words} -- words that have an
6838: arbitrary combination of interpretation and compilation semantics.
6839:
6840:
6841: doc-interpret/compile:
6842:
6843:
6844: This feature was introduced for implementing @code{TO} and @code{S"}. I
6845: recommend that you do not define such words, as cute as they may be:
6846: they make it hard to get at both parts of the word in some contexts.
6847: E.g., assume you want to get an execution token for the compilation
6848: part. Instead, define two words, one that embodies the interpretation
6849: part, and one that embodies the compilation part. Once you have done
6850: that, you can define a combined word with @code{interpret/compile:} for
6851: the convenience of your users.
6852:
6853: You might try to use this feature to provide an optimizing
6854: implementation of the default compilation semantics of a word. For
6855: example, by defining:
6856: @example
6857: :noname
6858: foo bar ;
6859: :noname
6860: POSTPONE foo POSTPONE bar ;
6861: interpret/compile: opti-foobar
6862: @end example
6863:
6864: @noindent
6865: as an optimizing version of:
6866:
6867: @example
6868: : foobar
6869: foo bar ;
6870: @end example
6871:
6872: Unfortunately, this does not work correctly with @code{[compile]},
6873: because @code{[compile]} assumes that the compilation semantics of all
6874: @code{interpret/compile:} words are non-default. I.e., @code{[compile]
6875: opti-foobar} would compile compilation semantics, whereas
6876: @code{[compile] foobar} would compile interpretation semantics.
6877:
6878: @cindex state-smart words (are a bad idea)
6879: Some people try to use @dfn{state-smart} words to emulate the feature provided
6880: by @code{interpret/compile:} (words are state-smart if they check
6881: @code{STATE} during execution). E.g., they would try to code
6882: @code{foobar} like this:
6883:
6884: @example
6885: : foobar
6886: STATE @@
6887: IF ( compilation state )
6888: POSTPONE foo POSTPONE bar
6889: ELSE
6890: foo bar
6891: ENDIF ; immediate
6892: @end example
6893:
6894: Although this works if @code{foobar} is only processed by the text
6895: interpreter, it does not work in other contexts (like @code{'} or
6896: @code{POSTPONE}). E.g., @code{' foobar} will produce an execution token
6897: for a state-smart word, not for the interpretation semantics of the
6898: original @code{foobar}; when you execute this execution token (directly
6899: with @code{EXECUTE} or indirectly through @code{COMPILE,}) in compile
6900: state, the result will not be what you expected (i.e., it will not
6901: perform @code{foo bar}). State-smart words are a bad idea. Simply don't
6902: write them@footnote{For a more detailed discussion of this topic, see
6903: M. Anton Ertl,
6904: @cite{@uref{http://www.complang.tuwien.ac.at/papers/ertl98.ps.gz,@code{State}-smartness---Why
6905: it is Evil and How to Exorcise it}}, EuroForth '98.}!
6906:
6907: @cindex defining words with arbitrary semantics combinations
6908: It is also possible to write defining words that define words with
6909: arbitrary combinations of interpretation and compilation semantics. In
6910: general, they look like this:
6911:
6912: @example
6913: : def-word
6914: create-interpret/compile
6915: @i{code1}
6916: interpretation>
6917: @i{code2}
6918: <interpretation
6919: compilation>
6920: @i{code3}
6921: <compilation ;
6922: @end example
6923:
6924: For a @i{word} defined with @code{def-word}, the interpretation
6925: semantics are to push the address of the body of @i{word} and perform
6926: @i{code2}, and the compilation semantics are to push the address of
6927: the body of @i{word} and perform @i{code3}. E.g., @code{constant}
6928: can also be defined like this (except that the defined constants don't
6929: behave correctly when @code{[compile]}d):
6930:
6931: @example
6932: : constant ( n "name" -- )
6933: create-interpret/compile
6934: ,
6935: interpretation> ( -- n )
6936: @@
6937: <interpretation
6938: compilation> ( compilation. -- ; run-time. -- n )
6939: @@ postpone literal
6940: <compilation ;
6941: @end example
6942:
6943:
6944: doc-create-interpret/compile
6945: doc-interpretation>
6946: doc-<interpretation
6947: doc-compilation>
6948: doc-<compilation
6949:
6950:
6951: Words defined with @code{interpret/compile:} and
6952: @code{create-interpret/compile} have an extended header structure that
6953: differs from other words; however, unless you try to access them with
6954: plain address arithmetic, you should not notice this. Words for
6955: accessing the header structure usually know how to deal with this; e.g.,
6956: @code{'} @i{word} @code{>body} also gives you the body of a word created
6957: with @code{create-interpret/compile}.
6958:
6959:
6960: doc-postpone
6961:
6962: @comment TODO -- expand glossary text for POSTPONE
6963:
6964:
6965: @c -------------------------------------------------------------
6966: @node Tokens for Words, The Text Interpreter, Interpretation and Compilation Semantics, Words
6967: @section Tokens for Words
6968: @cindex tokens for words
6969:
6970: This section describes the creation and use of tokens that represent
6971: words.
6972:
6973: Named words have information stored in their header space entries to
6974: indicate any non-default semantics (@pxref{Interpretation and
6975: Compilation Semantics}). The semantics can be modified, using
6976: @code{immediate} and/or @code{compile-only}, at the time that the words
6977: are defined. Unnamed words have (by definition) no header space
6978: entry, and therefore must have default semantics.
6979:
6980: Named words have interpretation and compilation semantics. Unnamed words
6981: just have execution semantics.
6982:
6983: @cindex xt
6984: @cindex execution token
6985: The execution semantics of an unnamed word are represented by an
6986: @dfn{execution token} (@i{xt}). As explained in @ref{Supplying names},
6987: the execution token of the last word defined can be produced with
6988: @code{lastxt}.
6989:
6990: The interpretation semantics of a named word are also represented by an
6991: execution token. You can produce the execution token using @code{'} or
6992: @code{[']}. A simple example shows the difference between the two:
6993:
6994: @example
6995: : greet ( -- ) ." Hello" ;
6996: : foo ( -- xt ) ['] greet execute ; \ ['] parses greet at compile-time
6997: : bar ( -- ) ' execute ; \ ' parses at run-time
6998:
6999: \ the next four lines all do the same thing
7000: foo
7001: bar greet
7002: greet
7003: ' greet EXECUTE
7004: @end example
7005:
7006: An execution token occupies one cell.
7007: @cindex code field address
7008: @cindex CFA
7009: In Gforth, the abstract data type @i{execution token} is implemented
7010: as a code field address (CFA).
7011: @comment TODO note that the standard does not say what it represents..
7012: @comment and you cannot necessarily compile it in all Forths (eg native
7013: @comment compilers?).
7014:
7015: For literals, use @code{'} in interpreted code and @code{[']} in
7016: compiled code. Gforth's @code{'} and @code{[']} behave somewhat
7017: unusually by complaining about compile-only words. To get the execution
7018: token for a compile-only word @i{name}, use @code{COMP' @i{name} DROP}
7019: or @code{[COMP'] @i{name} DROP}.
7020:
7021: @cindex compilation token
7022: The compilation semantics of a named word are represented by a
7023: @dfn{compilation token} consisting of two cells: @i{w xt}. The top cell
7024: @i{xt} is an execution token. The compilation semantics represented by
7025: the compilation token can be performed with @code{execute}, which
7026: consumes the whole compilation token, with an additional stack effect
7027: determined by the represented compilation semantics.
7028:
7029: At present, the @i{w} part of a compilation token is an execution token,
7030: and the @i{xt} part represents either @code{execute} or
7031: @code{compile,}@footnote{Depending upon the compilation semantics of the
7032: word. If the word has default compilation semantics, the @i{xt} will
7033: represent @code{compile,}. Otherwise (e.g., for immediate words), the
7034: @i{xt} will represent @code{execute}.}. However, don't rely on that
7035: knowledge, unless necessary; future versions of Gforth may introduce
7036: unusual compilation tokens (e.g., a compilation token that represents
7037: the compilation semantics of a literal).
7038:
7039: You can compile the compilation semantics with @code{postpone,}. I.e.,
7040: @code{COMP' @i{word} postpone,} is equivalent to @code{postpone
7041: @i{word}}.
7042:
7043: @cindex name token
7044: @cindex name field address
7045: @cindex NFA
7046: Named words are also represented by the @dfn{name token}, (@i{nt}). In
7047: Gforth, the abstract data type @emph{name token} is implemented as a
7048: name field address (NFA).
7049:
7050:
7051: doc-execute
7052: doc-perform
7053: doc-compile,
7054: doc-[']
7055: doc-'
7056: doc-[comp']
7057: doc-comp'
7058: doc-postpone,
7059:
7060: doc-find-name
7061: doc-name>int
7062: doc-name?int
7063: doc-name>comp
7064: doc-name>string
7065:
7066:
7067: @c ----------------------------------------------------------
7068: @node The Text Interpreter, Word Lists, Tokens for Words, Words
7069: @section The Text Interpreter
7070: @cindex interpreter - outer
7071: @cindex text interpreter
7072: @cindex outer interpreter
7073:
7074: @c Should we really describe all these ugly details? IMO the text
7075: @c interpreter should be much cleaner, but that may not be possible within
7076: @c ANS Forth. - anton
7077: @c nac-> I wanted to explain how it works to show how you can exploit
7078: @c it in your own programs. When I was writing a cross-compiler, figuring out
7079: @c some of these gory details was very helpful to me. None of the textbooks
7080: @c I've seen cover it, and the most modern Forth textbook -- Forth Inc's,
7081: @c seems to positively avoid going into too much detail for some of
7082: @c the internals.
7083:
7084: The text interpreter@footnote{This is an expanded version of the
7085: material in @ref{Introducing the Text Interpreter}.} is an endless loop
7086: that processes input from the current input device. It is also called
7087: the outer interpreter, in contrast to the inner interpreter
7088: (@pxref{Engine}) which executes the compiled Forth code on interpretive
7089: implementations.
7090:
7091: @cindex interpret state
7092: @cindex compile state
7093: The text interpreter operates in one of two states: @dfn{interpret
7094: state} and @dfn{compile state}. The current state is defined by the
7095: aptly-named variable, @code{state}.
7096:
7097: This section starts by describing how the text interpreter behaves when
7098: it is in interpret state, processing input from the user input device --
7099: the keyboard. This is the mode that a Forth system is in after it starts
7100: up.
7101:
7102: @cindex input buffer
7103: @cindex terminal input buffer
7104: The text interpreter works from an area of memory called the @dfn{input
7105: buffer}@footnote{When the text interpreter is processing input from the
7106: keyboard, this area of memory is called the @dfn{terminal input buffer}
7107: (TIB) and is addressed by the (obsolescent) words @code{TIB} and
7108: @code{#TIB}.}, which stores your keyboard input when you press the
7109: @key{RET} key. Starting at the beginning of the input buffer, it skips
7110: leading spaces (called @dfn{delimiters}) then parses a string (a
7111: sequence of non-space characters) until it reaches either a space
7112: character or the end of the buffer. Having parsed a string, it makes two
7113: attempts to process it:
7114:
7115: @cindex dictionary
7116: @itemize @bullet
7117: @item
7118: It looks for the string in a @dfn{dictionary} of definitions. If the
7119: string is found, the string names a @dfn{definition} (also known as a
7120: @dfn{word}) and the dictionary search returns information that allows
7121: the text interpreter to perform the word's @dfn{interpretation
7122: semantics}. In most cases, this simply means that the word will be
7123: executed.
7124: @item
7125: If the string is not found in the dictionary, the text interpreter
7126: attempts to treat it as a number, using the rules described in
7127: @ref{Number Conversion}. If the string represents a legal number in the
7128: current radix, the number is pushed onto a parameter stack (the data
7129: stack for integers, the floating-point stack for floating-point
7130: numbers).
7131: @end itemize
7132:
7133: If both attempts fail, or if the word is found in the dictionary but has
7134: no interpretation semantics@footnote{This happens if the word was
7135: defined as @code{COMPILE-ONLY}.} the text interpreter discards the
7136: remainder of the input buffer, issues an error message and waits for
7137: more input. If one of the attempts succeeds, the text interpreter
7138: repeats the parsing process until the whole of the input buffer has been
7139: processed, at which point it prints the status message ``@code{ ok}''
7140: and waits for more input.
7141:
7142: @cindex parse area
7143: The text interpreter keeps track of its position in the input buffer by
7144: updating a variable called @code{>IN} (pronounced ``to-in''). The value
7145: of @code{>IN} starts out as 0, indicating an offset of 0 from the start
7146: of the input buffer. The region from offset @code{>IN @@} to the end of
7147: the input buffer is called the @dfn{parse area}@footnote{In other words,
7148: the text interpreter processes the contents of the input buffer by
7149: parsing strings from the parse area until the parse area is empty.}.
7150: This example shows how @code{>IN} changes as the text interpreter parses
7151: the input buffer:
7152:
7153: @example
7154: : remaining >IN @@ SOURCE 2 PICK - -ROT + SWAP
7155: CR ." ->" TYPE ." <-" ; IMMEDIATE
7156:
7157: 1 2 3 remaining + remaining .
7158:
7159: : foo 1 2 3 remaining SWAP remaining ;
7160: @end example
7161:
7162: @noindent
7163: The result is:
7164:
7165: @example
7166: ->+ remaining .<-
7167: ->.<-5 ok
7168:
7169: ->SWAP remaining ;-<
7170: ->;<- ok
7171: @end example
7172:
7173: @cindex parsing words
7174: The value of @code{>IN} can also be modified by a word in the input
7175: buffer that is executed by the text interpreter. This means that a word
7176: can ``trick'' the text interpreter into either skipping a section of the
7177: input buffer@footnote{This is how parsing words work.} or into parsing a
7178: section twice. For example:
7179:
7180: @example
7181: : lat ." <<lat>>" ;
7182: : flat ." <<flat>>" >IN DUP @@ 3 - SWAP ! ;
7183: @end example
7184:
7185: @noindent
7186: When @code{flat} is executed, this output is produced@footnote{Exercise
7187: for the reader: what would happen if the @code{3} were replaced with
7188: @code{4}?}:
7189:
7190: @example
7191: <<flat>><<lat>>
7192: @end example
7193:
7194: @noindent
7195: Two important notes about the behaviour of the text interpreter:
7196:
7197: @itemize @bullet
7198: @item
7199: It processes each input string to completion before parsing additional
7200: characters from the input buffer.
7201: @item
7202: It treats the input buffer as a read-only region (and so must your code).
7203: @end itemize
7204:
7205: @noindent
7206: When the text interpreter is in compile state, its behaviour changes in
7207: these ways:
7208:
7209: @itemize @bullet
7210: @item
7211: If a parsed string is found in the dictionary, the text interpreter will
7212: perform the word's @dfn{compilation semantics}. In most cases, this
7213: simply means that the execution semantics of the word will be appended
7214: to the current definition.
7215: @item
7216: When a number is encountered, it is compiled into the current definition
7217: (as a literal) rather than being pushed onto a parameter stack.
7218: @item
7219: If an error occurs, @code{state} is modified to put the text interpreter
7220: back into interpret state.
7221: @item
7222: Each time a line is entered from the keyboard, Gforth prints
7223: ``@code{ compiled}'' rather than `` @code{ok}''.
7224: @end itemize
7225:
7226: @cindex text interpreter - input sources
7227: When the text interpreter is using an input device other than the
7228: keyboard, its behaviour changes in these ways:
7229:
7230: @itemize @bullet
7231: @item
7232: When the parse area is empty, the text interpreter attempts to refill
7233: the input buffer from the input source. When the input source is
7234: exhausted, the input source is set back to the user input device.
7235: @item
7236: It doesn't print out ``@code{ ok}'' or ``@code{ compiled}'' messages each
7237: time the parse area is emptied.
7238: @item
7239: If an error occurs, the input source is set back to the user input
7240: device.
7241: @end itemize
7242:
7243: You can read about this in more detail in @ref{Input Sources}.
7244:
7245: doc->in
7246: doc-source
7247:
7248: doc-tib
7249: doc-#tib
7250:
7251:
7252: @menu
7253: * Input Sources::
7254: * Number Conversion::
7255: * Interpret/Compile states::
7256: * Literals::
7257: * Interpreter Directives::
7258: @end menu
7259:
7260: @node Input Sources, Number Conversion, The Text Interpreter, The Text Interpreter
7261: @subsection Input Sources
7262: @cindex input sources
7263: @cindex text interpreter - input sources
7264:
7265: By default, the text interpreter processes input from the user input
7266: device (the keyboard) when Forth starts up. The text interpreter can
7267: process input from any of these sources:
7268:
7269: @itemize @bullet
7270: @item
7271: The user input device -- the keyboard.
7272: @item
7273: A file, using the words described in @ref{Forth source files}.
7274: @item
7275: A block, using the words described in @ref{Blocks}.
7276: @item
7277: A text string, using @code{evaluate}.
7278: @end itemize
7279:
7280: A program can identify the current input device from the values of
7281: @code{source-id} and @code{blk}.
7282:
7283:
7284: doc-source-id
7285: doc-blk
7286:
7287: doc-save-input
7288: doc-restore-input
7289:
7290: doc-evaluate
7291:
7292:
7293:
7294: @node Number Conversion, Interpret/Compile states, Input Sources, The Text Interpreter
7295: @subsection Number Conversion
7296: @cindex number conversion
7297: @cindex double-cell numbers, input format
7298: @cindex input format for double-cell numbers
7299: @cindex single-cell numbers, input format
7300: @cindex input format for single-cell numbers
7301: @cindex floating-point numbers, input format
7302: @cindex input format for floating-point numbers
7303:
7304: This section describes the rules that the text interpreter uses when it
7305: tries to convert a string into a number.
7306:
7307: Let <digit> represent any character that is a legal digit in the current
7308: number base@footnote{For example, 0-9 when the number base is decimal or
7309: 0-9, A-F when the number base is hexadecimal.}.
7310:
7311: Let <decimal digit> represent any character in the range 0-9.
7312:
7313: Let @{@i{a b}@} represent the @i{optional} presence of any of the characters
7314: in the braces (@i{a} or @i{b} or neither).
7315:
7316: Let * represent any number of instances of the previous character
7317: (including none).
7318:
7319: Let any other character represent itself.
7320:
7321: @noindent
7322: Now, the conversion rules are:
7323:
7324: @itemize @bullet
7325: @item
7326: A string of the form <digit><digit>* is treated as a single-precision
7327: (cell-sized) positive integer. Examples are 0 123 6784532 32343212343456 42
7328: @item
7329: A string of the form -<digit><digit>* is treated as a single-precision
7330: (cell-sized) negative integer, and is represented using 2's-complement
7331: arithmetic. Examples are -45 -5681 -0
7332: @item
7333: A string of the form <digit><digit>*.<digit>* is treated as a double-precision
7334: (double-cell-sized) positive integer. Examples are 3465. 3.465 34.65
7335: (all three of these represent the same number).
7336: @item
7337: A string of the form -<digit><digit>*.<digit>* is treated as a
7338: double-precision (double-cell-sized) negative integer, and is
7339: represented using 2's-complement arithmetic. Examples are -3465. -3.465
7340: -34.65 (all three of these represent the same number).
7341: @item
7342: A string of the form @{+ -@}<decimal digit>@{.@}<decimal digit>*@{e
7343: E@}@{+ -@}<decimal digit><decimal digit>* is treated as a floating-point
7344: number. Examples are 1e 1e0 1.e 1.e0 +1e+0 (which all represent the same
7345: number) +12.E-4
7346: @end itemize
7347:
7348: By default, the number base used for integer number conversion is given
7349: by the contents of the variable @code{base}. Note that a lot of
7350: confusion can result from unexpected values of @code{base}. If you
7351: change @code{base} anywhere, make sure to save the old value and restore
7352: it afterwards. In general I recommend keeping @code{base} decimal, and
7353: using the prefixes described below for the popular non-decimal bases.
7354:
7355: doc-dpl
7356: doc-base
7357: doc-hex
7358: doc-decimal
7359:
7360:
7361: @cindex '-prefix for character strings
7362: @cindex &-prefix for decimal numbers
7363: @cindex %-prefix for binary numbers
7364: @cindex $-prefix for hexadecimal numbers
7365: Gforth allows you to override the value of @code{base} by using a
7366: prefix@footnote{Some Forth implementations provide a similar scheme by
7367: implementing @code{$} etc. as parsing words that process the subsequent
7368: number in the input stream and push it onto the stack. For example, see
7369: @cite{Number Conversion and Literals}, by Wil Baden; Forth Dimensions
7370: 20(3) pages 26--27. In such implementations, unlike in Gforth, a space
7371: is required between the prefix and the number.} before the first digit
7372: of an (integer) number. Four prefixes are supported:
7373:
7374: @itemize @bullet
7375: @item
7376: @code{&} -- decimal
7377: @item
7378: @code{%} -- binary
7379: @item
7380: @code{$} -- hexadecimal
7381: @item
7382: @code{'} -- base @code{max-char+1}
7383: @end itemize
7384:
7385: Here are some examples, with the equivalent decimal number shown after
7386: in braces:
7387:
7388: -$41 (-65), %1001101 (205), %1001.0001 (145 - a double-precision number),
7389: 'AB (16706; ascii A is 65, ascii B is 66, number is 65*256 + 66),
7390: 'ab (24930; ascii a is 97, ascii B is 98, number is 97*256 + 98),
7391: &905 (905), $abc (2478), $ABC (2478).
7392:
7393: @cindex number conversion - traps for the unwary
7394: @noindent
7395: Number conversion has a number of traps for the unwary:
7396:
7397: @itemize @bullet
7398: @item
7399: You cannot determine the current number base using the code sequence
7400: @code{base @@ .} -- the number base is always 10 in the current number
7401: base. Instead, use something like @code{base @@ dec.}
7402: @item
7403: If the number base is set to a value greater than 14 (for example,
7404: hexadecimal), the number 123E4 is ambiguous; the conversion rules allow
7405: it to be intepreted as either a single-precision integer or a
7406: floating-point number (Gforth treats it as an integer). The ambiguity
7407: can be resolved by explicitly stating the sign of the mantissa and/or
7408: exponent: 123E+4 or +123E4 -- if the number base is decimal, no
7409: ambiguity arises; either representation will be treated as a
7410: floating-point number.
7411: @item
7412: There is a word @code{bin} but it does @i{not} set the number base!
7413: It is used to specify file types.
7414: @item
7415: ANS Forth requires the @code{.} of a double-precision number to
7416: be the final character in the string. Allowing the @code{.} to be
7417: anywhere after the first digit is a Gforth extension.
7418: @item
7419: The number conversion process does not check for overflow.
7420: @item
7421: In Gforth, number conversion to floating-point numbers always use base
7422: 10, irrespective of the value of @code{base}. In ANS Forth,
7423: conversion to floating-point numbers whilst the value of
7424: @code{base} is not 10 is an ambiguous condition.
7425: @end itemize
7426:
7427: You can read numbers into your programs with the words described in
7428: @ref{Input}.
7429:
7430: @node Interpret/Compile states, Literals, Number Conversion, The Text Interpreter
7431: @subsection Interpret/Compile states
7432: @cindex Interpret/Compile states
7433:
7434: A standard program is not permitted to change @code{state}
7435: explicitly. However, it can change @code{state} implicitly, using the
7436: words @code{[} and @code{]}. When @code{[} is executed it switches
7437: @code{state} to interpret state, and therefore the text interpreter
7438: starts interpreting. When @code{]} is executed it switches @code{state}
7439: to compile state and therefore the text interpreter starts
7440: compiling. The most common usage for these words is for switching into
7441: interpret state and back from within a colon definition; this technique
7442: can be used to compile a literal (for an example, @pxref{Literals}) or
7443: for conditional compilation (for an example, @pxref{Interpreter
7444: Directives}).
7445:
7446:
7447: @c This is a bad example: It's non-standard, and it's not necessary.
7448: @c However, I can't think of a good example for switching into compile
7449: @c state when there is no current word (@code{state}-smart words are not a
7450: @c good reason). So maybe we should use an example for switching into
7451: @c interpret @code{state} in a colon def. - anton
7452: @c nac-> I agree. I started out by putting in the example, then realised
7453: @c that it was non-ANS, so wrote more words around it. I hope this
7454: @c re-written version is acceptable to you. I do want to keep the example
7455: @c as it is helpful for showing what is and what is not portable, particularly
7456: @c where it outlaws a style in common use.
7457:
7458:
7459: @code{[} and @code{]} also give you the ability to switch into compile
7460: state and back, but we cannot think of any useful Standard application
7461: for this ability. Pre-ANS Forth textbooks have examples like this:
7462:
7463: @example
7464: : AA ." this is A" ;
7465: : BB ." this is B" ;
7466: : CC ." this is C" ;
7467:
7468: create table ] aa bb cc [
7469:
7470: : go ( n -- ) \ n is offset into table.. 0 for 1st entry
7471: cells table + @ execute ;
7472: @end example
7473:
7474: This example builds a jump table; @code{0 go} will display ``@code{this
7475: is A}''. Using @code{[} and @code{]} in this example is equivalent to
7476: defining @code{table} like this:
7477:
7478: @example
7479: create table ' aa COMPILE, ' bb COMPILE, ' cc COMPILE,
7480: @end example
7481:
7482: The problem with this code is that the definition of @code{table} is not
7483: portable -- it @i{compile}s execution tokens into code space. Whilst it
7484: @i{may} work on systems where code space and data space co-incide, the
7485: Standard only allows data space to be assigned for a @code{CREATE}d
7486: word. In addition, the Standard only allows @code{@@} to access data
7487: space, whilst this example is using it to access code space. The only
7488: portable, Standard way to build this table is to build it in data space,
7489: like this:
7490:
7491: @example
7492: create table ' aa , ' bb , ' cc ,
7493: @end example
7494:
7495: doc-state
7496: doc-[
7497: doc-]
7498:
7499:
7500: @node Literals, Interpreter Directives, Interpret/Compile states, The Text Interpreter
7501: @subsection Literals
7502: @cindex Literals
7503:
7504: Often, you want to use a number within a colon definition. When you do
7505: this, the text interpreter automatically compiles the number as a
7506: @i{literal}. A literal is a number whose run-time effect is to be pushed
7507: onto the stack. If you had to do some maths to generate the number, you
7508: might write it like this:
7509:
7510: @example
7511: : HOUR-TO-SEC ( n1 -- n2 )
7512: 60 * \ to minutes
7513: 60 * ; \ to seconds
7514: @end example
7515:
7516: It is very clear what this definition is doing, but it's inefficient
7517: since it is performing 2 multiples at run-time. An alternative would be
7518: to write:
7519:
7520: @example
7521: : HOUR-TO-SEC ( n1 -- n2 )
7522: 3600 * ; \ to seconds
7523: @end example
7524:
7525: Which does the same thing, and has the advantage of using a single
7526: multiply. Ideally, we'd like the efficiency of the second with the
7527: readability of the first.
7528:
7529: @code{Literal} allows us to achieve that. It takes a number from the
7530: stack and lays it down in the current definition just as though the
7531: number had been typed directly into the definition. Our first attempt
7532: might look like this:
7533:
7534: @example
7535: 60 \ mins per hour
7536: 60 * \ seconds per minute
7537: : HOUR-TO-SEC ( n1 -- n2 )
7538: Literal * ; \ to seconds
7539: @end example
7540:
7541: But this produces the error message @code{unstructured}. What happened?
7542: The stack notation for @code{:} is (@i{ -- colon-sys}) and the size of
7543: @i{colon-sys} is implementation-defined. In other words, once we start a
7544: colon definition we can't portably access anything that was on the stack
7545: before the definition began@footnote{@cite{Two Problems in ANS Forth},
7546: by Thomas Worthington; Forth Dimensions 20(2) pages 32--34 describes
7547: some situations where you might want to access stack items above
7548: colon-sys, and provides a solution to the problem.}. The correct way of
7549: solving this problem in this instance is to use @code{[ ]} like this:
7550:
7551: @example
7552: : HOUR-TO-SEC ( n1 -- n2 )
7553: [ 60 \ minutes per hour
7554: 60 * ] \ seconds per minute
7555: LITERAL * ; \ to seconds
7556: @end example
7557:
7558:
7559: doc-literal
7560: doc-]L
7561: doc-2literal
7562: doc-fliteral
7563:
7564:
7565: @node Interpreter Directives, , Literals, The Text Interpreter
7566: @subsection Interpreter Directives
7567: @cindex interpreter directives
7568:
7569: These words are usually used in interpret state; typically to control
7570: which parts of a source file are processed by the text
7571: interpreter. There are only a few ANS Forth Standard words, but Gforth
7572: supplements these with a rich set of immediate control structure words
7573: to compensate for the fact that the non-immediate versions can only be
7574: used in compile state (@pxref{Control Structures}). Typical usages:
7575:
7576: @example
7577: FALSE Constant ASSEMBLER
7578: .
7579: .
7580: ASSEMBLER [IF]
7581: : ASSEMBLER-FEATURE
7582: ...
7583: ;
7584: [ENDIF]
7585: .
7586: .
7587: : SEE
7588: ... \ general-purpose SEE code
7589: [ ASSEMBLER [IF] ]
7590: ... \ assembler-specific SEE code
7591: [ [ENDIF] ]
7592: ;
7593: @end example
7594:
7595:
7596: doc-[IF]
7597: doc-[ELSE]
7598: doc-[THEN]
7599: doc-[ENDIF]
7600:
7601: doc-[IFDEF]
7602: doc-[IFUNDEF]
7603:
7604: doc-[?DO]
7605: doc-[DO]
7606: doc-[FOR]
7607: doc-[LOOP]
7608: doc-[+LOOP]
7609: doc-[NEXT]
7610:
7611: doc-[BEGIN]
7612: doc-[UNTIL]
7613: doc-[AGAIN]
7614: doc-[WHILE]
7615: doc-[REPEAT]
7616:
7617:
7618: @c -------------------------------------------------------------
7619: @node Word Lists, Environmental Queries, The Text Interpreter, Words
7620: @section Word Lists
7621: @cindex word lists
7622: @cindex header space
7623:
7624: A wordlist is a list of named words; you can add new words and look up
7625: words by name (and you can remove words in a restricted way with
7626: markers). Every named (and @code{reveal}ed) word is in one wordlist.
7627:
7628: @cindex search order stack
7629: The text interpreter searches the wordlists present in the search order
7630: (a stack of wordlists), from the top to the bottom. Within each
7631: wordlist, the search starts conceptually at the newest word; i.e., if
7632: two words in a wordlist have the same name, the newer word is found.
7633:
7634: @cindex compilation word list
7635: New words are added to the @dfn{compilation wordlist} (aka current
7636: wordlist).
7637:
7638: @cindex wid
7639: A word list is identified by a cell-sized word list identifier (@i{wid})
7640: in much the same way as a file is identified by a file handle. The
7641: numerical value of the wid has no (portable) meaning, and might change
7642: from session to session.
7643:
7644: The ANS Forth ``Search order'' word set is intended to provide a set of
7645: low-level tools that allow various different schemes to be
7646: implemented. Gforth provides @code{vocabulary}, a traditional Forth
7647: word. @file{compat/vocabulary.fs} provides an implementation in ANS
7648: Forth.
7649:
7650: @comment TODO: locals section refers to here, saying that every word list (aka
7651: @comment vocabulary) has its own methods for searching etc. Need to document that.
7652:
7653: @comment TODO: document markers, reveal, tables, mappedwordlist
7654:
7655: @comment the gforthman- prefix is used to pick out the true definition of a
7656: @comment word from the source files, rather than some alias.
7657:
7658: doc-forth-wordlist
7659: doc-definitions
7660: doc-get-current
7661: doc-set-current
7662: doc-get-order
7663: doc---gforthman-set-order
7664: doc-wordlist
7665: doc-table
7666: doc-push-order
7667: doc-previous
7668: doc-also
7669: doc---gforthman-forth
7670: doc-only
7671: doc---gforthman-order
7672:
7673: doc-find
7674: doc-search-wordlist
7675:
7676: doc-words
7677: doc-vlist
7678: @c doc-words-deferred
7679:
7680: doc-mappedwordlist
7681: doc-root
7682: doc-vocabulary
7683: doc-seal
7684: doc-vocs
7685: doc-current
7686: doc-context
7687:
7688:
7689: @menu
7690: * Why use word lists?::
7691: * Word list examples::
7692: @end menu
7693:
7694: @node Why use word lists?, Word list examples, Word Lists, Word Lists
7695: @subsection Why use word lists?
7696: @cindex word lists - why use them?
7697:
7698: Here are some reasons for using multiple word lists:
7699:
7700: @itemize @bullet
7701: @item
7702: To improve compilation speed by reducing the number of header space
7703: entries that must be searched. This is achieved by creating a new
7704: word list that contains all of the definitions that are used in the
7705: definition of a Forth system but which would not usually be used by
7706: programs running on that system. That word list would be on the search
7707: list when the Forth system was compiled but would be removed from the
7708: search list for normal operation. This can be a useful technique for
7709: low-performance systems (for example, 8-bit processors in embedded
7710: systems) but is unlikely to be necessary in high-performance desktop
7711: systems.
7712: @item
7713: To prevent a set of words from being used outside the context in which
7714: they are valid. Two classic examples of this are an integrated editor
7715: (all of the edit commands are defined in a separate word list; the
7716: search order is set to the editor word list when the editor is invoked;
7717: the old search order is restored when the editor is terminated) and an
7718: integrated assembler (the op-codes for the machine are defined in a
7719: separate word list which is used when a @code{CODE} word is defined).
7720: @item
7721: To prevent a name-space clash between multiple definitions with the same
7722: name. For example, when building a cross-compiler you might have a word
7723: @code{IF} that generates conditional code for your target system. By
7724: placing this definition in a different word list you can control whether
7725: the host system's @code{IF} or the target system's @code{IF} get used in
7726: any particular context by controlling the order of the word lists on the
7727: search order stack.
7728: @end itemize
7729:
7730: @node Word list examples, , Why use word lists?, Word Lists
7731: @subsection Word list examples
7732: @cindex word lists - examples
7733:
7734: Here is an example of creating and using a new wordlist using ANS
7735: Forth Standard words:
7736:
7737: @example
7738: wordlist constant my-new-words-wordlist
7739: : my-new-words get-order nip my-new-words-wordlist swap set-order ;
7740:
7741: \ add it to the search order
7742: also my-new-words
7743:
7744: \ alternatively, add it to the search order and make it
7745: \ the compilation word list
7746: also my-new-words definitions
7747: \ type "order" to see the problem
7748: @end example
7749:
7750: The problem with this example is that @code{order} has no way to
7751: associate the name @code{my-new-words} with the wid of the word list (in
7752: Gforth, @code{order} and @code{vocs} will display @code{???} for a wid
7753: that has no associated name). There is no Standard way of associating a
7754: name with a wid.
7755:
7756: In Gforth, this example can be re-coded using @code{vocabulary}, which
7757: associates a name with a wid:
7758:
7759: @example
7760: vocabulary my-new-words
7761:
7762: \ add it to the search order
7763: also my-new-words
7764:
7765: \ alternatively, add it to the search order and make it
7766: \ the compilation word list
7767: my-new-words definitions
7768: \ type "order" to see that the problem is solved
7769: @end example
7770:
7771: @c -------------------------------------------------------------
7772: @node Environmental Queries, Files, Word Lists, Words
7773: @section Environmental Queries
7774: @cindex environmental queries
7775:
7776: ANS Forth introduced the idea of ``environmental queries'' as a way
7777: for a program running on a system to determine certain characteristics of the system.
7778: The Standard specifies a number of strings that might be recognised by a system.
7779:
7780: The Standard requires that the header space used for environmental queries
7781: be distinct from the header space used for definitions.
7782:
7783: Typically, environmental queries are supported by creating a set of
7784: definitions in a word list that is @i{only} used during environmental
7785: queries; that is what Gforth does. There is no Standard way of adding
7786: definitions to the set of recognised environmental queries, but any
7787: implementation that supports the loading of optional word sets must have
7788: some mechanism for doing this (after loading the word set, the
7789: associated environmental query string must return @code{true}). In
7790: Gforth, the word list used to honour environmental queries can be
7791: manipulated just like any other word list.
7792:
7793:
7794: doc-environment?
7795: doc-environment-wordlist
7796:
7797: doc-gforth
7798: doc-os-class
7799:
7800:
7801: Note that, whilst the documentation for (e.g.) @code{gforth} shows it
7802: returning two items on the stack, querying it using @code{environment?}
7803: will return an additional item; the @code{true} flag that shows that the
7804: string was recognised.
7805:
7806: @comment TODO Document the standard strings or note where they are documented herein
7807:
7808: Here are some examples of using environmental queries:
7809:
7810: @example
7811: s" address-unit-bits" environment? 0=
7812: [IF]
7813: cr .( environmental attribute address-units-bits unknown... ) cr
7814: [THEN]
7815:
7816: s" block" environment? [IF] DROP include block.fs [THEN]
7817:
7818: s" gforth" environment? [IF] 2DROP include compat/vocabulary.fs [THEN]
7819:
7820: s" gforth" environment? [IF] .( Gforth version ) TYPE
7821: [ELSE] .( Not Gforth..) [THEN]
7822: @end example
7823:
7824:
7825: Here is an example of adding a definition to the environment word list:
7826:
7827: @example
7828: get-current environment-wordlist set-current
7829: true constant block
7830: true constant block-ext
7831: set-current
7832: @end example
7833:
7834: You can see what definitions are in the environment word list like this:
7835:
7836: @example
7837: get-order 1+ environment-wordlist swap set-order words previous
7838: @end example
7839:
7840:
7841: @c -------------------------------------------------------------
7842: @node Files, Blocks, Environmental Queries, Words
7843: @section Files
7844: @cindex files
7845: @cindex I/O - file-handling
7846:
7847: Gforth provides facilities for accessing files that are stored in the
7848: host operating system's file-system. Files that are processed by Gforth
7849: can be divided into two categories:
7850:
7851: @itemize @bullet
7852: @item
7853: Files that are processed by the Text Interpreter (@dfn{Forth source files}).
7854: @item
7855: Files that are processed by some other program (@dfn{general files}).
7856: @end itemize
7857:
7858: doc-loadfilename
7859: doc-sourcefilename
7860: doc-sourceline#
7861:
7862: @menu
7863: * Forth source files::
7864: * General files::
7865: * Search Paths::
7866: @end menu
7867:
7868:
7869: @c -------------------------------------------------------------
7870: @node Forth source files, General files, Files, Files
7871: @subsection Forth source files
7872: @cindex including files
7873: @cindex Forth source files
7874:
7875: The simplest way to interpret the contents of a file is to use one of
7876: these two formats:
7877:
7878: @example
7879: include mysource.fs
7880: s" mysource.fs" included
7881: @end example
7882:
7883: Sometimes you want to include a file only if it is not included already
7884: (by, say, another source file). In that case, you can use one of these
7885: three formats:
7886:
7887: @example
7888: require mysource.fs
7889: needs mysource.fs
7890: s" mysource.fs" required
7891: @end example
7892:
7893: @cindex stack effect of included files
7894: @cindex including files, stack effect
7895: It is good practice to write your source files such that interpreting them
7896: does not change the stack. Source files designed in this way can be used with
7897: @code{required} and friends without complications. For example:
7898:
7899: @example
7900: 1 require foo.fs drop
7901: @end example
7902:
7903:
7904: doc-include-file
7905: doc-included
7906: doc-included?
7907: doc-include
7908: doc-required
7909: doc-require
7910: doc-needs
7911: doc-init-included-files
7912:
7913:
7914: A definition in ANS Forth for @code{required} is provided in
7915: @file{compat/required.fs}.
7916:
7917: @c -------------------------------------------------------------
7918: @node General files, Search Paths, Forth source files, Files
7919: @subsection General files
7920: @cindex general files
7921: @cindex file-handling
7922:
7923: Files are opened/created by name and type. The following types are
7924: recognised:
7925:
7926:
7927: doc-r/o
7928: doc-r/w
7929: doc-w/o
7930: doc-bin
7931:
7932:
7933: When a file is opened/created, it returns a file identifier,
7934: @i{wfileid} that is used for all other file commands. All file
7935: commands also return a status value, @i{wior}, that is 0 for a
7936: successful operation and an implementation-defined non-zero value in the
7937: case of an error.
7938:
7939:
7940: doc-open-file
7941: doc-create-file
7942:
7943: doc-close-file
7944: doc-delete-file
7945: doc-rename-file
7946: doc-read-file
7947: doc-read-line
7948: doc-write-file
7949: doc-write-line
7950: doc-emit-file
7951: doc-flush-file
7952:
7953: doc-file-status
7954: doc-file-position
7955: doc-reposition-file
7956: doc-file-size
7957: doc-resize-file
7958:
7959:
7960: @c ---------------------------------------------------------
7961: @node Search Paths, , General files, Files
7962: @subsection Search Paths
7963: @cindex path for @code{included}
7964: @cindex file search path
7965: @cindex @code{include} search path
7966: @cindex search path for files
7967:
7968: If you specify an absolute filename (i.e., a filename starting with
7969: @file{/} or @file{~}, or with @file{:} in the second position (as in
7970: @samp{C:...})) for @code{included} and friends, that file is included
7971: just as you would expect.
7972:
7973: For relative filenames, Gforth uses a search path similar to Forth's
7974: search order (@pxref{Word Lists}). It tries to find the given filename
7975: in the directories present in the path, and includes the first one it
7976: finds. There are separate search paths for Forth source files and
7977: general files.
7978:
7979: If the search path contains the directory @file{.} (as it should), this
7980: refers to the directory that the present file was @code{included}
7981: from. This allows files to include other files relative to their own
7982: position (irrespective of the current working directory or the absolute
7983: position). This feature is essential for libraries consisting of
7984: several files, where a file may include other files from the library.
7985: It corresponds to @code{#include "..."} in C. If the current input
7986: source is not a file, @file{.} refers to the directory of the innermost
7987: file being included, or, if there is no file being included, to the
7988: current working directory.
7989:
7990: Use @file{~+} to refer to the current working directory (as in the
7991: @code{bash}).
7992:
7993: If the filename starts with @file{./}, the search path is not searched
7994: (just as with absolute filenames), and the @file{.} has the same meaning
7995: as described above.
7996:
7997: @menu
7998: * Forth Search Paths::
7999: * General Search Paths::
8000: @end menu
8001:
8002: @c ---------------------------------------------------------
8003: @node Forth Search Paths, General Search Paths, Search Paths, Search Paths
8004: @subsubsection Forth Search Paths
8005: @cindex search path control - Forth
8006:
8007: The search path is initialized when you start Gforth (@pxref{Invoking
8008: Gforth}). You can display it and change it using these words:
8009:
8010:
8011: doc-.fpath
8012: doc-fpath+
8013: doc-fpath=
8014: doc-open-fpath-file
8015:
8016:
8017: @noindent
8018: Here is an example of using @code{fpath} and @code{require}:
8019:
8020: @example
8021: fpath= /usr/lib/forth/|./
8022: require timer.fs
8023: @end example
8024:
8025: @c ---------------------------------------------------------
8026: @node General Search Paths, , Forth Search Paths, Search Paths
8027: @subsubsection General Search Paths
8028: @cindex search path control - for user applications
8029:
8030: Your application may need to search files in several directories, like
8031: @code{included} does. To facilitate this, Gforth allows you to define
8032: and use your own search paths, by providing generic equivalents of the
8033: Forth search path words:
8034:
8035:
8036: doc-.path
8037: doc-path+
8038: doc-path=
8039: doc-open-path-file
8040:
8041:
8042: Here's an example of creating a search path:
8043:
8044: @example
8045: \ Make a buffer for the path:
8046: create mypath 100 chars , \ maximum length (is checked)
8047: 0 , \ real len
8048: 100 chars allot \ space for path
8049: @end example
8050:
8051: @c -------------------------------------------------------------
8052: @node Blocks, Other I/O, Files, Words
8053: @section Blocks
8054: @cindex I/O - blocks
8055: @cindex blocks
8056:
8057: When you run Gforth on a modern desk-top computer, it runs under the
8058: control of an operating system which provides certain services. One of
8059: these services is @var{file services}, which allows Forth source code
8060: and data to be stored in files and read into Gforth (@pxref{Files}).
8061:
8062: Traditionally, Forth has been an important programming language on
8063: systems where it has interfaced directly to the underlying hardware with
8064: no intervening operating system. Forth provides a mechanism, called
8065: @dfn{blocks}, for accessing mass storage on such systems.
8066:
8067: A block is a 1024-byte data area, which can be used to hold data or
8068: Forth source code. No structure is imposed on the contents of the
8069: block. A block is identified by its number; blocks are numbered
8070: contiguously from 1 to an implementation-defined maximum.
8071:
8072: A typical system that used blocks but no operating system might use a
8073: single floppy-disk drive for mass storage, with the disks formatted to
8074: provide 256-byte sectors. Blocks would be implemented by assigning the
8075: first four sectors of the disk to block 1, the second four sectors to
8076: block 2 and so on, up to the limit of the capacity of the disk. The disk
8077: would not contain any file system information, just the set of blocks.
8078:
8079: @cindex blocks file
8080: On systems that do provide file services, blocks are typically
8081: implemented by storing a sequence of blocks within a single @dfn{blocks
8082: file}. The size of the blocks file will be an exact multiple of 1024
8083: bytes, corresponding to the number of blocks it contains. This is the
8084: mechanism that Gforth uses.
8085:
8086: @cindex @file{blocks.fb}
8087: Only 1 blocks file can be open at a time. If you use block words without
8088: having specified a blocks file, Gforth defaults to the blocks file
8089: @file{blocks.fb}. Gforth uses the Forth search path when attempting to
8090: locate a blocks file (@pxref{Forth Search Paths}).
8091:
8092: @cindex block buffers
8093: When you read and write blocks under program control, Gforth uses a
8094: number of @dfn{block buffers} as intermediate storage. These buffers are
8095: not used when you use @code{load} to interpret the contents of a block.
8096:
8097: The behaviour of the block buffers is directly analagous to that of a
8098: cache. Each block buffer has three states:
8099:
8100: @itemize @bullet
8101: @item
8102: Unassigned
8103: @item
8104: Assigned-clean
8105: @item
8106: Assigned-dirty
8107: @end itemize
8108:
8109: Initially, all block buffers are @i{unassigned}. In order to access a
8110: block, the block (specified by its block number) must be assigned to a
8111: block buffer.
8112:
8113: The assignment of a block to a block buffer is performed by @code{block}
8114: or @code{buffer}. Use @code{block} when you wish to modify the existing
8115: contents of a block. Use @code{buffer} when you don't care about the
8116: existing contents of the block@footnote{The ANS Forth definition of
8117: @code{buffer} is intended not to cause disk I/O; if the data associated
8118: with the particular block is already stored in a block buffer due to an
8119: earlier @code{block} command, @code{buffer} will return that block
8120: buffer and the existing contents of the block will be
8121: available. Otherwise, @code{buffer} will simply assign a new, empty
8122: block buffer for the block.}.
8123:
8124: Once a block has been assigned to a block buffer using @code{block} or
8125: @code{buffer}, that block buffer becomes the @i{current block buffer}
8126: and its state changes to @i{assigned-clean}. Data may only be
8127: manipulated (read or written) within the current block buffer.
8128:
8129: When the contents of the current block buffer has been modified it is
8130: necessary, @emph{before calling @code{block} or @code{buffer} again}, to
8131: either abandon the changes (by doing nothing) or commit the changes,
8132: using @code{update}. Using @code{update} does not change the blocks
8133: file; it simply changes a block buffer's state to @i{assigned-dirty}.
8134:
8135: The word @code{flush} causes all @i{assigned-dirty} blocks to be
8136: written back to the blocks file on disk. Leaving Gforth using @code{bye}
8137: also causes a @code{flush} to be performed.
8138:
8139: In Gforth, @code{block} and @code{buffer} use a @i{direct-mapped}
8140: algorithm to assign a block buffer to a block. That means that any
8141: particular block can only be assigned to one specific block buffer,
8142: called (for the particular operation) the @i{victim buffer}. If the
8143: victim buffer is @i{unassigned} or @i{assigned-clean} it is allocated to
8144: the new block immediately. If it is @i{assigned-dirty} its current
8145: contents are written back to the blocks file on disk before it is
8146: allocated to the new block.
8147:
8148: Although no structure is imposed on the contents of a block, it is
8149: traditional to display the contents as 16 lines each of 64 characters. A
8150: block provides a single, continuous stream of input (for example, it
8151: acts as a single parse area) -- there are no end-of-line characters
8152: within a block, and no end-of-file character at the end of a
8153: block. There are two consequences of this:
8154:
8155: @itemize @bullet
8156: @item
8157: The last character of one line wraps straight into the first character
8158: of the following line
8159: @item
8160: The word @code{\} -- comment to end of line -- requires special
8161: treatment; in the context of a block it causes all characters until the
8162: end of the current 64-character ``line'' to be ignored.
8163: @end itemize
8164:
8165: In Gforth, when you use @code{block} with a non-existent block number,
8166: the current blocks file will be extended to the appropriate size and the
8167: block buffer will be initialised with spaces.
8168:
8169: Gforth includes a simple block editor (type @code{use blocked.fb 0 list}
8170: for details) but doesn't encourage the use of blocks; the mechanism is
8171: only provided for backward compatibility -- ANS Forth requires blocks to
8172: be available when files are.
8173:
8174: Common techniques that are used when working with blocks include:
8175:
8176: @itemize @bullet
8177: @item
8178: A screen editor that allows you to edit blocks without leaving the Forth
8179: environment.
8180: @item
8181: Shadow screens; where every code block has an associated block
8182: containing comments (for example: code in odd block numbers, comments in
8183: even block numbers). Typically, the block editor provides a convenient
8184: mechanism to toggle between code and comments.
8185: @item
8186: Load blocks; a single block (typically block 1) contains a number of
8187: @code{thru} commands which @code{load} the whole of the application.
8188: @end itemize
8189:
8190: See Frank Sergeant's Pygmy Forth to see just how well blocks can be
8191: integrated into a Forth programming environment.
8192:
8193: @comment TODO what about errors on open-blocks?
8194:
8195: doc-open-blocks
8196: doc-use
8197: doc-get-block-fid
8198: doc-block-position
8199:
8200: doc-scr
8201: doc-list
8202:
8203: doc---gforthman-block
8204: doc-buffer
8205:
8206: doc-update
8207: doc-updated?
8208: doc-save-buffers
8209: doc-empty-buffers
8210: doc-empty-buffer
8211: doc-flush
8212:
8213: doc-load
8214: doc-thru
8215: doc-+load
8216: doc-+thru
8217: doc---gforthman--->
8218: doc-block-included
8219:
8220:
8221: @c -------------------------------------------------------------
8222: @node Other I/O, Programming Tools, Blocks, Words
8223: @section Other I/O
8224: @cindex I/O - keyboard and display
8225:
8226: @menu
8227: * Simple numeric output:: Predefined formats
8228: * Formatted numeric output:: Formatted (pictured) output
8229: * String Formats:: How Forth stores strings in memory
8230: * Displaying characters and strings:: Other stuff
8231: * Input:: Input
8232: @end menu
8233:
8234: @node Simple numeric output, Formatted numeric output, Other I/O, Other I/O
8235: @subsection Simple numeric output
8236: @cindex numeric output - simple/free-format
8237:
8238: The simplest output functions are those that display numbers from the
8239: data or floating-point stacks. Floating-point output is always displayed
8240: using base 10. Numbers displayed from the data stack use the value stored
8241: in @code{base}.
8242:
8243:
8244: doc-.
8245: doc-dec.
8246: doc-hex.
8247: doc-u.
8248: doc-.r
8249: doc-u.r
8250: doc-d.
8251: doc-ud.
8252: doc-d.r
8253: doc-ud.r
8254: doc-f.
8255: doc-fe.
8256: doc-fs.
8257:
8258:
8259: Examples of printing the number 1234.5678E23 in the different floating-point output
8260: formats are shown below:
8261:
8262: @example
8263: f. 123456779999999000000000000.
8264: fe. 123.456779999999E24
8265: fs. 1.23456779999999E26
8266: @end example
8267:
8268:
8269: @node Formatted numeric output, String Formats, Simple numeric output, Other I/O
8270: @subsection Formatted numeric output
8271: @cindex formatted numeric output
8272: @cindex pictured numeric output
8273: @cindex numeric output - formatted
8274:
8275: Forth traditionally uses a technique called @dfn{pictured numeric
8276: output} for formatted printing of integers. In this technique, digits
8277: are extracted from the number (using the current output radix defined by
8278: @code{base}), converted to ASCII codes and appended to a string that is
8279: built in a scratch-pad area of memory (@pxref{core-idef,
8280: Implementation-defined options, Implementation-defined
8281: options}). Arbitrary characters can be appended to the string during the
8282: extraction process. The completed string is specified by an address
8283: and length and can be manipulated (@code{TYPE}ed, copied, modified)
8284: under program control.
8285:
8286: All of the words described in the previous section for simple numeric
8287: output are implemented in Gforth using pictured numeric output.
8288:
8289: Three important things to remember about pictured numeric output:
8290:
8291: @itemize @bullet
8292: @item
8293: It always operates on double-precision numbers; to display a
8294: single-precision number, convert it first (for ways of doing this
8295: @pxref{Double precision}).
8296: @item
8297: It always treats the double-precision number as though it were
8298: unsigned. The examples below show ways of printing signed numbers.
8299: @item
8300: The string is built up from right to left; least significant digit first.
8301: @end itemize
8302:
8303:
8304: doc-<#
8305: doc-<<#
8306: doc-#
8307: doc-#s
8308: doc-hold
8309: doc-sign
8310: doc-#>
8311: doc-#>>
8312:
8313: doc-represent
8314:
8315:
8316: @noindent
8317: Here are some examples of using pictured numeric output:
8318:
8319: @example
8320: : my-u. ( u -- )
8321: \ Simplest use of pns.. behaves like Standard u.
8322: 0 \ convert to unsigned double
8323: <# \ start conversion
8324: #s \ convert all digits
8325: #> \ complete conversion
8326: TYPE SPACE ; \ display, with trailing space
8327:
8328: : cents-only ( u -- )
8329: 0 \ convert to unsigned double
8330: <# \ start conversion
8331: # # \ convert two least-significant digits
8332: #> \ complete conversion, discard other digits
8333: TYPE SPACE ; \ display, with trailing space
8334:
8335: : dollars-and-cents ( u -- )
8336: 0 \ convert to unsigned double
8337: <# \ start conversion
8338: # # \ convert two least-significant digits
8339: [char] . hold \ insert decimal point
8340: #s \ convert remaining digits
8341: [char] $ hold \ append currency symbol
8342: #> \ complete conversion
8343: TYPE SPACE ; \ display, with trailing space
8344:
8345: : my-. ( n -- )
8346: \ handling negatives.. behaves like Standard .
8347: s>d \ convert to signed double
8348: swap over dabs \ leave sign byte followed by unsigned double
8349: <# \ start conversion
8350: #s \ convert all digits
8351: rot sign \ get at sign byte, append "-" if needed
8352: #> \ complete conversion
8353: TYPE SPACE ; \ display, with trailing space
8354:
8355: : account. ( n -- )
8356: \ accountants don't like minus signs, they use braces
8357: \ for negative numbers
8358: s>d \ convert to signed double
8359: swap over dabs \ leave sign byte followed by unsigned double
8360: <# \ start conversion
8361: 2 pick \ get copy of sign byte
8362: 0< IF [char] ) hold THEN \ right-most character of output
8363: #s \ convert all digits
8364: rot \ get at sign byte
8365: 0< IF [char] ( hold THEN
8366: #> \ complete conversion
8367: TYPE SPACE ; \ display, with trailing space
8368: @end example
8369:
8370: Here are some examples of using these words:
8371:
8372: @example
8373: 1 my-u. 1
8374: hex -1 my-u. decimal FFFFFFFF
8375: 1 cents-only 01
8376: 1234 cents-only 34
8377: 2 dollars-and-cents $0.02
8378: 1234 dollars-and-cents $12.34
8379: 123 my-. 123
8380: -123 my. -123
8381: 123 account. 123
8382: -456 account. (456)
8383: @end example
8384:
8385:
8386: @node String Formats, Displaying characters and strings, Formatted numeric output, Other I/O
8387: @subsection String Formats
8388: @cindex strings - see character strings
8389: @cindex character strings - formats
8390: @cindex I/O - see character strings
8391:
8392: Forth commonly uses two different methods for representing character
8393: strings:
8394:
8395: @itemize @bullet
8396: @item
8397: @cindex address of counted string
8398: @cindex counted string
8399: As a @dfn{counted string}, represented by a @i{c-addr}. The char
8400: addressed by @i{c-addr} contains a character-count, @i{n}, of the
8401: string and the string occupies the subsequent @i{n} char addresses in
8402: memory.
8403: @item
8404: As cell pair on the stack; @i{c-addr u}, where @i{u} is the length
8405: of the string in characters, and @i{c-addr} is the address of the
8406: first byte of the string.
8407: @end itemize
8408:
8409: ANS Forth encourages the use of the second format when representing
8410: strings on the stack, whilst conceeding that the counted string format
8411: remains useful as a way of storing strings in memory.
8412:
8413:
8414: doc-count
8415:
8416:
8417: For words that move, copy and search for strings see @ref{Memory
8418: Blocks}. For words that display characters and strings see
8419: @ref{Displaying characters and strings}.
8420:
8421: @node Displaying characters and strings, Input, String Formats, Other I/O
8422: @subsection Displaying characters and strings
8423: @cindex characters - compiling and displaying
8424: @cindex character strings - compiling and displaying
8425:
8426: This section starts with a glossary of Forth words and ends with a set
8427: of examples.
8428:
8429:
8430: doc-bl
8431: doc-space
8432: doc-spaces
8433: doc-emit
8434: doc-toupper
8435: doc-."
8436: doc-.(
8437: doc-type
8438: doc-typewhite
8439: doc-cr
8440: @cindex cursor control
8441: doc-at-xy
8442: doc-page
8443: doc-s"
8444: doc-c"
8445: doc-char
8446: doc-[char]
8447: doc-sliteral
8448:
8449:
8450: @noindent
8451: As an example, consider the following text, stored in a file @file{test.fs}:
8452:
8453: @example
8454: .( text-1)
8455: : my-word
8456: ." text-2" cr
8457: .( text-3)
8458: ;
8459:
8460: ." text-4"
8461:
8462: : my-char
8463: [char] ALPHABET emit
8464: char emit
8465: ;
8466: @end example
8467:
8468: When you load this code into Gforth, the following output is generated:
8469:
8470: @example
8471: @kbd{include test.fs @key{RET}} text-1text-3text-4 ok
8472: @end example
8473:
8474: @itemize @bullet
8475: @item
8476: Messages @code{text-1} and @code{text-3} are displayed because @code{.(}
8477: is an immediate word; it behaves in the same way whether it is used inside
8478: or outside a colon definition.
8479: @item
8480: Message @code{text-4} is displayed because of Gforth's added interpretation
8481: semantics for @code{."}.
8482: @item
8483: Message @code{text-2} is @i{not} displayed, because the text interpreter
8484: performs the compilation semantics for @code{."} within the definition of
8485: @code{my-word}.
8486: @end itemize
8487:
8488: Here are some examples of executing @code{my-word} and @code{my-char}:
8489:
8490: @example
8491: @kbd{my-word @key{RET}} text-2
8492: ok
8493: @kbd{my-char fred @key{RET}} Af ok
8494: @kbd{my-char jim @key{RET}} Aj ok
8495: @end example
8496:
8497: @itemize @bullet
8498: @item
8499: Message @code{text-2} is displayed because of the run-time behaviour of
8500: @code{."}.
8501: @item
8502: @code{[char]} compiles the ``A'' from ``ALPHABET'' and puts its display code
8503: on the stack at run-time. @code{emit} always displays the character
8504: when @code{my-char} is executed.
8505: @item
8506: @code{char} parses a string at run-time and the second @code{emit} displays
8507: the first character of the string.
8508: @item
8509: If you type @code{see my-char} you can see that @code{[char]} discarded
8510: the text ``LPHABET'' and only compiled the display code for ``A'' into the
8511: definition of @code{my-char}.
8512: @end itemize
8513:
8514:
8515:
8516: @node Input, , Displaying characters and strings, Other I/O
8517: @subsection Input
8518: @cindex input
8519: @cindex I/O - see input
8520: @cindex parsing a string
8521:
8522: For ways of storing character strings in memory see @ref{String Formats}.
8523:
8524: @comment TODO examples for >number >float accept key key? pad parse word refill
8525: @comment then index them
8526:
8527:
8528: doc-key
8529: doc-key?
8530: doc-ekey
8531: doc-ekey?
8532: doc-ekey>char
8533: doc->number
8534: doc->float
8535: doc-accept
8536: doc-pad
8537: doc-parse
8538: doc-word
8539: doc-sword
8540: doc-(name)
8541: doc-refill
8542: @comment obsolescent words..
8543: doc-convert
8544: doc-query
8545: doc-expect
8546: doc-span
8547:
8548:
8549:
8550: @c -------------------------------------------------------------
8551: @node Programming Tools, Assembler and Code Words, Other I/O, Words
8552: @section Programming Tools
8553: @cindex programming tools
8554:
8555: @menu
8556: * Debugging:: Simple and quick.
8557: * Assertions:: Making your programs self-checking.
8558: * Singlestep Debugger:: Executing your program word by word.
8559: @end menu
8560:
8561: @node Debugging, Assertions, Programming Tools, Programming Tools
8562: @subsection Debugging
8563: @cindex debugging
8564:
8565: Languages with a slow edit/compile/link/test development loop tend to
8566: require sophisticated tracing/stepping debuggers to facilate
8567: productive debugging.
8568:
8569: A much better (faster) way in fast-compiling languages is to add
8570: printing code at well-selected places, let the program run, look at
8571: the output, see where things went wrong, add more printing code, etc.,
8572: until the bug is found.
8573:
8574: The simple debugging aids provided in @file{debugs.fs}
8575: are meant to support this style of debugging. In addition, there are
8576: words for non-destructively inspecting the stack and memory:
8577:
8578:
8579: doc-.s
8580: doc-f.s
8581:
8582:
8583: There is a word @code{.r} but it does @i{not} display the return
8584: stack! It is used for formatted numeric output.
8585:
8586:
8587: doc-depth
8588: doc-fdepth
8589: doc-clearstack
8590: doc-?
8591: doc-dump
8592:
8593:
8594: The word @code{~~} prints debugging information (by default the source
8595: location and the stack contents). It is easy to insert. If you use Emacs
8596: it is also easy to remove (@kbd{C-x ~} in the Emacs Forth mode to
8597: query-replace them with nothing). The deferred words
8598: @code{printdebugdata} and @code{printdebugline} control the output of
8599: @code{~~}. The default source location output format works well with
8600: Emacs' compilation mode, so you can step through the program at the
8601: source level using @kbd{C-x `} (the advantage over a stepping debugger
8602: is that you can step in any direction and you know where the crash has
8603: happened or where the strange data has occurred).
8604:
8605: The default actions of @code{~~} clobber the contents of the pictured
8606: numeric output string, so you should not use @code{~~}, e.g., between
8607: @code{<#} and @code{#>}.
8608:
8609:
8610: doc-~~
8611: doc-printdebugdata
8612: doc-printdebugline
8613:
8614: doc-see
8615: doc-marker
8616:
8617:
8618: Here's an example of using @code{marker} at the start of a source file
8619: that you are debugging; it ensures that you only ever have one copy of
8620: the file's definitions compiled at any time:
8621:
8622: @example
8623: [IFDEF] my-code
8624: my-code
8625: [ENDIF]
8626:
8627: marker my-code
8628: init-included-files
8629:
8630: \ .. definitions start here
8631: \ .
8632: \ .
8633: \ end
8634: @end example
8635:
8636:
8637:
8638: @node Assertions, Singlestep Debugger, Debugging, Programming Tools
8639: @subsection Assertions
8640: @cindex assertions
8641:
8642: It is a good idea to make your programs self-checking, especially if you
8643: make an assumption that may become invalid during maintenance (for
8644: example, that a certain field of a data structure is never zero). Gforth
8645: supports @dfn{assertions} for this purpose. They are used like this:
8646:
8647: @example
8648: assert( @i{flag} )
8649: @end example
8650:
8651: The code between @code{assert(} and @code{)} should compute a flag, that
8652: should be true if everything is alright and false otherwise. It should
8653: not change anything else on the stack. The overall stack effect of the
8654: assertion is @code{( -- )}. E.g.
8655:
8656: @example
8657: assert( 1 1 + 2 = ) \ what we learn in school
8658: assert( dup 0<> ) \ assert that the top of stack is not zero
8659: assert( false ) \ this code should not be reached
8660: @end example
8661:
8662: The need for assertions is different at different times. During
8663: debugging, we want more checking, in production we sometimes care more
8664: for speed. Therefore, assertions can be turned off, i.e., the assertion
8665: becomes a comment. Depending on the importance of an assertion and the
8666: time it takes to check it, you may want to turn off some assertions and
8667: keep others turned on. Gforth provides several levels of assertions for
8668: this purpose:
8669:
8670:
8671: doc-assert0(
8672: doc-assert1(
8673: doc-assert2(
8674: doc-assert3(
8675: doc-assert(
8676: doc-)
8677:
8678:
8679: The variable @code{assert-level} specifies the highest assertions that
8680: are turned on. I.e., at the default @code{assert-level} of one,
8681: @code{assert0(} and @code{assert1(} assertions perform checking, while
8682: @code{assert2(} and @code{assert3(} assertions are treated as comments.
8683:
8684: The value of @code{assert-level} is evaluated at compile-time, not at
8685: run-time. Therefore you cannot turn assertions on or off at run-time;
8686: you have to set the @code{assert-level} appropriately before compiling a
8687: piece of code. You can compile different pieces of code at different
8688: @code{assert-level}s (e.g., a trusted library at level 1 and
8689: newly-written code at level 3).
8690:
8691:
8692: doc-assert-level
8693:
8694:
8695: If an assertion fails, a message compatible with Emacs' compilation mode
8696: is produced and the execution is aborted (currently with @code{ABORT"}.
8697: If there is interest, we will introduce a special throw code. But if you
8698: intend to @code{catch} a specific condition, using @code{throw} is
8699: probably more appropriate than an assertion).
8700:
8701: Definitions in ANS Forth for these assertion words are provided
8702: in @file{compat/assert.fs}.
8703:
8704:
8705: @node Singlestep Debugger, , Assertions, Programming Tools
8706: @subsection Singlestep Debugger
8707: @cindex singlestep Debugger
8708: @cindex debugging Singlestep
8709:
8710: When you create a new word there's often the need to check whether it
8711: behaves correctly or not. You can do this by typing @code{dbg
8712: badword}. A debug session might look like this:
8713:
8714: @example
8715: : badword 0 DO i . LOOP ; ok
8716: 2 dbg badword
8717: : badword
8718: Scanning code...
8719:
8720: Nesting debugger ready!
8721:
8722: 400D4738 8049BC4 0 -> [ 2 ] 00002 00000
8723: 400D4740 8049F68 DO -> [ 0 ]
8724: 400D4744 804A0C8 i -> [ 1 ] 00000
8725: 400D4748 400C5E60 . -> 0 [ 0 ]
8726: 400D474C 8049D0C LOOP -> [ 0 ]
8727: 400D4744 804A0C8 i -> [ 1 ] 00001
8728: 400D4748 400C5E60 . -> 1 [ 0 ]
8729: 400D474C 8049D0C LOOP -> [ 0 ]
8730: 400D4758 804B384 ; -> ok
8731: @end example
8732:
8733: Each line displayed is one step. You always have to hit return to
8734: execute the next word that is displayed. If you don't want to execute
8735: the next word in a whole, you have to type @kbd{n} for @code{nest}. Here is
8736: an overview what keys are available:
8737:
8738: @table @i
8739:
8740: @item @key{RET}
8741: Next; Execute the next word.
8742:
8743: @item n
8744: Nest; Single step through next word.
8745:
8746: @item u
8747: Unnest; Stop debugging and execute rest of word. If we got to this word
8748: with nest, continue debugging with the calling word.
8749:
8750: @item d
8751: Done; Stop debugging and execute rest.
8752:
8753: @item s
8754: Stop; Abort immediately.
8755:
8756: @end table
8757:
8758: Debugging large application with this mechanism is very difficult, because
8759: you have to nest very deeply into the program before the interesting part
8760: begins. This takes a lot of time.
8761:
8762: To do it more directly put a @code{BREAK:} command into your source code.
8763: When program execution reaches @code{BREAK:} the single step debugger is
8764: invoked and you have all the features described above.
8765:
8766: If you have more than one part to debug it is useful to know where the
8767: program has stopped at the moment. You can do this by the
8768: @code{BREAK" string"} command. This behaves like @code{BREAK:} except that
8769: string is typed out when the ``breakpoint'' is reached.
8770:
8771:
8772: doc-dbg
8773: doc-break:
8774: doc-break"
8775:
8776:
8777:
8778: @c -------------------------------------------------------------
8779: @node Assembler and Code Words, Threading Words, Programming Tools, Words
8780: @section Assembler and Code Words
8781: @cindex assembler
8782: @cindex code words
8783:
8784: @menu
8785: * Code and ;code::
8786: * Common Assembler:: Assembler Syntax
8787: * Common Disassembler::
8788: * 386 Assembler:: Deviations and special cases
8789: * Alpha Assembler:: Deviations and special cases
8790: * MIPS assembler:: Deviations and special cases
8791: * Other assemblers:: How to write them
8792: @end menu
8793:
8794: @node Code and ;code, Common Assembler, Assembler and Code Words, Assembler and Code Words
8795: @subsection @code{Code} and @code{;code}
8796:
8797: Gforth provides some words for defining primitives (words written in
8798: machine code), and for defining the machine-code equivalent of
8799: @code{DOES>}-based defining words. However, the machine-independent
8800: nature of Gforth poses a few problems: First of all, Gforth runs on
8801: several architectures, so it can provide no standard assembler. What's
8802: worse is that the register allocation not only depends on the processor,
8803: but also on the @code{gcc} version and options used.
8804:
8805: The words that Gforth offers encapsulate some system dependences (e.g.,
8806: the header structure), so a system-independent assembler may be used in
8807: Gforth. If you do not have an assembler, you can compile machine code
8808: directly with @code{,} and @code{c,}@footnote{This isn't portable,
8809: because these words emit stuff in @i{data} space; it works because
8810: Gforth has unified code/data spaces. Assembler isn't likely to be
8811: portable anyway.}.
8812:
8813:
8814: doc-assembler
8815: doc-init-asm
8816: doc-code
8817: doc-end-code
8818: doc-;code
8819: doc-flush-icache
8820:
8821:
8822: If @code{flush-icache} does not work correctly, @code{code} words
8823: etc. will not work (reliably), either.
8824:
8825: The typical usage of these @code{code} words can be shown most easily by
8826: analogy to the equivalent high-level defining words:
8827:
8828: @example
8829: : foo code foo
8830: <high-level Forth words> <assembler>
8831: ; end-code
8832:
8833: : bar : bar
8834: <high-level Forth words> <high-level Forth words>
8835: CREATE CREATE
8836: <high-level Forth words> <high-level Forth words>
8837: DOES> ;code
8838: <high-level Forth words> <assembler>
8839: ; end-code
8840: @end example
8841:
8842: @code{flush-icache} is always present. The other words are rarely used
8843: and reside in @code{code.fs}, which is usually not loaded. You can load
8844: it with @code{require code.fs}.
8845:
8846: @cindex registers of the inner interpreter
8847: In the assembly code you will want to refer to the inner interpreter's
8848: registers (e.g., the data stack pointer) and you may want to use other
8849: registers for temporary storage. Unfortunately, the register allocation
8850: is installation-dependent.
8851:
8852: The easiest solution is to use explicit register declarations
8853: (@pxref{Explicit Reg Vars, , Variables in Specified Registers, gcc.info,
8854: GNU C Manual}) for all of the inner interpreter's registers: You have to
8855: compile Gforth with @code{-DFORCE_REG} (configure option
8856: @code{--enable-force-reg}) and the appropriate declarations must be
8857: present in the @code{machine.h} file (see @code{mips.h} for an example;
8858: you can find a full list of all declarable register symbols with
8859: @code{grep register engine.c}). If you give explicit registers to all
8860: variables that are declared at the beginning of @code{engine()}, you
8861: should be able to use the other caller-saved registers for temporary
8862: storage. Alternatively, you can use the @code{gcc} option
8863: @code{-ffixed-REG} (@pxref{Code Gen Options, , Options for Code
8864: Generation Conventions, gcc.info, GNU C Manual}) to reserve a register
8865: (however, this restriction on register allocation may slow Gforth
8866: significantly).
8867:
8868: If this solution is not viable (e.g., because @code{gcc} does not allow
8869: you to explicitly declare all the registers you need), you have to find
8870: out by looking at the code where the inner interpreter's registers
8871: reside and which registers can be used for temporary storage. You can
8872: get an assembly listing of the engine's code with @code{make engine.s}.
8873:
8874: In any case, it is good practice to abstract your assembly code from the
8875: actual register allocation. E.g., if the data stack pointer resides in
8876: register @code{$17}, create an alias for this register called @code{sp},
8877: and use that in your assembly code.
8878:
8879: @cindex code words, portable
8880: Another option for implementing normal and defining words efficiently
8881: is to add the desired functionality to the source of Gforth. For normal
8882: words you just have to edit @file{primitives} (@pxref{Automatic
8883: Generation}). Defining words (equivalent to @code{;CODE} words, for fast
8884: defined words) may require changes in @file{engine.c}, @file{kernel.fs},
8885: @file{prims2x.fs}, and possibly @file{cross.fs}.
8886:
8887: @node Common Assembler, Common Disassembler, Code and ;code, Assembler and Code Words
8888: @subsection Common Assembler
8889:
8890: The assemblers in Gforth generally use a postfix syntax, i.e., the
8891: instruction name follows the operands.
8892:
8893: The operands are passed in the usual order (the same that is used in the
8894: manual of the architecture). Since they all are Forth words, they have
8895: to be separated by spaces; you can also use Forth words to compute the
8896: operands.
8897:
8898: The instruction names usually end with a @code{,}. This makes it easier
8899: to visually separate instructions if you put several of them on one
8900: line; it also avoids shadowing other Forth words (e.g., @code{and}).
8901:
8902: Registers are usually specified by number; e.g., (decimal) @code{11}
8903: specifies registers R11 and F11 on the Alpha architecture (which one,
8904: depends on the instruction). The usual names are also available, e.g.,
8905: @code{s2} for R11 on Alpha.
8906:
8907: Control flow is specified similar to normal Forth code (@pxref{Arbitrary
8908: control structures}), with @code{if,}, @code{ahead,}, @code{then,},
8909: @code{begin,}, @code{until,}, @code{again,}, @code{cs-roll},
8910: @code{cs-pick}, @code{else,}, @code{while,}, and @code{repeat,}. The
8911: conditions are specified in a way specific to each assembler.
8912:
8913: Note that the register assignments of the Gforth engine can change
8914: between Gforth versions, or even between different compilations of the
8915: same Gforth version (e.g., if you use a different GCC version). So if
8916: you want to refer to Gforth's registers (e.g., the stack pointer or
8917: TOS), I recommend defining your own words for refering to these
8918: registers, and using them later on; then you can easily adapt to a
8919: changed register assignment. The stability of the register assignment
8920: is usually better if you build Gforth with @code{--enable-force-reg}.
8921:
8922: In particular, the resturn stack pointer and the instruction pointer are
8923: in memory in @code{gforth}, and usually in registers in
8924: @code{gforth-fast}. The most common use of these registers is to
8925: dispatch to the next word (the @code{next} routine). A portable way to
8926: do this is to jump to @code{' noop >code-address} (of course, this is
8927: less efficient than integrating the @code{next} code and scheduling it
8928: well).
8929:
8930: @node Common Disassembler, 386 Assembler, Common Assembler, Assembler and Code Words
8931: @subsection Common Disassembler
8932:
8933: You can disassemble a @code{code} word with @code{see}
8934: (@pxref{Debugging}). You can disassemble a section of memory with
8935:
8936: doc-disasm
8937:
8938: The disassembler generally produces output that can be fed into the
8939: assembler (i.e., same syntax, etc.). It also includes additional
8940: information in comments. In particular, the address of the instruction
8941: is given in a comment before the instruction.
8942:
8943: @code{See} may display more or less than the actual code of the word,
8944: because the recognition of the end of the code is unreliable. You can
8945: use @code{disasm} if it did not display enough. It may display more, if
8946: the code word is not immediately followed by a named word. If you have
8947: something else there, you can follow the word with @code{align last @ ,}
8948: to ensure that the end is recognized.
8949:
8950: @node 386 Assembler, Alpha Assembler, Common Disassembler, Assembler and Code Words
8951: @subsection 386 Assembler
8952:
8953: The 386 assembler included in Gforth was written by Bernd Paysan, it's
8954: available under GPL, and originally part of bigFORTH.
8955:
8956: The 386 disassembler included in Gforth was written by Andrew McKewan
8957: and is in the public domain.
8958:
8959: The disassembler displays code in prefix Intel syntax.
8960:
8961: The assembler uses a postfix syntax with reversed parameters.
8962:
8963: The assembler includes all instruction of the Athlon, i.e. 486 core
8964: instructions, Pentium and PPro extensions, floating point, MMX, 3Dnow!,
8965: but not ISSE. It's an integrated 16- and 32-bit assembler. Default is 32
8966: bit, you can switch to 16 bit with .86 and back to 32 bit with .386.
8967:
8968: There are several prefixes to switch between different operation sizes,
8969: @code{.b} for byte accesses, @code{.w} for word accesses, @code{.d} for
8970: double-word accesses. Addressing modes can be switched with @code{.wa}
8971: for 16 bit addresses, and @code{.da} for 32 bit addresses. You don't
8972: need a prefix for byte register names (@code{AL} et al).
8973:
8974: For floating point operations, the prefixes are @code{.fs} (IEEE
8975: single), @code{.fl} (IEEE double), @code{.fx} (extended), @code{.fw}
8976: (word), @code{.fd} (double-word), and @code{.fq} (quad-word).
8977:
8978: The MMX opcodes don't have size prefixes, they are spelled out like in
8979: the Intel assembler. Instead of move from and to memory, there are
8980: PLDQ/PLDD and PSTQ/PSTD.
8981:
8982: The registers lack the 'e' prefix; even in 32 bit mode, eax is called
8983: ax. Immediate values are indicated by postfixing them with @code{#},
8984: e.g., @code{3 #}. Here are some examples of addressing modes:
8985:
8986: @example
8987: 3 # \ immediate
8988: ax \ register
8989: 100 di d) \ 100[edi]
8990: 4 bx cx di) \ 4[ebx][ecx]
8991: di ax *4 i) \ [edi][eax*4]
8992: 20 ax *4 i#) \ 20[eax*4]
8993: @end example
8994:
8995: Some example of instructions are:
8996:
8997: @example
8998: ax bx mov \ move ebx,eax
8999: 3 # ax mov \ mov eax,3
9000: 100 di ) ax mov \ mov eax,100[edi]
9001: 4 bx cx di) ax mov \ mov eax,4[ebx][ecx]
9002: .w ax bx mov \ mov bx,ax
9003: @end example
9004:
9005: The following forms are supported for binary instructions:
9006:
9007: @example
9008: <reg> <reg> <inst>
9009: <n> # <reg> <inst>
9010: <mem> <reg> <inst>
9011: <reg> <mem> <inst>
9012: @end example
9013:
9014: Immediate to memory is not supported. The shift/rotate syntax is:
9015:
9016: @example
9017: <reg/mem> 1 # shl \ shortens to shift without immediate
9018: <reg/mem> 4 # shl
9019: <reg/mem> cl shl
9020: @end example
9021:
9022: Precede string instructions (@code{movs} etc.) with @code{.b} to get
9023: the byte version.
9024:
9025: The control structure words @code{IF} @code{UNTIL} etc. must be preceded
9026: by one of these conditions: @code{vs vc u< u>= 0= 0<> u<= u> 0< 0>= ps
9027: pc < >= <= >}. (Note that most of these words shadow some Forth words
9028: when @code{assembler} is in front of @code{forth} in the search path,
9029: e.g., in @code{code} words). Currently the control structure words use
9030: one stack item, so you have to use @code{roll} instead of @code{cs-roll}
9031: to shuffle them (you can also use @code{swap} etc.).
9032:
9033: Here is an example of a @code{code} word (assumes that the stack pointer
9034: is in esi and the TOS is in ebx):
9035:
9036: @example
9037: code my+ ( n1 n2 -- n )
9038: 4 si D) bx add
9039: 4 # si add
9040: Next
9041: end-code
9042: @end example
9043:
9044: @node Alpha Assembler, MIPS assembler, 386 Assembler, Assembler and Code Words
9045: @subsection Alpha Assembler
9046:
9047: The Alpha assembler and disassembler were originally written by Bernd
9048: Thallner.
9049:
9050: The register names @code{a0}--@code{a5} are not available to avoid
9051: shadowing hex numbers.
9052:
9053: Immediate forms of arithmetic instructions are distinguished by a
9054: @code{#} just before the @code{,}, e.g., @code{and#,} (note: @code{lda,}
9055: does not count as arithmetic instruction).
9056:
9057: You have to specify all operands to an instruction, even those that
9058: other assemblers consider optional, e.g., the destination register for
9059: @code{br,}, or the destination register and hint for @code{jmp,}.
9060:
9061: You can specify conditions for @code{if,} by removing the first @code{b}
9062: and the trailing @code{,} from a branch with a corresponding name; e.g.,
9063:
9064: @example
9065: 11 fgt if, \ if F11>0e
9066: ...
9067: endif,
9068: @end example
9069:
9070: @code{fbgt,} gives @code{fgt}.
9071:
9072: @node MIPS assembler, Other assemblers, Alpha Assembler, Assembler and Code Words
9073: @subsection MIPS assembler
9074:
9075: The MIPS assembler was originally written by Christian Pirker.
9076:
9077: Currently the assembler and disassembler only cover the MIPS-I
9078: architecture (R3000), and don't support FP instructions.
9079:
9080: The register names @code{$a0}--@code{$a3} are not available to avoid
9081: shadowing hex numbers.
9082:
9083: Because there is no way to distinguish registers from immediate values,
9084: you have to explicitly use the immediate forms of instructions, i.e.,
9085: @code{addiu,}, not just @code{addu,} (@command{as} does this
9086: implicitly).
9087:
9088: If the architecture manual specifies several formats for the instruction
9089: (e.g., for @code{jalr,}), you usually have to use the one with more
9090: arguments (i.e., two for @code{jalr,}). When in doubt, see
9091: @code{arch/mips/testasm.fs} for an example of correct use.
9092:
9093: Branches and jumps in the MIPS architecture have a delay slot. You have
9094: to fill it yourself (the simplest way is to use @code{nop,}), the
9095: assembler does not do it for you (unlike @command{as}). Even
9096: @code{if,}, @code{ahead,}, @code{until,}, @code{again,}, @code{while,},
9097: @code{else,} and @code{repeat,} need a delay slot. Since @code{begin,}
9098: and @code{then,} just specify branch targets, they are not affected.
9099:
9100: Note that you must not put branches, jumps, or @code{li,} into the delay
9101: slot: @code{li,} may expand to several instructions, and control flow
9102: instructions may not be put into the branch delay slot in any case.
9103:
9104: For branches the argument specifying the target is a relative address;
9105: You have to add the address of the delay slot to get the absolute
9106: address.
9107:
9108: The MIPS architecture also has load delay slots and restrictions on
9109: using @code{mfhi,} and @code{mflo,}; you have to order the instructions
9110: yourself to satisfy these restrictions, the assembler does not do it for
9111: you.
9112:
9113: You can specify the conditions for @code{if,} etc. by taking a
9114: conditional branch and leaving away the @code{b} at the start and the
9115: @code{,} at the end. E.g.,
9116:
9117: @example
9118: 4 5 eq if,
9119: ... \ do something if $4 equals $5
9120: then,
9121: @end example
9122:
9123: @node Other assemblers, , MIPS assembler, Assembler and Code Words
9124: @subsection Other assemblers
9125:
9126: If you want to contribute another assembler/disassembler, please contact
9127: us (@email{bug-gforth@@gnu.org}) to check if we have such an assembler
9128: already. If you are writing them from scratch, please use a similar
9129: syntax style as the one we use (i.e., postfix, commas at the end of the
9130: instruction names, @pxref{Common Assembler}); make the output of the
9131: disassembler be valid input for the assembler, and keep the style
9132: similar to the style we used.
9133:
9134: Hints on implementation: The most important part is to have a good test
9135: suite that contains all instructions. Once you have that, the rest is
9136: easy. For actual coding you can take a look at
9137: @file{arch/mips/disasm.fs} to get some ideas on how to use data for both
9138: the assembler and disassembler, avoiding redundancy and some potential
9139: bugs. You can also look at that file (and @pxref{Advanced does> usage
9140: example}) to get ideas how to factor a disassembler.
9141:
9142: Start with the disassembler, because it's easier to reuse data from the
9143: disassembler for the assembler than the other way round.
9144:
9145: For the assembler, take a look at @file{arch/alpha/asm.fs}, which shows
9146: how simple it can be.
9147:
9148: @c -------------------------------------------------------------
9149: @node Threading Words, Locals, Assembler and Code Words, Words
9150: @section Threading Words
9151: @cindex threading words
9152:
9153: @cindex code address
9154: These words provide access to code addresses and other threading stuff
9155: in Gforth (and, possibly, other interpretive Forths). It more or less
9156: abstracts away the differences between direct and indirect threading
9157: (and, for direct threading, the machine dependences). However, at
9158: present this wordset is still incomplete. It is also pretty low-level;
9159: some day it will hopefully be made unnecessary by an internals wordset
9160: that abstracts implementation details away completely.
9161:
9162:
9163: doc-threading-method
9164: doc->code-address
9165: doc->does-code
9166: doc-code-address!
9167: doc-does-code!
9168: doc-does-handler!
9169: doc-/does-handler
9170:
9171:
9172: The code addresses produced by various defining words are produced by
9173: the following words:
9174:
9175:
9176: doc-docol:
9177: doc-docon:
9178: doc-dovar:
9179: doc-douser:
9180: doc-dodefer:
9181: doc-dofield:
9182:
9183:
9184: You can recognize words defined by a @code{CREATE}...@code{DOES>} word
9185: with @code{>does-code}. If the word was defined in that way, the value
9186: returned is non-zero and identifies the @code{DOES>} used by the
9187: defining word.
9188: @comment TODO should that be ``identifies the xt of the DOES> ??''
9189:
9190: @c -------------------------------------------------------------
9191: @node Locals, Structures, Threading Words, Words
9192: @section Locals
9193: @cindex locals
9194:
9195: Local variables can make Forth programming more enjoyable and Forth
9196: programs easier to read. Unfortunately, the locals of ANS Forth are
9197: laden with restrictions. Therefore, we provide not only the ANS Forth
9198: locals wordset, but also our own, more powerful locals wordset (we
9199: implemented the ANS Forth locals wordset through our locals wordset).
9200:
9201: The ideas in this section have also been published in M. Anton Ertl,
9202: @cite{@uref{http://www.complang.tuwien.ac.at/papers/ertl94l.ps.gz,
9203: Automatic Scoping of Local Variables}}, EuroForth '94.
9204:
9205: @menu
9206: * Gforth locals::
9207: * ANS Forth locals::
9208: @end menu
9209:
9210: @node Gforth locals, ANS Forth locals, Locals, Locals
9211: @subsection Gforth locals
9212: @cindex Gforth locals
9213: @cindex locals, Gforth style
9214:
9215: Locals can be defined with
9216:
9217: @example
9218: @{ local1 local2 ... -- comment @}
9219: @end example
9220: or
9221: @example
9222: @{ local1 local2 ... @}
9223: @end example
9224:
9225: E.g.,
9226: @example
9227: : max @{ n1 n2 -- n3 @}
9228: n1 n2 > if
9229: n1
9230: else
9231: n2
9232: endif ;
9233: @end example
9234:
9235: The similarity of locals definitions with stack comments is intended. A
9236: locals definition often replaces the stack comment of a word. The order
9237: of the locals corresponds to the order in a stack comment and everything
9238: after the @code{--} is really a comment.
9239:
9240: This similarity has one disadvantage: It is too easy to confuse locals
9241: declarations with stack comments, causing bugs and making them hard to
9242: find. However, this problem can be avoided by appropriate coding
9243: conventions: Do not use both notations in the same program. If you do,
9244: they should be distinguished using additional means, e.g. by position.
9245:
9246: @cindex types of locals
9247: @cindex locals types
9248: The name of the local may be preceded by a type specifier, e.g.,
9249: @code{F:} for a floating point value:
9250:
9251: @example
9252: : CX* @{ F: Ar F: Ai F: Br F: Bi -- Cr Ci @}
9253: \ complex multiplication
9254: Ar Br f* Ai Bi f* f-
9255: Ar Bi f* Ai Br f* f+ ;
9256: @end example
9257:
9258: @cindex flavours of locals
9259: @cindex locals flavours
9260: @cindex value-flavoured locals
9261: @cindex variable-flavoured locals
9262: Gforth currently supports cells (@code{W:}, @code{W^}), doubles
9263: (@code{D:}, @code{D^}), floats (@code{F:}, @code{F^}) and characters
9264: (@code{C:}, @code{C^}) in two flavours: a value-flavoured local (defined
9265: with @code{W:}, @code{D:} etc.) produces its value and can be changed
9266: with @code{TO}. A variable-flavoured local (defined with @code{W^} etc.)
9267: produces its address (which becomes invalid when the variable's scope is
9268: left). E.g., the standard word @code{emit} can be defined in terms of
9269: @code{type} like this:
9270:
9271: @example
9272: : emit @{ C^ char* -- @}
9273: char* 1 type ;
9274: @end example
9275:
9276: @cindex default type of locals
9277: @cindex locals, default type
9278: A local without type specifier is a @code{W:} local. Both flavours of
9279: locals are initialized with values from the data or FP stack.
9280:
9281: Currently there is no way to define locals with user-defined data
9282: structures, but we are working on it.
9283:
9284: Gforth allows defining locals everywhere in a colon definition. This
9285: poses the following questions:
9286:
9287: @menu
9288: * Where are locals visible by name?::
9289: * How long do locals live?::
9290: * Programming Style::
9291: * Implementation::
9292: @end menu
9293:
9294: @node Where are locals visible by name?, How long do locals live?, Gforth locals, Gforth locals
9295: @subsubsection Where are locals visible by name?
9296: @cindex locals visibility
9297: @cindex visibility of locals
9298: @cindex scope of locals
9299:
9300: Basically, the answer is that locals are visible where you would expect
9301: it in block-structured languages, and sometimes a little longer. If you
9302: want to restrict the scope of a local, enclose its definition in
9303: @code{SCOPE}...@code{ENDSCOPE}.
9304:
9305:
9306: doc-scope
9307: doc-endscope
9308:
9309:
9310: These words behave like control structure words, so you can use them
9311: with @code{CS-PICK} and @code{CS-ROLL} to restrict the scope in
9312: arbitrary ways.
9313:
9314: If you want a more exact answer to the visibility question, here's the
9315: basic principle: A local is visible in all places that can only be
9316: reached through the definition of the local@footnote{In compiler
9317: construction terminology, all places dominated by the definition of the
9318: local.}. In other words, it is not visible in places that can be reached
9319: without going through the definition of the local. E.g., locals defined
9320: in @code{IF}...@code{ENDIF} are visible until the @code{ENDIF}, locals
9321: defined in @code{BEGIN}...@code{UNTIL} are visible after the
9322: @code{UNTIL} (until, e.g., a subsequent @code{ENDSCOPE}).
9323:
9324: The reasoning behind this solution is: We want to have the locals
9325: visible as long as it is meaningful. The user can always make the
9326: visibility shorter by using explicit scoping. In a place that can
9327: only be reached through the definition of a local, the meaning of a
9328: local name is clear. In other places it is not: How is the local
9329: initialized at the control flow path that does not contain the
9330: definition? Which local is meant, if the same name is defined twice in
9331: two independent control flow paths?
9332:
9333: This should be enough detail for nearly all users, so you can skip the
9334: rest of this section. If you really must know all the gory details and
9335: options, read on.
9336:
9337: In order to implement this rule, the compiler has to know which places
9338: are unreachable. It knows this automatically after @code{AHEAD},
9339: @code{AGAIN}, @code{EXIT} and @code{LEAVE}; in other cases (e.g., after
9340: most @code{THROW}s), you can use the word @code{UNREACHABLE} to tell the
9341: compiler that the control flow never reaches that place. If
9342: @code{UNREACHABLE} is not used where it could, the only consequence is
9343: that the visibility of some locals is more limited than the rule above
9344: says. If @code{UNREACHABLE} is used where it should not (i.e., if you
9345: lie to the compiler), buggy code will be produced.
9346:
9347:
9348: doc-unreachable
9349:
9350:
9351: Another problem with this rule is that at @code{BEGIN}, the compiler
9352: does not know which locals will be visible on the incoming
9353: back-edge. All problems discussed in the following are due to this
9354: ignorance of the compiler (we discuss the problems using @code{BEGIN}
9355: loops as examples; the discussion also applies to @code{?DO} and other
9356: loops). Perhaps the most insidious example is:
9357: @example
9358: AHEAD
9359: BEGIN
9360: x
9361: [ 1 CS-ROLL ] THEN
9362: @{ x @}
9363: ...
9364: UNTIL
9365: @end example
9366:
9367: This should be legal according to the visibility rule. The use of
9368: @code{x} can only be reached through the definition; but that appears
9369: textually below the use.
9370:
9371: From this example it is clear that the visibility rules cannot be fully
9372: implemented without major headaches. Our implementation treats common
9373: cases as advertised and the exceptions are treated in a safe way: The
9374: compiler makes a reasonable guess about the locals visible after a
9375: @code{BEGIN}; if it is too pessimistic, the
9376: user will get a spurious error about the local not being defined; if the
9377: compiler is too optimistic, it will notice this later and issue a
9378: warning. In the case above the compiler would complain about @code{x}
9379: being undefined at its use. You can see from the obscure examples in
9380: this section that it takes quite unusual control structures to get the
9381: compiler into trouble, and even then it will often do fine.
9382:
9383: If the @code{BEGIN} is reachable from above, the most optimistic guess
9384: is that all locals visible before the @code{BEGIN} will also be
9385: visible after the @code{BEGIN}. This guess is valid for all loops that
9386: are entered only through the @code{BEGIN}, in particular, for normal
9387: @code{BEGIN}...@code{WHILE}...@code{REPEAT} and
9388: @code{BEGIN}...@code{UNTIL} loops and it is implemented in our
9389: compiler. When the branch to the @code{BEGIN} is finally generated by
9390: @code{AGAIN} or @code{UNTIL}, the compiler checks the guess and
9391: warns the user if it was too optimistic:
9392: @example
9393: IF
9394: @{ x @}
9395: BEGIN
9396: \ x ?
9397: [ 1 cs-roll ] THEN
9398: ...
9399: UNTIL
9400: @end example
9401:
9402: Here, @code{x} lives only until the @code{BEGIN}, but the compiler
9403: optimistically assumes that it lives until the @code{THEN}. It notices
9404: this difference when it compiles the @code{UNTIL} and issues a
9405: warning. The user can avoid the warning, and make sure that @code{x}
9406: is not used in the wrong area by using explicit scoping:
9407: @example
9408: IF
9409: SCOPE
9410: @{ x @}
9411: ENDSCOPE
9412: BEGIN
9413: [ 1 cs-roll ] THEN
9414: ...
9415: UNTIL
9416: @end example
9417:
9418: Since the guess is optimistic, there will be no spurious error messages
9419: about undefined locals.
9420:
9421: If the @code{BEGIN} is not reachable from above (e.g., after
9422: @code{AHEAD} or @code{EXIT}), the compiler cannot even make an
9423: optimistic guess, as the locals visible after the @code{BEGIN} may be
9424: defined later. Therefore, the compiler assumes that no locals are
9425: visible after the @code{BEGIN}. However, the user can use
9426: @code{ASSUME-LIVE} to make the compiler assume that the same locals are
9427: visible at the BEGIN as at the point where the top control-flow stack
9428: item was created.
9429:
9430:
9431: doc-assume-live
9432:
9433:
9434: @noindent
9435: E.g.,
9436: @example
9437: @{ x @}
9438: AHEAD
9439: ASSUME-LIVE
9440: BEGIN
9441: x
9442: [ 1 CS-ROLL ] THEN
9443: ...
9444: UNTIL
9445: @end example
9446:
9447: Other cases where the locals are defined before the @code{BEGIN} can be
9448: handled by inserting an appropriate @code{CS-ROLL} before the
9449: @code{ASSUME-LIVE} (and changing the control-flow stack manipulation
9450: behind the @code{ASSUME-LIVE}).
9451:
9452: Cases where locals are defined after the @code{BEGIN} (but should be
9453: visible immediately after the @code{BEGIN}) can only be handled by
9454: rearranging the loop. E.g., the ``most insidious'' example above can be
9455: arranged into:
9456: @example
9457: BEGIN
9458: @{ x @}
9459: ... 0=
9460: WHILE
9461: x
9462: REPEAT
9463: @end example
9464:
9465: @node How long do locals live?, Programming Style, Where are locals visible by name?, Gforth locals
9466: @subsubsection How long do locals live?
9467: @cindex locals lifetime
9468: @cindex lifetime of locals
9469:
9470: The right answer for the lifetime question would be: A local lives at
9471: least as long as it can be accessed. For a value-flavoured local this
9472: means: until the end of its visibility. However, a variable-flavoured
9473: local could be accessed through its address far beyond its visibility
9474: scope. Ultimately, this would mean that such locals would have to be
9475: garbage collected. Since this entails un-Forth-like implementation
9476: complexities, I adopted the same cowardly solution as some other
9477: languages (e.g., C): The local lives only as long as it is visible;
9478: afterwards its address is invalid (and programs that access it
9479: afterwards are erroneous).
9480:
9481: @node Programming Style, Implementation, How long do locals live?, Gforth locals
9482: @subsubsection Programming Style
9483: @cindex locals programming style
9484: @cindex programming style, locals
9485:
9486: The freedom to define locals anywhere has the potential to change
9487: programming styles dramatically. In particular, the need to use the
9488: return stack for intermediate storage vanishes. Moreover, all stack
9489: manipulations (except @code{PICK}s and @code{ROLL}s with run-time
9490: determined arguments) can be eliminated: If the stack items are in the
9491: wrong order, just write a locals definition for all of them; then
9492: write the items in the order you want.
9493:
9494: This seems a little far-fetched and eliminating stack manipulations is
9495: unlikely to become a conscious programming objective. Still, the number
9496: of stack manipulations will be reduced dramatically if local variables
9497: are used liberally (e.g., compare @code{max} (@pxref{Gforth locals}) with
9498: a traditional implementation of @code{max}).
9499:
9500: This shows one potential benefit of locals: making Forth programs more
9501: readable. Of course, this benefit will only be realized if the
9502: programmers continue to honour the principle of factoring instead of
9503: using the added latitude to make the words longer.
9504:
9505: @cindex single-assignment style for locals
9506: Using @code{TO} can and should be avoided. Without @code{TO},
9507: every value-flavoured local has only a single assignment and many
9508: advantages of functional languages apply to Forth. I.e., programs are
9509: easier to analyse, to optimize and to read: It is clear from the
9510: definition what the local stands for, it does not turn into something
9511: different later.
9512:
9513: E.g., a definition using @code{TO} might look like this:
9514: @example
9515: : strcmp @{ addr1 u1 addr2 u2 -- n @}
9516: u1 u2 min 0
9517: ?do
9518: addr1 c@@ addr2 c@@ -
9519: ?dup-if
9520: unloop exit
9521: then
9522: addr1 char+ TO addr1
9523: addr2 char+ TO addr2
9524: loop
9525: u1 u2 - ;
9526: @end example
9527: Here, @code{TO} is used to update @code{addr1} and @code{addr2} at
9528: every loop iteration. @code{strcmp} is a typical example of the
9529: readability problems of using @code{TO}. When you start reading
9530: @code{strcmp}, you think that @code{addr1} refers to the start of the
9531: string. Only near the end of the loop you realize that it is something
9532: else.
9533:
9534: This can be avoided by defining two locals at the start of the loop that
9535: are initialized with the right value for the current iteration.
9536: @example
9537: : strcmp @{ addr1 u1 addr2 u2 -- n @}
9538: addr1 addr2
9539: u1 u2 min 0
9540: ?do @{ s1 s2 @}
9541: s1 c@@ s2 c@@ -
9542: ?dup-if
9543: unloop exit
9544: then
9545: s1 char+ s2 char+
9546: loop
9547: 2drop
9548: u1 u2 - ;
9549: @end example
9550: Here it is clear from the start that @code{s1} has a different value
9551: in every loop iteration.
9552:
9553: @node Implementation, , Programming Style, Gforth locals
9554: @subsubsection Implementation
9555: @cindex locals implementation
9556: @cindex implementation of locals
9557:
9558: @cindex locals stack
9559: Gforth uses an extra locals stack. The most compelling reason for
9560: this is that the return stack is not float-aligned; using an extra stack
9561: also eliminates the problems and restrictions of using the return stack
9562: as locals stack. Like the other stacks, the locals stack grows toward
9563: lower addresses. A few primitives allow an efficient implementation:
9564:
9565:
9566: doc-@local#
9567: doc-f@local#
9568: doc-laddr#
9569: doc-lp+!#
9570: doc-lp!
9571: doc->l
9572: doc-f>l
9573:
9574:
9575: In addition to these primitives, some specializations of these
9576: primitives for commonly occurring inline arguments are provided for
9577: efficiency reasons, e.g., @code{@@local0} as specialization of
9578: @code{@@local#} for the inline argument 0. The following compiling words
9579: compile the right specialized version, or the general version, as
9580: appropriate:
9581:
9582:
9583: doc-compile-@local
9584: doc-compile-f@local
9585: doc-compile-lp+!
9586:
9587:
9588: Combinations of conditional branches and @code{lp+!#} like
9589: @code{?branch-lp+!#} (the locals pointer is only changed if the branch
9590: is taken) are provided for efficiency and correctness in loops.
9591:
9592: A special area in the dictionary space is reserved for keeping the
9593: local variable names. @code{@{} switches the dictionary pointer to this
9594: area and @code{@}} switches it back and generates the locals
9595: initializing code. @code{W:} etc.@ are normal defining words. This
9596: special area is cleared at the start of every colon definition.
9597:
9598: @cindex word list for defining locals
9599: A special feature of Gforth's dictionary is used to implement the
9600: definition of locals without type specifiers: every word list (aka
9601: vocabulary) has its own methods for searching
9602: etc. (@pxref{Word Lists}). For the present purpose we defined a word list
9603: with a special search method: When it is searched for a word, it
9604: actually creates that word using @code{W:}. @code{@{} changes the search
9605: order to first search the word list containing @code{@}}, @code{W:} etc.,
9606: and then the word list for defining locals without type specifiers.
9607:
9608: The lifetime rules support a stack discipline within a colon
9609: definition: The lifetime of a local is either nested with other locals
9610: lifetimes or it does not overlap them.
9611:
9612: At @code{BEGIN}, @code{IF}, and @code{AHEAD} no code for locals stack
9613: pointer manipulation is generated. Between control structure words
9614: locals definitions can push locals onto the locals stack. @code{AGAIN}
9615: is the simplest of the other three control flow words. It has to
9616: restore the locals stack depth of the corresponding @code{BEGIN}
9617: before branching. The code looks like this:
9618: @format
9619: @code{lp+!#} current-locals-size @minus{} dest-locals-size
9620: @code{branch} <begin>
9621: @end format
9622:
9623: @code{UNTIL} is a little more complicated: If it branches back, it
9624: must adjust the stack just like @code{AGAIN}. But if it falls through,
9625: the locals stack must not be changed. The compiler generates the
9626: following code:
9627: @format
9628: @code{?branch-lp+!#} <begin> current-locals-size @minus{} dest-locals-size
9629: @end format
9630: The locals stack pointer is only adjusted if the branch is taken.
9631:
9632: @code{THEN} can produce somewhat inefficient code:
9633: @format
9634: @code{lp+!#} current-locals-size @minus{} orig-locals-size
9635: <orig target>:
9636: @code{lp+!#} orig-locals-size @minus{} new-locals-size
9637: @end format
9638: The second @code{lp+!#} adjusts the locals stack pointer from the
9639: level at the @i{orig} point to the level after the @code{THEN}. The
9640: first @code{lp+!#} adjusts the locals stack pointer from the current
9641: level to the level at the orig point, so the complete effect is an
9642: adjustment from the current level to the right level after the
9643: @code{THEN}.
9644:
9645: @cindex locals information on the control-flow stack
9646: @cindex control-flow stack items, locals information
9647: In a conventional Forth implementation a dest control-flow stack entry
9648: is just the target address and an orig entry is just the address to be
9649: patched. Our locals implementation adds a word list to every orig or dest
9650: item. It is the list of locals visible (or assumed visible) at the point
9651: described by the entry. Our implementation also adds a tag to identify
9652: the kind of entry, in particular to differentiate between live and dead
9653: (reachable and unreachable) orig entries.
9654:
9655: A few unusual operations have to be performed on locals word lists:
9656:
9657:
9658: doc-common-list
9659: doc-sub-list?
9660: doc-list-size
9661:
9662:
9663: Several features of our locals word list implementation make these
9664: operations easy to implement: The locals word lists are organised as
9665: linked lists; the tails of these lists are shared, if the lists
9666: contain some of the same locals; and the address of a name is greater
9667: than the address of the names behind it in the list.
9668:
9669: Another important implementation detail is the variable
9670: @code{dead-code}. It is used by @code{BEGIN} and @code{THEN} to
9671: determine if they can be reached directly or only through the branch
9672: that they resolve. @code{dead-code} is set by @code{UNREACHABLE},
9673: @code{AHEAD}, @code{EXIT} etc., and cleared at the start of a colon
9674: definition, by @code{BEGIN} and usually by @code{THEN}.
9675:
9676: Counted loops are similar to other loops in most respects, but
9677: @code{LEAVE} requires special attention: It performs basically the same
9678: service as @code{AHEAD}, but it does not create a control-flow stack
9679: entry. Therefore the information has to be stored elsewhere;
9680: traditionally, the information was stored in the target fields of the
9681: branches created by the @code{LEAVE}s, by organizing these fields into a
9682: linked list. Unfortunately, this clever trick does not provide enough
9683: space for storing our extended control flow information. Therefore, we
9684: introduce another stack, the leave stack. It contains the control-flow
9685: stack entries for all unresolved @code{LEAVE}s.
9686:
9687: Local names are kept until the end of the colon definition, even if
9688: they are no longer visible in any control-flow path. In a few cases
9689: this may lead to increased space needs for the locals name area, but
9690: usually less than reclaiming this space would cost in code size.
9691:
9692:
9693: @node ANS Forth locals, , Gforth locals, Locals
9694: @subsection ANS Forth locals
9695: @cindex locals, ANS Forth style
9696:
9697: The ANS Forth locals wordset does not define a syntax for locals, but
9698: words that make it possible to define various syntaxes. One of the
9699: possible syntaxes is a subset of the syntax we used in the Gforth locals
9700: wordset, i.e.:
9701:
9702: @example
9703: @{ local1 local2 ... -- comment @}
9704: @end example
9705: @noindent
9706: or
9707: @example
9708: @{ local1 local2 ... @}
9709: @end example
9710:
9711: The order of the locals corresponds to the order in a stack comment. The
9712: restrictions are:
9713:
9714: @itemize @bullet
9715: @item
9716: Locals can only be cell-sized values (no type specifiers are allowed).
9717: @item
9718: Locals can be defined only outside control structures.
9719: @item
9720: Locals can interfere with explicit usage of the return stack. For the
9721: exact (and long) rules, see the standard. If you don't use return stack
9722: accessing words in a definition using locals, you will be all right. The
9723: purpose of this rule is to make locals implementation on the return
9724: stack easier.
9725: @item
9726: The whole definition must be in one line.
9727: @end itemize
9728:
9729: Locals defined in this way behave like @code{VALUE}s
9730: (@pxref{Values}). I.e., they are initialized from the stack. Using their
9731: name produces their value. Their value can be changed using @code{TO}.
9732:
9733: Since this syntax is supported by Gforth directly, you need not do
9734: anything to use it. If you want to port a program using this syntax to
9735: another ANS Forth system, use @file{compat/anslocal.fs} to implement the
9736: syntax on the other system.
9737:
9738: Note that a syntax shown in the standard, section A.13 looks
9739: similar, but is quite different in having the order of locals
9740: reversed. Beware!
9741:
9742: The ANS Forth locals wordset itself consists of a word:
9743:
9744:
9745: doc-(local)
9746:
9747:
9748: The ANS Forth locals extension wordset defines a syntax using @code{locals|}, but it is so
9749: awful that we strongly recommend not to use it. We have implemented this
9750: syntax to make porting to Gforth easy, but do not document it here. The
9751: problem with this syntax is that the locals are defined in an order
9752: reversed with respect to the standard stack comment notation, making
9753: programs harder to read, and easier to misread and miswrite. The only
9754: merit of this syntax is that it is easy to implement using the ANS Forth
9755: locals wordset.
9756:
9757:
9758: @c ----------------------------------------------------------
9759: @node Structures, Object-oriented Forth, Locals, Words
9760: @section Structures
9761: @cindex structures
9762: @cindex records
9763:
9764: This section presents the structure package that comes with Gforth. A
9765: version of the package implemented in ANS Forth is available in
9766: @file{compat/struct.fs}. This package was inspired by a posting on
9767: comp.lang.forth in 1989 (unfortunately I don't remember, by whom;
9768: possibly John Hayes). A version of this section has been published in
9769: ???. Marcel Hendrix provided helpful comments.
9770:
9771: @menu
9772: * Why explicit structure support?::
9773: * Structure Usage::
9774: * Structure Naming Convention::
9775: * Structure Implementation::
9776: * Structure Glossary::
9777: @end menu
9778:
9779: @node Why explicit structure support?, Structure Usage, Structures, Structures
9780: @subsection Why explicit structure support?
9781:
9782: @cindex address arithmetic for structures
9783: @cindex structures using address arithmetic
9784: If we want to use a structure containing several fields, we could simply
9785: reserve memory for it, and access the fields using address arithmetic
9786: (@pxref{Address arithmetic}). As an example, consider a structure with
9787: the following fields
9788:
9789: @table @code
9790: @item a
9791: is a float
9792: @item b
9793: is a cell
9794: @item c
9795: is a float
9796: @end table
9797:
9798: Given the (float-aligned) base address of the structure we get the
9799: address of the field
9800:
9801: @table @code
9802: @item a
9803: without doing anything further.
9804: @item b
9805: with @code{float+}
9806: @item c
9807: with @code{float+ cell+ faligned}
9808: @end table
9809:
9810: It is easy to see that this can become quite tiring.
9811:
9812: Moreover, it is not very readable, because seeing a
9813: @code{cell+} tells us neither which kind of structure is
9814: accessed nor what field is accessed; we have to somehow infer the kind
9815: of structure, and then look up in the documentation, which field of
9816: that structure corresponds to that offset.
9817:
9818: Finally, this kind of address arithmetic also causes maintenance
9819: troubles: If you add or delete a field somewhere in the middle of the
9820: structure, you have to find and change all computations for the fields
9821: afterwards.
9822:
9823: So, instead of using @code{cell+} and friends directly, how
9824: about storing the offsets in constants:
9825:
9826: @example
9827: 0 constant a-offset
9828: 0 float+ constant b-offset
9829: 0 float+ cell+ faligned c-offset
9830: @end example
9831:
9832: Now we can get the address of field @code{x} with @code{x-offset
9833: +}. This is much better in all respects. Of course, you still
9834: have to change all later offset definitions if you add a field. You can
9835: fix this by declaring the offsets in the following way:
9836:
9837: @example
9838: 0 constant a-offset
9839: a-offset float+ constant b-offset
9840: b-offset cell+ faligned constant c-offset
9841: @end example
9842:
9843: Since we always use the offsets with @code{+}, we could use a defining
9844: word @code{cfield} that includes the @code{+} in the action of the
9845: defined word:
9846:
9847: @example
9848: : cfield ( n "name" -- )
9849: create ,
9850: does> ( name execution: addr1 -- addr2 )
9851: @@ + ;
9852:
9853: 0 cfield a
9854: 0 a float+ cfield b
9855: 0 b cell+ faligned cfield c
9856: @end example
9857:
9858: Instead of @code{x-offset +}, we now simply write @code{x}.
9859:
9860: The structure field words now can be used quite nicely. However,
9861: their definition is still a bit cumbersome: We have to repeat the
9862: name, the information about size and alignment is distributed before
9863: and after the field definitions etc. The structure package presented
9864: here addresses these problems.
9865:
9866: @node Structure Usage, Structure Naming Convention, Why explicit structure support?, Structures
9867: @subsection Structure Usage
9868: @cindex structure usage
9869:
9870: @cindex @code{field} usage
9871: @cindex @code{struct} usage
9872: @cindex @code{end-struct} usage
9873: You can define a structure for a (data-less) linked list with:
9874: @example
9875: struct
9876: cell% field list-next
9877: end-struct list%
9878: @end example
9879:
9880: With the address of the list node on the stack, you can compute the
9881: address of the field that contains the address of the next node with
9882: @code{list-next}. E.g., you can determine the length of a list
9883: with:
9884:
9885: @example
9886: : list-length ( list -- n )
9887: \ "list" is a pointer to the first element of a linked list
9888: \ "n" is the length of the list
9889: 0 BEGIN ( list1 n1 )
9890: over
9891: WHILE ( list1 n1 )
9892: 1+ swap list-next @@ swap
9893: REPEAT
9894: nip ;
9895: @end example
9896:
9897: You can reserve memory for a list node in the dictionary with
9898: @code{list% %allot}, which leaves the address of the list node on the
9899: stack. For the equivalent allocation on the heap you can use @code{list%
9900: %alloc} (or, for an @code{allocate}-like stack effect (i.e., with ior),
9901: use @code{list% %allocate}). You can get the the size of a list
9902: node with @code{list% %size} and its alignment with @code{list%
9903: %alignment}.
9904:
9905: Note that in ANS Forth the body of a @code{create}d word is
9906: @code{aligned} but not necessarily @code{faligned};
9907: therefore, if you do a:
9908: @example
9909: create @emph{name} foo% %allot
9910: @end example
9911:
9912: @noindent
9913: then the memory alloted for @code{foo%} is
9914: guaranteed to start at the body of @code{@emph{name}} only if
9915: @code{foo%} contains only character, cell and double fields.
9916:
9917: @cindex structures containing structures
9918: You can include a structure @code{foo%} as a field of
9919: another structure, like this:
9920: @example
9921: struct
9922: ...
9923: foo% field ...
9924: ...
9925: end-struct ...
9926: @end example
9927:
9928: @cindex structure extension
9929: @cindex extended records
9930: Instead of starting with an empty structure, you can extend an
9931: existing structure. E.g., a plain linked list without data, as defined
9932: above, is hardly useful; You can extend it to a linked list of integers,
9933: like this:@footnote{This feature is also known as @emph{extended
9934: records}. It is the main innovation in the Oberon language; in other
9935: words, adding this feature to Modula-2 led Wirth to create a new
9936: language, write a new compiler etc. Adding this feature to Forth just
9937: required a few lines of code.}
9938:
9939: @example
9940: list%
9941: cell% field intlist-int
9942: end-struct intlist%
9943: @end example
9944:
9945: @code{intlist%} is a structure with two fields:
9946: @code{list-next} and @code{intlist-int}.
9947:
9948: @cindex structures containing arrays
9949: You can specify an array type containing @emph{n} elements of
9950: type @code{foo%} like this:
9951:
9952: @example
9953: foo% @emph{n} *
9954: @end example
9955:
9956: You can use this array type in any place where you can use a normal
9957: type, e.g., when defining a @code{field}, or with
9958: @code{%allot}.
9959:
9960: @cindex first field optimization
9961: The first field is at the base address of a structure and the word
9962: for this field (e.g., @code{list-next}) actually does not change
9963: the address on the stack. You may be tempted to leave it away in the
9964: interest of run-time and space efficiency. This is not necessary,
9965: because the structure package optimizes this case and compiling such
9966: words does not generate any code. So, in the interest of readability
9967: and maintainability you should include the word for the field when
9968: accessing the field.
9969:
9970: @node Structure Naming Convention, Structure Implementation, Structure Usage, Structures
9971: @subsection Structure Naming Convention
9972: @cindex structure naming convention
9973:
9974: The field names that come to (my) mind are often quite generic, and,
9975: if used, would cause frequent name clashes. E.g., many structures
9976: probably contain a @code{counter} field. The structure names
9977: that come to (my) mind are often also the logical choice for the names
9978: of words that create such a structure.
9979:
9980: Therefore, I have adopted the following naming conventions:
9981:
9982: @itemize @bullet
9983: @cindex field naming convention
9984: @item
9985: The names of fields are of the form
9986: @code{@emph{struct}-@emph{field}}, where
9987: @code{@emph{struct}} is the basic name of the structure, and
9988: @code{@emph{field}} is the basic name of the field. You can
9989: think of field words as converting the (address of the)
9990: structure into the (address of the) field.
9991:
9992: @cindex structure naming convention
9993: @item
9994: The names of structures are of the form
9995: @code{@emph{struct}%}, where
9996: @code{@emph{struct}} is the basic name of the structure.
9997: @end itemize
9998:
9999: This naming convention does not work that well for fields of extended
10000: structures; e.g., the integer list structure has a field
10001: @code{intlist-int}, but has @code{list-next}, not
10002: @code{intlist-next}.
10003:
10004: @node Structure Implementation, Structure Glossary, Structure Naming Convention, Structures
10005: @subsection Structure Implementation
10006: @cindex structure implementation
10007: @cindex implementation of structures
10008:
10009: The central idea in the implementation is to pass the data about the
10010: structure being built on the stack, not in some global
10011: variable. Everything else falls into place naturally once this design
10012: decision is made.
10013:
10014: The type description on the stack is of the form @emph{align
10015: size}. Keeping the size on the top-of-stack makes dealing with arrays
10016: very simple.
10017:
10018: @code{field} is a defining word that uses @code{Create}
10019: and @code{DOES>}. The body of the field contains the offset
10020: of the field, and the normal @code{DOES>} action is simply:
10021:
10022: @example
10023: @@ +
10024: @end example
10025:
10026: @noindent
10027: i.e., add the offset to the address, giving the stack effect
10028: @i{addr1 -- addr2} for a field.
10029:
10030: @cindex first field optimization, implementation
10031: This simple structure is slightly complicated by the optimization
10032: for fields with offset 0, which requires a different
10033: @code{DOES>}-part (because we cannot rely on there being
10034: something on the stack if such a field is invoked during
10035: compilation). Therefore, we put the different @code{DOES>}-parts
10036: in separate words, and decide which one to invoke based on the
10037: offset. For a zero offset, the field is basically a noop; it is
10038: immediate, and therefore no code is generated when it is compiled.
10039:
10040: @node Structure Glossary, , Structure Implementation, Structures
10041: @subsection Structure Glossary
10042: @cindex structure glossary
10043:
10044:
10045: doc-%align
10046: doc-%alignment
10047: doc-%alloc
10048: doc-%allocate
10049: doc-%allot
10050: doc-cell%
10051: doc-char%
10052: doc-dfloat%
10053: doc-double%
10054: doc-end-struct
10055: doc-field
10056: doc-float%
10057: doc-naligned
10058: doc-sfloat%
10059: doc-%size
10060: doc-struct
10061:
10062:
10063: @c -------------------------------------------------------------
10064: @node Object-oriented Forth, Passing Commands to the OS, Structures, Words
10065: @section Object-oriented Forth
10066:
10067: Gforth comes with three packages for object-oriented programming:
10068: @file{objects.fs}, @file{oof.fs}, and @file{mini-oof.fs}; none of them
10069: is preloaded, so you have to @code{include} them before use. The most
10070: important differences between these packages (and others) are discussed
10071: in @ref{Comparison with other object models}. All packages are written
10072: in ANS Forth and can be used with any other ANS Forth.
10073:
10074: @menu
10075: * Why object-oriented programming?::
10076: * Object-Oriented Terminology::
10077: * Objects::
10078: * OOF::
10079: * Mini-OOF::
10080: * Comparison with other object models::
10081: @end menu
10082:
10083: @c ----------------------------------------------------------------
10084: @node Why object-oriented programming?, Object-Oriented Terminology, Object-oriented Forth, Object-oriented Forth
10085: @subsection Why object-oriented programming?
10086: @cindex object-oriented programming motivation
10087: @cindex motivation for object-oriented programming
10088:
10089: Often we have to deal with several data structures (@emph{objects}),
10090: that have to be treated similarly in some respects, but differently in
10091: others. Graphical objects are the textbook example: circles, triangles,
10092: dinosaurs, icons, and others, and we may want to add more during program
10093: development. We want to apply some operations to any graphical object,
10094: e.g., @code{draw} for displaying it on the screen. However, @code{draw}
10095: has to do something different for every kind of object.
10096: @comment TODO add some other operations eg perimeter, area
10097: @comment and tie in to concrete examples later..
10098:
10099: We could implement @code{draw} as a big @code{CASE}
10100: control structure that executes the appropriate code depending on the
10101: kind of object to be drawn. This would be not be very elegant, and,
10102: moreover, we would have to change @code{draw} every time we add
10103: a new kind of graphical object (say, a spaceship).
10104:
10105: What we would rather do is: When defining spaceships, we would tell
10106: the system: ``Here's how you @code{draw} a spaceship; you figure
10107: out the rest''.
10108:
10109: This is the problem that all systems solve that (rightfully) call
10110: themselves object-oriented; the object-oriented packages presented here
10111: solve this problem (and not much else).
10112: @comment TODO ?list properties of oo systems.. oo vs o-based?
10113:
10114: @c ------------------------------------------------------------------------
10115: @node Object-Oriented Terminology, Objects, Why object-oriented programming?, Object-oriented Forth
10116: @subsection Object-Oriented Terminology
10117: @cindex object-oriented terminology
10118: @cindex terminology for object-oriented programming
10119:
10120: This section is mainly for reference, so you don't have to understand
10121: all of it right away. The terminology is mainly Smalltalk-inspired. In
10122: short:
10123:
10124: @table @emph
10125: @cindex class
10126: @item class
10127: a data structure definition with some extras.
10128:
10129: @cindex object
10130: @item object
10131: an instance of the data structure described by the class definition.
10132:
10133: @cindex instance variables
10134: @item instance variables
10135: fields of the data structure.
10136:
10137: @cindex selector
10138: @cindex method selector
10139: @cindex virtual function
10140: @item selector
10141: (or @emph{method selector}) a word (e.g.,
10142: @code{draw}) that performs an operation on a variety of data
10143: structures (classes). A selector describes @emph{what} operation to
10144: perform. In C++ terminology: a (pure) virtual function.
10145:
10146: @cindex method
10147: @item method
10148: the concrete definition that performs the operation
10149: described by the selector for a specific class. A method specifies
10150: @emph{how} the operation is performed for a specific class.
10151:
10152: @cindex selector invocation
10153: @cindex message send
10154: @cindex invoking a selector
10155: @item selector invocation
10156: a call of a selector. One argument of the call (the TOS (top-of-stack))
10157: is used for determining which method is used. In Smalltalk terminology:
10158: a message (consisting of the selector and the other arguments) is sent
10159: to the object.
10160:
10161: @cindex receiving object
10162: @item receiving object
10163: the object used for determining the method executed by a selector
10164: invocation. In the @file{objects.fs} model, it is the object that is on
10165: the TOS when the selector is invoked. (@emph{Receiving} comes from
10166: the Smalltalk @emph{message} terminology.)
10167:
10168: @cindex child class
10169: @cindex parent class
10170: @cindex inheritance
10171: @item child class
10172: a class that has (@emph{inherits}) all properties (instance variables,
10173: selectors, methods) from a @emph{parent class}. In Smalltalk
10174: terminology: The subclass inherits from the superclass. In C++
10175: terminology: The derived class inherits from the base class.
10176:
10177: @end table
10178:
10179: @c If you wonder about the message sending terminology, it comes from
10180: @c a time when each object had it's own task and objects communicated via
10181: @c message passing; eventually the Smalltalk developers realized that
10182: @c they can do most things through simple (indirect) calls. They kept the
10183: @c terminology.
10184:
10185: @c --------------------------------------------------------------
10186: @node Objects, OOF, Object-Oriented Terminology, Object-oriented Forth
10187: @subsection The @file{objects.fs} model
10188: @cindex objects
10189: @cindex object-oriented programming
10190:
10191: @cindex @file{objects.fs}
10192: @cindex @file{oof.fs}
10193:
10194: This section describes the @file{objects.fs} package. This material also
10195: has been published in M. Anton Ertl,
10196: @cite{@uref{http://www.complang.tuwien.ac.at/forth/objects/objects.html,
10197: Yet Another Forth Objects Package}}, Forth Dimensions 19(2), pages
10198: 37--43.
10199: @c McKewan's and Zsoter's packages
10200:
10201: This section assumes that you have read @ref{Structures}.
10202:
10203: The techniques on which this model is based have been used to implement
10204: the parser generator, Gray, and have also been used in Gforth for
10205: implementing the various flavours of word lists (hashed or not,
10206: case-sensitive or not, special-purpose word lists for locals etc.).
10207:
10208:
10209: @menu
10210: * Properties of the Objects model::
10211: * Basic Objects Usage::
10212: * The Objects base class::
10213: * Creating objects::
10214: * Object-Oriented Programming Style::
10215: * Class Binding::
10216: * Method conveniences::
10217: * Classes and Scoping::
10218: * Dividing classes::
10219: * Object Interfaces::
10220: * Objects Implementation::
10221: * Objects Glossary::
10222: @end menu
10223:
10224: Marcel Hendrix provided helpful comments on this section. Andras Zsoter
10225: and Bernd Paysan helped me with the related works section.
10226:
10227: @node Properties of the Objects model, Basic Objects Usage, Objects, Objects
10228: @subsubsection Properties of the @file{objects.fs} model
10229: @cindex @file{objects.fs} properties
10230:
10231: @itemize @bullet
10232: @item
10233: It is straightforward to pass objects on the stack. Passing
10234: selectors on the stack is a little less convenient, but possible.
10235:
10236: @item
10237: Objects are just data structures in memory, and are referenced by their
10238: address. You can create words for objects with normal defining words
10239: like @code{constant}. Likewise, there is no difference between instance
10240: variables that contain objects and those that contain other data.
10241:
10242: @item
10243: Late binding is efficient and easy to use.
10244:
10245: @item
10246: It avoids parsing, and thus avoids problems with state-smartness
10247: and reduced extensibility; for convenience there are a few parsing
10248: words, but they have non-parsing counterparts. There are also a few
10249: defining words that parse. This is hard to avoid, because all standard
10250: defining words parse (except @code{:noname}); however, such
10251: words are not as bad as many other parsing words, because they are not
10252: state-smart.
10253:
10254: @item
10255: It does not try to incorporate everything. It does a few things and does
10256: them well (IMO). In particular, this model was not designed to support
10257: information hiding (although it has features that may help); you can use
10258: a separate package for achieving this.
10259:
10260: @item
10261: It is layered; you don't have to learn and use all features to use this
10262: model. Only a few features are necessary (@pxref{Basic Objects Usage},
10263: @pxref{The Objects base class}, @pxref{Creating objects}.), the others
10264: are optional and independent of each other.
10265:
10266: @item
10267: An implementation in ANS Forth is available.
10268:
10269: @end itemize
10270:
10271:
10272: @node Basic Objects Usage, The Objects base class, Properties of the Objects model, Objects
10273: @subsubsection Basic @file{objects.fs} Usage
10274: @cindex basic objects usage
10275: @cindex objects, basic usage
10276:
10277: You can define a class for graphical objects like this:
10278:
10279: @cindex @code{class} usage
10280: @cindex @code{end-class} usage
10281: @cindex @code{selector} usage
10282: @example
10283: object class \ "object" is the parent class
10284: selector draw ( x y graphical -- )
10285: end-class graphical
10286: @end example
10287:
10288: This code defines a class @code{graphical} with an
10289: operation @code{draw}. We can perform the operation
10290: @code{draw} on any @code{graphical} object, e.g.:
10291:
10292: @example
10293: 100 100 t-rex draw
10294: @end example
10295:
10296: @noindent
10297: where @code{t-rex} is a word (say, a constant) that produces a
10298: graphical object.
10299:
10300: @comment TODO add a 2nd operation eg perimeter.. and use for
10301: @comment a concrete example
10302:
10303: @cindex abstract class
10304: How do we create a graphical object? With the present definitions,
10305: we cannot create a useful graphical object. The class
10306: @code{graphical} describes graphical objects in general, but not
10307: any concrete graphical object type (C++ users would call it an
10308: @emph{abstract class}); e.g., there is no method for the selector
10309: @code{draw} in the class @code{graphical}.
10310:
10311: For concrete graphical objects, we define child classes of the
10312: class @code{graphical}, e.g.:
10313:
10314: @cindex @code{overrides} usage
10315: @cindex @code{field} usage in class definition
10316: @example
10317: graphical class \ "graphical" is the parent class
10318: cell% field circle-radius
10319:
10320: :noname ( x y circle -- )
10321: circle-radius @@ draw-circle ;
10322: overrides draw
10323:
10324: :noname ( n-radius circle -- )
10325: circle-radius ! ;
10326: overrides construct
10327:
10328: end-class circle
10329: @end example
10330:
10331: Here we define a class @code{circle} as a child of @code{graphical},
10332: with field @code{circle-radius} (which behaves just like a field
10333: (@pxref{Structures}); it defines (using @code{overrides}) new methods
10334: for the selectors @code{draw} and @code{construct} (@code{construct} is
10335: defined in @code{object}, the parent class of @code{graphical}).
10336:
10337: Now we can create a circle on the heap (i.e.,
10338: @code{allocate}d memory) with:
10339:
10340: @cindex @code{heap-new} usage
10341: @example
10342: 50 circle heap-new constant my-circle
10343: @end example
10344:
10345: @noindent
10346: @code{heap-new} invokes @code{construct}, thus
10347: initializing the field @code{circle-radius} with 50. We can draw
10348: this new circle at (100,100) with:
10349:
10350: @example
10351: 100 100 my-circle draw
10352: @end example
10353:
10354: @cindex selector invocation, restrictions
10355: @cindex class definition, restrictions
10356: Note: You can only invoke a selector if the object on the TOS
10357: (the receiving object) belongs to the class where the selector was
10358: defined or one of its descendents; e.g., you can invoke
10359: @code{draw} only for objects belonging to @code{graphical}
10360: or its descendents (e.g., @code{circle}). Immediately before
10361: @code{end-class}, the search order has to be the same as
10362: immediately after @code{class}.
10363:
10364: @node The Objects base class, Creating objects, Basic Objects Usage, Objects
10365: @subsubsection The @file{object.fs} base class
10366: @cindex @code{object} class
10367:
10368: When you define a class, you have to specify a parent class. So how do
10369: you start defining classes? There is one class available from the start:
10370: @code{object}. It is ancestor for all classes and so is the
10371: only class that has no parent. It has two selectors: @code{construct}
10372: and @code{print}.
10373:
10374: @node Creating objects, Object-Oriented Programming Style, The Objects base class, Objects
10375: @subsubsection Creating objects
10376: @cindex creating objects
10377: @cindex object creation
10378: @cindex object allocation options
10379:
10380: @cindex @code{heap-new} discussion
10381: @cindex @code{dict-new} discussion
10382: @cindex @code{construct} discussion
10383: You can create and initialize an object of a class on the heap with
10384: @code{heap-new} ( ... class -- object ) and in the dictionary
10385: (allocation with @code{allot}) with @code{dict-new} (
10386: ... class -- object ). Both words invoke @code{construct}, which
10387: consumes the stack items indicated by "..." above.
10388:
10389: @cindex @code{init-object} discussion
10390: @cindex @code{class-inst-size} discussion
10391: If you want to allocate memory for an object yourself, you can get its
10392: alignment and size with @code{class-inst-size 2@@} ( class --
10393: align size ). Once you have memory for an object, you can initialize
10394: it with @code{init-object} ( ... class object -- );
10395: @code{construct} does only a part of the necessary work.
10396:
10397: @node Object-Oriented Programming Style, Class Binding, Creating objects, Objects
10398: @subsubsection Object-Oriented Programming Style
10399: @cindex object-oriented programming style
10400: @cindex programming style, object-oriented
10401:
10402: This section is not exhaustive.
10403:
10404: @cindex stack effects of selectors
10405: @cindex selectors and stack effects
10406: In general, it is a good idea to ensure that all methods for the
10407: same selector have the same stack effect: when you invoke a selector,
10408: you often have no idea which method will be invoked, so, unless all
10409: methods have the same stack effect, you will not know the stack effect
10410: of the selector invocation.
10411:
10412: One exception to this rule is methods for the selector
10413: @code{construct}. We know which method is invoked, because we
10414: specify the class to be constructed at the same place. Actually, I
10415: defined @code{construct} as a selector only to give the users a
10416: convenient way to specify initialization. The way it is used, a
10417: mechanism different from selector invocation would be more natural
10418: (but probably would take more code and more space to explain).
10419:
10420: @node Class Binding, Method conveniences, Object-Oriented Programming Style, Objects
10421: @subsubsection Class Binding
10422: @cindex class binding
10423: @cindex early binding
10424:
10425: @cindex late binding
10426: Normal selector invocations determine the method at run-time depending
10427: on the class of the receiving object. This run-time selection is called
10428: @i{late binding}.
10429:
10430: Sometimes it's preferable to invoke a different method. For example,
10431: you might want to use the simple method for @code{print}ing
10432: @code{object}s instead of the possibly long-winded @code{print} method
10433: of the receiver class. You can achieve this by replacing the invocation
10434: of @code{print} with:
10435:
10436: @cindex @code{[bind]} usage
10437: @example
10438: [bind] object print
10439: @end example
10440:
10441: @noindent
10442: in compiled code or:
10443:
10444: @cindex @code{bind} usage
10445: @example
10446: bind object print
10447: @end example
10448:
10449: @cindex class binding, alternative to
10450: @noindent
10451: in interpreted code. Alternatively, you can define the method with a
10452: name (e.g., @code{print-object}), and then invoke it through the
10453: name. Class binding is just a (often more convenient) way to achieve
10454: the same effect; it avoids name clutter and allows you to invoke
10455: methods directly without naming them first.
10456:
10457: @cindex superclass binding
10458: @cindex parent class binding
10459: A frequent use of class binding is this: When we define a method
10460: for a selector, we often want the method to do what the selector does
10461: in the parent class, and a little more. There is a special word for
10462: this purpose: @code{[parent]}; @code{[parent]
10463: @emph{selector}} is equivalent to @code{[bind] @emph{parent
10464: selector}}, where @code{@emph{parent}} is the parent
10465: class of the current class. E.g., a method definition might look like:
10466:
10467: @cindex @code{[parent]} usage
10468: @example
10469: :noname
10470: dup [parent] foo \ do parent's foo on the receiving object
10471: ... \ do some more
10472: ; overrides foo
10473: @end example
10474:
10475: @cindex class binding as optimization
10476: In @cite{Object-oriented programming in ANS Forth} (Forth Dimensions,
10477: March 1997), Andrew McKewan presents class binding as an optimization
10478: technique. I recommend not using it for this purpose unless you are in
10479: an emergency. Late binding is pretty fast with this model anyway, so the
10480: benefit of using class binding is small; the cost of using class binding
10481: where it is not appropriate is reduced maintainability.
10482:
10483: While we are at programming style questions: You should bind
10484: selectors only to ancestor classes of the receiving object. E.g., say,
10485: you know that the receiving object is of class @code{foo} or its
10486: descendents; then you should bind only to @code{foo} and its
10487: ancestors.
10488:
10489: @node Method conveniences, Classes and Scoping, Class Binding, Objects
10490: @subsubsection Method conveniences
10491: @cindex method conveniences
10492:
10493: In a method you usually access the receiving object pretty often. If
10494: you define the method as a plain colon definition (e.g., with
10495: @code{:noname}), you may have to do a lot of stack
10496: gymnastics. To avoid this, you can define the method with @code{m:
10497: ... ;m}. E.g., you could define the method for
10498: @code{draw}ing a @code{circle} with
10499:
10500: @cindex @code{this} usage
10501: @cindex @code{m:} usage
10502: @cindex @code{;m} usage
10503: @example
10504: m: ( x y circle -- )
10505: ( x y ) this circle-radius @@ draw-circle ;m
10506: @end example
10507:
10508: @cindex @code{exit} in @code{m: ... ;m}
10509: @cindex @code{exitm} discussion
10510: @cindex @code{catch} in @code{m: ... ;m}
10511: When this method is executed, the receiver object is removed from the
10512: stack; you can access it with @code{this} (admittedly, in this
10513: example the use of @code{m: ... ;m} offers no advantage). Note
10514: that I specify the stack effect for the whole method (i.e. including
10515: the receiver object), not just for the code between @code{m:}
10516: and @code{;m}. You cannot use @code{exit} in
10517: @code{m:...;m}; instead, use
10518: @code{exitm}.@footnote{Moreover, for any word that calls
10519: @code{catch} and was defined before loading
10520: @code{objects.fs}, you have to redefine it like I redefined
10521: @code{catch}: @code{: catch this >r catch r> to-this ;}}
10522:
10523: @cindex @code{inst-var} usage
10524: You will frequently use sequences of the form @code{this
10525: @emph{field}} (in the example above: @code{this
10526: circle-radius}). If you use the field only in this way, you can
10527: define it with @code{inst-var} and eliminate the
10528: @code{this} before the field name. E.g., the @code{circle}
10529: class above could also be defined with:
10530:
10531: @example
10532: graphical class
10533: cell% inst-var radius
10534:
10535: m: ( x y circle -- )
10536: radius @@ draw-circle ;m
10537: overrides draw
10538:
10539: m: ( n-radius circle -- )
10540: radius ! ;m
10541: overrides construct
10542:
10543: end-class circle
10544: @end example
10545:
10546: @code{radius} can only be used in @code{circle} and its
10547: descendent classes and inside @code{m:...;m}.
10548:
10549: @cindex @code{inst-value} usage
10550: You can also define fields with @code{inst-value}, which is
10551: to @code{inst-var} what @code{value} is to
10552: @code{variable}. You can change the value of such a field with
10553: @code{[to-inst]}. E.g., we could also define the class
10554: @code{circle} like this:
10555:
10556: @example
10557: graphical class
10558: inst-value radius
10559:
10560: m: ( x y circle -- )
10561: radius draw-circle ;m
10562: overrides draw
10563:
10564: m: ( n-radius circle -- )
10565: [to-inst] radius ;m
10566: overrides construct
10567:
10568: end-class circle
10569: @end example
10570:
10571: Finally, you can define named methods with @code{:m}. One use of this
10572: feature is the definition of words that occur only in one class and are
10573: not intended to be overridden, but which still need method context
10574: (e.g., for accessing @code{inst-var}s). Another use is for methods that
10575: would be bound frequently, if defined anonymously.
10576:
10577:
10578: @node Classes and Scoping, Dividing classes, Method conveniences, Objects
10579: @subsubsection Classes and Scoping
10580: @cindex classes and scoping
10581: @cindex scoping and classes
10582:
10583: Inheritance is frequent, unlike structure extension. This exacerbates
10584: the problem with the field name convention (@pxref{Structure Naming
10585: Convention}): One always has to remember in which class the field was
10586: originally defined; changing a part of the class structure would require
10587: changes for renaming in otherwise unaffected code.
10588:
10589: @cindex @code{inst-var} visibility
10590: @cindex @code{inst-value} visibility
10591: To solve this problem, I added a scoping mechanism (which was not in my
10592: original charter): A field defined with @code{inst-var} (or
10593: @code{inst-value}) is visible only in the class where it is defined and in
10594: the descendent classes of this class. Using such fields only makes
10595: sense in @code{m:}-defined methods in these classes anyway.
10596:
10597: This scoping mechanism allows us to use the unadorned field name,
10598: because name clashes with unrelated words become much less likely.
10599:
10600: @cindex @code{protected} discussion
10601: @cindex @code{private} discussion
10602: Once we have this mechanism, we can also use it for controlling the
10603: visibility of other words: All words defined after
10604: @code{protected} are visible only in the current class and its
10605: descendents. @code{public} restores the compilation
10606: (i.e. @code{current}) word list that was in effect before. If you
10607: have several @code{protected}s without an intervening
10608: @code{public} or @code{set-current}, @code{public}
10609: will restore the compilation word list in effect before the first of
10610: these @code{protected}s.
10611:
10612: @node Dividing classes, Object Interfaces, Classes and Scoping, Objects
10613: @subsubsection Dividing classes
10614: @cindex Dividing classes
10615: @cindex @code{methods}...@code{end-methods}
10616:
10617: You may want to do the definition of methods separate from the
10618: definition of the class, its selectors, fields, and instance variables,
10619: i.e., separate the implementation from the definition. You can do this
10620: in the following way:
10621:
10622: @example
10623: graphical class
10624: inst-value radius
10625: end-class circle
10626:
10627: ... \ do some other stuff
10628:
10629: circle methods \ now we are ready
10630:
10631: m: ( x y circle -- )
10632: radius draw-circle ;m
10633: overrides draw
10634:
10635: m: ( n-radius circle -- )
10636: [to-inst] radius ;m
10637: overrides construct
10638:
10639: end-methods
10640: @end example
10641:
10642: You can use several @code{methods}...@code{end-methods} sections. The
10643: only things you can do to the class in these sections are: defining
10644: methods, and overriding the class's selectors. You must not define new
10645: selectors or fields.
10646:
10647: Note that you often have to override a selector before using it. In
10648: particular, you usually have to override @code{construct} with a new
10649: method before you can invoke @code{heap-new} and friends. E.g., you
10650: must not create a circle before the @code{overrides construct} sequence
10651: in the example above.
10652:
10653: @node Object Interfaces, Objects Implementation, Dividing classes, Objects
10654: @subsubsection Object Interfaces
10655: @cindex object interfaces
10656: @cindex interfaces for objects
10657:
10658: In this model you can only call selectors defined in the class of the
10659: receiving objects or in one of its ancestors. If you call a selector
10660: with a receiving object that is not in one of these classes, the
10661: result is undefined; if you are lucky, the program crashes
10662: immediately.
10663:
10664: @cindex selectors common to hardly-related classes
10665: Now consider the case when you want to have a selector (or several)
10666: available in two classes: You would have to add the selector to a
10667: common ancestor class, in the worst case to @code{object}. You
10668: may not want to do this, e.g., because someone else is responsible for
10669: this ancestor class.
10670:
10671: The solution for this problem is interfaces. An interface is a
10672: collection of selectors. If a class implements an interface, the
10673: selectors become available to the class and its descendents. A class
10674: can implement an unlimited number of interfaces. For the problem
10675: discussed above, we would define an interface for the selector(s), and
10676: both classes would implement the interface.
10677:
10678: As an example, consider an interface @code{storage} for
10679: writing objects to disk and getting them back, and a class
10680: @code{foo} that implements it. The code would look like this:
10681:
10682: @cindex @code{interface} usage
10683: @cindex @code{end-interface} usage
10684: @cindex @code{implementation} usage
10685: @example
10686: interface
10687: selector write ( file object -- )
10688: selector read1 ( file object -- )
10689: end-interface storage
10690:
10691: bar class
10692: storage implementation
10693:
10694: ... overrides write
10695: ... overrides read1
10696: ...
10697: end-class foo
10698: @end example
10699:
10700: @noindent
10701: (I would add a word @code{read} @i{( file -- object )} that uses
10702: @code{read1} internally, but that's beyond the point illustrated
10703: here.)
10704:
10705: Note that you cannot use @code{protected} in an interface; and
10706: of course you cannot define fields.
10707:
10708: In the Neon model, all selectors are available for all classes;
10709: therefore it does not need interfaces. The price you pay in this model
10710: is slower late binding, and therefore, added complexity to avoid late
10711: binding.
10712:
10713: @node Objects Implementation, Objects Glossary, Object Interfaces, Objects
10714: @subsubsection @file{objects.fs} Implementation
10715: @cindex @file{objects.fs} implementation
10716:
10717: @cindex @code{object-map} discussion
10718: An object is a piece of memory, like one of the data structures
10719: described with @code{struct...end-struct}. It has a field
10720: @code{object-map} that points to the method map for the object's
10721: class.
10722:
10723: @cindex method map
10724: @cindex virtual function table
10725: The @emph{method map}@footnote{This is Self terminology; in C++
10726: terminology: virtual function table.} is an array that contains the
10727: execution tokens (@i{xt}s) of the methods for the object's class. Each
10728: selector contains an offset into a method map.
10729:
10730: @cindex @code{selector} implementation, class
10731: @code{selector} is a defining word that uses
10732: @code{CREATE} and @code{DOES>}. The body of the
10733: selector contains the offset; the @code{DOES>} action for a
10734: class selector is, basically:
10735:
10736: @example
10737: ( object addr ) @@ over object-map @@ + @@ execute
10738: @end example
10739:
10740: Since @code{object-map} is the first field of the object, it
10741: does not generate any code. As you can see, calling a selector has a
10742: small, constant cost.
10743:
10744: @cindex @code{current-interface} discussion
10745: @cindex class implementation and representation
10746: A class is basically a @code{struct} combined with a method
10747: map. During the class definition the alignment and size of the class
10748: are passed on the stack, just as with @code{struct}s, so
10749: @code{field} can also be used for defining class
10750: fields. However, passing more items on the stack would be
10751: inconvenient, so @code{class} builds a data structure in memory,
10752: which is accessed through the variable
10753: @code{current-interface}. After its definition is complete, the
10754: class is represented on the stack by a pointer (e.g., as parameter for
10755: a child class definition).
10756:
10757: A new class starts off with the alignment and size of its parent,
10758: and a copy of the parent's method map. Defining new fields extends the
10759: size and alignment; likewise, defining new selectors extends the
10760: method map. @code{overrides} just stores a new @i{xt} in the method
10761: map at the offset given by the selector.
10762:
10763: @cindex class binding, implementation
10764: Class binding just gets the @i{xt} at the offset given by the selector
10765: from the class's method map and @code{compile,}s (in the case of
10766: @code{[bind]}) it.
10767:
10768: @cindex @code{this} implementation
10769: @cindex @code{catch} and @code{this}
10770: @cindex @code{this} and @code{catch}
10771: I implemented @code{this} as a @code{value}. At the
10772: start of an @code{m:...;m} method the old @code{this} is
10773: stored to the return stack and restored at the end; and the object on
10774: the TOS is stored @code{TO this}. This technique has one
10775: disadvantage: If the user does not leave the method via
10776: @code{;m}, but via @code{throw} or @code{exit},
10777: @code{this} is not restored (and @code{exit} may
10778: crash). To deal with the @code{throw} problem, I have redefined
10779: @code{catch} to save and restore @code{this}; the same
10780: should be done with any word that can catch an exception. As for
10781: @code{exit}, I simply forbid it (as a replacement, there is
10782: @code{exitm}).
10783:
10784: @cindex @code{inst-var} implementation
10785: @code{inst-var} is just the same as @code{field}, with
10786: a different @code{DOES>} action:
10787: @example
10788: @@ this +
10789: @end example
10790: Similar for @code{inst-value}.
10791:
10792: @cindex class scoping implementation
10793: Each class also has a word list that contains the words defined with
10794: @code{inst-var} and @code{inst-value}, and its protected
10795: words. It also has a pointer to its parent. @code{class} pushes
10796: the word lists of the class and all its ancestors onto the search order stack,
10797: and @code{end-class} drops them.
10798:
10799: @cindex interface implementation
10800: An interface is like a class without fields, parent and protected
10801: words; i.e., it just has a method map. If a class implements an
10802: interface, its method map contains a pointer to the method map of the
10803: interface. The positive offsets in the map are reserved for class
10804: methods, therefore interface map pointers have negative
10805: offsets. Interfaces have offsets that are unique throughout the
10806: system, unlike class selectors, whose offsets are only unique for the
10807: classes where the selector is available (invokable).
10808:
10809: This structure means that interface selectors have to perform one
10810: indirection more than class selectors to find their method. Their body
10811: contains the interface map pointer offset in the class method map, and
10812: the method offset in the interface method map. The
10813: @code{does>} action for an interface selector is, basically:
10814:
10815: @example
10816: ( object selector-body )
10817: 2dup selector-interface @@ ( object selector-body object interface-offset )
10818: swap object-map @@ + @@ ( object selector-body map )
10819: swap selector-offset @@ + @@ execute
10820: @end example
10821:
10822: where @code{object-map} and @code{selector-offset} are
10823: first fields and generate no code.
10824:
10825: As a concrete example, consider the following code:
10826:
10827: @example
10828: interface
10829: selector if1sel1
10830: selector if1sel2
10831: end-interface if1
10832:
10833: object class
10834: if1 implementation
10835: selector cl1sel1
10836: cell% inst-var cl1iv1
10837:
10838: ' m1 overrides construct
10839: ' m2 overrides if1sel1
10840: ' m3 overrides if1sel2
10841: ' m4 overrides cl1sel2
10842: end-class cl1
10843:
10844: create obj1 object dict-new drop
10845: create obj2 cl1 dict-new drop
10846: @end example
10847:
10848: The data structure created by this code (including the data structure
10849: for @code{object}) is shown in the <a
10850: href="objects-implementation.eps">figure</a>, assuming a cell size of 4.
10851: @comment TODO add this diagram..
10852:
10853: @node Objects Glossary, , Objects Implementation, Objects
10854: @subsubsection @file{objects.fs} Glossary
10855: @cindex @file{objects.fs} Glossary
10856:
10857:
10858: doc---objects-bind
10859: doc---objects-<bind>
10860: doc---objects-bind'
10861: doc---objects-[bind]
10862: doc---objects-class
10863: doc---objects-class->map
10864: doc---objects-class-inst-size
10865: doc---objects-class-override!
10866: doc---objects-construct
10867: doc---objects-current'
10868: doc---objects-[current]
10869: doc---objects-current-interface
10870: doc---objects-dict-new
10871: doc---objects-drop-order
10872: doc---objects-end-class
10873: doc---objects-end-class-noname
10874: doc---objects-end-interface
10875: doc---objects-end-interface-noname
10876: doc---objects-end-methods
10877: doc---objects-exitm
10878: doc---objects-heap-new
10879: doc---objects-implementation
10880: doc---objects-init-object
10881: doc---objects-inst-value
10882: doc---objects-inst-var
10883: doc---objects-interface
10884: doc---objects-m:
10885: doc---objects-:m
10886: doc---objects-;m
10887: doc---objects-method
10888: doc---objects-methods
10889: doc---objects-object
10890: doc---objects-overrides
10891: doc---objects-[parent]
10892: doc---objects-print
10893: doc---objects-protected
10894: doc---objects-public
10895: doc---objects-push-order
10896: doc---objects-selector
10897: doc---objects-this
10898: doc---objects-<to-inst>
10899: doc---objects-[to-inst]
10900: doc---objects-to-this
10901: doc---objects-xt-new
10902:
10903:
10904: @c -------------------------------------------------------------
10905: @node OOF, Mini-OOF, Objects, Object-oriented Forth
10906: @subsection The @file{oof.fs} model
10907: @cindex oof
10908: @cindex object-oriented programming
10909:
10910: @cindex @file{objects.fs}
10911: @cindex @file{oof.fs}
10912:
10913: This section describes the @file{oof.fs} package.
10914:
10915: The package described in this section has been used in bigFORTH since 1991, and
10916: used for two large applications: a chromatographic system used to
10917: create new medicaments, and a graphic user interface library (MINOS).
10918:
10919: You can find a description (in German) of @file{oof.fs} in @cite{Object
10920: oriented bigFORTH} by Bernd Paysan, published in @cite{Vierte Dimension}
10921: 10(2), 1994.
10922:
10923: @menu
10924: * Properties of the OOF model::
10925: * Basic OOF Usage::
10926: * The OOF base class::
10927: * Class Declaration::
10928: * Class Implementation::
10929: @end menu
10930:
10931: @node Properties of the OOF model, Basic OOF Usage, OOF, OOF
10932: @subsubsection Properties of the @file{oof.fs} model
10933: @cindex @file{oof.fs} properties
10934:
10935: @itemize @bullet
10936: @item
10937: This model combines object oriented programming with information
10938: hiding. It helps you writing large application, where scoping is
10939: necessary, because it provides class-oriented scoping.
10940:
10941: @item
10942: Named objects, object pointers, and object arrays can be created,
10943: selector invocation uses the ``object selector'' syntax. Selector invocation
10944: to objects and/or selectors on the stack is a bit less convenient, but
10945: possible.
10946:
10947: @item
10948: Selector invocation and instance variable usage of the active object is
10949: straightforward, since both make use of the active object.
10950:
10951: @item
10952: Late binding is efficient and easy to use.
10953:
10954: @item
10955: State-smart objects parse selectors. However, extensibility is provided
10956: using a (parsing) selector @code{postpone} and a selector @code{'}.
10957:
10958: @item
10959: An implementation in ANS Forth is available.
10960:
10961: @end itemize
10962:
10963:
10964: @node Basic OOF Usage, The OOF base class, Properties of the OOF model, OOF
10965: @subsubsection Basic @file{oof.fs} Usage
10966: @cindex @file{oof.fs} usage
10967:
10968: This section uses the same example as for @code{objects} (@pxref{Basic Objects Usage}).
10969:
10970: You can define a class for graphical objects like this:
10971:
10972: @cindex @code{class} usage
10973: @cindex @code{class;} usage
10974: @cindex @code{method} usage
10975: @example
10976: object class graphical \ "object" is the parent class
10977: method draw ( x y graphical -- )
10978: class;
10979: @end example
10980:
10981: This code defines a class @code{graphical} with an
10982: operation @code{draw}. We can perform the operation
10983: @code{draw} on any @code{graphical} object, e.g.:
10984:
10985: @example
10986: 100 100 t-rex draw
10987: @end example
10988:
10989: @noindent
10990: where @code{t-rex} is an object or object pointer, created with e.g.
10991: @code{graphical : t-rex}.
10992:
10993: @cindex abstract class
10994: How do we create a graphical object? With the present definitions,
10995: we cannot create a useful graphical object. The class
10996: @code{graphical} describes graphical objects in general, but not
10997: any concrete graphical object type (C++ users would call it an
10998: @emph{abstract class}); e.g., there is no method for the selector
10999: @code{draw} in the class @code{graphical}.
11000:
11001: For concrete graphical objects, we define child classes of the
11002: class @code{graphical}, e.g.:
11003:
11004: @example
11005: graphical class circle \ "graphical" is the parent class
11006: cell var circle-radius
11007: how:
11008: : draw ( x y -- )
11009: circle-radius @@ draw-circle ;
11010:
11011: : init ( n-radius -- (
11012: circle-radius ! ;
11013: class;
11014: @end example
11015:
11016: Here we define a class @code{circle} as a child of @code{graphical},
11017: with a field @code{circle-radius}; it defines new methods for the
11018: selectors @code{draw} and @code{init} (@code{init} is defined in
11019: @code{object}, the parent class of @code{graphical}).
11020:
11021: Now we can create a circle in the dictionary with:
11022:
11023: @example
11024: 50 circle : my-circle
11025: @end example
11026:
11027: @noindent
11028: @code{:} invokes @code{init}, thus initializing the field
11029: @code{circle-radius} with 50. We can draw this new circle at (100,100)
11030: with:
11031:
11032: @example
11033: 100 100 my-circle draw
11034: @end example
11035:
11036: @cindex selector invocation, restrictions
11037: @cindex class definition, restrictions
11038: Note: You can only invoke a selector if the receiving object belongs to
11039: the class where the selector was defined or one of its descendents;
11040: e.g., you can invoke @code{draw} only for objects belonging to
11041: @code{graphical} or its descendents (e.g., @code{circle}). The scoping
11042: mechanism will check if you try to invoke a selector that is not
11043: defined in this class hierarchy, so you'll get an error at compilation
11044: time.
11045:
11046:
11047: @node The OOF base class, Class Declaration, Basic OOF Usage, OOF
11048: @subsubsection The @file{oof.fs} base class
11049: @cindex @file{oof.fs} base class
11050:
11051: When you define a class, you have to specify a parent class. So how do
11052: you start defining classes? There is one class available from the start:
11053: @code{object}. You have to use it as ancestor for all classes. It is the
11054: only class that has no parent. Classes are also objects, except that
11055: they don't have instance variables; class manipulation such as
11056: inheritance or changing definitions of a class is handled through
11057: selectors of the class @code{object}.
11058:
11059: @code{object} provides a number of selectors:
11060:
11061: @itemize @bullet
11062: @item
11063: @code{class} for subclassing, @code{definitions} to add definitions
11064: later on, and @code{class?} to get type informations (is the class a
11065: subclass of the class passed on the stack?).
11066:
11067: doc---object-class
11068: doc---object-definitions
11069: doc---object-class?
11070:
11071:
11072: @item
11073: @code{init} and @code{dispose} as constructor and destructor of the
11074: object. @code{init} is invocated after the object's memory is allocated,
11075: while @code{dispose} also handles deallocation. Thus if you redefine
11076: @code{dispose}, you have to call the parent's dispose with @code{super
11077: dispose}, too.
11078:
11079: doc---object-init
11080: doc---object-dispose
11081:
11082:
11083: @item
11084: @code{new}, @code{new[]}, @code{:}, @code{ptr}, @code{asptr}, and
11085: @code{[]} to create named and unnamed objects and object arrays or
11086: object pointers.
11087:
11088: doc---object-new
11089: doc---object-new[]
11090: doc---object-:
11091: doc---object-ptr
11092: doc---object-asptr
11093: doc---object-[]
11094:
11095:
11096: @item
11097: @code{::} and @code{super} for explicit scoping. You should use explicit
11098: scoping only for super classes or classes with the same set of instance
11099: variables. Explicitly-scoped selectors use early binding.
11100:
11101: doc---object-::
11102: doc---object-super
11103:
11104:
11105: @item
11106: @code{self} to get the address of the object
11107:
11108: doc---object-self
11109:
11110:
11111: @item
11112: @code{bind}, @code{bound}, @code{link}, and @code{is} to assign object
11113: pointers and instance defers.
11114:
11115: doc---object-bind
11116: doc---object-bound
11117: doc---object-link
11118: doc---object-is
11119:
11120:
11121: @item
11122: @code{'} to obtain selector tokens, @code{send} to invocate selectors
11123: form the stack, and @code{postpone} to generate selector invocation code.
11124:
11125: doc---object-'
11126: doc---object-postpone
11127:
11128:
11129: @item
11130: @code{with} and @code{endwith} to select the active object from the
11131: stack, and enable its scope. Using @code{with} and @code{endwith}
11132: also allows you to create code using selector @code{postpone} without being
11133: trapped by the state-smart objects.
11134:
11135: doc---object-with
11136: doc---object-endwith
11137:
11138:
11139: @end itemize
11140:
11141: @node Class Declaration, Class Implementation, The OOF base class, OOF
11142: @subsubsection Class Declaration
11143: @cindex class declaration
11144:
11145: @itemize @bullet
11146: @item
11147: Instance variables
11148:
11149: doc---oof-var
11150:
11151:
11152: @item
11153: Object pointers
11154:
11155: doc---oof-ptr
11156: doc---oof-asptr
11157:
11158:
11159: @item
11160: Instance defers
11161:
11162: doc---oof-defer
11163:
11164:
11165: @item
11166: Method selectors
11167:
11168: doc---oof-early
11169: doc---oof-method
11170:
11171:
11172: @item
11173: Class-wide variables
11174:
11175: doc---oof-static
11176:
11177:
11178: @item
11179: End declaration
11180:
11181: doc---oof-how:
11182: doc---oof-class;
11183:
11184:
11185: @end itemize
11186:
11187: @c -------------------------------------------------------------
11188: @node Class Implementation, , Class Declaration, OOF
11189: @subsubsection Class Implementation
11190: @cindex class implementation
11191:
11192: @c -------------------------------------------------------------
11193: @node Mini-OOF, Comparison with other object models, OOF, Object-oriented Forth
11194: @subsection The @file{mini-oof.fs} model
11195: @cindex mini-oof
11196:
11197: Gforth's third object oriented Forth package is a 12-liner. It uses a
11198: mixture of the @file{object.fs} and the @file{oof.fs} syntax,
11199: and reduces to the bare minimum of features. This is based on a posting
11200: of Bernd Paysan in comp.lang.forth.
11201:
11202: @menu
11203: * Basic Mini-OOF Usage::
11204: * Mini-OOF Example::
11205: * Mini-OOF Implementation::
11206: @end menu
11207:
11208: @c -------------------------------------------------------------
11209: @node Basic Mini-OOF Usage, Mini-OOF Example, Mini-OOF, Mini-OOF
11210: @subsubsection Basic @file{mini-oof.fs} Usage
11211: @cindex mini-oof usage
11212:
11213: There is a base class (@code{class}, which allocates one cell for the
11214: object pointer) plus seven other words: to define a method, a variable,
11215: a class; to end a class, to resolve binding, to allocate an object and
11216: to compile a class method.
11217: @comment TODO better description of the last one
11218:
11219:
11220: doc-object
11221: doc-method
11222: doc-var
11223: doc-class
11224: doc-end-class
11225: doc-defines
11226: doc-new
11227: doc-::
11228:
11229:
11230:
11231: @c -------------------------------------------------------------
11232: @node Mini-OOF Example, Mini-OOF Implementation, Basic Mini-OOF Usage, Mini-OOF
11233: @subsubsection Mini-OOF Example
11234: @cindex mini-oof example
11235:
11236: A short example shows how to use this package. This example, in slightly
11237: extended form, is supplied as @file{moof-exm.fs}
11238: @comment TODO could flesh this out with some comments from the Forthwrite article
11239:
11240: @example
11241: object class
11242: method init
11243: method draw
11244: end-class graphical
11245: @end example
11246:
11247: This code defines a class @code{graphical} with an
11248: operation @code{draw}. We can perform the operation
11249: @code{draw} on any @code{graphical} object, e.g.:
11250:
11251: @example
11252: 100 100 t-rex draw
11253: @end example
11254:
11255: where @code{t-rex} is an object or object pointer, created with e.g.
11256: @code{graphical new Constant t-rex}.
11257:
11258: For concrete graphical objects, we define child classes of the
11259: class @code{graphical}, e.g.:
11260:
11261: @example
11262: graphical class
11263: cell var circle-radius
11264: end-class circle \ "graphical" is the parent class
11265:
11266: :noname ( x y -- )
11267: circle-radius @@ draw-circle ; circle defines draw
11268: :noname ( r -- )
11269: circle-radius ! ; circle defines init
11270: @end example
11271:
11272: There is no implicit init method, so we have to define one. The creation
11273: code of the object now has to call init explicitely.
11274:
11275: @example
11276: circle new Constant my-circle
11277: 50 my-circle init
11278: @end example
11279:
11280: It is also possible to add a function to create named objects with
11281: automatic call of @code{init}, given that all objects have @code{init}
11282: on the same place:
11283:
11284: @example
11285: : new: ( .. o "name" -- )
11286: new dup Constant init ;
11287: 80 circle new: large-circle
11288: @end example
11289:
11290: We can draw this new circle at (100,100) with:
11291:
11292: @example
11293: 100 100 my-circle draw
11294: @end example
11295:
11296: @node Mini-OOF Implementation, , Mini-OOF Example, Mini-OOF
11297: @subsubsection @file{mini-oof.fs} Implementation
11298:
11299: Object-oriented systems with late binding typically use a
11300: ``vtable''-approach: the first variable in each object is a pointer to a
11301: table, which contains the methods as function pointers. The vtable
11302: may also contain other information.
11303:
11304: So first, let's declare methods:
11305:
11306: @example
11307: : method ( m v -- m' v ) Create over , swap cell+ swap
11308: DOES> ( ... o -- ... ) @ over @ + @ execute ;
11309: @end example
11310:
11311: During method declaration, the number of methods and instance
11312: variables is on the stack (in address units). @code{method} creates
11313: one method and increments the method number. To execute a method, it
11314: takes the object, fetches the vtable pointer, adds the offset, and
11315: executes the @i{xt} stored there. Each method takes the object it is
11316: invoked from as top of stack parameter. The method itself should
11317: consume that object.
11318:
11319: Now, we also have to declare instance variables
11320:
11321: @example
11322: : var ( m v size -- m v' ) Create over , +
11323: DOES> ( o -- addr ) @ + ;
11324: @end example
11325:
11326: As before, a word is created with the current offset. Instance
11327: variables can have different sizes (cells, floats, doubles, chars), so
11328: all we do is take the size and add it to the offset. If your machine
11329: has alignment restrictions, put the proper @code{aligned} or
11330: @code{faligned} before the variable, to adjust the variable
11331: offset. That's why it is on the top of stack.
11332:
11333: We need a starting point (the base object) and some syntactic sugar:
11334:
11335: @example
11336: Create object 1 cells , 2 cells ,
11337: : class ( class -- class methods vars ) dup 2@ ;
11338: @end example
11339:
11340: For inheritance, the vtable of the parent object has to be
11341: copied when a new, derived class is declared. This gives all the
11342: methods of the parent class, which can be overridden, though.
11343:
11344: @example
11345: : end-class ( class methods vars -- )
11346: Create here >r , dup , 2 cells ?DO ['] noop , 1 cells +LOOP
11347: cell+ dup cell+ r> rot @ 2 cells /string move ;
11348: @end example
11349:
11350: The first line creates the vtable, initialized with
11351: @code{noop}s. The second line is the inheritance mechanism, it
11352: copies the xts from the parent vtable.
11353:
11354: We still have no way to define new methods, let's do that now:
11355:
11356: @example
11357: : defines ( xt class -- ) ' >body @ + ! ;
11358: @end example
11359:
11360: To allocate a new object, we need a word, too:
11361:
11362: @example
11363: : new ( class -- o ) here over @ allot swap over ! ;
11364: @end example
11365:
11366: Sometimes derived classes want to access the method of the
11367: parent object. There are two ways to achieve this with Mini-OOF:
11368: first, you could use named words, and second, you could look up the
11369: vtable of the parent object.
11370:
11371: @example
11372: : :: ( class "name" -- ) ' >body @ + @ compile, ;
11373: @end example
11374:
11375:
11376: Nothing can be more confusing than a good example, so here is
11377: one. First let's declare a text object (called
11378: @code{button}), that stores text and position:
11379:
11380: @example
11381: object class
11382: cell var text
11383: cell var len
11384: cell var x
11385: cell var y
11386: method init
11387: method draw
11388: end-class button
11389: @end example
11390:
11391: @noindent
11392: Now, implement the two methods, @code{draw} and @code{init}:
11393:
11394: @example
11395: :noname ( o -- )
11396: >r r@ x @ r@ y @ at-xy r@ text @ r> len @ type ;
11397: button defines draw
11398: :noname ( addr u o -- )
11399: >r 0 r@ x ! 0 r@ y ! r@ len ! r> text ! ;
11400: button defines init
11401: @end example
11402:
11403: @noindent
11404: To demonstrate inheritance, we define a class @code{bold-button}, with no
11405: new data and no new methods:
11406:
11407: @example
11408: button class
11409: end-class bold-button
11410:
11411: : bold 27 emit ." [1m" ;
11412: : normal 27 emit ." [0m" ;
11413: @end example
11414:
11415: @noindent
11416: The class @code{bold-button} has a different draw method to
11417: @code{button}, but the new method is defined in terms of the draw method
11418: for @code{button}:
11419:
11420: @example
11421: :noname bold [ button :: draw ] normal ; bold-button defines draw
11422: @end example
11423:
11424: @noindent
11425: Finally, create two objects and apply methods:
11426:
11427: @example
11428: button new Constant foo
11429: s" thin foo" foo init
11430: page
11431: foo draw
11432: bold-button new Constant bar
11433: s" fat bar" bar init
11434: 1 bar y !
11435: bar draw
11436: @end example
11437:
11438:
11439: @node Comparison with other object models, , Mini-OOF, Object-oriented Forth
11440: @subsection Comparison with other object models
11441: @cindex comparison of object models
11442: @cindex object models, comparison
11443:
11444: Many object-oriented Forth extensions have been proposed (@cite{A survey
11445: of object-oriented Forths} (SIGPLAN Notices, April 1996) by Bradford
11446: J. Rodriguez and W. F. S. Poehlman lists 17). This section discusses the
11447: relation of the object models described here to two well-known and two
11448: closely-related (by the use of method maps) models.
11449:
11450: @cindex Neon model
11451: The most popular model currently seems to be the Neon model (see
11452: @cite{Object-oriented programming in ANS Forth} (Forth Dimensions, March
11453: 1997) by Andrew McKewan) but this model has a number of limitations
11454: @footnote{A longer version of this critique can be
11455: found in @cite{On Standardizing Object-Oriented Forth Extensions} (Forth
11456: Dimensions, May 1997) by Anton Ertl.}:
11457:
11458: @itemize @bullet
11459: @item
11460: It uses a @code{@emph{selector object}} syntax, which makes it unnatural
11461: to pass objects on the stack.
11462:
11463: @item
11464: It requires that the selector parses the input stream (at
11465: compile time); this leads to reduced extensibility and to bugs that are+
11466: hard to find.
11467:
11468: @item
11469: It allows using every selector to every object;
11470: this eliminates the need for classes, but makes it harder to create
11471: efficient implementations.
11472: @end itemize
11473:
11474: @cindex Pountain's object-oriented model
11475: Another well-known publication is @cite{Object-Oriented Forth} (Academic
11476: Press, London, 1987) by Dick Pountain. However, it is not really about
11477: object-oriented programming, because it hardly deals with late
11478: binding. Instead, it focuses on features like information hiding and
11479: overloading that are characteristic of modular languages like Ada (83).
11480:
11481: @cindex Zsoter's object-oriented model
11482: In @cite{Does late binding have to be slow?} (Forth Dimensions 18(1)
11483: 1996, pages 31-35) Andras Zsoter describes a model that makes heavy use
11484: of an active object (like @code{this} in @file{objects.fs}): The active
11485: object is not only used for accessing all fields, but also specifies the
11486: receiving object of every selector invocation; you have to change the
11487: active object explicitly with @code{@{ ... @}}, whereas in
11488: @file{objects.fs} it changes more or less implicitly at @code{m:
11489: ... ;m}. Such a change at the method entry point is unnecessary with the
11490: Zsoter's model, because the receiving object is the active object
11491: already. On the other hand, the explicit change is absolutely necessary
11492: in that model, because otherwise no one could ever change the active
11493: object. An ANS Forth implementation of this model is available at
11494: @uref{http://www.forth.org/fig/oopf.html}.
11495:
11496: @cindex @file{oof.fs}, differences to other models
11497: The @file{oof.fs} model combines information hiding and overloading
11498: resolution (by keeping names in various word lists) with object-oriented
11499: programming. It sets the active object implicitly on method entry, but
11500: also allows explicit changing (with @code{>o...o>} or with
11501: @code{with...endwith}). It uses parsing and state-smart objects and
11502: classes for resolving overloading and for early binding: the object or
11503: class parses the selector and determines the method from this. If the
11504: selector is not parsed by an object or class, it performs a call to the
11505: selector for the active object (late binding), like Zsoter's model.
11506: Fields are always accessed through the active object. The big
11507: disadvantage of this model is the parsing and the state-smartness, which
11508: reduces extensibility and increases the opportunities for subtle bugs;
11509: essentially, you are only safe if you never tick or @code{postpone} an
11510: object or class (Bernd disagrees, but I (Anton) am not convinced).
11511:
11512: @cindex @file{mini-oof.fs}, differences to other models
11513: The @file{mini-oof.fs} model is quite similar to a very stripped-down
11514: version of the @file{objects.fs} model, but syntactically it is a
11515: mixture of the @file{objects.fs} and @file{oof.fs} models.
11516:
11517: @c -------------------------------------------------------------
11518: @node Passing Commands to the OS, Keeping track of Time, Object-oriented Forth, Words
11519: @section Passing Commands to the Operating System
11520: @cindex operating system - passing commands
11521: @cindex shell commands
11522:
11523: Gforth allows you to pass an arbitrary string to the host operating
11524: system shell (if such a thing exists) for execution.
11525:
11526:
11527: doc-sh
11528: doc-system
11529: doc-$?
11530: doc-getenv
11531:
11532:
11533: @c -------------------------------------------------------------
11534: @node Keeping track of Time, Miscellaneous Words, Passing Commands to the OS, Words
11535: @section Keeping track of Time
11536: @cindex time-related words
11537:
11538: Gforth implements time-related operations by making calls to the C
11539: library function, @code{gettimeofday}.
11540:
11541: doc-ms
11542: doc-time&date
11543:
11544:
11545:
11546: @c -------------------------------------------------------------
11547: @node Miscellaneous Words, , Keeping track of Time, Words
11548: @section Miscellaneous Words
11549: @cindex miscellaneous words
11550:
11551: @comment TODO find homes for these
11552:
11553: These section lists the ANS Forth words that are not documented
11554: elsewhere in this manual. Ultimately, they all need proper homes.
11555:
11556: doc-[compile]
11557: doc-quit
11558:
11559: The following ANS Forth words are not currently supported by Gforth
11560: (@pxref{ANS conformance}):
11561:
11562: @code{EDITOR}
11563: @code{EMIT?}
11564: @code{FORGET}
11565:
11566: @c ******************************************************************
11567: @node Error messages, Tools, Words, Top
11568: @chapter Error messages
11569: @cindex error messages
11570: @cindex backtrace
11571:
11572: A typical Gforth error message looks like this:
11573:
11574: @example
11575: in file included from :-1
11576: in file included from ./yyy.fs:1
11577: ./xxx.fs:4: Invalid memory address
11578: bar
11579: ^^^
11580: $400E664C @@
11581: $400E6664 foo
11582: @end example
11583:
11584: The message identifying the error is @code{Invalid memory address}. The
11585: error happened when text-interpreting line 4 of the file
11586: @file{./xxx.fs}. This line is given (it contains @code{bar}), and the
11587: word on the line where the error happened, is pointed out (with
11588: @code{^^^}).
11589:
11590: The file containing the error was included in line 1 of @file{./yyy.fs},
11591: and @file{yyy.fs} was included from a non-file (in this case, by giving
11592: @file{yyy.fs} as command-line parameter to Gforth).
11593:
11594: At the end of the error message you find a return stack dump that can be
11595: interpreted as a backtrace (possibly empty). On top you find the top of
11596: the return stack when the @code{throw} happened, and at the bottom you
11597: find the return stack entry just above the return stack of the topmost
11598: text interpreter.
11599:
11600: To the right of most return stack entries you see a guess for the word
11601: that pushed that return stack entry as its return address. This gives a
11602: backtrace. In our case we see that @code{bar} called @code{foo}, and
11603: @code{foo} called @code{@@} (and @code{@@} had an @emph{Invalid memory
11604: address} exception).
11605:
11606: Note that the backtrace is not perfect: We don't know which return stack
11607: entries are return addresses (so we may get false positives); and in
11608: some cases (e.g., for @code{abort"}) we cannot determine from the return
11609: address the word that pushed the return address, so for some return
11610: addresses you see no names in the return stack dump.
11611:
11612: @cindex @code{catch} and backtraces
11613: The return stack dump represents the return stack at the time when a
11614: specific @code{throw} was executed. In programs that make use of
11615: @code{catch}, it is not necessarily clear which @code{throw} should be
11616: used for the return stack dump (e.g., consider one @code{throw} that
11617: indicates an error, which is caught, and during recovery another error
11618: happens; which @code{throw} should be used for the stack dump?). Gforth
11619: presents the return stack dump for the first @code{throw} after the last
11620: executed (not returned-to) @code{catch}; this works well in the usual
11621: case.
11622:
11623: @cindex @code{gforth-fast} and backtraces
11624: @cindex @code{gforth-fast}, difference from @code{gforth}
11625: @cindex backtraces with @code{gforth-fast}
11626: @cindex return stack dump with @code{gforth-fast}
11627: @code{gforth} is able to do a return stack dump for throws generated
11628: from primitives (e.g., invalid memory address, stack empty etc.);
11629: @code{gforth-fast} is only able to do a return stack dump from a
11630: directly called @code{throw} (including @code{abort} etc.). This is the
11631: only difference (apart from a speed factor of between 1.15 (K6-2) and
11632: 1.6 (21164A)) between @code{gforth} and @code{gforth-fast}. Given an
11633: exception caused by a primitive in @code{gforth-fast}, you will
11634: typically see no return stack dump at all; however, if the exception is
11635: caught by @code{catch} (e.g., for restoring some state), and then
11636: @code{throw}n again, the return stack dump will be for the first such
11637: @code{throw}.
11638:
11639: @c ******************************************************************
11640: @node Tools, ANS conformance, Error messages, Top
11641: @chapter Tools
11642:
11643: @menu
11644: * ANS Report:: Report the words used, sorted by wordset.
11645: @end menu
11646:
11647: See also @ref{Emacs and Gforth}.
11648:
11649: @node ANS Report, , Tools, Tools
11650: @section @file{ans-report.fs}: Report the words used, sorted by wordset
11651: @cindex @file{ans-report.fs}
11652: @cindex report the words used in your program
11653: @cindex words used in your program
11654:
11655: If you want to label a Forth program as ANS Forth Program, you must
11656: document which wordsets the program uses; for extension wordsets, it is
11657: helpful to list the words the program requires from these wordsets
11658: (because Forth systems are allowed to provide only some words of them).
11659:
11660: The @file{ans-report.fs} tool makes it easy for you to determine which
11661: words from which wordset and which non-ANS words your application
11662: uses. You simply have to include @file{ans-report.fs} before loading the
11663: program you want to check. After loading your program, you can get the
11664: report with @code{print-ans-report}. A typical use is to run this as
11665: batch job like this:
11666: @example
11667: gforth ans-report.fs myprog.fs -e "print-ans-report bye"
11668: @end example
11669:
11670: The output looks like this (for @file{compat/control.fs}):
11671: @example
11672: The program uses the following words
11673: from CORE :
11674: : POSTPONE THEN ; immediate ?dup IF 0=
11675: from BLOCK-EXT :
11676: \
11677: from FILE :
11678: (
11679: @end example
11680:
11681: @subsection Caveats
11682:
11683: Note that @file{ans-report.fs} just checks which words are used, not whether
11684: they are used in an ANS Forth conforming way!
11685:
11686: Some words are defined in several wordsets in the
11687: standard. @file{ans-report.fs} reports them for only one of the
11688: wordsets, and not necessarily the one you expect. It depends on usage
11689: which wordset is the right one to specify. E.g., if you only use the
11690: compilation semantics of @code{S"}, it is a Core word; if you also use
11691: its interpretation semantics, it is a File word.
11692:
11693: @c ******************************************************************
11694: @node ANS conformance, Standard vs Extensions, Tools, Top
11695: @chapter ANS conformance
11696: @cindex ANS conformance of Gforth
11697:
11698: To the best of our knowledge, Gforth is an
11699:
11700: ANS Forth System
11701: @itemize @bullet
11702: @item providing the Core Extensions word set
11703: @item providing the Block word set
11704: @item providing the Block Extensions word set
11705: @item providing the Double-Number word set
11706: @item providing the Double-Number Extensions word set
11707: @item providing the Exception word set
11708: @item providing the Exception Extensions word set
11709: @item providing the Facility word set
11710: @item providing @code{EKEY}, @code{EKEY>CHAR}, @code{EKEY?}, @code{MS} and @code{TIME&DATE} from the Facility Extensions word set
11711: @item providing the File Access word set
11712: @item providing the File Access Extensions word set
11713: @item providing the Floating-Point word set
11714: @item providing the Floating-Point Extensions word set
11715: @item providing the Locals word set
11716: @item providing the Locals Extensions word set
11717: @item providing the Memory-Allocation word set
11718: @item providing the Memory-Allocation Extensions word set (that one's easy)
11719: @item providing the Programming-Tools word set
11720: @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
11721: @item providing the Search-Order word set
11722: @item providing the Search-Order Extensions word set
11723: @item providing the String word set
11724: @item providing the String Extensions word set (another easy one)
11725: @end itemize
11726:
11727: @cindex system documentation
11728: In addition, ANS Forth systems are required to document certain
11729: implementation choices. This chapter tries to meet these
11730: requirements. In many cases it gives a way to ask the system for the
11731: information instead of providing the information directly, in
11732: particular, if the information depends on the processor, the operating
11733: system or the installation options chosen, or if they are likely to
11734: change during the maintenance of Gforth.
11735:
11736: @comment The framework for the rest has been taken from pfe.
11737:
11738: @menu
11739: * The Core Words::
11740: * The optional Block word set::
11741: * The optional Double Number word set::
11742: * The optional Exception word set::
11743: * The optional Facility word set::
11744: * The optional File-Access word set::
11745: * The optional Floating-Point word set::
11746: * The optional Locals word set::
11747: * The optional Memory-Allocation word set::
11748: * The optional Programming-Tools word set::
11749: * The optional Search-Order word set::
11750: @end menu
11751:
11752:
11753: @c =====================================================================
11754: @node The Core Words, The optional Block word set, ANS conformance, ANS conformance
11755: @comment node-name, next, previous, up
11756: @section The Core Words
11757: @c =====================================================================
11758: @cindex core words, system documentation
11759: @cindex system documentation, core words
11760:
11761: @menu
11762: * core-idef:: Implementation Defined Options
11763: * core-ambcond:: Ambiguous Conditions
11764: * core-other:: Other System Documentation
11765: @end menu
11766:
11767: @c ---------------------------------------------------------------------
11768: @node core-idef, core-ambcond, The Core Words, The Core Words
11769: @subsection Implementation Defined Options
11770: @c ---------------------------------------------------------------------
11771: @cindex core words, implementation-defined options
11772: @cindex implementation-defined options, core words
11773:
11774:
11775: @table @i
11776: @item (Cell) aligned addresses:
11777: @cindex cell-aligned addresses
11778: @cindex aligned addresses
11779: processor-dependent. Gforth's alignment words perform natural alignment
11780: (e.g., an address aligned for a datum of size 8 is divisible by
11781: 8). Unaligned accesses usually result in a @code{-23 THROW}.
11782:
11783: @item @code{EMIT} and non-graphic characters:
11784: @cindex @code{EMIT} and non-graphic characters
11785: @cindex non-graphic characters and @code{EMIT}
11786: The character is output using the C library function (actually, macro)
11787: @code{putc}.
11788:
11789: @item character editing of @code{ACCEPT} and @code{EXPECT}:
11790: @cindex character editing of @code{ACCEPT} and @code{EXPECT}
11791: @cindex editing in @code{ACCEPT} and @code{EXPECT}
11792: @cindex @code{ACCEPT}, editing
11793: @cindex @code{EXPECT}, editing
11794: This is modeled on the GNU readline library (@pxref{Readline
11795: Interaction, , Command Line Editing, readline, The GNU Readline
11796: Library}) with Emacs-like key bindings. @kbd{Tab} deviates a little by
11797: producing a full word completion every time you type it (instead of
11798: producing the common prefix of all completions). @xref{Command-line editing}.
11799:
11800: @item character set:
11801: @cindex character set
11802: The character set of your computer and display device. Gforth is
11803: 8-bit-clean (but some other component in your system may make trouble).
11804:
11805: @item Character-aligned address requirements:
11806: @cindex character-aligned address requirements
11807: installation-dependent. Currently a character is represented by a C
11808: @code{unsigned char}; in the future we might switch to @code{wchar_t}
11809: (Comments on that requested).
11810:
11811: @item character-set extensions and matching of names:
11812: @cindex character-set extensions and matching of names
11813: @cindex case-sensitivity for name lookup
11814: @cindex name lookup, case-sensitivity
11815: @cindex locale and case-sensitivity
11816: Any character except the ASCII NUL character can be used in a
11817: name. Matching is case-insensitive (except in @code{TABLE}s). The
11818: matching is performed using the C library function @code{strncasecmp}, whose
11819: function is probably influenced by the locale. E.g., the @code{C} locale
11820: does not know about accents and umlauts, so they are matched
11821: case-sensitively in that locale. For portability reasons it is best to
11822: write programs such that they work in the @code{C} locale. Then one can
11823: use libraries written by a Polish programmer (who might use words
11824: containing ISO Latin-2 encoded characters) and by a French programmer
11825: (ISO Latin-1) in the same program (of course, @code{WORDS} will produce
11826: funny results for some of the words (which ones, depends on the font you
11827: are using)). Also, the locale you prefer may not be available in other
11828: operating systems. Hopefully, Unicode will solve these problems one day.
11829:
11830: @item conditions under which control characters match a space delimiter:
11831: @cindex space delimiters
11832: @cindex control characters as delimiters
11833: If @code{WORD} is called with the space character as a delimiter, all
11834: white-space characters (as identified by the C macro @code{isspace()})
11835: are delimiters. @code{PARSE}, on the other hand, treats space like other
11836: delimiters. @code{SWORD} treats space like @code{WORD}, but behaves
11837: like @code{PARSE} otherwise. @code{(NAME)}, which is used by the outer
11838: interpreter (aka text interpreter) by default, treats all white-space
11839: characters as delimiters.
11840:
11841: @item format of the control-flow stack:
11842: @cindex control-flow stack, format
11843: The data stack is used as control-flow stack. The size of a control-flow
11844: stack item in cells is given by the constant @code{cs-item-size}. At the
11845: time of this writing, an item consists of a (pointer to a) locals list
11846: (third), an address in the code (second), and a tag for identifying the
11847: item (TOS). The following tags are used: @code{defstart},
11848: @code{live-orig}, @code{dead-orig}, @code{dest}, @code{do-dest},
11849: @code{scopestart}.
11850:
11851: @item conversion of digits > 35
11852: @cindex digits > 35
11853: The characters @code{[\]^_'} are the digits with the decimal value
11854: 36@minus{}41. There is no way to input many of the larger digits.
11855:
11856: @item display after input terminates in @code{ACCEPT} and @code{EXPECT}:
11857: @cindex @code{EXPECT}, display after end of input
11858: @cindex @code{ACCEPT}, display after end of input
11859: The cursor is moved to the end of the entered string. If the input is
11860: terminated using the @kbd{Return} key, a space is typed.
11861:
11862: @item exception abort sequence of @code{ABORT"}:
11863: @cindex exception abort sequence of @code{ABORT"}
11864: @cindex @code{ABORT"}, exception abort sequence
11865: The error string is stored into the variable @code{"error} and a
11866: @code{-2 throw} is performed.
11867:
11868: @item input line terminator:
11869: @cindex input line terminator
11870: @cindex line terminator on input
11871: @cindex newline character on input
11872: For interactive input, @kbd{C-m} (CR) and @kbd{C-j} (LF) terminate
11873: lines. One of these characters is typically produced when you type the
11874: @kbd{Enter} or @kbd{Return} key.
11875:
11876: @item maximum size of a counted string:
11877: @cindex maximum size of a counted string
11878: @cindex counted string, maximum size
11879: @code{s" /counted-string" environment? drop .}. Currently 255 characters
11880: on all ports, but this may change.
11881:
11882: @item maximum size of a parsed string:
11883: @cindex maximum size of a parsed string
11884: @cindex parsed string, maximum size
11885: Given by the constant @code{/line}. Currently 255 characters.
11886:
11887: @item maximum size of a definition name, in characters:
11888: @cindex maximum size of a definition name, in characters
11889: @cindex name, maximum length
11890: 31
11891:
11892: @item maximum string length for @code{ENVIRONMENT?}, in characters:
11893: @cindex maximum string length for @code{ENVIRONMENT?}, in characters
11894: @cindex @code{ENVIRONMENT?} string length, maximum
11895: 31
11896:
11897: @item method of selecting the user input device:
11898: @cindex user input device, method of selecting
11899: The user input device is the standard input. There is currently no way to
11900: change it from within Gforth. However, the input can typically be
11901: redirected in the command line that starts Gforth.
11902:
11903: @item method of selecting the user output device:
11904: @cindex user output device, method of selecting
11905: @code{EMIT} and @code{TYPE} output to the file-id stored in the value
11906: @code{outfile-id} (@code{stdout} by default). Gforth uses unbuffered
11907: output when the user output device is a terminal, otherwise the output
11908: is buffered.
11909:
11910: @item methods of dictionary compilation:
11911: What are we expected to document here?
11912:
11913: @item number of bits in one address unit:
11914: @cindex number of bits in one address unit
11915: @cindex address unit, size in bits
11916: @code{s" address-units-bits" environment? drop .}. 8 in all current
11917: ports.
11918:
11919: @item number representation and arithmetic:
11920: @cindex number representation and arithmetic
11921: Processor-dependent. Binary two's complement on all current ports.
11922:
11923: @item ranges for integer types:
11924: @cindex ranges for integer types
11925: @cindex integer types, ranges
11926: Installation-dependent. Make environmental queries for @code{MAX-N},
11927: @code{MAX-U}, @code{MAX-D} and @code{MAX-UD}. The lower bounds for
11928: unsigned (and positive) types is 0. The lower bound for signed types on
11929: two's complement and one's complement machines machines can be computed
11930: by adding 1 to the upper bound.
11931:
11932: @item read-only data space regions:
11933: @cindex read-only data space regions
11934: @cindex data-space, read-only regions
11935: The whole Forth data space is writable.
11936:
11937: @item size of buffer at @code{WORD}:
11938: @cindex size of buffer at @code{WORD}
11939: @cindex @code{WORD} buffer size
11940: @code{PAD HERE - .}. 104 characters on 32-bit machines. The buffer is
11941: shared with the pictured numeric output string. If overwriting
11942: @code{PAD} is acceptable, it is as large as the remaining dictionary
11943: space, although only as much can be sensibly used as fits in a counted
11944: string.
11945:
11946: @item size of one cell in address units:
11947: @cindex cell size
11948: @code{1 cells .}.
11949:
11950: @item size of one character in address units:
11951: @cindex char size
11952: @code{1 chars .}. 1 on all current ports.
11953:
11954: @item size of the keyboard terminal buffer:
11955: @cindex size of the keyboard terminal buffer
11956: @cindex terminal buffer, size
11957: Varies. You can determine the size at a specific time using @code{lp@@
11958: tib - .}. It is shared with the locals stack and TIBs of files that
11959: include the current file. You can change the amount of space for TIBs
11960: and locals stack at Gforth startup with the command line option
11961: @code{-l}.
11962:
11963: @item size of the pictured numeric output buffer:
11964: @cindex size of the pictured numeric output buffer
11965: @cindex pictured numeric output buffer, size
11966: @code{PAD HERE - .}. 104 characters on 32-bit machines. The buffer is
11967: shared with @code{WORD}.
11968:
11969: @item size of the scratch area returned by @code{PAD}:
11970: @cindex size of the scratch area returned by @code{PAD}
11971: @cindex @code{PAD} size
11972: The remainder of dictionary space. @code{unused pad here - - .}.
11973:
11974: @item system case-sensitivity characteristics:
11975: @cindex case-sensitivity characteristics
11976: Dictionary searches are case-insensitive (except in
11977: @code{TABLE}s). However, as explained above under @i{character-set
11978: extensions}, the matching for non-ASCII characters is determined by the
11979: locale you are using. In the default @code{C} locale all non-ASCII
11980: characters are matched case-sensitively.
11981:
11982: @item system prompt:
11983: @cindex system prompt
11984: @cindex prompt
11985: @code{ ok} in interpret state, @code{ compiled} in compile state.
11986:
11987: @item division rounding:
11988: @cindex division rounding
11989: installation dependent. @code{s" floored" environment? drop .}. We leave
11990: the choice to @code{gcc} (what to use for @code{/}) and to you (whether
11991: to use @code{fm/mod}, @code{sm/rem} or simply @code{/}).
11992:
11993: @item values of @code{STATE} when true:
11994: @cindex @code{STATE} values
11995: -1.
11996:
11997: @item values returned after arithmetic overflow:
11998: On two's complement machines, arithmetic is performed modulo
11999: 2**bits-per-cell for single arithmetic and 4**bits-per-cell for double
12000: arithmetic (with appropriate mapping for signed types). Division by zero
12001: typically results in a @code{-55 throw} (Floating-point unidentified
12002: fault), although a @code{-10 throw} (divide by zero) would be more
12003: appropriate.
12004:
12005: @item whether the current definition can be found after @t{DOES>}:
12006: @cindex @t{DOES>}, visibility of current definition
12007: No.
12008:
12009: @end table
12010:
12011: @c ---------------------------------------------------------------------
12012: @node core-ambcond, core-other, core-idef, The Core Words
12013: @subsection Ambiguous conditions
12014: @c ---------------------------------------------------------------------
12015: @cindex core words, ambiguous conditions
12016: @cindex ambiguous conditions, core words
12017:
12018: @table @i
12019:
12020: @item a name is neither a word nor a number:
12021: @cindex name not found
12022: @cindex undefined word
12023: @code{-13 throw} (Undefined word). Actually, @code{-13 bounce}, which
12024: preserves the data and FP stack, so you don't lose more work than
12025: necessary.
12026:
12027: @item a definition name exceeds the maximum length allowed:
12028: @cindex word name too long
12029: @code{-19 throw} (Word name too long)
12030:
12031: @item addressing a region not inside the various data spaces of the forth system:
12032: @cindex Invalid memory address
12033: The stacks, code space and header space are accessible. Machine code space is
12034: typically readable. Accessing other addresses gives results dependent on
12035: the operating system. On decent systems: @code{-9 throw} (Invalid memory
12036: address).
12037:
12038: @item argument type incompatible with parameter:
12039: @cindex argument type mismatch
12040: This is usually not caught. Some words perform checks, e.g., the control
12041: flow words, and issue a @code{ABORT"} or @code{-12 THROW} (Argument type
12042: mismatch).
12043:
12044: @item attempting to obtain the execution token of a word with undefined execution semantics:
12045: @cindex Interpreting a compile-only word, for @code{'} etc.
12046: @cindex execution token of words with undefined execution semantics
12047: @code{-14 throw} (Interpreting a compile-only word). In some cases, you
12048: get an execution token for @code{compile-only-error} (which performs a
12049: @code{-14 throw} when executed).
12050:
12051: @item dividing by zero:
12052: @cindex dividing by zero
12053: @cindex floating point unidentified fault, integer division
12054: On better platforms, this produces a @code{-10 throw} (Division by
12055: zero); on other systems, this typically results in a @code{-55 throw}
12056: (Floating-point unidentified fault).
12057:
12058: @item insufficient data stack or return stack space:
12059: @cindex insufficient data stack or return stack space
12060: @cindex stack overflow
12061: @cindex address alignment exception, stack overflow
12062: @cindex Invalid memory address, stack overflow
12063: Depending on the operating system, the installation, and the invocation
12064: of Gforth, this is either checked by the memory management hardware, or
12065: it is not checked. If it is checked, you typically get a @code{-3 throw}
12066: (Stack overflow), @code{-5 throw} (Return stack overflow), or @code{-9
12067: throw} (Invalid memory address) (depending on the platform and how you
12068: achieved the overflow) as soon as the overflow happens. If it is not
12069: checked, overflows typically result in mysterious illegal memory
12070: accesses, producing @code{-9 throw} (Invalid memory address) or
12071: @code{-23 throw} (Address alignment exception); they might also destroy
12072: the internal data structure of @code{ALLOCATE} and friends, resulting in
12073: various errors in these words.
12074:
12075: @item insufficient space for loop control parameters:
12076: @cindex insufficient space for loop control parameters
12077: like other return stack overflows.
12078:
12079: @item insufficient space in the dictionary:
12080: @cindex insufficient space in the dictionary
12081: @cindex dictionary overflow
12082: If you try to allot (either directly with @code{allot}, or indirectly
12083: with @code{,}, @code{create} etc.) more memory than available in the
12084: dictionary, you get a @code{-8 throw} (Dictionary overflow). If you try
12085: to access memory beyond the end of the dictionary, the results are
12086: similar to stack overflows.
12087:
12088: @item interpreting a word with undefined interpretation semantics:
12089: @cindex interpreting a word with undefined interpretation semantics
12090: @cindex Interpreting a compile-only word
12091: For some words, we have defined interpretation semantics. For the
12092: others: @code{-14 throw} (Interpreting a compile-only word).
12093:
12094: @item modifying the contents of the input buffer or a string literal:
12095: @cindex modifying the contents of the input buffer or a string literal
12096: These are located in writable memory and can be modified.
12097:
12098: @item overflow of the pictured numeric output string:
12099: @cindex overflow of the pictured numeric output string
12100: @cindex pictured numeric output string, overflow
12101: @code{-17 throw} (Pictured numeric ouput string overflow).
12102:
12103: @item parsed string overflow:
12104: @cindex parsed string overflow
12105: @code{PARSE} cannot overflow. @code{WORD} does not check for overflow.
12106:
12107: @item producing a result out of range:
12108: @cindex result out of range
12109: On two's complement machines, arithmetic is performed modulo
12110: 2**bits-per-cell for single arithmetic and 4**bits-per-cell for double
12111: arithmetic (with appropriate mapping for signed types). Division by zero
12112: typically results in a @code{-10 throw} (divide by zero) or @code{-55
12113: throw} (floating point unidentified fault). @code{convert} and
12114: @code{>number} currently overflow silently.
12115:
12116: @item reading from an empty data or return stack:
12117: @cindex stack empty
12118: @cindex stack underflow
12119: @cindex return stack underflow
12120: The data stack is checked by the outer (aka text) interpreter after
12121: every word executed. If it has underflowed, a @code{-4 throw} (Stack
12122: underflow) is performed. Apart from that, stacks may be checked or not,
12123: depending on operating system, installation, and invocation. If they are
12124: caught by a check, they typically result in @code{-4 throw} (Stack
12125: underflow), @code{-6 throw} (Return stack underflow) or @code{-9 throw}
12126: (Invalid memory address), depending on the platform and which stack
12127: underflows and by how much. Note that even if the system uses checking
12128: (through the MMU), your program may have to underflow by a significant
12129: number of stack items to trigger the reaction (the reason for this is
12130: that the MMU, and therefore the checking, works with a page-size
12131: granularity). If there is no checking, the symptoms resulting from an
12132: underflow are similar to those from an overflow. Unbalanced return
12133: stack errors result in a variaty of symptoms, including @code{-9 throw}
12134: (Invalid memory address) and Illegal Instruction (typically @code{-260
12135: throw}).
12136:
12137: @item unexpected end of the input buffer, resulting in an attempt to use a zero-length string as a name:
12138: @cindex unexpected end of the input buffer
12139: @cindex zero-length string as a name
12140: @cindex Attempt to use zero-length string as a name
12141: @code{Create} and its descendants perform a @code{-16 throw} (Attempt to
12142: use zero-length string as a name). Words like @code{'} probably will not
12143: find what they search. Note that it is possible to create zero-length
12144: names with @code{nextname} (should it not?).
12145:
12146: @item @code{>IN} greater than input buffer:
12147: @cindex @code{>IN} greater than input buffer
12148: The next invocation of a parsing word returns a string with length 0.
12149:
12150: @item @code{RECURSE} appears after @code{DOES>}:
12151: @cindex @code{RECURSE} appears after @code{DOES>}
12152: Compiles a recursive call to the defining word, not to the defined word.
12153:
12154: @item argument input source different than current input source for @code{RESTORE-INPUT}:
12155: @cindex argument input source different than current input source for @code{RESTORE-INPUT}
12156: @cindex argument type mismatch, @code{RESTORE-INPUT}
12157: @cindex @code{RESTORE-INPUT}, Argument type mismatch
12158: @code{-12 THROW}. Note that, once an input file is closed (e.g., because
12159: the end of the file was reached), its source-id may be
12160: reused. Therefore, restoring an input source specification referencing a
12161: closed file may lead to unpredictable results instead of a @code{-12
12162: THROW}.
12163:
12164: In the future, Gforth may be able to restore input source specifications
12165: from other than the current input source.
12166:
12167: @item data space containing definitions gets de-allocated:
12168: @cindex data space containing definitions gets de-allocated
12169: Deallocation with @code{allot} is not checked. This typically results in
12170: memory access faults or execution of illegal instructions.
12171:
12172: @item data space read/write with incorrect alignment:
12173: @cindex data space read/write with incorrect alignment
12174: @cindex alignment faults
12175: @cindex address alignment exception
12176: Processor-dependent. Typically results in a @code{-23 throw} (Address
12177: alignment exception). Under Linux-Intel on a 486 or later processor with
12178: alignment turned on, incorrect alignment results in a @code{-9 throw}
12179: (Invalid memory address). There are reportedly some processors with
12180: alignment restrictions that do not report violations.
12181:
12182: @item data space pointer not properly aligned, @code{,}, @code{C,}:
12183: @cindex data space pointer not properly aligned, @code{,}, @code{C,}
12184: Like other alignment errors.
12185:
12186: @item less than u+2 stack items (@code{PICK} and @code{ROLL}):
12187: Like other stack underflows.
12188:
12189: @item loop control parameters not available:
12190: @cindex loop control parameters not available
12191: Not checked. The counted loop words simply assume that the top of return
12192: stack items are loop control parameters and behave accordingly.
12193:
12194: @item most recent definition does not have a name (@code{IMMEDIATE}):
12195: @cindex most recent definition does not have a name (@code{IMMEDIATE})
12196: @cindex last word was headerless
12197: @code{abort" last word was headerless"}.
12198:
12199: @item name not defined by @code{VALUE} used by @code{TO}:
12200: @cindex name not defined by @code{VALUE} used by @code{TO}
12201: @cindex @code{TO} on non-@code{VALUE}s
12202: @cindex Invalid name argument, @code{TO}
12203: @code{-32 throw} (Invalid name argument) (unless name is a local or was
12204: defined by @code{CONSTANT}; in the latter case it just changes the constant).
12205:
12206: @item name not found (@code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]}):
12207: @cindex name not found (@code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]})
12208: @cindex undefined word, @code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]}
12209: @code{-13 throw} (Undefined word)
12210:
12211: @item parameters are not of the same type (@code{DO}, @code{?DO}, @code{WITHIN}):
12212: @cindex parameters are not of the same type (@code{DO}, @code{?DO}, @code{WITHIN})
12213: Gforth behaves as if they were of the same type. I.e., you can predict
12214: the behaviour by interpreting all parameters as, e.g., signed.
12215:
12216: @item @code{POSTPONE} or @code{[COMPILE]} applied to @code{TO}:
12217: @cindex @code{POSTPONE} or @code{[COMPILE]} applied to @code{TO}
12218: Assume @code{: X POSTPONE TO ; IMMEDIATE}. @code{X} performs the
12219: compilation semantics of @code{TO}.
12220:
12221: @item String longer than a counted string returned by @code{WORD}:
12222: @cindex string longer than a counted string returned by @code{WORD}
12223: @cindex @code{WORD}, string overflow
12224: Not checked. The string will be ok, but the count will, of course,
12225: contain only the least significant bits of the length.
12226:
12227: @item u greater than or equal to the number of bits in a cell (@code{LSHIFT}, @code{RSHIFT}):
12228: @cindex @code{LSHIFT}, large shift counts
12229: @cindex @code{RSHIFT}, large shift counts
12230: Processor-dependent. Typical behaviours are returning 0 and using only
12231: the low bits of the shift count.
12232:
12233: @item word not defined via @code{CREATE}:
12234: @cindex @code{>BODY} of non-@code{CREATE}d words
12235: @code{>BODY} produces the PFA of the word no matter how it was defined.
12236:
12237: @cindex @code{DOES>} of non-@code{CREATE}d words
12238: @code{DOES>} changes the execution semantics of the last defined word no
12239: matter how it was defined. E.g., @code{CONSTANT DOES>} is equivalent to
12240: @code{CREATE , DOES>}.
12241:
12242: @item words improperly used outside @code{<#} and @code{#>}:
12243: Not checked. As usual, you can expect memory faults.
12244:
12245: @end table
12246:
12247:
12248: @c ---------------------------------------------------------------------
12249: @node core-other, , core-ambcond, The Core Words
12250: @subsection Other system documentation
12251: @c ---------------------------------------------------------------------
12252: @cindex other system documentation, core words
12253: @cindex core words, other system documentation
12254:
12255: @table @i
12256: @item nonstandard words using @code{PAD}:
12257: @cindex @code{PAD} use by nonstandard words
12258: None.
12259:
12260: @item operator's terminal facilities available:
12261: @cindex operator's terminal facilities available
12262: After processing the command line, Gforth goes into interactive mode,
12263: and you can give commands to Gforth interactively. The actual facilities
12264: available depend on how you invoke Gforth.
12265:
12266: @item program data space available:
12267: @cindex program data space available
12268: @cindex data space available
12269: @code{UNUSED .} gives the remaining dictionary space. The total
12270: dictionary space can be specified with the @code{-m} switch
12271: (@pxref{Invoking Gforth}) when Gforth starts up.
12272:
12273: @item return stack space available:
12274: @cindex return stack space available
12275: You can compute the total return stack space in cells with
12276: @code{s" RETURN-STACK-CELLS" environment? drop .}. You can specify it at
12277: startup time with the @code{-r} switch (@pxref{Invoking Gforth}).
12278:
12279: @item stack space available:
12280: @cindex stack space available
12281: You can compute the total data stack space in cells with
12282: @code{s" STACK-CELLS" environment? drop .}. You can specify it at
12283: startup time with the @code{-d} switch (@pxref{Invoking Gforth}).
12284:
12285: @item system dictionary space required, in address units:
12286: @cindex system dictionary space required, in address units
12287: Type @code{here forthstart - .} after startup. At the time of this
12288: writing, this gives 80080 (bytes) on a 32-bit system.
12289: @end table
12290:
12291:
12292: @c =====================================================================
12293: @node The optional Block word set, The optional Double Number word set, The Core Words, ANS conformance
12294: @section The optional Block word set
12295: @c =====================================================================
12296: @cindex system documentation, block words
12297: @cindex block words, system documentation
12298:
12299: @menu
12300: * block-idef:: Implementation Defined Options
12301: * block-ambcond:: Ambiguous Conditions
12302: * block-other:: Other System Documentation
12303: @end menu
12304:
12305:
12306: @c ---------------------------------------------------------------------
12307: @node block-idef, block-ambcond, The optional Block word set, The optional Block word set
12308: @subsection Implementation Defined Options
12309: @c ---------------------------------------------------------------------
12310: @cindex implementation-defined options, block words
12311: @cindex block words, implementation-defined options
12312:
12313: @table @i
12314: @item the format for display by @code{LIST}:
12315: @cindex @code{LIST} display format
12316: First the screen number is displayed, then 16 lines of 64 characters,
12317: each line preceded by the line number.
12318:
12319: @item the length of a line affected by @code{\}:
12320: @cindex length of a line affected by @code{\}
12321: @cindex @code{\}, line length in blocks
12322: 64 characters.
12323: @end table
12324:
12325:
12326: @c ---------------------------------------------------------------------
12327: @node block-ambcond, block-other, block-idef, The optional Block word set
12328: @subsection Ambiguous conditions
12329: @c ---------------------------------------------------------------------
12330: @cindex block words, ambiguous conditions
12331: @cindex ambiguous conditions, block words
12332:
12333: @table @i
12334: @item correct block read was not possible:
12335: @cindex block read not possible
12336: Typically results in a @code{throw} of some OS-derived value (between
12337: -512 and -2048). If the blocks file was just not long enough, blanks are
12338: supplied for the missing portion.
12339:
12340: @item I/O exception in block transfer:
12341: @cindex I/O exception in block transfer
12342: @cindex block transfer, I/O exception
12343: Typically results in a @code{throw} of some OS-derived value (between
12344: -512 and -2048).
12345:
12346: @item invalid block number:
12347: @cindex invalid block number
12348: @cindex block number invalid
12349: @code{-35 throw} (Invalid block number)
12350:
12351: @item a program directly alters the contents of @code{BLK}:
12352: @cindex @code{BLK}, altering @code{BLK}
12353: The input stream is switched to that other block, at the same
12354: position. If the storing to @code{BLK} happens when interpreting
12355: non-block input, the system will get quite confused when the block ends.
12356:
12357: @item no current block buffer for @code{UPDATE}:
12358: @cindex @code{UPDATE}, no current block buffer
12359: @code{UPDATE} has no effect.
12360:
12361: @end table
12362:
12363: @c ---------------------------------------------------------------------
12364: @node block-other, , block-ambcond, The optional Block word set
12365: @subsection Other system documentation
12366: @c ---------------------------------------------------------------------
12367: @cindex other system documentation, block words
12368: @cindex block words, other system documentation
12369:
12370: @table @i
12371: @item any restrictions a multiprogramming system places on the use of buffer addresses:
12372: No restrictions (yet).
12373:
12374: @item the number of blocks available for source and data:
12375: depends on your disk space.
12376:
12377: @end table
12378:
12379:
12380: @c =====================================================================
12381: @node The optional Double Number word set, The optional Exception word set, The optional Block word set, ANS conformance
12382: @section The optional Double Number word set
12383: @c =====================================================================
12384: @cindex system documentation, double words
12385: @cindex double words, system documentation
12386:
12387: @menu
12388: * double-ambcond:: Ambiguous Conditions
12389: @end menu
12390:
12391:
12392: @c ---------------------------------------------------------------------
12393: @node double-ambcond, , The optional Double Number word set, The optional Double Number word set
12394: @subsection Ambiguous conditions
12395: @c ---------------------------------------------------------------------
12396: @cindex double words, ambiguous conditions
12397: @cindex ambiguous conditions, double words
12398:
12399: @table @i
12400: @item @i{d} outside of range of @i{n} in @code{D>S}:
12401: @cindex @code{D>S}, @i{d} out of range of @i{n}
12402: The least significant cell of @i{d} is produced.
12403:
12404: @end table
12405:
12406:
12407: @c =====================================================================
12408: @node The optional Exception word set, The optional Facility word set, The optional Double Number word set, ANS conformance
12409: @section The optional Exception word set
12410: @c =====================================================================
12411: @cindex system documentation, exception words
12412: @cindex exception words, system documentation
12413:
12414: @menu
12415: * exception-idef:: Implementation Defined Options
12416: @end menu
12417:
12418:
12419: @c ---------------------------------------------------------------------
12420: @node exception-idef, , The optional Exception word set, The optional Exception word set
12421: @subsection Implementation Defined Options
12422: @c ---------------------------------------------------------------------
12423: @cindex implementation-defined options, exception words
12424: @cindex exception words, implementation-defined options
12425:
12426: @table @i
12427: @item @code{THROW}-codes used in the system:
12428: @cindex @code{THROW}-codes used in the system
12429: The codes -256@minus{}-511 are used for reporting signals. The mapping
12430: from OS signal numbers to throw codes is -256@minus{}@i{signal}. The
12431: codes -512@minus{}-2047 are used for OS errors (for file and memory
12432: allocation operations). The mapping from OS error numbers to throw codes
12433: is -512@minus{}@code{errno}. One side effect of this mapping is that
12434: undefined OS errors produce a message with a strange number; e.g.,
12435: @code{-1000 THROW} results in @code{Unknown error 488} on my system.
12436: @end table
12437:
12438: @c =====================================================================
12439: @node The optional Facility word set, The optional File-Access word set, The optional Exception word set, ANS conformance
12440: @section The optional Facility word set
12441: @c =====================================================================
12442: @cindex system documentation, facility words
12443: @cindex facility words, system documentation
12444:
12445: @menu
12446: * facility-idef:: Implementation Defined Options
12447: * facility-ambcond:: Ambiguous Conditions
12448: @end menu
12449:
12450:
12451: @c ---------------------------------------------------------------------
12452: @node facility-idef, facility-ambcond, The optional Facility word set, The optional Facility word set
12453: @subsection Implementation Defined Options
12454: @c ---------------------------------------------------------------------
12455: @cindex implementation-defined options, facility words
12456: @cindex facility words, implementation-defined options
12457:
12458: @table @i
12459: @item encoding of keyboard events (@code{EKEY}):
12460: @cindex keyboard events, encoding in @code{EKEY}
12461: @cindex @code{EKEY}, encoding of keyboard events
12462: Keys corresponding to ASCII characters are encoded as ASCII characters.
12463: Other keys are encoded with the constants @code{k-left}, @code{k-right},
12464: @code{k-up}, @code{k-down}, @code{k-home}, @code{k-end}, @code{k1},
12465: @code{k2}, @code{k3}, @code{k4}, @code{k5}, @code{k6}, @code{k7},
12466: @code{k8}, @code{k9}, @code{k10}, @code{k11}, @code{k12}.
12467:
12468:
12469: @item duration of a system clock tick:
12470: @cindex duration of a system clock tick
12471: @cindex clock tick duration
12472: System dependent. With respect to @code{MS}, the time is specified in
12473: microseconds. How well the OS and the hardware implement this, is
12474: another question.
12475:
12476: @item repeatability to be expected from the execution of @code{MS}:
12477: @cindex repeatability to be expected from the execution of @code{MS}
12478: @cindex @code{MS}, repeatability to be expected
12479: System dependent. On Unix, a lot depends on load. If the system is
12480: lightly loaded, and the delay is short enough that Gforth does not get
12481: swapped out, the performance should be acceptable. Under MS-DOS and
12482: other single-tasking systems, it should be good.
12483:
12484: @end table
12485:
12486:
12487: @c ---------------------------------------------------------------------
12488: @node facility-ambcond, , facility-idef, The optional Facility word set
12489: @subsection Ambiguous conditions
12490: @c ---------------------------------------------------------------------
12491: @cindex facility words, ambiguous conditions
12492: @cindex ambiguous conditions, facility words
12493:
12494: @table @i
12495: @item @code{AT-XY} can't be performed on user output device:
12496: @cindex @code{AT-XY} can't be performed on user output device
12497: Largely terminal dependent. No range checks are done on the arguments.
12498: No errors are reported. You may see some garbage appearing, you may see
12499: simply nothing happen.
12500:
12501: @end table
12502:
12503:
12504: @c =====================================================================
12505: @node The optional File-Access word set, The optional Floating-Point word set, The optional Facility word set, ANS conformance
12506: @section The optional File-Access word set
12507: @c =====================================================================
12508: @cindex system documentation, file words
12509: @cindex file words, system documentation
12510:
12511: @menu
12512: * file-idef:: Implementation Defined Options
12513: * file-ambcond:: Ambiguous Conditions
12514: @end menu
12515:
12516: @c ---------------------------------------------------------------------
12517: @node file-idef, file-ambcond, The optional File-Access word set, The optional File-Access word set
12518: @subsection Implementation Defined Options
12519: @c ---------------------------------------------------------------------
12520: @cindex implementation-defined options, file words
12521: @cindex file words, implementation-defined options
12522:
12523: @table @i
12524: @item file access methods used:
12525: @cindex file access methods used
12526: @code{R/O}, @code{R/W} and @code{BIN} work as you would
12527: expect. @code{W/O} translates into the C file opening mode @code{w} (or
12528: @code{wb}): The file is cleared, if it exists, and created, if it does
12529: not (with both @code{open-file} and @code{create-file}). Under Unix
12530: @code{create-file} creates a file with 666 permissions modified by your
12531: umask.
12532:
12533: @item file exceptions:
12534: @cindex file exceptions
12535: The file words do not raise exceptions (except, perhaps, memory access
12536: faults when you pass illegal addresses or file-ids).
12537:
12538: @item file line terminator:
12539: @cindex file line terminator
12540: System-dependent. Gforth uses C's newline character as line
12541: terminator. What the actual character code(s) of this are is
12542: system-dependent.
12543:
12544: @item file name format:
12545: @cindex file name format
12546: System dependent. Gforth just uses the file name format of your OS.
12547:
12548: @item information returned by @code{FILE-STATUS}:
12549: @cindex @code{FILE-STATUS}, returned information
12550: @code{FILE-STATUS} returns the most powerful file access mode allowed
12551: for the file: Either @code{R/O}, @code{W/O} or @code{R/W}. If the file
12552: cannot be accessed, @code{R/O BIN} is returned. @code{BIN} is applicable
12553: along with the returned mode.
12554:
12555: @item input file state after an exception when including source:
12556: @cindex exception when including source
12557: All files that are left via the exception are closed.
12558:
12559: @item @i{ior} values and meaning:
12560: @cindex @i{ior} values and meaning
12561: @cindex @i{wior} values and meaning
12562: The @i{ior}s returned by the file and memory allocation words are
12563: intended as throw codes. They typically are in the range
12564: -512@minus{}-2047 of OS errors. The mapping from OS error numbers to
12565: @i{ior}s is -512@minus{}@i{errno}.
12566:
12567: @item maximum depth of file input nesting:
12568: @cindex maximum depth of file input nesting
12569: @cindex file input nesting, maximum depth
12570: limited by the amount of return stack, locals/TIB stack, and the number
12571: of open files available. This should not give you troubles.
12572:
12573: @item maximum size of input line:
12574: @cindex maximum size of input line
12575: @cindex input line size, maximum
12576: @code{/line}. Currently 255.
12577:
12578: @item methods of mapping block ranges to files:
12579: @cindex mapping block ranges to files
12580: @cindex files containing blocks
12581: @cindex blocks in files
12582: By default, blocks are accessed in the file @file{blocks.fb} in the
12583: current working directory. The file can be switched with @code{USE}.
12584:
12585: @item number of string buffers provided by @code{S"}:
12586: @cindex @code{S"}, number of string buffers
12587: 1
12588:
12589: @item size of string buffer used by @code{S"}:
12590: @cindex @code{S"}, size of string buffer
12591: @code{/line}. currently 255.
12592:
12593: @end table
12594:
12595: @c ---------------------------------------------------------------------
12596: @node file-ambcond, , file-idef, The optional File-Access word set
12597: @subsection Ambiguous conditions
12598: @c ---------------------------------------------------------------------
12599: @cindex file words, ambiguous conditions
12600: @cindex ambiguous conditions, file words
12601:
12602: @table @i
12603: @item attempting to position a file outside its boundaries:
12604: @cindex @code{REPOSITION-FILE}, outside the file's boundaries
12605: @code{REPOSITION-FILE} is performed as usual: Afterwards,
12606: @code{FILE-POSITION} returns the value given to @code{REPOSITION-FILE}.
12607:
12608: @item attempting to read from file positions not yet written:
12609: @cindex reading from file positions not yet written
12610: End-of-file, i.e., zero characters are read and no error is reported.
12611:
12612: @item @i{file-id} is invalid (@code{INCLUDE-FILE}):
12613: @cindex @code{INCLUDE-FILE}, @i{file-id} is invalid
12614: An appropriate exception may be thrown, but a memory fault or other
12615: problem is more probable.
12616:
12617: @item I/O exception reading or closing @i{file-id} (@code{INCLUDE-FILE}, @code{INCLUDED}):
12618: @cindex @code{INCLUDE-FILE}, I/O exception reading or closing @i{file-id}
12619: @cindex @code{INCLUDED}, I/O exception reading or closing @i{file-id}
12620: The @i{ior} produced by the operation, that discovered the problem, is
12621: thrown.
12622:
12623: @item named file cannot be opened (@code{INCLUDED}):
12624: @cindex @code{INCLUDED}, named file cannot be opened
12625: The @i{ior} produced by @code{open-file} is thrown.
12626:
12627: @item requesting an unmapped block number:
12628: @cindex unmapped block numbers
12629: There are no unmapped legal block numbers. On some operating systems,
12630: writing a block with a large number may overflow the file system and
12631: have an error message as consequence.
12632:
12633: @item using @code{source-id} when @code{blk} is non-zero:
12634: @cindex @code{SOURCE-ID}, behaviour when @code{BLK} is non-zero
12635: @code{source-id} performs its function. Typically it will give the id of
12636: the source which loaded the block. (Better ideas?)
12637:
12638: @end table
12639:
12640:
12641: @c =====================================================================
12642: @node The optional Floating-Point word set, The optional Locals word set, The optional File-Access word set, ANS conformance
12643: @section The optional Floating-Point word set
12644: @c =====================================================================
12645: @cindex system documentation, floating-point words
12646: @cindex floating-point words, system documentation
12647:
12648: @menu
12649: * floating-idef:: Implementation Defined Options
12650: * floating-ambcond:: Ambiguous Conditions
12651: @end menu
12652:
12653:
12654: @c ---------------------------------------------------------------------
12655: @node floating-idef, floating-ambcond, The optional Floating-Point word set, The optional Floating-Point word set
12656: @subsection Implementation Defined Options
12657: @c ---------------------------------------------------------------------
12658: @cindex implementation-defined options, floating-point words
12659: @cindex floating-point words, implementation-defined options
12660:
12661: @table @i
12662: @item format and range of floating point numbers:
12663: @cindex format and range of floating point numbers
12664: @cindex floating point numbers, format and range
12665: System-dependent; the @code{double} type of C.
12666:
12667: @item results of @code{REPRESENT} when @i{float} is out of range:
12668: @cindex @code{REPRESENT}, results when @i{float} is out of range
12669: System dependent; @code{REPRESENT} is implemented using the C library
12670: function @code{ecvt()} and inherits its behaviour in this respect.
12671:
12672: @item rounding or truncation of floating-point numbers:
12673: @cindex rounding of floating-point numbers
12674: @cindex truncation of floating-point numbers
12675: @cindex floating-point numbers, rounding or truncation
12676: System dependent; the rounding behaviour is inherited from the hosting C
12677: compiler. IEEE-FP-based (i.e., most) systems by default round to
12678: nearest, and break ties by rounding to even (i.e., such that the last
12679: bit of the mantissa is 0).
12680:
12681: @item size of floating-point stack:
12682: @cindex floating-point stack size
12683: @code{s" FLOATING-STACK" environment? drop .} gives the total size of
12684: the floating-point stack (in floats). You can specify this on startup
12685: with the command-line option @code{-f} (@pxref{Invoking Gforth}).
12686:
12687: @item width of floating-point stack:
12688: @cindex floating-point stack width
12689: @code{1 floats}.
12690:
12691: @end table
12692:
12693:
12694: @c ---------------------------------------------------------------------
12695: @node floating-ambcond, , floating-idef, The optional Floating-Point word set
12696: @subsection Ambiguous conditions
12697: @c ---------------------------------------------------------------------
12698: @cindex floating-point words, ambiguous conditions
12699: @cindex ambiguous conditions, floating-point words
12700:
12701: @table @i
12702: @item @code{df@@} or @code{df!} used with an address that is not double-float aligned:
12703: @cindex @code{df@@} or @code{df!} used with an address that is not double-float aligned
12704: System-dependent. Typically results in a @code{-23 THROW} like other
12705: alignment violations.
12706:
12707: @item @code{f@@} or @code{f!} used with an address that is not float aligned:
12708: @cindex @code{f@@} used with an address that is not float aligned
12709: @cindex @code{f!} used with an address that is not float aligned
12710: System-dependent. Typically results in a @code{-23 THROW} like other
12711: alignment violations.
12712:
12713: @item floating-point result out of range:
12714: @cindex floating-point result out of range
12715: System-dependent. Can result in a @code{-55 THROW} (Floating-point
12716: unidentified fault), or can produce a special value representing, e.g.,
12717: Infinity.
12718:
12719: @item @code{sf@@} or @code{sf!} used with an address that is not single-float aligned:
12720: @cindex @code{sf@@} or @code{sf!} used with an address that is not single-float aligned
12721: System-dependent. Typically results in an alignment fault like other
12722: alignment violations.
12723:
12724: @item @code{base} is not decimal (@code{REPRESENT}, @code{F.}, @code{FE.}, @code{FS.}):
12725: @cindex @code{base} is not decimal (@code{REPRESENT}, @code{F.}, @code{FE.}, @code{FS.})
12726: The floating-point number is converted into decimal nonetheless.
12727:
12728: @item Both arguments are equal to zero (@code{FATAN2}):
12729: @cindex @code{FATAN2}, both arguments are equal to zero
12730: System-dependent. @code{FATAN2} is implemented using the C library
12731: function @code{atan2()}.
12732:
12733: @item Using @code{FTAN} on an argument @i{r1} where cos(@i{r1}) is zero:
12734: @cindex @code{FTAN} on an argument @i{r1} where cos(@i{r1}) is zero
12735: System-dependent. Anyway, typically the cos of @i{r1} will not be zero
12736: because of small errors and the tan will be a very large (or very small)
12737: but finite number.
12738:
12739: @item @i{d} cannot be presented precisely as a float in @code{D>F}:
12740: @cindex @code{D>F}, @i{d} cannot be presented precisely as a float
12741: The result is rounded to the nearest float.
12742:
12743: @item dividing by zero:
12744: @cindex dividing by zero, floating-point
12745: @cindex floating-point dividing by zero
12746: @cindex floating-point unidentified fault, FP divide-by-zero
12747: @code{-55 throw} (Floating-point unidentified fault)
12748:
12749: @item exponent too big for conversion (@code{DF!}, @code{DF@@}, @code{SF!}, @code{SF@@}):
12750: @cindex exponent too big for conversion (@code{DF!}, @code{DF@@}, @code{SF!}, @code{SF@@})
12751: System dependent. On IEEE-FP based systems the number is converted into
12752: an infinity.
12753:
12754: @item @i{float}<1 (@code{FACOSH}):
12755: @cindex @code{FACOSH}, @i{float}<1
12756: @cindex floating-point unidentified fault, @code{FACOSH}
12757: @code{-55 throw} (Floating-point unidentified fault)
12758:
12759: @item @i{float}=<-1 (@code{FLNP1}):
12760: @cindex @code{FLNP1}, @i{float}=<-1
12761: @cindex floating-point unidentified fault, @code{FLNP1}
12762: @code{-55 throw} (Floating-point unidentified fault). On IEEE-FP systems
12763: negative infinity is typically produced for @i{float}=-1.
12764:
12765: @item @i{float}=<0 (@code{FLN}, @code{FLOG}):
12766: @cindex @code{FLN}, @i{float}=<0
12767: @cindex @code{FLOG}, @i{float}=<0
12768: @cindex floating-point unidentified fault, @code{FLN} or @code{FLOG}
12769: @code{-55 throw} (Floating-point unidentified fault). On IEEE-FP systems
12770: negative infinity is typically produced for @i{float}=0.
12771:
12772: @item @i{float}<0 (@code{FASINH}, @code{FSQRT}):
12773: @cindex @code{FASINH}, @i{float}<0
12774: @cindex @code{FSQRT}, @i{float}<0
12775: @cindex floating-point unidentified fault, @code{FASINH} or @code{FSQRT}
12776: @code{-55 throw} (Floating-point unidentified fault). @code{fasinh}
12777: produces values for these inputs on my Linux box (Bug in the C library?)
12778:
12779: @item |@i{float}|>1 (@code{FACOS}, @code{FASIN}, @code{FATANH}):
12780: @cindex @code{FACOS}, |@i{float}|>1
12781: @cindex @code{FASIN}, |@i{float}|>1
12782: @cindex @code{FATANH}, |@i{float}|>1
12783: @cindex floating-point unidentified fault, @code{FACOS}, @code{FASIN} or @code{FATANH}
12784: @code{-55 throw} (Floating-point unidentified fault).
12785:
12786: @item integer part of float cannot be represented by @i{d} in @code{F>D}:
12787: @cindex @code{F>D}, integer part of float cannot be represented by @i{d}
12788: @cindex floating-point unidentified fault, @code{F>D}
12789: @code{-55 throw} (Floating-point unidentified fault).
12790:
12791: @item string larger than pictured numeric output area (@code{f.}, @code{fe.}, @code{fs.}):
12792: @cindex string larger than pictured numeric output area (@code{f.}, @code{fe.}, @code{fs.})
12793: This does not happen.
12794: @end table
12795:
12796: @c =====================================================================
12797: @node The optional Locals word set, The optional Memory-Allocation word set, The optional Floating-Point word set, ANS conformance
12798: @section The optional Locals word set
12799: @c =====================================================================
12800: @cindex system documentation, locals words
12801: @cindex locals words, system documentation
12802:
12803: @menu
12804: * locals-idef:: Implementation Defined Options
12805: * locals-ambcond:: Ambiguous Conditions
12806: @end menu
12807:
12808:
12809: @c ---------------------------------------------------------------------
12810: @node locals-idef, locals-ambcond, The optional Locals word set, The optional Locals word set
12811: @subsection Implementation Defined Options
12812: @c ---------------------------------------------------------------------
12813: @cindex implementation-defined options, locals words
12814: @cindex locals words, implementation-defined options
12815:
12816: @table @i
12817: @item maximum number of locals in a definition:
12818: @cindex maximum number of locals in a definition
12819: @cindex locals, maximum number in a definition
12820: @code{s" #locals" environment? drop .}. Currently 15. This is a lower
12821: bound, e.g., on a 32-bit machine there can be 41 locals of up to 8
12822: characters. The number of locals in a definition is bounded by the size
12823: of locals-buffer, which contains the names of the locals.
12824:
12825: @end table
12826:
12827:
12828: @c ---------------------------------------------------------------------
12829: @node locals-ambcond, , locals-idef, The optional Locals word set
12830: @subsection Ambiguous conditions
12831: @c ---------------------------------------------------------------------
12832: @cindex locals words, ambiguous conditions
12833: @cindex ambiguous conditions, locals words
12834:
12835: @table @i
12836: @item executing a named local in interpretation state:
12837: @cindex local in interpretation state
12838: @cindex Interpreting a compile-only word, for a local
12839: Locals have no interpretation semantics. If you try to perform the
12840: interpretation semantics, you will get a @code{-14 throw} somewhere
12841: (Interpreting a compile-only word). If you perform the compilation
12842: semantics, the locals access will be compiled (irrespective of state).
12843:
12844: @item @i{name} not defined by @code{VALUE} or @code{(LOCAL)} (@code{TO}):
12845: @cindex name not defined by @code{VALUE} or @code{(LOCAL)} used by @code{TO}
12846: @cindex @code{TO} on non-@code{VALUE}s and non-locals
12847: @cindex Invalid name argument, @code{TO}
12848: @code{-32 throw} (Invalid name argument)
12849:
12850: @end table
12851:
12852:
12853: @c =====================================================================
12854: @node The optional Memory-Allocation word set, The optional Programming-Tools word set, The optional Locals word set, ANS conformance
12855: @section The optional Memory-Allocation word set
12856: @c =====================================================================
12857: @cindex system documentation, memory-allocation words
12858: @cindex memory-allocation words, system documentation
12859:
12860: @menu
12861: * memory-idef:: Implementation Defined Options
12862: @end menu
12863:
12864:
12865: @c ---------------------------------------------------------------------
12866: @node memory-idef, , The optional Memory-Allocation word set, The optional Memory-Allocation word set
12867: @subsection Implementation Defined Options
12868: @c ---------------------------------------------------------------------
12869: @cindex implementation-defined options, memory-allocation words
12870: @cindex memory-allocation words, implementation-defined options
12871:
12872: @table @i
12873: @item values and meaning of @i{ior}:
12874: @cindex @i{ior} values and meaning
12875: The @i{ior}s returned by the file and memory allocation words are
12876: intended as throw codes. They typically are in the range
12877: -512@minus{}-2047 of OS errors. The mapping from OS error numbers to
12878: @i{ior}s is -512@minus{}@i{errno}.
12879:
12880: @end table
12881:
12882: @c =====================================================================
12883: @node The optional Programming-Tools word set, The optional Search-Order word set, The optional Memory-Allocation word set, ANS conformance
12884: @section The optional Programming-Tools word set
12885: @c =====================================================================
12886: @cindex system documentation, programming-tools words
12887: @cindex programming-tools words, system documentation
12888:
12889: @menu
12890: * programming-idef:: Implementation Defined Options
12891: * programming-ambcond:: Ambiguous Conditions
12892: @end menu
12893:
12894:
12895: @c ---------------------------------------------------------------------
12896: @node programming-idef, programming-ambcond, The optional Programming-Tools word set, The optional Programming-Tools word set
12897: @subsection Implementation Defined Options
12898: @c ---------------------------------------------------------------------
12899: @cindex implementation-defined options, programming-tools words
12900: @cindex programming-tools words, implementation-defined options
12901:
12902: @table @i
12903: @item ending sequence for input following @code{;CODE} and @code{CODE}:
12904: @cindex @code{;CODE} ending sequence
12905: @cindex @code{CODE} ending sequence
12906: @code{END-CODE}
12907:
12908: @item manner of processing input following @code{;CODE} and @code{CODE}:
12909: @cindex @code{;CODE}, processing input
12910: @cindex @code{CODE}, processing input
12911: The @code{ASSEMBLER} vocabulary is pushed on the search order stack, and
12912: the input is processed by the text interpreter, (starting) in interpret
12913: state.
12914:
12915: @item search order capability for @code{EDITOR} and @code{ASSEMBLER}:
12916: @cindex @code{ASSEMBLER}, search order capability
12917: The ANS Forth search order word set.
12918:
12919: @item source and format of display by @code{SEE}:
12920: @cindex @code{SEE}, source and format of output
12921: The source for @code{see} is the intermediate code used by the inner
12922: interpreter. The current @code{see} tries to output Forth source code
12923: as well as possible.
12924:
12925: @end table
12926:
12927: @c ---------------------------------------------------------------------
12928: @node programming-ambcond, , programming-idef, The optional Programming-Tools word set
12929: @subsection Ambiguous conditions
12930: @c ---------------------------------------------------------------------
12931: @cindex programming-tools words, ambiguous conditions
12932: @cindex ambiguous conditions, programming-tools words
12933:
12934: @table @i
12935:
12936: @item deleting the compilation word list (@code{FORGET}):
12937: @cindex @code{FORGET}, deleting the compilation word list
12938: Not implemented (yet).
12939:
12940: @item fewer than @i{u}+1 items on the control-flow stack (@code{CS-PICK}, @code{CS-ROLL}):
12941: @cindex @code{CS-PICK}, fewer than @i{u}+1 items on the control flow-stack
12942: @cindex @code{CS-ROLL}, fewer than @i{u}+1 items on the control flow-stack
12943: @cindex control-flow stack underflow
12944: This typically results in an @code{abort"} with a descriptive error
12945: message (may change into a @code{-22 throw} (Control structure mismatch)
12946: in the future). You may also get a memory access error. If you are
12947: unlucky, this ambiguous condition is not caught.
12948:
12949: @item @i{name} can't be found (@code{FORGET}):
12950: @cindex @code{FORGET}, @i{name} can't be found
12951: Not implemented (yet).
12952:
12953: @item @i{name} not defined via @code{CREATE}:
12954: @cindex @code{;CODE}, @i{name} not defined via @code{CREATE}
12955: @code{;CODE} behaves like @code{DOES>} in this respect, i.e., it changes
12956: the execution semantics of the last defined word no matter how it was
12957: defined.
12958:
12959: @item @code{POSTPONE} applied to @code{[IF]}:
12960: @cindex @code{POSTPONE} applied to @code{[IF]}
12961: @cindex @code{[IF]} and @code{POSTPONE}
12962: After defining @code{: X POSTPONE [IF] ; IMMEDIATE}. @code{X} is
12963: equivalent to @code{[IF]}.
12964:
12965: @item reaching the end of the input source before matching @code{[ELSE]} or @code{[THEN]}:
12966: @cindex @code{[IF]}, end of the input source before matching @code{[ELSE]} or @code{[THEN]}
12967: Continue in the same state of conditional compilation in the next outer
12968: input source. Currently there is no warning to the user about this.
12969:
12970: @item removing a needed definition (@code{FORGET}):
12971: @cindex @code{FORGET}, removing a needed definition
12972: Not implemented (yet).
12973:
12974: @end table
12975:
12976:
12977: @c =====================================================================
12978: @node The optional Search-Order word set, , The optional Programming-Tools word set, ANS conformance
12979: @section The optional Search-Order word set
12980: @c =====================================================================
12981: @cindex system documentation, search-order words
12982: @cindex search-order words, system documentation
12983:
12984: @menu
12985: * search-idef:: Implementation Defined Options
12986: * search-ambcond:: Ambiguous Conditions
12987: @end menu
12988:
12989:
12990: @c ---------------------------------------------------------------------
12991: @node search-idef, search-ambcond, The optional Search-Order word set, The optional Search-Order word set
12992: @subsection Implementation Defined Options
12993: @c ---------------------------------------------------------------------
12994: @cindex implementation-defined options, search-order words
12995: @cindex search-order words, implementation-defined options
12996:
12997: @table @i
12998: @item maximum number of word lists in search order:
12999: @cindex maximum number of word lists in search order
13000: @cindex search order, maximum depth
13001: @code{s" wordlists" environment? drop .}. Currently 16.
13002:
13003: @item minimum search order:
13004: @cindex minimum search order
13005: @cindex search order, minimum
13006: @code{root root}.
13007:
13008: @end table
13009:
13010: @c ---------------------------------------------------------------------
13011: @node search-ambcond, , search-idef, The optional Search-Order word set
13012: @subsection Ambiguous conditions
13013: @c ---------------------------------------------------------------------
13014: @cindex search-order words, ambiguous conditions
13015: @cindex ambiguous conditions, search-order words
13016:
13017: @table @i
13018: @item changing the compilation word list (during compilation):
13019: @cindex changing the compilation word list (during compilation)
13020: @cindex compilation word list, change before definition ends
13021: The word is entered into the word list that was the compilation word list
13022: at the start of the definition. Any changes to the name field (e.g.,
13023: @code{immediate}) or the code field (e.g., when executing @code{DOES>})
13024: are applied to the latest defined word (as reported by @code{last} or
13025: @code{lastxt}), if possible, irrespective of the compilation word list.
13026:
13027: @item search order empty (@code{previous}):
13028: @cindex @code{previous}, search order empty
13029: @cindex vocstack empty, @code{previous}
13030: @code{abort" Vocstack empty"}.
13031:
13032: @item too many word lists in search order (@code{also}):
13033: @cindex @code{also}, too many word lists in search order
13034: @cindex vocstack full, @code{also}
13035: @code{abort" Vocstack full"}.
13036:
13037: @end table
13038:
13039: @c ***************************************************************
13040: @node Standard vs Extensions, Model, ANS conformance, Top
13041: @chapter Should I use Gforth extensions?
13042: @cindex Gforth extensions
13043:
13044: As you read through the rest of this manual, you will see documentation
13045: for @i{Standard} words, and documentation for some appealing Gforth
13046: @i{extensions}. You might ask yourself the question: @i{``Should I
13047: restrict myself to the standard, or should I use the extensions?''}
13048:
13049: The answer depends on the goals you have for the program you are working
13050: on:
13051:
13052: @itemize @bullet
13053:
13054: @item Is it just for yourself or do you want to share it with others?
13055:
13056: @item
13057: If you want to share it, do the others all use Gforth?
13058:
13059: @item
13060: If it is just for yourself, do you want to restrict yourself to Gforth?
13061:
13062: @end itemize
13063:
13064: If restricting the program to Gforth is ok, then there is no reason not
13065: to use extensions. It is still a good idea to keep to the standard
13066: where it is easy, in case you want to reuse these parts in another
13067: program that you want to be portable.
13068:
13069: If you want to be able to port the program to other Forth systems, there
13070: are the following points to consider:
13071:
13072: @itemize @bullet
13073:
13074: @item
13075: Most Forth systems that are being maintained support the ANS Forth
13076: standard. So if your program complies with the standard, it will be
13077: portable among many systems.
13078:
13079: @item
13080: A number of the Gforth extensions can be implemented in ANS Forth using
13081: public-domain files provided in the @file{compat/} directory. These are
13082: mentioned in the text in passing. There is no reason not to use these
13083: extensions, your program will still be ANS Forth compliant; just include
13084: the appropriate compat files with your program.
13085:
13086: @item
13087: The tool @file{ans-report.fs} (@pxref{ANS Report}) makes it easy to
13088: analyse your program and determine what non-Standard words it relies
13089: upon. However, it does not check whether you use standard words in a
13090: non-standard way.
13091:
13092: @item
13093: Some techniques are not standardized by ANS Forth, and are hard or
13094: impossible to implement in a standard way, but can be implemented in
13095: most Forth systems easily, and usually in similar ways (e.g., accessing
13096: word headers). Forth has a rich historical precedent for programmers
13097: taking advantage of implementation-dependent features of their tools
13098: (for example, relying on a knowledge of the dictionary
13099: structure). Sometimes these techniques are necessary to extract every
13100: last bit of performance from the hardware, sometimes they are just a
13101: programming shorthand.
13102:
13103: @item
13104: Does using a Gforth extension save more work than the porting this part
13105: to other Forth systems (if any) will cost?
13106:
13107: @item
13108: Is the additional functionality worth the reduction in portability and
13109: the additional porting problems?
13110:
13111: @end itemize
13112:
13113: In order to perform these consideratios, you need to know what's
13114: standard and what's not. This manual generally states if something is
13115: non-standard, but the authoritative source is the standard document.
13116: Appendix A of the Standard (@var{Rationale}) provides a valuable insight
13117: into the thought processes of the technical committee.
13118:
13119: Note also that portability between Forth systems is not the only
13120: portability issue; there is also the issue of portability between
13121: different platforms (processor/OS combinations).
13122:
13123: @c ***************************************************************
13124: @node Model, Integrating Gforth, Standard vs Extensions, Top
13125: @chapter Model
13126:
13127: This chapter has yet to be written. It will contain information, on
13128: which internal structures you can rely.
13129:
13130: @c ***************************************************************
13131: @node Integrating Gforth, Emacs and Gforth, Model, Top
13132: @chapter Integrating Gforth into C programs
13133:
13134: This is not yet implemented.
13135:
13136: Several people like to use Forth as scripting language for applications
13137: that are otherwise written in C, C++, or some other language.
13138:
13139: The Forth system ATLAST provides facilities for embedding it into
13140: applications; unfortunately it has several disadvantages: most
13141: importantly, it is not based on ANS Forth, and it is apparently dead
13142: (i.e., not developed further and not supported). The facilities
13143: provided by Gforth in this area are inspired by ATLAST's facilities, so
13144: making the switch should not be hard.
13145:
13146: We also tried to design the interface such that it can easily be
13147: implemented by other Forth systems, so that we may one day arrive at a
13148: standardized interface. Such a standard interface would allow you to
13149: replace the Forth system without having to rewrite C code.
13150:
13151: You embed the Gforth interpreter by linking with the library
13152: @code{libgforth.a} (give the compiler the option @code{-lgforth}). All
13153: global symbols in this library that belong to the interface, have the
13154: prefix @code{forth_}. (Global symbols that are used internally have the
13155: prefix @code{gforth_}).
13156:
13157: You can include the declarations of Forth types and the functions and
13158: variables of the interface with @code{#include <forth.h>}.
13159:
13160: Types.
13161:
13162: Variables.
13163:
13164: Data and FP Stack pointer. Area sizes.
13165:
13166: functions.
13167:
13168: forth_init(imagefile)
13169: forth_evaluate(string) exceptions?
13170: forth_goto(address) (or forth_execute(xt)?)
13171: forth_continue() (a corountining mechanism)
13172:
13173: Adding primitives.
13174:
13175: No checking.
13176:
13177: Signals?
13178:
13179: Accessing the Stacks
13180:
13181: @c ******************************************************************
13182: @node Emacs and Gforth, Image Files, Integrating Gforth, Top
13183: @chapter Emacs and Gforth
13184: @cindex Emacs and Gforth
13185:
13186: @cindex @file{gforth.el}
13187: @cindex @file{forth.el}
13188: @cindex Rydqvist, Goran
13189: @cindex comment editing commands
13190: @cindex @code{\}, editing with Emacs
13191: @cindex debug tracer editing commands
13192: @cindex @code{~~}, removal with Emacs
13193: @cindex Forth mode in Emacs
13194: Gforth comes with @file{gforth.el}, an improved version of
13195: @file{forth.el} by Goran Rydqvist (included in the TILE package). The
13196: improvements are:
13197:
13198: @itemize @bullet
13199: @item
13200: A better (but still not perfect) handling of indentation.
13201: @item
13202: Comment paragraph filling (@kbd{M-q})
13203: @item
13204: Commenting (@kbd{C-x \}) and uncommenting (@kbd{C-u C-x \}) of regions
13205: @item
13206: Removal of debugging tracers (@kbd{C-x ~}, @pxref{Debugging}).
13207: @item
13208: Support of the @code{info-lookup} feature for looking up the
13209: documentation of a word.
13210: @end itemize
13211:
13212: I left the stuff I do not use alone, even though some of it only makes
13213: sense for TILE. To get a description of these features, enter Forth mode
13214: and type @kbd{C-h m}.
13215:
13216: @cindex source location of error or debugging output in Emacs
13217: @cindex error output, finding the source location in Emacs
13218: @cindex debugging output, finding the source location in Emacs
13219: In addition, Gforth supports Emacs quite well: The source code locations
13220: given in error messages, debugging output (from @code{~~}) and failed
13221: assertion messages are in the right format for Emacs' compilation mode
13222: (@pxref{Compilation, , Running Compilations under Emacs, emacs, Emacs
13223: Manual}) so the source location corresponding to an error or other
13224: message is only a few keystrokes away (@kbd{C-x `} for the next error,
13225: @kbd{C-c C-c} for the error under the cursor).
13226:
13227: @cindex @file{TAGS} file
13228: @cindex @file{etags.fs}
13229: @cindex viewing the source of a word in Emacs
13230: @cindex @code{require}, placement in files
13231: @cindex @code{include}, placement in files
13232: Also, if you @code{require} @file{etags.fs}, a new @file{TAGS} file will
13233: be produced (@pxref{Tags, , Tags Tables, emacs, Emacs Manual}) that
13234: contains the definitions of all words defined afterwards. You can then
13235: find the source for a word using @kbd{M-.}. Note that emacs can use
13236: several tags files at the same time (e.g., one for the Gforth sources
13237: and one for your program, @pxref{Select Tags Table,,Selecting a Tags
13238: Table,emacs, Emacs Manual}). The TAGS file for the preloaded words is
13239: @file{$(datadir)/gforth/$(VERSION)/TAGS} (e.g.,
13240: @file{/usr/local/share/gforth/0.2.0/TAGS}). To get the best behaviour
13241: with @file{etags.fs}, you should avoid putting definitions both before
13242: and after @code{require} etc., otherwise you will see the same file
13243: visited several times by commands like @code{tags-search}.
13244:
13245: @cindex viewing the documentation of a word in Emacs
13246: @cindex context-sensitive help
13247: Moreover, for words documented in this manual, you can look up the
13248: glossary entry quickly by using @kbd{C-h TAB}
13249: (@code{info-lookup-symbol}, see @pxref{Documentation, ,Documentation
13250: Commands, emacs, Emacs Manual}). This feature requires Emacs 20.3 or
13251: later and does not work for words containing @code{:}.
13252:
13253:
13254: @cindex @file{.emacs}
13255: To get all these benefits, add the following lines to your @file{.emacs}
13256: file:
13257:
13258: @example
13259: (autoload 'forth-mode "gforth.el")
13260: (setq auto-mode-alist (cons '("\\.fs\\'" . forth-mode) auto-mode-alist))
13261: @end example
13262:
13263: @c ******************************************************************
13264: @node Image Files, Engine, Emacs and Gforth, Top
13265: @chapter Image Files
13266: @cindex image file
13267: @cindex @file{.fi} files
13268: @cindex precompiled Forth code
13269: @cindex dictionary in persistent form
13270: @cindex persistent form of dictionary
13271:
13272: An image file is a file containing an image of the Forth dictionary,
13273: i.e., compiled Forth code and data residing in the dictionary. By
13274: convention, we use the extension @code{.fi} for image files.
13275:
13276: @menu
13277: * Image Licensing Issues:: Distribution terms for images.
13278: * Image File Background:: Why have image files?
13279: * Non-Relocatable Image Files:: don't always work.
13280: * Data-Relocatable Image Files:: are better.
13281: * Fully Relocatable Image Files:: better yet.
13282: * Stack and Dictionary Sizes:: Setting the default sizes for an image.
13283: * Running Image Files:: @code{gforth -i @i{file}} or @i{file}.
13284: * Modifying the Startup Sequence:: and turnkey applications.
13285: @end menu
13286:
13287: @node Image Licensing Issues, Image File Background, Image Files, Image Files
13288: @section Image Licensing Issues
13289: @cindex license for images
13290: @cindex image license
13291:
13292: An image created with @code{gforthmi} (@pxref{gforthmi}) or
13293: @code{savesystem} (@pxref{Non-Relocatable Image Files}) includes the
13294: original image; i.e., according to copyright law it is a derived work of
13295: the original image.
13296:
13297: Since Gforth is distributed under the GNU GPL, the newly created image
13298: falls under the GNU GPL, too. In particular, this means that if you
13299: distribute the image, you have to make all of the sources for the image
13300: available, including those you wrote. For details see @ref{License, ,
13301: GNU General Public License (Section 3)}.
13302:
13303: If you create an image with @code{cross} (@pxref{cross.fs}), the image
13304: contains only code compiled from the sources you gave it; if none of
13305: these sources is under the GPL, the terms discussed above do not apply
13306: to the image. However, if your image needs an engine (a gforth binary)
13307: that is under the GPL, you should make sure that you distribute both in
13308: a way that is at most a @emph{mere aggregation}, if you don't want the
13309: terms of the GPL to apply to the image.
13310:
13311: @node Image File Background, Non-Relocatable Image Files, Image Licensing Issues, Image Files
13312: @section Image File Background
13313: @cindex image file background
13314:
13315: Our Forth system consists not only of primitives, but also of
13316: definitions written in Forth. Since the Forth compiler itself belongs to
13317: those definitions, it is not possible to start the system with the
13318: primitives and the Forth source alone. Therefore we provide the Forth
13319: code as an image file in nearly executable form. When Gforth starts up,
13320: a C routine loads the image file into memory, optionally relocates the
13321: addresses, then sets up the memory (stacks etc.) according to
13322: information in the image file, and (finally) starts executing Forth
13323: code.
13324:
13325: The image file variants represent different compromises between the
13326: goals of making it easy to generate image files and making them
13327: portable.
13328:
13329: @cindex relocation at run-time
13330: Win32Forth 3.4 and Mitch Bradley's @code{cforth} use relocation at
13331: run-time. This avoids many of the complications discussed below (image
13332: files are data relocatable without further ado), but costs performance
13333: (one addition per memory access).
13334:
13335: @cindex relocation at load-time
13336: By contrast, the Gforth loader performs relocation at image load time. The
13337: loader also has to replace tokens that represent primitive calls with the
13338: appropriate code-field addresses (or code addresses in the case of
13339: direct threading).
13340:
13341: There are three kinds of image files, with different degrees of
13342: relocatability: non-relocatable, data-relocatable, and fully relocatable
13343: image files.
13344:
13345: @cindex image file loader
13346: @cindex relocating loader
13347: @cindex loader for image files
13348: These image file variants have several restrictions in common; they are
13349: caused by the design of the image file loader:
13350:
13351: @itemize @bullet
13352: @item
13353: There is only one segment; in particular, this means, that an image file
13354: cannot represent @code{ALLOCATE}d memory chunks (and pointers to
13355: them). The contents of the stacks are not represented, either.
13356:
13357: @item
13358: The only kinds of relocation supported are: adding the same offset to
13359: all cells that represent data addresses; and replacing special tokens
13360: with code addresses or with pieces of machine code.
13361:
13362: If any complex computations involving addresses are performed, the
13363: results cannot be represented in the image file. Several applications that
13364: use such computations come to mind:
13365: @itemize @minus
13366: @item
13367: Hashing addresses (or data structures which contain addresses) for table
13368: lookup. If you use Gforth's @code{table}s or @code{wordlist}s for this
13369: purpose, you will have no problem, because the hash tables are
13370: recomputed automatically when the system is started. If you use your own
13371: hash tables, you will have to do something similar.
13372:
13373: @item
13374: There's a cute implementation of doubly-linked lists that uses
13375: @code{XOR}ed addresses. You could represent such lists as singly-linked
13376: in the image file, and restore the doubly-linked representation on
13377: startup.@footnote{In my opinion, though, you should think thrice before
13378: using a doubly-linked list (whatever implementation).}
13379:
13380: @item
13381: The code addresses of run-time routines like @code{docol:} cannot be
13382: represented in the image file (because their tokens would be replaced by
13383: machine code in direct threaded implementations). As a workaround,
13384: compute these addresses at run-time with @code{>code-address} from the
13385: executions tokens of appropriate words (see the definitions of
13386: @code{docol:} and friends in @file{kernel.fs}).
13387:
13388: @item
13389: On many architectures addresses are represented in machine code in some
13390: shifted or mangled form. You cannot put @code{CODE} words that contain
13391: absolute addresses in this form in a relocatable image file. Workarounds
13392: are representing the address in some relative form (e.g., relative to
13393: the CFA, which is present in some register), or loading the address from
13394: a place where it is stored in a non-mangled form.
13395: @end itemize
13396: @end itemize
13397:
13398: @node Non-Relocatable Image Files, Data-Relocatable Image Files, Image File Background, Image Files
13399: @section Non-Relocatable Image Files
13400: @cindex non-relocatable image files
13401: @cindex image file, non-relocatable
13402:
13403: These files are simple memory dumps of the dictionary. They are specific
13404: to the executable (i.e., @file{gforth} file) they were created
13405: with. What's worse, they are specific to the place on which the
13406: dictionary resided when the image was created. Now, there is no
13407: guarantee that the dictionary will reside at the same place the next
13408: time you start Gforth, so there's no guarantee that a non-relocatable
13409: image will work the next time (Gforth will complain instead of crashing,
13410: though).
13411:
13412: You can create a non-relocatable image file with
13413:
13414:
13415: doc-savesystem
13416:
13417:
13418: @node Data-Relocatable Image Files, Fully Relocatable Image Files, Non-Relocatable Image Files, Image Files
13419: @section Data-Relocatable Image Files
13420: @cindex data-relocatable image files
13421: @cindex image file, data-relocatable
13422:
13423: These files contain relocatable data addresses, but fixed code addresses
13424: (instead of tokens). They are specific to the executable (i.e.,
13425: @file{gforth} file) they were created with. For direct threading on some
13426: architectures (e.g., the i386), data-relocatable images do not work. You
13427: get a data-relocatable image, if you use @file{gforthmi} with a
13428: Gforth binary that is not doubly indirect threaded (@pxref{Fully
13429: Relocatable Image Files}).
13430:
13431: @node Fully Relocatable Image Files, Stack and Dictionary Sizes, Data-Relocatable Image Files, Image Files
13432: @section Fully Relocatable Image Files
13433: @cindex fully relocatable image files
13434: @cindex image file, fully relocatable
13435:
13436: @cindex @file{kern*.fi}, relocatability
13437: @cindex @file{gforth.fi}, relocatability
13438: These image files have relocatable data addresses, and tokens for code
13439: addresses. They can be used with different binaries (e.g., with and
13440: without debugging) on the same machine, and even across machines with
13441: the same data formats (byte order, cell size, floating point
13442: format). However, they are usually specific to the version of Gforth
13443: they were created with. The files @file{gforth.fi} and @file{kernl*.fi}
13444: are fully relocatable.
13445:
13446: There are two ways to create a fully relocatable image file:
13447:
13448: @menu
13449: * gforthmi:: The normal way
13450: * cross.fs:: The hard way
13451: @end menu
13452:
13453: @node gforthmi, cross.fs, Fully Relocatable Image Files, Fully Relocatable Image Files
13454: @subsection @file{gforthmi}
13455: @cindex @file{comp-i.fs}
13456: @cindex @file{gforthmi}
13457:
13458: You will usually use @file{gforthmi}. If you want to create an
13459: image @i{file} that contains everything you would load by invoking
13460: Gforth with @code{gforth @i{options}}, you simply say:
13461: @example
13462: gforthmi @i{file} @i{options}
13463: @end example
13464:
13465: E.g., if you want to create an image @file{asm.fi} that has the file
13466: @file{asm.fs} loaded in addition to the usual stuff, you could do it
13467: like this:
13468:
13469: @example
13470: gforthmi asm.fi asm.fs
13471: @end example
13472:
13473: @file{gforthmi} is implemented as a sh script and works like this: It
13474: produces two non-relocatable images for different addresses and then
13475: compares them. Its output reflects this: first you see the output (if
13476: any) of the two Gforth invocations that produce the non-relocatable image
13477: files, then you see the output of the comparing program: It displays the
13478: offset used for data addresses and the offset used for code addresses;
13479: moreover, for each cell that cannot be represented correctly in the
13480: image files, it displays a line like this:
13481:
13482: @example
13483: 78DC BFFFFA50 BFFFFA40
13484: @end example
13485:
13486: This means that at offset $78dc from @code{forthstart}, one input image
13487: contains $bffffa50, and the other contains $bffffa40. Since these cells
13488: cannot be represented correctly in the output image, you should examine
13489: these places in the dictionary and verify that these cells are dead
13490: (i.e., not read before they are written).
13491:
13492: @cindex --application, @code{gforthmi} option
13493: If you insert the option @code{--application} in front of the image file
13494: name, you will get an image that uses the @code{--appl-image} option
13495: instead of the @code{--image-file} option (@pxref{Invoking
13496: Gforth}). When you execute such an image on Unix (by typing the image
13497: name as command), the Gforth engine will pass all options to the image
13498: instead of trying to interpret them as engine options.
13499:
13500: If you type @file{gforthmi} with no arguments, it prints some usage
13501: instructions.
13502:
13503: @cindex @code{savesystem} during @file{gforthmi}
13504: @cindex @code{bye} during @file{gforthmi}
13505: @cindex doubly indirect threaded code
13506: @cindex environment variables
13507: @cindex @code{GFORTHD} -- environment variable
13508: @cindex @code{GFORTH} -- environment variable
13509: @cindex @code{gforth-ditc}
13510: There are a few wrinkles: After processing the passed @i{options}, the
13511: words @code{savesystem} and @code{bye} must be visible. A special doubly
13512: indirect threaded version of the @file{gforth} executable is used for
13513: creating the non-relocatable images; you can pass the exact filename of
13514: this executable through the environment variable @code{GFORTHD}
13515: (default: @file{gforth-ditc}); if you pass a version that is not doubly
13516: indirect threaded, you will not get a fully relocatable image, but a
13517: data-relocatable image (because there is no code address offset). The
13518: normal @file{gforth} executable is used for creating the relocatable
13519: image; you can pass the exact filename of this executable through the
13520: environment variable @code{GFORTH}.
13521:
13522: @node cross.fs, , gforthmi, Fully Relocatable Image Files
13523: @subsection @file{cross.fs}
13524: @cindex @file{cross.fs}
13525: @cindex cross-compiler
13526: @cindex metacompiler
13527: @cindex target compiler
13528:
13529: You can also use @code{cross}, a batch compiler that accepts a Forth-like
13530: programming language (@pxref{Cross Compiler}).
13531:
13532: @code{cross} allows you to create image files for machines with
13533: different data sizes and data formats than the one used for generating
13534: the image file. You can also use it to create an application image that
13535: does not contain a Forth compiler. These features are bought with
13536: restrictions and inconveniences in programming. E.g., addresses have to
13537: be stored in memory with special words (@code{A!}, @code{A,}, etc.) in
13538: order to make the code relocatable.
13539:
13540:
13541: @node Stack and Dictionary Sizes, Running Image Files, Fully Relocatable Image Files, Image Files
13542: @section Stack and Dictionary Sizes
13543: @cindex image file, stack and dictionary sizes
13544: @cindex dictionary size default
13545: @cindex stack size default
13546:
13547: If you invoke Gforth with a command line flag for the size
13548: (@pxref{Invoking Gforth}), the size you specify is stored in the
13549: dictionary. If you save the dictionary with @code{savesystem} or create
13550: an image with @file{gforthmi}, this size will become the default
13551: for the resulting image file. E.g., the following will create a
13552: fully relocatable version of @file{gforth.fi} with a 1MB dictionary:
13553:
13554: @example
13555: gforthmi gforth.fi -m 1M
13556: @end example
13557:
13558: In other words, if you want to set the default size for the dictionary
13559: and the stacks of an image, just invoke @file{gforthmi} with the
13560: appropriate options when creating the image.
13561:
13562: @cindex stack size, cache-friendly
13563: Note: For cache-friendly behaviour (i.e., good performance), you should
13564: make the sizes of the stacks modulo, say, 2K, somewhat different. E.g.,
13565: the default stack sizes are: data: 16k (mod 2k=0); fp: 15.5k (mod
13566: 2k=1.5k); return: 15k(mod 2k=1k); locals: 14.5k (mod 2k=0.5k).
13567:
13568: @node Running Image Files, Modifying the Startup Sequence, Stack and Dictionary Sizes, Image Files
13569: @section Running Image Files
13570: @cindex running image files
13571: @cindex invoking image files
13572: @cindex image file invocation
13573:
13574: @cindex -i, invoke image file
13575: @cindex --image file, invoke image file
13576: You can invoke Gforth with an image file @i{image} instead of the
13577: default @file{gforth.fi} with the @code{-i} flag (@pxref{Invoking Gforth}):
13578: @example
13579: gforth -i @i{image}
13580: @end example
13581:
13582: @cindex executable image file
13583: @cindex image file, executable
13584: If your operating system supports starting scripts with a line of the
13585: form @code{#! ...}, you just have to type the image file name to start
13586: Gforth with this image file (note that the file extension @code{.fi} is
13587: just a convention). I.e., to run Gforth with the image file @i{image},
13588: you can just type @i{image} instead of @code{gforth -i @i{image}}.
13589: This works because every @code{.fi} file starts with a line of this
13590: format:
13591:
13592: @example
13593: #! /usr/local/bin/gforth-0.4.0 -i
13594: @end example
13595:
13596: The file and pathname for the Gforth engine specified on this line is
13597: the specific Gforth executable that it was built against; i.e. the value
13598: of the environment variable @code{GFORTH} at the time that
13599: @file{gforthmi} was executed.
13600:
13601: You can make use of the same shell capability to make a Forth source
13602: file into an executable. For example, if you place this text in a file:
13603:
13604: @example
13605: #! /usr/local/bin/gforth
13606:
13607: ." Hello, world" CR
13608: bye
13609: @end example
13610:
13611: @noindent
13612: and then make the file executable (chmod +x in Unix), you can run it
13613: directly from the command line. The sequence @code{#!} is used in two
13614: ways; firstly, it is recognised as a ``magic sequence'' by the operating
13615: system@footnote{The Unix kernel actually recognises two types of files:
13616: executable files and files of data, where the data is processed by an
13617: interpreter that is specified on the ``interpreter line'' -- the first
13618: line of the file, starting with the sequence #!. There may be a small
13619: limit (e.g., 32) on the number of characters that may be specified on
13620: the interpreter line.} secondly it is treated as a comment character by
13621: Gforth. Because of the second usage, a space is required between
13622: @code{#!} and the path to the executable.
13623:
13624: The disadvantage of this latter technique, compared with using
13625: @file{gforthmi}, is that it is slower; the Forth source code is compiled
13626: on-the-fly, each time the program is invoked.
13627:
13628:
13629: doc-#!
13630:
13631:
13632: @node Modifying the Startup Sequence, , Running Image Files, Image Files
13633: @section Modifying the Startup Sequence
13634: @cindex startup sequence for image file
13635: @cindex image file initialization sequence
13636: @cindex initialization sequence of image file
13637:
13638: You can add your own initialization to the startup sequence through the
13639: deferred word @code{'cold}. @code{'cold} is invoked just before the
13640: image-specific command line processing (by default, loading files and
13641: evaluating (@code{-e}) strings) starts.
13642:
13643: A sequence for adding your initialization usually looks like this:
13644:
13645: @example
13646: :noname
13647: Defers 'cold \ do other initialization stuff (e.g., rehashing wordlists)
13648: ... \ your stuff
13649: ; IS 'cold
13650: @end example
13651:
13652: @cindex turnkey image files
13653: @cindex image file, turnkey applications
13654: You can make a turnkey image by letting @code{'cold} execute a word
13655: (your turnkey application) that never returns; instead, it exits Gforth
13656: via @code{bye} or @code{throw}.
13657:
13658: @cindex command-line arguments, access
13659: @cindex arguments on the command line, access
13660: You can access the (image-specific) command-line arguments through the
13661: variables @code{argc} and @code{argv}. @code{arg} provides convenient
13662: access to @code{argv}.
13663:
13664: If @code{'cold} exits normally, Gforth processes the command-line
13665: arguments as files to be loaded and strings to be evaluated. Therefore,
13666: @code{'cold} should remove the arguments it has used in this case.
13667:
13668:
13669:
13670: doc-'cold
13671: doc-argc
13672: doc-argv
13673: doc-arg
13674:
13675:
13676:
13677: @c ******************************************************************
13678: @node Engine, Binding to System Library, Image Files, Top
13679: @chapter Engine
13680: @cindex engine
13681: @cindex virtual machine
13682:
13683: Reading this chapter is not necessary for programming with Gforth. It
13684: may be helpful for finding your way in the Gforth sources.
13685:
13686: The ideas in this section have also been published in Bernd Paysan,
13687: @cite{ANS fig/GNU/??? Forth} (in German), Forth-Tagung '93 and M. Anton
13688: Ertl, @cite{@uref{http://www.complang.tuwien.ac.at/papers/ertl93.ps.Z, A
13689: Portable Forth Engine}}, EuroForth '93.
13690:
13691: @menu
13692: * Portability::
13693: * Threading::
13694: * Primitives::
13695: * Performance::
13696: @end menu
13697:
13698: @node Portability, Threading, Engine, Engine
13699: @section Portability
13700: @cindex engine portability
13701:
13702: An important goal of the Gforth Project is availability across a wide
13703: range of personal machines. fig-Forth, and, to a lesser extent, F83,
13704: achieved this goal by manually coding the engine in assembly language
13705: for several then-popular processors. This approach is very
13706: labor-intensive and the results are short-lived due to progress in
13707: computer architecture.
13708:
13709: @cindex C, using C for the engine
13710: Others have avoided this problem by coding in C, e.g., Mitch Bradley
13711: (cforth), Mikael Patel (TILE) and Dirk Zoller (pfe). This approach is
13712: particularly popular for UNIX-based Forths due to the large variety of
13713: architectures of UNIX machines. Unfortunately an implementation in C
13714: does not mix well with the goals of efficiency and with using
13715: traditional techniques: Indirect or direct threading cannot be expressed
13716: in C, and switch threading, the fastest technique available in C, is
13717: significantly slower. Another problem with C is that it is very
13718: cumbersome to express double integer arithmetic.
13719:
13720: @cindex GNU C for the engine
13721: @cindex long long
13722: Fortunately, there is a portable language that does not have these
13723: limitations: GNU C, the version of C processed by the GNU C compiler
13724: (@pxref{C Extensions, , Extensions to the C Language Family, gcc.info,
13725: GNU C Manual}). Its labels as values feature (@pxref{Labels as Values, ,
13726: Labels as Values, gcc.info, GNU C Manual}) makes direct and indirect
13727: threading possible, its @code{long long} type (@pxref{Long Long, ,
13728: Double-Word Integers, gcc.info, GNU C Manual}) corresponds to Forth's
13729: double numbers@footnote{Unfortunately, long longs are not implemented
13730: properly on all machines (e.g., on alpha-osf1, long longs are only 64
13731: bits, the same size as longs (and pointers), but they should be twice as
13732: long according to @pxref{Long Long, , Double-Word Integers, gcc.info, GNU
13733: C Manual}). So, we had to implement doubles in C after all. Still, on
13734: most machines we can use long longs and achieve better performance than
13735: with the emulation package.}. GNU C is available for free on all
13736: important (and many unimportant) UNIX machines, VMS, 80386s running
13737: MS-DOS, the Amiga, and the Atari ST, so a Forth written in GNU C can run
13738: on all these machines.
13739:
13740: Writing in a portable language has the reputation of producing code that
13741: is slower than assembly. For our Forth engine we repeatedly looked at
13742: the code produced by the compiler and eliminated most compiler-induced
13743: inefficiencies by appropriate changes in the source code.
13744:
13745: @cindex explicit register declarations
13746: @cindex --enable-force-reg, configuration flag
13747: @cindex -DFORCE_REG
13748: However, register allocation cannot be portably influenced by the
13749: programmer, leading to some inefficiencies on register-starved
13750: machines. We use explicit register declarations (@pxref{Explicit Reg
13751: Vars, , Variables in Specified Registers, gcc.info, GNU C Manual}) to
13752: improve the speed on some machines. They are turned on by using the
13753: configuration flag @code{--enable-force-reg} (@code{gcc} switch
13754: @code{-DFORCE_REG}). Unfortunately, this feature not only depends on the
13755: machine, but also on the compiler version: On some machines some
13756: compiler versions produce incorrect code when certain explicit register
13757: declarations are used. So by default @code{-DFORCE_REG} is not used.
13758:
13759: @node Threading, Primitives, Portability, Engine
13760: @section Threading
13761: @cindex inner interpreter implementation
13762: @cindex threaded code implementation
13763:
13764: @cindex labels as values
13765: GNU C's labels as values extension (available since @code{gcc-2.0},
13766: @pxref{Labels as Values, , Labels as Values, gcc.info, GNU C Manual})
13767: makes it possible to take the address of @i{label} by writing
13768: @code{&&@i{label}}. This address can then be used in a statement like
13769: @code{goto *@i{address}}. I.e., @code{goto *&&x} is the same as
13770: @code{goto x}.
13771:
13772: @cindex @code{NEXT}, indirect threaded
13773: @cindex indirect threaded inner interpreter
13774: @cindex inner interpreter, indirect threaded
13775: With this feature an indirect threaded @code{NEXT} looks like:
13776: @example
13777: cfa = *ip++;
13778: ca = *cfa;
13779: goto *ca;
13780: @end example
13781: @cindex instruction pointer
13782: For those unfamiliar with the names: @code{ip} is the Forth instruction
13783: pointer; the @code{cfa} (code-field address) corresponds to ANS Forths
13784: execution token and points to the code field of the next word to be
13785: executed; The @code{ca} (code address) fetched from there points to some
13786: executable code, e.g., a primitive or the colon definition handler
13787: @code{docol}.
13788:
13789: @cindex @code{NEXT}, direct threaded
13790: @cindex direct threaded inner interpreter
13791: @cindex inner interpreter, direct threaded
13792: Direct threading is even simpler:
13793: @example
13794: ca = *ip++;
13795: goto *ca;
13796: @end example
13797:
13798: Of course we have packaged the whole thing neatly in macros called
13799: @code{NEXT} and @code{NEXT1} (the part of @code{NEXT} after fetching the cfa).
13800:
13801: @menu
13802: * Scheduling::
13803: * Direct or Indirect Threaded?::
13804: * DOES>::
13805: @end menu
13806:
13807: @node Scheduling, Direct or Indirect Threaded?, Threading, Threading
13808: @subsection Scheduling
13809: @cindex inner interpreter optimization
13810:
13811: There is a little complication: Pipelined and superscalar processors,
13812: i.e., RISC and some modern CISC machines can process independent
13813: instructions while waiting for the results of an instruction. The
13814: compiler usually reorders (schedules) the instructions in a way that
13815: achieves good usage of these delay slots. However, on our first tries
13816: the compiler did not do well on scheduling primitives. E.g., for
13817: @code{+} implemented as
13818: @example
13819: n=sp[0]+sp[1];
13820: sp++;
13821: sp[0]=n;
13822: NEXT;
13823: @end example
13824: the @code{NEXT} comes strictly after the other code, i.e., there is nearly no
13825: scheduling. After a little thought the problem becomes clear: The
13826: compiler cannot know that @code{sp} and @code{ip} point to different
13827: addresses (and the version of @code{gcc} we used would not know it even
13828: if it was possible), so it could not move the load of the cfa above the
13829: store to the TOS. Indeed the pointers could be the same, if code on or
13830: very near the top of stack were executed. In the interest of speed we
13831: chose to forbid this probably unused ``feature'' and helped the compiler
13832: in scheduling: @code{NEXT} is divided into the loading part (@code{NEXT_P1})
13833: and the goto part (@code{NEXT_P2}). @code{+} now looks like:
13834: @example
13835: n=sp[0]+sp[1];
13836: sp++;
13837: NEXT_P1;
13838: sp[0]=n;
13839: NEXT_P2;
13840: @end example
13841: This can be scheduled optimally by the compiler.
13842:
13843: This division can be turned off with the switch @code{-DCISC_NEXT}. This
13844: switch is on by default on machines that do not profit from scheduling
13845: (e.g., the 80386), in order to preserve registers.
13846:
13847: @node Direct or Indirect Threaded?, DOES>, Scheduling, Threading
13848: @subsection Direct or Indirect Threaded?
13849: @cindex threading, direct or indirect?
13850:
13851: @cindex -DDIRECT_THREADED
13852: Both! After packaging the nasty details in macro definitions we
13853: realized that we could switch between direct and indirect threading by
13854: simply setting a compilation flag (@code{-DDIRECT_THREADED}) and
13855: defining a few machine-specific macros for the direct-threading case.
13856: On the Forth level we also offer access words that hide the
13857: differences between the threading methods (@pxref{Threading Words}).
13858:
13859: Indirect threading is implemented completely machine-independently.
13860: Direct threading needs routines for creating jumps to the executable
13861: code (e.g. to @code{docol} or @code{dodoes}). These routines are inherently
13862: machine-dependent, but they do not amount to many source lines. Therefore,
13863: even porting direct threading to a new machine requires little effort.
13864:
13865: @cindex --enable-indirect-threaded, configuration flag
13866: @cindex --enable-direct-threaded, configuration flag
13867: The default threading method is machine-dependent. You can enforce a
13868: specific threading method when building Gforth with the configuration
13869: flag @code{--enable-direct-threaded} or
13870: @code{--enable-indirect-threaded}. Note that direct threading is not
13871: supported on all machines.
13872:
13873: @node DOES>, , Direct or Indirect Threaded?, Threading
13874: @subsection DOES>
13875: @cindex @code{DOES>} implementation
13876:
13877: @cindex @code{dodoes} routine
13878: @cindex @code{DOES>}-code
13879: One of the most complex parts of a Forth engine is @code{dodoes}, i.e.,
13880: the chunk of code executed by every word defined by a
13881: @code{CREATE}...@code{DOES>} pair. The main problem here is: How to find
13882: the Forth code to be executed, i.e. the code after the
13883: @code{DOES>} (the @code{DOES>}-code)? There are two solutions:
13884:
13885: In fig-Forth the code field points directly to the @code{dodoes} and the
13886: @code{DOES>}-code address is stored in the cell after the code address (i.e. at
13887: @code{@i{CFA} cell+}). It may seem that this solution is illegal in
13888: the Forth-79 and all later standards, because in fig-Forth this address
13889: lies in the body (which is illegal in these standards). However, by
13890: making the code field larger for all words this solution becomes legal
13891: again. We use this approach for the indirect threaded version and for
13892: direct threading on some machines. Leaving a cell unused in most words
13893: is a bit wasteful, but on the machines we are targeting this is hardly a
13894: problem. The other reason for having a code field size of two cells is
13895: to avoid having different image files for direct and indirect threaded
13896: systems (direct threaded systems require two-cell code fields on many
13897: machines).
13898:
13899: @cindex @code{DOES>}-handler
13900: The other approach is that the code field points or jumps to the cell
13901: after @code{DOES>}. In this variant there is a jump to @code{dodoes} at
13902: this address (the @code{DOES>}-handler). @code{dodoes} can then get the
13903: @code{DOES>}-code address by computing the code address, i.e., the address of
13904: the jump to @code{dodoes}, and add the length of that jump field. A variant of
13905: this is to have a call to @code{dodoes} after the @code{DOES>}; then the
13906: return address (which can be found in the return register on RISCs) is
13907: the @code{DOES>}-code address. Since the two cells available in the code field
13908: are used up by the jump to the code address in direct threading on many
13909: architectures, we use this approach for direct threading on these
13910: architectures. We did not want to add another cell to the code field.
13911:
13912: @node Primitives, Performance, Threading, Engine
13913: @section Primitives
13914: @cindex primitives, implementation
13915: @cindex virtual machine instructions, implementation
13916:
13917: @menu
13918: * Automatic Generation::
13919: * TOS Optimization::
13920: * Produced code::
13921: @end menu
13922:
13923: @node Automatic Generation, TOS Optimization, Primitives, Primitives
13924: @subsection Automatic Generation
13925: @cindex primitives, automatic generation
13926:
13927: @cindex @file{prims2x.fs}
13928: Since the primitives are implemented in a portable language, there is no
13929: longer any need to minimize the number of primitives. On the contrary,
13930: having many primitives has an advantage: speed. In order to reduce the
13931: number of errors in primitives and to make programming them easier, we
13932: provide a tool, the primitive generator (@file{prims2x.fs}), that
13933: automatically generates most (and sometimes all) of the C code for a
13934: primitive from the stack effect notation. The source for a primitive
13935: has the following form:
13936:
13937: @cindex primitive source format
13938: @format
13939: @i{Forth-name} ( @i{stack-effect} ) @i{category} [@i{pronounc.}]
13940: [@code{""}@i{glossary entry}@code{""}]
13941: @i{C code}
13942: [@code{:}
13943: @i{Forth code}]
13944: @end format
13945:
13946: The items in brackets are optional. The category and glossary fields
13947: are there for generating the documentation, the Forth code is there
13948: for manual implementations on machines without GNU C. E.g., the source
13949: for the primitive @code{+} is:
13950: @example
13951: + ( n1 n2 -- n ) core plus
13952: n = n1+n2;
13953: @end example
13954:
13955: This looks like a specification, but in fact @code{n = n1+n2} is C
13956: code. Our primitive generation tool extracts a lot of information from
13957: the stack effect notations@footnote{We use a one-stack notation, even
13958: though we have separate data and floating-point stacks; The separate
13959: notation can be generated easily from the unified notation.}: The number
13960: of items popped from and pushed on the stack, their type, and by what
13961: name they are referred to in the C code. It then generates a C code
13962: prelude and postlude for each primitive. The final C code for @code{+}
13963: looks like this:
13964:
13965: @example
13966: I_plus: /* + ( n1 n2 -- n ) */ /* label, stack effect */
13967: /* */ /* documentation */
13968: @{
13969: DEF_CA /* definition of variable ca (indirect threading) */
13970: Cell n1; /* definitions of variables */
13971: Cell n2;
13972: Cell n;
13973: n1 = (Cell) sp[1]; /* input */
13974: n2 = (Cell) TOS;
13975: sp += 1; /* stack adjustment */
13976: NAME("+") /* debugging output (with -DDEBUG) */
13977: @{
13978: n = n1+n2; /* C code taken from the source */
13979: @}
13980: NEXT_P1; /* NEXT part 1 */
13981: TOS = (Cell)n; /* output */
13982: NEXT_P2; /* NEXT part 2 */
13983: @}
13984: @end example
13985:
13986: This looks long and inefficient, but the GNU C compiler optimizes quite
13987: well and produces optimal code for @code{+} on, e.g., the R3000 and the
13988: HP RISC machines: Defining the @code{n}s does not produce any code, and
13989: using them as intermediate storage also adds no cost.
13990:
13991: There are also other optimizations that are not illustrated by this
13992: example: assignments between simple variables are usually for free (copy
13993: propagation). If one of the stack items is not used by the primitive
13994: (e.g. in @code{drop}), the compiler eliminates the load from the stack
13995: (dead code elimination). On the other hand, there are some things that
13996: the compiler does not do, therefore they are performed by
13997: @file{prims2x.fs}: The compiler does not optimize code away that stores
13998: a stack item to the place where it just came from (e.g., @code{over}).
13999:
14000: While programming a primitive is usually easy, there are a few cases
14001: where the programmer has to take the actions of the generator into
14002: account, most notably @code{?dup}, but also words that do not (always)
14003: fall through to @code{NEXT}.
14004:
14005: @node TOS Optimization, Produced code, Automatic Generation, Primitives
14006: @subsection TOS Optimization
14007: @cindex TOS optimization for primitives
14008: @cindex primitives, keeping the TOS in a register
14009:
14010: An important optimization for stack machine emulators, e.g., Forth
14011: engines, is keeping one or more of the top stack items in
14012: registers. If a word has the stack effect @i{in1}...@i{inx} @code{--}
14013: @i{out1}...@i{outy}, keeping the top @i{n} items in registers
14014: @itemize @bullet
14015: @item
14016: is better than keeping @i{n-1} items, if @i{x>=n} and @i{y>=n},
14017: due to fewer loads from and stores to the stack.
14018: @item is slower than keeping @i{n-1} items, if @i{x<>y} and @i{x<n} and
14019: @i{y<n}, due to additional moves between registers.
14020: @end itemize
14021:
14022: @cindex -DUSE_TOS
14023: @cindex -DUSE_NO_TOS
14024: In particular, keeping one item in a register is never a disadvantage,
14025: if there are enough registers. Keeping two items in registers is a
14026: disadvantage for frequent words like @code{?branch}, constants,
14027: variables, literals and @code{i}. Therefore our generator only produces
14028: code that keeps zero or one items in registers. The generated C code
14029: covers both cases; the selection between these alternatives is made at
14030: C-compile time using the switch @code{-DUSE_TOS}. @code{TOS} in the C
14031: code for @code{+} is just a simple variable name in the one-item case,
14032: otherwise it is a macro that expands into @code{sp[0]}. Note that the
14033: GNU C compiler tries to keep simple variables like @code{TOS} in
14034: registers, and it usually succeeds, if there are enough registers.
14035:
14036: @cindex -DUSE_FTOS
14037: @cindex -DUSE_NO_FTOS
14038: The primitive generator performs the TOS optimization for the
14039: floating-point stack, too (@code{-DUSE_FTOS}). For floating-point
14040: operations the benefit of this optimization is even larger:
14041: floating-point operations take quite long on most processors, but can be
14042: performed in parallel with other operations as long as their results are
14043: not used. If the FP-TOS is kept in a register, this works. If
14044: it is kept on the stack, i.e., in memory, the store into memory has to
14045: wait for the result of the floating-point operation, lengthening the
14046: execution time of the primitive considerably.
14047:
14048: The TOS optimization makes the automatic generation of primitives a
14049: bit more complicated. Just replacing all occurrences of @code{sp[0]} by
14050: @code{TOS} is not sufficient. There are some special cases to
14051: consider:
14052: @itemize @bullet
14053: @item In the case of @code{dup ( w -- w w )} the generator must not
14054: eliminate the store to the original location of the item on the stack,
14055: if the TOS optimization is turned on.
14056: @item Primitives with stack effects of the form @code{--}
14057: @i{out1}...@i{outy} must store the TOS to the stack at the start.
14058: Likewise, primitives with the stack effect @i{in1}...@i{inx} @code{--}
14059: must load the TOS from the stack at the end. But for the null stack
14060: effect @code{--} no stores or loads should be generated.
14061: @end itemize
14062:
14063: @node Produced code, , TOS Optimization, Primitives
14064: @subsection Produced code
14065: @cindex primitives, assembly code listing
14066:
14067: @cindex @file{engine.s}
14068: To see what assembly code is produced for the primitives on your machine
14069: with your compiler and your flag settings, type @code{make engine.s} and
14070: look at the resulting file @file{engine.s}.
14071:
14072: @node Performance, , Primitives, Engine
14073: @section Performance
14074: @cindex performance of some Forth interpreters
14075: @cindex engine performance
14076: @cindex benchmarking Forth systems
14077: @cindex Gforth performance
14078:
14079: On RISCs the Gforth engine is very close to optimal; i.e., it is usually
14080: impossible to write a significantly faster engine.
14081:
14082: On register-starved machines like the 386 architecture processors
14083: improvements are possible, because @code{gcc} does not utilize the
14084: registers as well as a human, even with explicit register declarations;
14085: e.g., Bernd Beuster wrote a Forth system fragment in assembly language
14086: and hand-tuned it for the 486; this system is 1.19 times faster on the
14087: Sieve benchmark on a 486DX2/66 than Gforth compiled with
14088: @code{gcc-2.6.3} with @code{-DFORCE_REG}. The situation has improved
14089: with gcc-2.95 and gforth-0.4.9; now the most important virtual machine
14090: registers fit in real registers (and we can even afford to use the TOS
14091: optimization), resulting in a speedup of 1.14 on the sieve over the
14092: earlier results.
14093:
14094: @cindex Win32Forth performance
14095: @cindex NT Forth performance
14096: @cindex eforth performance
14097: @cindex ThisForth performance
14098: @cindex PFE performance
14099: @cindex TILE performance
14100: The potential advantage of assembly language implementations
14101: is not necessarily realized in complete Forth systems: We compared
14102: Gforth-0.4.9 (direct threaded, compiled with @code{gcc-2.95.1} and
14103: @code{-DFORCE_REG}) with Win32Forth 1.2093, LMI's NT Forth (Beta, May
14104: 1994) and Eforth (with and without peephole (aka pinhole) optimization
14105: of the threaded code); all these systems were written in assembly
14106: language. We also compared Gforth with three systems written in C:
14107: PFE-0.9.14 (compiled with @code{gcc-2.6.3} with the default
14108: configuration for Linux: @code{-O2 -fomit-frame-pointer -DUSE_REGS
14109: -DUNROLL_NEXT}), ThisForth Beta (compiled with @code{gcc-2.6.3 -O3
14110: -fomit-frame-pointer}; ThisForth employs peephole optimization of the
14111: threaded code) and TILE (compiled with @code{make opt}). We benchmarked
14112: Gforth, PFE, ThisForth and TILE on a 486DX2/66 under Linux. Kenneth
14113: O'Heskin kindly provided the results for Win32Forth and NT Forth on a
14114: 486DX2/66 with similar memory performance under Windows NT. Marcel
14115: Hendrix ported Eforth to Linux, then extended it to run the benchmarks,
14116: added the peephole optimizer, ran the benchmarks and reported the
14117: results.
14118:
14119: We used four small benchmarks: the ubiquitous Sieve; bubble-sorting and
14120: matrix multiplication come from the Stanford integer benchmarks and have
14121: been translated into Forth by Martin Fraeman; we used the versions
14122: included in the TILE Forth package, but with bigger data set sizes; and
14123: a recursive Fibonacci number computation for benchmarking calling
14124: performance. The following table shows the time taken for the benchmarks
14125: scaled by the time taken by Gforth (in other words, it shows the speedup
14126: factor that Gforth achieved over the other systems).
14127:
14128: @example
14129: relative Win32- NT eforth This-
14130: time Gforth Forth Forth eforth +opt PFE Forth TILE
14131: sieve 1.00 1.58 1.30 1.58 0.97 1.80 3.63 9.79
14132: bubble 1.00 1.55 1.67 1.75 1.04 1.78 4.59
14133: matmul 1.00 1.67 1.53 1.66 0.84 1.79 4.63
14134: fib 1.00 1.75 1.53 1.40 0.99 1.99 3.43 4.93
14135: @end example
14136:
14137: You may be quite surprised by the good performance of Gforth when
14138: compared with systems written in assembly language. One important reason
14139: for the disappointing performance of these other systems is probably
14140: that they are not written optimally for the 486 (e.g., they use the
14141: @code{lods} instruction). In addition, Win32Forth uses a comfortable,
14142: but costly method for relocating the Forth image: like @code{cforth}, it
14143: computes the actual addresses at run time, resulting in two address
14144: computations per @code{NEXT} (@pxref{Image File Background}).
14145:
14146: Only Eforth with the peephole optimizer performs comparable to
14147: Gforth. The speedups achieved with peephole optimization of threaded
14148: code are quite remarkable. Adding a peephole optimizer to Gforth should
14149: cause similar speedups.
14150:
14151: The speedup of Gforth over PFE, ThisForth and TILE can be easily
14152: explained with the self-imposed restriction of the latter systems to
14153: standard C, which makes efficient threading impossible (however, the
14154: measured implementation of PFE uses a GNU C extension: @pxref{Global Reg
14155: Vars, , Defining Global Register Variables, gcc.info, GNU C Manual}).
14156: Moreover, current C compilers have a hard time optimizing other aspects
14157: of the ThisForth and the TILE source.
14158:
14159: The performance of Gforth on 386 architecture processors varies widely
14160: with the version of @code{gcc} used. E.g., @code{gcc-2.5.8} failed to
14161: allocate any of the virtual machine registers into real machine
14162: registers by itself and would not work correctly with explicit register
14163: declarations, giving a 1.5 times slower engine (on a 486DX2/66 running
14164: the Sieve) than the one measured above.
14165:
14166: Note that there have been several releases of Win32Forth since the
14167: release presented here, so the results presented above may have little
14168: predictive value for the performance of Win32Forth today (results for
14169: the current release on an i486DX2/66 are welcome).
14170:
14171: @cindex @file{Benchres}
14172: In
14173: @cite{@uref{http://www.complang.tuwien.ac.at/papers/ertl&maierhofer95.ps.gz,
14174: Translating Forth to Efficient C}} by M. Anton Ertl and Martin
14175: Maierhofer (presented at EuroForth '95), an indirect threaded version of
14176: Gforth is compared with Win32Forth, NT Forth, PFE, ThisForth, and
14177: several native code systems; that version of Gforth is slower on a 486
14178: than the direct threaded version used here. You can find a newer version
14179: of these measurements at
14180: @uref{http://www.complang.tuwien.ac.at/forth/performance.html}. You can
14181: find numbers for Gforth on various machines in @file{Benchres}.
14182:
14183: @c ******************************************************************
14184: @node Binding to System Library, Cross Compiler, Engine, Top
14185: @chapter Binding to System Library
14186:
14187: @node Cross Compiler, Bugs, Binding to System Library, Top
14188: @chapter Cross Compiler
14189: @cindex @file{cross.fs}
14190: @cindex cross-compiler
14191: @cindex metacompiler
14192: @cindex target compiler
14193:
14194: The cross compiler is used to bootstrap a Forth kernel. Since Gforth is
14195: mostly written in Forth, including crucial parts like the outer
14196: interpreter and compiler, it needs compiled Forth code to get
14197: started. The cross compiler allows to create new images for other
14198: architectures, even running under another Forth system.
14199:
14200: @menu
14201: * Using the Cross Compiler::
14202: * How the Cross Compiler Works::
14203: @end menu
14204:
14205: @node Using the Cross Compiler, How the Cross Compiler Works, Cross Compiler, Cross Compiler
14206: @section Using the Cross Compiler
14207:
14208: The cross compiler uses a language that resembles Forth, but isn't. The
14209: main difference is that you can execute Forth code after definition,
14210: while you usually can't execute the code compiled by cross, because the
14211: code you are compiling is typically for a different computer than the
14212: one you are compiling on.
14213:
14214: The Makefile is already set up to allow you to create kernels for new
14215: architectures with a simple make command. The generic kernels using the
14216: GCC compiled virtual machine are created in the normal build process
14217: with @code{make}. To create a embedded Gforth executable for e.g. the
14218: 8086 processor (running on a DOS machine), type
14219:
14220: @example
14221: make kernl-8086.fi
14222: @end example
14223:
14224: This will use the machine description from the @file{arch/8086}
14225: directory to create a new kernel. A machine file may look like that:
14226:
14227: @example
14228: \ Parameter for target systems 06oct92py
14229:
14230: 4 Constant cell \ cell size in bytes
14231: 2 Constant cell<< \ cell shift to bytes
14232: 5 Constant cell>bit \ cell shift to bits
14233: 8 Constant bits/char \ bits per character
14234: 8 Constant bits/byte \ bits per byte [default: 8]
14235: 8 Constant float \ bytes per float
14236: 8 Constant /maxalign \ maximum alignment in bytes
14237: false Constant bigendian \ byte order
14238: ( true=big, false=little )
14239:
14240: include machpc.fs \ feature list
14241: @end example
14242:
14243: This part is obligatory for the cross compiler itself, the feature list
14244: is used by the kernel to conditionally compile some features in and out,
14245: depending on whether the target supports these features.
14246:
14247: There are some optional features, if you define your own primitives,
14248: have an assembler, or need special, nonstandard preparation to make the
14249: boot process work. @code{asm-include} include an assembler,
14250: @code{prims-include} includes primitives, and @code{>boot} prepares for
14251: booting.
14252:
14253: @example
14254: : asm-include ." Include assembler" cr
14255: s" arch/8086/asm.fs" included ;
14256:
14257: : prims-include ." Include primitives" cr
14258: s" arch/8086/prim.fs" included ;
14259:
14260: : >boot ." Prepare booting" cr
14261: s" ' boot >body into-forth 1+ !" evaluate ;
14262: @end example
14263:
14264: These words are used as sort of macro during the cross compilation in
14265: the file @file{kernel/main.fs}. Instead of using this macros, it would
14266: be possible --- but more complicated --- to write a new kernel project
14267: file, too.
14268:
14269: @file{kernel/main.fs} expects the machine description file name on the
14270: stack; the cross compiler itself (@file{cross.fs}) assumes that either
14271: @code{mach-file} leaves a counted string on the stack, or
14272: @code{machine-file} leaves an address, count pair of the filename on the
14273: stack.
14274:
14275: The feature list is typically controlled using @code{SetValue}, generic
14276: files that are used by several projects can use @code{DefaultValue}
14277: instead. Both functions work like @code{Value}, when the value isn't
14278: defined, but @code{SetValue} works like @code{to} if the value is
14279: defined, and @code{DefaultValue} doesn't set anything, if the value is
14280: defined.
14281:
14282: @example
14283: \ generic mach file for pc gforth 03sep97jaw
14284:
14285: true DefaultValue NIL \ relocating
14286:
14287: >ENVIRON
14288:
14289: true DefaultValue file \ controls the presence of the
14290: \ file access wordset
14291: true DefaultValue OS \ flag to indicate a operating system
14292:
14293: true DefaultValue prims \ true: primitives are c-code
14294:
14295: true DefaultValue floating \ floating point wordset is present
14296:
14297: true DefaultValue glocals \ gforth locals are present
14298: \ will be loaded
14299: true DefaultValue dcomps \ double number comparisons
14300:
14301: true DefaultValue hash \ hashing primitives are loaded/present
14302:
14303: true DefaultValue xconds \ used together with glocals,
14304: \ special conditionals supporting gforths'
14305: \ local variables
14306: true DefaultValue header \ save a header information
14307:
14308: true DefaultValue backtrace \ enables backtrace code
14309:
14310: false DefaultValue ec
14311: false DefaultValue crlf
14312:
14313: cell 2 = [IF] &32 [ELSE] &256 [THEN] KB DefaultValue kernel-size
14314:
14315: &16 KB DefaultValue stack-size
14316: &15 KB &512 + DefaultValue fstack-size
14317: &15 KB DefaultValue rstack-size
14318: &14 KB &512 + DefaultValue lstack-size
14319: @end example
14320:
14321: @node How the Cross Compiler Works, , Using the Cross Compiler, Cross Compiler
14322: @section How the Cross Compiler Works
14323:
14324: @node Bugs, Origin, Cross Compiler, Top
14325: @appendix Bugs
14326: @cindex bug reporting
14327:
14328: Known bugs are described in the file @file{BUGS} in the Gforth distribution.
14329:
14330: If you find a bug, please send a bug report to
14331: @email{bug-gforth@@gnu.org}. A bug report should include this
14332: information:
14333:
14334: @itemize @bullet
14335: @item
14336: The Gforth version used (it is announced at the start of an
14337: interactive Gforth session).
14338: @item
14339: The machine and operating system (on Unix
14340: systems @code{uname -a} will report this information).
14341: @item
14342: The installation options (send the file @file{config.status}).
14343: @item
14344: A complete list of changes (if any) you (or your installer) have made to the
14345: Gforth sources.
14346: @item
14347: A program (or a sequence of keyboard commands) that reproduces the bug.
14348: @item
14349: A description of what you think constitutes the buggy behaviour.
14350: @end itemize
14351:
14352: For a thorough guide on reporting bugs read @ref{Bug Reporting, , How
14353: to Report Bugs, gcc.info, GNU C Manual}.
14354:
14355:
14356: @node Origin, Forth-related information, Bugs, Top
14357: @appendix Authors and Ancestors of Gforth
14358:
14359: @section Authors and Contributors
14360: @cindex authors of Gforth
14361: @cindex contributors to Gforth
14362:
14363: The Gforth project was started in mid-1992 by Bernd Paysan and Anton
14364: Ertl. The third major author was Jens Wilke. Lennart Benschop (who was
14365: one of Gforth's first users, in mid-1993) and Stuart Ramsden inspired us
14366: with their continuous feedback. Lennart Benshop contributed
14367: @file{glosgen.fs}, while Stuart Ramsden has been working on automatic
14368: support for calling C libraries. Helpful comments also came from Paul
14369: Kleinrubatscher, Christian Pirker, Dirk Zoller, Marcel Hendrix, John
14370: Wavrik, Barrie Stott, Marc de Groot, Jorge Acerada, Bruce Hoyt, and
14371: Robert Epprecht. Since the release of Gforth-0.2.1 there were also
14372: helpful comments from many others; thank you all, sorry for not listing
14373: you here (but digging through my mailbox to extract your names is on my
14374: to-do list). Since the release of Gforth-0.4.0 Neal Crook worked on the
14375: manual.
14376:
14377: Gforth also owes a lot to the authors of the tools we used (GCC, CVS,
14378: and autoconf, among others), and to the creators of the Internet: Gforth
14379: was developed across the Internet, and its authors did not meet
14380: physically for the first 4 years of development.
14381:
14382: @section Pedigree
14383: @cindex pedigree of Gforth
14384:
14385: Gforth descends from bigFORTH (1993) and fig-Forth. Gforth and PFE (by
14386: Dirk Zoller) will cross-fertilize each other. Of course, a significant
14387: part of the design of Gforth was prescribed by ANS Forth.
14388:
14389: Bernd Paysan wrote bigFORTH, a descendent from TurboForth, an unreleased
14390: 32 bit native code version of VolksForth for the Atari ST, written
14391: mostly by Dietrich Weineck.
14392:
14393: VolksForth descends from F83. It was written by Klaus Schleisiek, Bernd
14394: Pennemann, Georg Rehfeld and Dietrich Weineck for the C64 (called
14395: UltraForth there) in the mid-80s and ported to the Atari ST in 1986.
14396:
14397: Henry Laxen and Mike Perry wrote F83 as a model implementation of the
14398: Forth-83 standard. !! Pedigree? When?
14399:
14400: A team led by Bill Ragsdale implemented fig-Forth on many processors in
14401: 1979. Robert Selzer and Bill Ragsdale developed the original
14402: implementation of fig-Forth for the 6502 based on microForth.
14403:
14404: The principal architect of microForth was Dean Sanderson. microForth was
14405: FORTH, Inc.'s first off-the-shelf product. It was developed in 1976 for
14406: the 1802, and subsequently implemented on the 8080, the 6800 and the
14407: Z80.
14408:
14409: All earlier Forth systems were custom-made, usually by Charles Moore,
14410: who discovered (as he puts it) Forth during the late 60s. The first full
14411: Forth existed in 1971.
14412:
14413: A part of the information in this section comes from @cite{The Evolution
14414: of Forth} by Elizabeth D. Rather, Donald R. Colburn and Charles
14415: H. Moore, presented at the HOPL-II conference and preprinted in SIGPLAN
14416: Notices 28(3), 1993. You can find more historical and genealogical
14417: information about Forth there.
14418:
14419: @node Forth-related information, Word Index, Origin, Top
14420: @appendix Other Forth-related information
14421: @cindex Forth-related information
14422:
14423: @menu
14424: * Internet resources::
14425: * Books::
14426: * The Forth Interest Group::
14427: * Conferences::
14428: @end menu
14429:
14430:
14431: @node Internet resources, Books, Forth-related information, Forth-related information
14432: @section Internet resources
14433: @cindex internet resources
14434:
14435: @cindex comp.lang.forth
14436: @cindex frequently asked questions
14437: There is an active news group (comp.lang.forth) discussing Forth and
14438: Forth-related issues. A frequently-asked-questions (FAQ) list
14439: is posted to the news group regularly, and archived at these sites:
14440:
14441: @itemize @bullet
14442: @item
14443: @uref{ftp://rtfm.mit.edu/pub/usenet-by-group/comp.lang.forth/}
14444: @item
14445: @uref{ftp://ftp.forth.org/pub/Forth/FAQ/}
14446: @end itemize
14447:
14448: The FAQ list should be considered mandatory reading before posting to
14449: the news group.
14450:
14451: Here are some other web sites holding Forth-related material:
14452:
14453: @itemize @bullet
14454: @item
14455: @uref{http://www.taygeta.com/forth.html} -- Skip Carter's Forth pages.
14456: @item
14457: @uref{http://www.jwdt.com/~paysan/gforth.html} -- the Gforth home page.
14458: @item
14459: @uref{http://www.minerva.com/uathena.htm} -- home of ANS Forth Standard.
14460: @item
14461: @uref{http://dec.bournemouth.ac.uk/forth/index.html} -- the Forth
14462: Research page, including links to the Journal of Forth Application and
14463: Research (JFAR) and a searchable Forth bibliography.
14464: @end itemize
14465:
14466:
14467: @node Books, The Forth Interest Group, Internet resources, Forth-related information
14468: @section Books
14469: @cindex books on Forth
14470:
14471: As the Standard is relatively new, there are not many books out yet. It
14472: is not recommended to learn Forth by using Gforth and a book that is not
14473: written for ANS Forth, as you will not know your mistakes from the
14474: deviations of the book. However, books based on the Forth-83 standard
14475: should be ok, because ANS Forth is primarily an extension of Forth-83.
14476: Refer to the Forth FAQ for details of Forth-related books.
14477:
14478: @cindex standard document for ANS Forth
14479: @cindex ANS Forth document
14480: The definite reference if you want to write ANS Forth programs is, of
14481: course, the ANS Forth document. It is available in printed form from the
14482: National Standards Institute Sales Department (Tel.: USA (212) 642-4900;
14483: Fax.: USA (212) 302-1286) as document @cite{X3.215-1994} for about
14484: $200. You can also get it from Global Engineering Documents (Tel.: USA
14485: (800) 854-7179; Fax.: (303) 843-9880) for about $300.
14486:
14487: @cite{dpANS6}, the last draft of the standard, which was then submitted
14488: to ANSI for publication is available electronically and for free in some
14489: MS Word format, and it has been converted to HTML
14490: (@uref{http://www.taygeta.com/forth/dpans.html}; this HTML version also
14491: includes the answers to Requests for Interpretation (RFIs). Some
14492: pointers to these versions can be found through
14493: @*@uref{http://www.complang.tuwien.ac.at/projects/forth.html}.
14494:
14495:
14496: @node The Forth Interest Group, Conferences, Books, Forth-related information
14497: @section The Forth Interest Group
14498: @cindex Forth interest group (FIG)
14499:
14500: The Forth Interest Group (FIG) is a world-wide, non-profit,
14501: member-supported organisation. It publishes a regular magazine,
14502: @var{FORTH Dimensions}, and offers other benefits of membership. You can
14503: contact the FIG through their office email address:
14504: @email{office@@forth.org} or by visiting their web site at
14505: @uref{http://www.forth.org/}. This web site also includes links to FIG
14506: chapters in other countries and American cities
14507: (@uref{http://www.forth.org/chapters.html}).
14508:
14509: @node Conferences, , The Forth Interest Group, Forth-related information
14510: @section Conferences
14511: @cindex Conferences
14512:
14513: There are several regular conferences related to Forth. They are all
14514: well-publicised in @var{FORTH Dimensions} and on the comp.lang.forth
14515: news group:
14516:
14517: @itemize @bullet
14518: @item
14519: FORML -- the Forth modification laboratory convenes every year near
14520: Monterey, California.
14521: @item
14522: The Rochester Forth Conference -- an annual conference traditionally
14523: held in Rochester, New York.
14524: @item
14525: EuroForth -- this European conference takes place annually.
14526: @end itemize
14527:
14528:
14529: @node Word Index, Name Index, Forth-related information, Top
14530: @unnumbered Word Index
14531:
14532: This index is a list of Forth words that have ``glossary'' entries
14533: within this manual. Each word is listed with its stack effect and
14534: wordset.
14535:
14536: @printindex fn
14537:
14538: @node Name Index, Concept Index, Word Index, Top
14539: @unnumbered Name Index
14540:
14541: This index is a list of Forth words that have ``glossary'' entries
14542: within this manual.
14543:
14544: @printindex ky
14545:
14546: @node Concept Index, , Name Index, Top
14547: @unnumbered Concept and Word Index
14548:
14549: Not all entries listed in this index are present verbatim in the
14550: text. This index also duplicates, in abbreviated form, all of the words
14551: listed in the Word Index (only the names are listed for the words here).
14552:
14553: @printindex cp
14554:
14555: @contents
14556: @bye
14557:
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